]> wimlib.net Git - wimlib/blob - src/extract.c
d4efc24e2ff826b4983b7bc3d1359010dddb58b9
[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);
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);
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         int ret;
557
558         if (dentry->was_hardlinked)
559                 return 0;
560
561 #ifdef ENABLE_DEBUG
562         if (lte_spec) {
563                 char sha1_str[100];
564                 char *p = sha1_str;
565                 for (unsigned i = 0; i < SHA1_HASH_SIZE; i++)
566                         p += sprintf(p, "%02x", lte_override->hash[i]);
567                 DEBUG("Extracting stream SHA1=%s to \"%"TS"\"",
568                       sha1_str, path, inode->i_ino);
569         } else {
570                 DEBUG("Extracting streams to \"%"TS"\"", path, inode->i_ino);
571         }
572 #endif
573
574         /* Unnamed data stream.  */
575         lte = inode_unnamed_lte_resolved(inode);
576         if (lte && (!lte_spec || lte == lte_spec)) {
577                 if (lte_spec)
578                         lte = lte_override;
579                 if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
580                                              FILE_ATTRIBUTE_REPARSE_POINT)))
581                 {
582                         if ((inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) &&
583                             ctx->supported_features.encrypted_files)
584                                 ret = ctx->ops->extract_encrypted_stream(path, lte, ctx);
585                         else
586                                 ret = ctx->ops->extract_unnamed_stream(path, lte, ctx);
587                         if (ret)
588                                 goto error;
589                         update_extract_progress(ctx, lte);
590                 }
591                 else if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT)
592                 {
593                         ret = 0;
594                         if (ctx->supported_features.reparse_points)
595                                 ret = extract_reparse_data(path, ctx, inode, lte);
596                 #ifndef __WIN32__
597                         else if ((inode_is_symlink(inode) &&
598                                   ctx->supported_features.symlink_reparse_points))
599                                 ret = extract_symlink(path, ctx, inode, lte);
600                 #endif
601                         if (ret)
602                                 return ret;
603                 }
604         }
605
606         /* Named data streams.  */
607         if (can_extract_named_data_streams(ctx)) {
608                 for (u16 i = 0; i < inode->i_num_ads; i++) {
609                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
610
611                         if (!ads_entry_is_named_stream(entry))
612                                 continue;
613                         lte = entry->lte;
614                         if (!lte)
615                                 continue;
616                         if (lte_spec && lte_spec != lte)
617                                 continue;
618                         if (lte_spec)
619                                 lte = lte_override;
620                         ret = ctx->ops->extract_named_stream(path, entry->stream_name,
621                                                              entry->stream_name_nbytes / 2,
622                                                              lte, ctx);
623                         if (ret)
624                                 goto error;
625                         update_extract_progress(ctx, lte);
626                 }
627         }
628         return 0;
629
630 error:
631         ERROR_WITH_ERRNO("Failed to extract data of \"%"TS"\"", path);
632         return ret;
633 }
634
635 /* Set attributes on an extracted file or directory if supported by the
636  * extraction mode.  */
637 static int
638 extract_file_attributes(const tchar *path, struct apply_ctx *ctx,
639                         struct wim_dentry *dentry)
640 {
641         int ret;
642
643         if (ctx->ops->set_file_attributes) {
644                 if (dentry == ctx->extract_root && ctx->root_dentry_is_special)
645                         return 0;
646                 ret = ctx->ops->set_file_attributes(path,
647                                                     dentry->d_inode->i_attributes,
648                                                     ctx);
649                 if (ret) {
650                         ERROR_WITH_ERRNO("Failed to set attributes on "
651                                          "\"%"TS"\"", path);
652                         return ret;
653                 }
654         }
655         return 0;
656 }
657
658
659 /* Set or remove the short (DOS) name on an extracted file or directory if
660  * supported by the extraction mode.  Since DOS names are unimportant and it's
661  * easy to run into problems setting them on Windows (SetFileShortName()
662  * requires SE_RESTORE privilege, which only the Administrator can request, and
663  * also requires DELETE access to the file), failure is ignored unless
664  * WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES is set.  */
665 static int
666 extract_short_name(const tchar *path, struct apply_ctx *ctx,
667                    struct wim_dentry *dentry)
668 {
669         int ret;
670
671         /* The root of the dentry tree being extracted may not be extracted to
672          * its original name, so its short name should be ignored.  */
673         if (dentry == ctx->extract_root)
674                 return 0;
675
676         if (ctx->supported_features.short_names) {
677                 ret = ctx->ops->set_short_name(path,
678                                                dentry->short_name,
679                                                dentry->short_name_nbytes / 2,
680                                                ctx);
681                 if (ret && (ctx->extract_flags &
682                             WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES))
683                 {
684                         ERROR_WITH_ERRNO("Failed to set short name of "
685                                          "\"%"TS"\"", path);
686                         return ret;
687                 }
688         }
689         return 0;
690 }
691
692 /* Set security descriptor, UNIX data, or neither on an extracted file, taking
693  * into account the current extraction mode and flags.  */
694 static int
695 extract_security(const tchar *path, struct apply_ctx *ctx,
696                  struct wim_dentry *dentry)
697 {
698         int ret;
699         struct wim_inode *inode = dentry->d_inode;
700
701         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS)
702                 return 0;
703
704         if ((ctx->extract_root == dentry) && ctx->root_dentry_is_special)
705                 return 0;
706
707 #ifndef __WIN32__
708         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
709                 struct wimlib_unix_data data;
710
711                 ret = inode_get_unix_data(inode, &data, NULL);
712                 if (ret < 0)
713                         ret = 0;
714                 else if (ret == 0)
715                         ret = ctx->ops->set_unix_data(path, &data, ctx);
716                 if (ret) {
717                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
718                                 ERROR_WITH_ERRNO("Failed to set UNIX owner, "
719                                                  "group, and/or mode on "
720                                                  "\"%"TS"\"", path);
721                                 return ret;
722                         } else {
723                                 WARNING_WITH_ERRNO("Failed to set UNIX owner, "
724                                                    "group, and/or/mode on "
725                                                    "\"%"TS"\"", path);
726                         }
727                 }
728         }
729         else
730 #endif /* __WIN32__ */
731         if (ctx->supported_features.security_descriptors &&
732             inode->i_security_id != -1)
733         {
734                 const struct wim_security_data *sd;
735                 const u8 *desc;
736                 size_t desc_size;
737
738                 sd = wim_const_security_data(ctx->wim);
739                 desc = sd->descriptors[inode->i_security_id];
740                 desc_size = sd->sizes[inode->i_security_id];
741
742                 ret = ctx->ops->set_security_descriptor(path, desc,
743                                                         desc_size, ctx);
744                 if (ret) {
745                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_ACLS) {
746                                 ERROR_WITH_ERRNO("Failed to set security "
747                                                  "descriptor on \"%"TS"\"", path);
748                                 return ret;
749                         } else {
750                                 if (errno != EACCES) {
751                                         WARNING_WITH_ERRNO("Failed to set "
752                                                            "security descriptor "
753                                                            "on \"%"TS"\"", path);
754                                 }
755                         }
756                 }
757         }
758         return 0;
759 }
760
761 /* Set timestamps on an extracted file.  Failure is warning-only unless
762  * WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS is set.  */
763 static int
764 extract_timestamps(const tchar *path, struct apply_ctx *ctx,
765                    struct wim_dentry *dentry)
766 {
767         struct wim_inode *inode = dentry->d_inode;
768         int ret;
769
770         if ((ctx->extract_root == dentry) && ctx->root_dentry_is_special)
771                 return 0;
772
773         if (ctx->ops->set_timestamps) {
774                 ret = ctx->ops->set_timestamps(path,
775                                                inode->i_creation_time,
776                                                inode->i_last_write_time,
777                                                inode->i_last_access_time,
778                                                ctx);
779                 if (ret) {
780                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) {
781                                 ERROR_WITH_ERRNO("Failed to set timestamps "
782                                                  "on \"%"TS"\"", path);
783                                 return ret;
784                         } else {
785                                 WARNING_WITH_ERRNO("Failed to set timestamps "
786                                                    "on \"%"TS"\"", path);
787                         }
788                 }
789         }
790         return 0;
791 }
792
793 /* Check whether the extraction of a dentry should be skipped completely.  */
794 static bool
795 dentry_is_supported(struct wim_dentry *dentry,
796                     const struct wim_features *supported_features)
797 {
798         struct wim_inode *inode = dentry->d_inode;
799
800         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
801                 if (supported_features->reparse_points)
802                         return true;
803                 if (supported_features->symlink_reparse_points &&
804                     inode_is_symlink(inode))
805                         return true;
806                 return false;
807         }
808         return true;
809 }
810
811 /* Given a WIM dentry to extract, build the path to which to extract it, in the
812  * format understood by the callbacks in the apply_operations being used.
813  *
814  * Write the resulting path into @path, which must have room for at least
815  * ctx->ops->max_path characters including the null-terminator.
816  *
817  * Return %true if successful; %false if this WIM dentry doesn't actually need
818  * to be extracted or if the calculated path exceeds ctx->ops->max_path
819  * characters.
820  *
821  * This function clobbers the tmp_list member of @dentry and its ancestors up
822  * until the extraction root.  */
823 static bool
824 build_extraction_path(tchar path[], struct wim_dentry *dentry,
825                       struct apply_ctx *ctx)
826 {
827         size_t path_nchars;
828         LIST_HEAD(ancestor_list);
829         tchar *p = path;
830         const tchar *target_prefix;
831         size_t target_prefix_nchars;
832         struct wim_dentry *d;
833
834         if (dentry->extraction_skipped)
835                 return false;
836
837         path_nchars = ctx->ops->path_prefix_nchars;
838
839         if (ctx->ops->requires_realtarget_in_paths) {
840                 target_prefix        = ctx->realtarget;
841                 target_prefix_nchars = ctx->realtarget_nchars;
842         } else if (ctx->ops->requires_target_in_paths) {
843                 target_prefix        = ctx->target;
844                 target_prefix_nchars = ctx->target_nchars;
845         } else {
846                 target_prefix        = NULL;
847                 target_prefix_nchars = 0;
848         }
849         path_nchars += target_prefix_nchars;
850
851         for (d = dentry; d != ctx->extract_root; d = d->parent) {
852                 path_nchars += d->extraction_name_nchars + 1;
853                 list_add(&d->tmp_list, &ancestor_list);
854         }
855
856         path_nchars++; /* null terminator */
857
858         if (path_nchars > ctx->ops->path_max) {
859                 WARNING("\"%"TS"\": Path too long to extract",
860                         dentry_full_path(dentry));
861                 return false;
862         }
863
864         p = tmempcpy(p, ctx->ops->path_prefix, ctx->ops->path_prefix_nchars);
865         p = tmempcpy(p, target_prefix, target_prefix_nchars);
866         list_for_each_entry(d, &ancestor_list, tmp_list) {
867                 *p++ = ctx->ops->path_separator;
868                 p = tmempcpy(p, d->extraction_name, d->extraction_name_nchars);
869         }
870         *p++ = T('\0');
871         wimlib_assert(p - path == path_nchars);
872         return true;
873 }
874
875 static unsigned
876 get_num_path_components(const tchar *path, tchar path_separator)
877 {
878         unsigned num_components = 0;
879
880         while (*path) {
881                 while (*path == path_separator)
882                         path++;
883                 if (*path)
884                         num_components++;
885                 while (*path && *path != path_separator)
886                         path++;
887         }
888         return num_components;
889 }
890
891 static int
892 extract_multiimage_symlink(const tchar *oldpath, const tchar *newpath,
893                            struct apply_ctx *ctx, struct wim_dentry *dentry)
894 {
895         size_t num_raw_path_components;
896         const struct wim_dentry *d;
897         size_t num_target_path_components;
898         tchar *p;
899         const tchar *p_old;
900         int ret;
901
902         num_raw_path_components = 0;
903         for (d = dentry; d != ctx->extract_root; d = d->parent)
904                 num_raw_path_components++;
905
906         if (ctx->ops->requires_realtarget_in_paths)
907                 num_target_path_components = get_num_path_components(ctx->realtarget,
908                                                                      ctx->ops->path_separator);
909         else if (ctx->ops->requires_target_in_paths)
910                 num_target_path_components = get_num_path_components(ctx->target,
911                                                                      ctx->ops->path_separator);
912         else
913                 num_target_path_components = 0;
914
915         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_MULTI_IMAGE) {
916                 wimlib_assert(num_target_path_components > 0);
917                 num_raw_path_components++;
918                 num_target_path_components--;
919         }
920
921         p_old = oldpath;
922         while (*p_old == ctx->ops->path_separator)
923                 p_old++;
924         while (--num_target_path_components) {
925                 while (*p_old != ctx->ops->path_separator)
926                         p_old++;
927                 while (*p_old == ctx->ops->path_separator)
928                         p_old++;
929         }
930
931         tchar symlink_target[tstrlen(p_old) + 3 * num_raw_path_components + 1];
932
933         p = &symlink_target[0];
934         while (num_raw_path_components--) {
935                 *p++ = '.';
936                 *p++ = '.';
937                 *p++ = ctx->ops->path_separator;
938         }
939         tstrcpy(p, p_old);
940         DEBUG("Creating symlink \"%"TS"\" => \"%"TS"\"",
941               newpath, symlink_target);
942         ret = ctx->ops->create_symlink(symlink_target, newpath, ctx);
943         if (ret) {
944                 ERROR_WITH_ERRNO("Failed to create symlink "
945                                  "\"%"TS"\" => \"%"TS"\"",
946                                  newpath, symlink_target);
947         }
948         return ret;
949 }
950
951 /* Create the "skeleton" of an extracted file or directory.  Don't yet extract
952  * data streams, reparse data (including symbolic links), timestamps, and
953  * security descriptors.  Basically, everything that doesn't require reading
954  * non-metadata resources from the WIM file and isn't delayed until the final
955  * pass.  */
956 static int
957 do_dentry_extract_skeleton(tchar path[], struct wim_dentry *dentry,
958                            struct apply_ctx *ctx)
959 {
960         struct wim_inode *inode = dentry->d_inode;
961         int ret;
962         const tchar *oldpath;
963
964         if (unlikely(is_linked_extraction(ctx))) {
965                 struct wim_lookup_table_entry *unnamed_lte;
966
967                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
968                 if (unnamed_lte && unnamed_lte->extracted_file) {
969                         oldpath = unnamed_lte->extracted_file;
970                         if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK)
971                                 goto hardlink;
972                         else
973                                 goto symlink;
974                 }
975         }
976
977         /* Create hard link if this dentry corresponds to an already-extracted
978          * inode.  */
979         if (inode->i_extracted_file) {
980                 oldpath = inode->i_extracted_file;
981                 goto hardlink;
982         }
983
984         /* Skip symlinks unless they can be extracted as reparse points rather
985          * than created directly.  */
986         if (inode_is_symlink(inode) && !ctx->supported_features.reparse_points)
987                 return 0;
988
989         /* Create this file or directory unless it's the extraction root, which
990          * was already created if necessary.  */
991         if (dentry != ctx->extract_root) {
992                 ret = extract_inode(path, ctx, inode);
993                 if (ret)
994                         return ret;
995         }
996
997         /* Create empty named data streams.  */
998         if (can_extract_named_data_streams(ctx)) {
999                 for (u16 i = 0; i < inode->i_num_ads; i++) {
1000                         struct wim_ads_entry *entry = &inode->i_ads_entries[i];
1001
1002                         if (!ads_entry_is_named_stream(entry))
1003                                 continue;
1004                         if (entry->lte)
1005                                 continue;
1006                         ret = ctx->ops->extract_named_stream(path,
1007                                                              entry->stream_name,
1008                                                              entry->stream_name_nbytes / 2,
1009                                                              entry->lte, ctx);
1010                         if (ret) {
1011                                 ERROR_WITH_ERRNO("\"%"TS"\": failed to create "
1012                                                  "empty named data stream",
1013                                                  path);
1014                                 return ret;
1015                         }
1016                 }
1017         }
1018
1019         /* Set file attributes (if supported).  */
1020         ret = extract_file_attributes(path, ctx, dentry);
1021         if (ret)
1022                 return ret;
1023
1024         /* Set or remove file short name (if supported).  */
1025         ret = extract_short_name(path, ctx, dentry);
1026         if (ret)
1027                 return ret;
1028
1029         /* If inode has multiple links and hard links are supported in this
1030          * extraction mode and volume, save the path to the extracted file in
1031          * case it's needed to create a hard link.  */
1032         if (unlikely(is_linked_extraction(ctx))) {
1033                 struct wim_lookup_table_entry *unnamed_lte;
1034
1035                 unnamed_lte = inode_unnamed_lte_resolved(dentry->d_inode);
1036                 if (unnamed_lte) {
1037                         unnamed_lte->extracted_file = TSTRDUP(path);
1038                         if (!unnamed_lte->extracted_file)
1039                                 return WIMLIB_ERR_NOMEM;
1040                 }
1041         } else if (inode->i_nlink > 1 && ctx->supported_features.hard_links) {
1042                 inode->i_extracted_file = TSTRDUP(path);
1043                 if (!inode->i_extracted_file)
1044                         return WIMLIB_ERR_NOMEM;
1045         }
1046         return 0;
1047
1048 symlink:
1049         ret = extract_multiimage_symlink(oldpath, path, ctx, dentry);
1050         if (ret)
1051                 return ret;
1052         dentry->was_hardlinked = 1;
1053         return 0;
1054
1055 hardlink:
1056         ret = extract_hardlink(oldpath, path, ctx);
1057         if (ret)
1058                 return ret;
1059         dentry->was_hardlinked = 1;
1060         return 0;
1061 }
1062
1063 static int
1064 dentry_extract_skeleton(struct wim_dentry *dentry, void *_ctx)
1065 {
1066         struct apply_ctx *ctx = _ctx;
1067         tchar path[ctx->ops->path_max];
1068         struct wim_dentry *orig_dentry;
1069         struct wim_dentry *other_dentry;
1070         int ret;
1071
1072         /* Here we may re-order the extraction of multiple names (hard links)
1073          * for the same file in the same directory in order to ensure the short
1074          * (DOS) name is set correctly.  A short name is always associated with
1075          * exactly one long name, and at least on NTFS, only one long name for a
1076          * file can have a short name associated with it.  (More specifically,
1077          * there can be unlimited names in the POSIX namespace, but only one
1078          * name can be in the Win32+DOS namespace, or one name in the Win32
1079          * namespace with a corresponding name in the DOS namespace.) To ensure
1080          * the short name of a file is associated with the correct long name in
1081          * a directory, we extract the long name with a corresponding short name
1082          * before any additional names.  This can affect NTFS-3g extraction
1083          * (which uses ntfs_set_ntfs_dos_name(), which doesn't allow specifying
1084          * the long name to associate with a short name) and may affect Win32
1085          * extraction as well (which uses SetFileShortName()).  */
1086
1087         if (dentry->skeleton_extracted)
1088                 return 0;
1089         orig_dentry = NULL;
1090         if (ctx->supported_features.short_names
1091             && !dentry_has_short_name(dentry)
1092             && !dentry->d_inode->i_dos_name_extracted)
1093         {
1094                 inode_for_each_dentry(other_dentry, dentry->d_inode) {
1095                         if (dentry_has_short_name(other_dentry)
1096                             && !other_dentry->skeleton_extracted
1097                             && other_dentry->parent == dentry->parent)
1098                         {
1099                                 DEBUG("Creating %"TS" before %"TS" "
1100                                       "to guarantee correct DOS name extraction",
1101                                       dentry_full_path(other_dentry),
1102                                       dentry_full_path(dentry));
1103                                 orig_dentry = dentry;
1104                                 dentry = other_dentry;
1105                                 break;
1106                         }
1107                 }
1108         }
1109 again:
1110         if (!build_extraction_path(path, dentry, ctx))
1111                 return 0;
1112         ret = do_dentry_extract_skeleton(path, dentry, ctx);
1113         if (ret)
1114                 return ret;
1115
1116         dentry->skeleton_extracted = 1;
1117
1118         if (orig_dentry) {
1119                 dentry = orig_dentry;
1120                 orig_dentry = NULL;
1121                 goto again;
1122         }
1123         dentry->d_inode->i_dos_name_extracted = 1;
1124         return 0;
1125 }
1126
1127 /* Create a file or directory, then immediately extract all streams.  This
1128  * assumes that WIMLIB_EXTRACT_FLAG_SEQUENTIAL is not specified, since the WIM
1129  * may not be read sequentially by this function.  */
1130 static int
1131 dentry_extract(struct wim_dentry *dentry, void *_ctx)
1132 {
1133         struct apply_ctx *ctx = _ctx;
1134         tchar path[ctx->ops->path_max];
1135         int ret;
1136
1137         ret = dentry_extract_skeleton(dentry, ctx);
1138         if (ret)
1139                 return ret;
1140
1141         if (!build_extraction_path(path, dentry, ctx))
1142                 return 0;
1143
1144         return extract_streams(path, ctx, dentry, NULL, NULL);
1145 }
1146
1147 /* Extract all instances of the stream @lte that are being extracted in this
1148  * call of extract_tree().  @can_seek specifies whether the WIM file descriptor
1149  * is seekable or not (e.g. is a pipe).  If not and the stream needs to be
1150  * extracted multiple times, it is extracted to a temporary file first.
1151  *
1152  * This is intended for use with sequential extraction of a WIM image
1153  * (WIMLIB_EXTRACT_FLAG_SEQUENTIAL specified).  */
1154 static int
1155 extract_stream_instances(struct wim_lookup_table_entry *lte,
1156                          struct apply_ctx *ctx, bool can_seek)
1157 {
1158         struct wim_dentry **lte_dentries;
1159         struct wim_lookup_table_entry *lte_tmp = NULL;
1160         struct wim_lookup_table_entry *lte_override;
1161         tchar *stream_tmp_filename = NULL;
1162         tchar path[ctx->ops->path_max];
1163         unsigned i;
1164         int ret;
1165
1166         if (lte->out_refcnt <= ARRAY_LEN(lte->inline_lte_dentries))
1167                 lte_dentries = lte->inline_lte_dentries;
1168         else
1169                 lte_dentries = lte->lte_dentries;
1170
1171         if (likely(can_seek || lte->out_refcnt < 2)) {
1172                 lte_override = lte;
1173         } else {
1174                 /* Need to extract stream to temporary file.  */
1175                 struct filedes fd;
1176                 int raw_fd;
1177
1178                 stream_tmp_filename = ttempnam(NULL, T("wimlib"));
1179                 if (!stream_tmp_filename) {
1180                         ERROR_WITH_ERRNO("Failed to create temporary filename");
1181                         ret = WIMLIB_ERR_OPEN;
1182                         goto out;
1183                 }
1184
1185                 lte_tmp = memdup(lte, sizeof(struct wim_lookup_table_entry));
1186                 if (!lte_tmp) {
1187                         ret = WIMLIB_ERR_NOMEM;
1188                         goto out_free_stream_tmp_filename;
1189                 }
1190                 lte_tmp->resource_location = RESOURCE_IN_FILE_ON_DISK;
1191                 lte_tmp->file_on_disk = stream_tmp_filename;
1192                 lte_override = lte_tmp;
1193
1194                 raw_fd = topen(stream_tmp_filename,
1195                                O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, 0600);
1196                 if (raw_fd < 0) {
1197                         ERROR_WITH_ERRNO("Failed to open temporary file");
1198                         ret = WIMLIB_ERR_OPEN;
1199                         goto out_free_lte_tmp;
1200                 }
1201                 filedes_init(&fd, raw_fd);
1202                 ret = extract_wim_resource_to_fd(lte, &fd,
1203                                                  wim_resource_size(lte));
1204                 if (filedes_close(&fd) && !ret)
1205                         ret = WIMLIB_ERR_WRITE;
1206                 if (ret)
1207                         goto out_unlink_stream_tmp_file;
1208         }
1209
1210         /* Extract all instances of the stream, reading either from the stream
1211          * in the WIM file or from the temporary file containing the stream.
1212          * dentry->tmp_flag is used to ensure that each dentry is processed only
1213          * once regardless of how many times this stream appears in the streams
1214          * of the corresponding inode.  */
1215         for (i = 0; i < lte->out_refcnt; i++) {
1216                 struct wim_dentry *dentry = lte_dentries[i];
1217
1218                 if (dentry->tmp_flag)
1219                         continue;
1220                 if (!build_extraction_path(path, dentry, ctx))
1221                         continue;
1222                 ret = extract_streams(path, ctx, dentry,
1223                                       lte, lte_override);
1224                 if (ret)
1225                         goto out_clear_tmp_flags;
1226                 dentry->tmp_flag = 1;
1227         }
1228         ret = 0;
1229 out_clear_tmp_flags:
1230         for (i = 0; i < lte->out_refcnt; i++)
1231                 lte_dentries[i]->tmp_flag = 0;
1232 out_unlink_stream_tmp_file:
1233         if (stream_tmp_filename)
1234                 tunlink(stream_tmp_filename);
1235 out_free_lte_tmp:
1236         FREE(lte_tmp);
1237 out_free_stream_tmp_filename:
1238         FREE(stream_tmp_filename);
1239 out:
1240         return ret;
1241 }
1242
1243 /* Extracts a list of streams (ctx.stream_list), assuming that the directory
1244  * structure and empty files were already created.  This relies on the
1245  * per-`struct wim_lookup_table_entry' list of dentries that reference each
1246  * stream that was constructed earlier.  Streams are extracted exactly in the
1247  * order of the stream list; however, unless the WIM's file descriptor is
1248  * detected to be non-seekable, streams may be read from the WIM file more than
1249  * one time if multiple copies need to be extracted.  */
1250 static int
1251 extract_stream_list(struct apply_ctx *ctx)
1252 {
1253         struct wim_lookup_table_entry *lte;
1254         bool can_seek;
1255         int ret;
1256
1257         can_seek = (lseek(ctx->wim->in_fd.fd, 0, SEEK_CUR) != -1);
1258         list_for_each_entry(lte, &ctx->stream_list, extraction_list) {
1259                 ret = extract_stream_instances(lte, ctx, can_seek);
1260                 if (ret)
1261                         return ret;
1262         }
1263         return 0;
1264 }
1265
1266 #define PWM_ALLOW_WIM_HDR 0x00001
1267 #define PWM_SILENT_EOF    0x00002
1268
1269 /* Read the header from a stream in a pipable WIM.  */
1270 static int
1271 read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
1272                        int flags, struct wim_header_disk *hdr_ret)
1273 {
1274         union {
1275                 struct pwm_stream_hdr stream_hdr;
1276                 struct wim_header_disk pwm_hdr;
1277         } buf;
1278         int ret;
1279
1280         ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
1281         if (ret)
1282                 goto read_error;
1283
1284         if ((flags & PWM_ALLOW_WIM_HDR) && buf.stream_hdr.magic == PWM_MAGIC) {
1285                 BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
1286                 ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
1287                                 sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
1288
1289                 if (ret)
1290                         goto read_error;
1291                 lte->resource_location = RESOURCE_NONEXISTENT;
1292                 memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
1293                 return 0;
1294         }
1295
1296         if (buf.stream_hdr.magic != PWM_STREAM_MAGIC) {
1297                 ERROR("Data read on pipe is invalid (expected stream header).");
1298                 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
1299         }
1300
1301         lte->resource_entry.original_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
1302         copy_hash(lte->hash, buf.stream_hdr.hash);
1303         lte->resource_entry.flags = le32_to_cpu(buf.stream_hdr.flags);
1304         lte->resource_entry.offset = pwm->in_fd.offset;
1305         lte->resource_location = RESOURCE_IN_WIM;
1306         lte->wim = pwm;
1307         if (lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED) {
1308                 lte->compression_type = pwm->compression_type;
1309                 lte->resource_entry.size = 0;
1310         } else {
1311                 lte->compression_type = WIMLIB_COMPRESSION_TYPE_NONE;
1312                 lte->resource_entry.size = lte->resource_entry.original_size;
1313         }
1314         lte->is_pipable = 1;
1315         return 0;
1316
1317 read_error:
1318         if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF))
1319                 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
1320         return ret;
1321 }
1322
1323 /* Skip over an unneeded stream in a pipable WIM being read from a pipe.  */
1324 static int
1325 skip_pwm_stream(struct wim_lookup_table_entry *lte)
1326 {
1327         return read_partial_wim_resource(lte, wim_resource_size(lte),
1328                                          NULL, NULL,
1329                                          WIMLIB_READ_RESOURCE_FLAG_SEEK_ONLY,
1330                                          0);
1331 }
1332
1333 static int
1334 extract_streams_from_pipe(struct apply_ctx *ctx)
1335 {
1336         struct wim_lookup_table_entry *found_lte;
1337         struct wim_lookup_table_entry *needed_lte;
1338         struct wim_lookup_table *lookup_table;
1339         struct wim_header_disk pwm_hdr;
1340         int ret;
1341         int pwm_flags;
1342
1343         ret = WIMLIB_ERR_NOMEM;
1344         found_lte = new_lookup_table_entry();
1345         if (!found_lte)
1346                 goto out;
1347
1348         lookup_table = ctx->wim->lookup_table;
1349         pwm_flags = PWM_ALLOW_WIM_HDR;
1350         if ((ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1351                 pwm_flags |= PWM_SILENT_EOF;
1352         memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GID_LEN);
1353         ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
1354         ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
1355         if (ctx->progress_func)
1356                 ctx->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1357                                    &ctx->progress);
1358         while (ctx->num_streams_remaining) {
1359                 ret = read_pwm_stream_header(ctx->wim, found_lte, pwm_flags,
1360                                              &pwm_hdr);
1361                 if (ret) {
1362                         if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE &&
1363                             (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_RESUME))
1364                         {
1365                                 goto resume_done;
1366                         }
1367                         goto out_free_found_lte;
1368                 }
1369
1370                 if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
1371                     && !(found_lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA)
1372                     && (needed_lte = __lookup_resource(lookup_table, found_lte->hash))
1373                     && (needed_lte->out_refcnt))
1374                 {
1375                         copy_resource_entry(&needed_lte->resource_entry,
1376                                             &found_lte->resource_entry);
1377                         needed_lte->resource_location = found_lte->resource_location;
1378                         needed_lte->wim               = found_lte->wim;
1379                         needed_lte->compression_type  = found_lte->compression_type;
1380                         needed_lte->is_pipable        = found_lte->is_pipable;
1381
1382                         ret = extract_stream_instances(needed_lte, ctx, false);
1383                         if (ret)
1384                                 goto out_free_found_lte;
1385                         ctx->num_streams_remaining--;
1386                 } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
1387                         ret = skip_pwm_stream(found_lte);
1388                         if (ret)
1389                                 goto out_free_found_lte;
1390                 } else {
1391                         u16 part_number = le16_to_cpu(pwm_hdr.part_number);
1392                         u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
1393
1394                         if (part_number != ctx->progress.extract.part_number ||
1395                             total_parts != ctx->progress.extract.total_parts ||
1396                             memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
1397                                    WIM_GID_LEN))
1398                         {
1399                                 ctx->progress.extract.part_number = part_number;
1400                                 ctx->progress.extract.total_parts = total_parts;
1401                                 memcpy(ctx->progress.extract.guid,
1402                                        pwm_hdr.guid, WIM_GID_LEN);
1403                                 if (ctx->progress_func) {
1404                                         ctx->progress_func(
1405                                                 WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
1406                                                            &ctx->progress);
1407                                 }
1408
1409                         }
1410                 }
1411         }
1412         ret = 0;
1413 out_free_found_lte:
1414         free_lookup_table_entry(found_lte);
1415 out:
1416         return ret;
1417
1418 resume_done:
1419         /* TODO */
1420         return 0;
1421 }
1422
1423 /* Finish extracting a file, directory, or symbolic link by setting file
1424  * security and timestamps.  */
1425 static int
1426 dentry_extract_final(struct wim_dentry *dentry, void *_ctx)
1427 {
1428         struct apply_ctx *ctx = _ctx;
1429         int ret;
1430         tchar path[ctx->ops->path_max];
1431
1432         if (!build_extraction_path(path, dentry, ctx))
1433                 return 0;
1434
1435         ret = extract_security(path, ctx, dentry);
1436         if (ret)
1437                 return ret;
1438
1439         return extract_timestamps(path, ctx, dentry);
1440 }
1441
1442 /*
1443  * Extract a WIM dentry to standard output.
1444  *
1445  * This obviously doesn't make sense in all cases.  We return an error if the
1446  * dentry does not correspond to a regular file.  Otherwise we extract the
1447  * unnamed data stream only.
1448  */
1449 static int
1450 extract_dentry_to_stdout(struct wim_dentry *dentry)
1451 {
1452         int ret = 0;
1453         if (dentry->d_inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
1454                                              FILE_ATTRIBUTE_DIRECTORY))
1455         {
1456                 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
1457                       "extracted to standard output", dentry_full_path(dentry));
1458                 ret = WIMLIB_ERR_NOT_A_REGULAR_FILE;
1459         } else {
1460                 struct wim_lookup_table_entry *lte;
1461
1462                 lte = inode_unnamed_lte_resolved(dentry->d_inode);
1463                 if (lte) {
1464                         struct filedes _stdout;
1465                         filedes_init(&_stdout, STDOUT_FILENO);
1466                         ret = extract_wim_resource_to_fd(lte, &_stdout,
1467                                                          wim_resource_size(lte));
1468                 }
1469         }
1470         return ret;
1471 }
1472
1473 #ifdef __WIN32__
1474 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
1475 #else
1476 static const utf16lechar replacement_char = cpu_to_le16('?');
1477 #endif
1478
1479 static bool
1480 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
1481 {
1482         size_t i;
1483
1484         if (num_chars == 0)
1485                 return true;
1486         for (i = 0; i < num_chars; i++) {
1487                 switch (name[i]) {
1488         #ifdef __WIN32__
1489                 case cpu_to_le16('\\'):
1490                 case cpu_to_le16(':'):
1491                 case cpu_to_le16('*'):
1492                 case cpu_to_le16('?'):
1493                 case cpu_to_le16('"'):
1494                 case cpu_to_le16('<'):
1495                 case cpu_to_le16('>'):
1496                 case cpu_to_le16('|'):
1497         #endif
1498                 case cpu_to_le16('/'):
1499                 case cpu_to_le16('\0'):
1500                         if (fix)
1501                                 name[i] = replacement_char;
1502                         else
1503                                 return false;
1504                 }
1505         }
1506
1507 #ifdef __WIN32__
1508         if (name[num_chars - 1] == cpu_to_le16(' ') ||
1509             name[num_chars - 1] == cpu_to_le16('.'))
1510         {
1511                 if (fix)
1512                         name[num_chars - 1] = replacement_char;
1513                 else
1514                         return false;
1515         }
1516 #endif
1517         return true;
1518 }
1519
1520 static bool
1521 dentry_is_dot_or_dotdot(const struct wim_dentry *dentry)
1522 {
1523         const utf16lechar *file_name = dentry->file_name;
1524         return file_name != NULL &&
1525                 file_name[0] == cpu_to_le16('.') &&
1526                 (file_name[1] == cpu_to_le16('\0') ||
1527                  (file_name[1] == cpu_to_le16('.') &&
1528                   file_name[2] == cpu_to_le16('\0')));
1529 }
1530
1531 static int
1532 dentry_mark_skipped(struct wim_dentry *dentry, void *_ignore)
1533 {
1534         dentry->extraction_skipped = 1;
1535         return 0;
1536 }
1537
1538 /*
1539  * dentry_calculate_extraction_path-
1540  *
1541  * Calculate the actual filename component at which a WIM dentry will be
1542  * extracted, handling invalid filenames "properly".
1543  *
1544  * dentry->extraction_name usually will be set the same as dentry->file_name (on
1545  * UNIX, converted into the platform's multibyte encoding).  However, if the
1546  * file name contains characters that are not valid on the current platform or
1547  * has some other format that is not valid, leave dentry->extraction_name as
1548  * NULL and set dentry->extraction_skipped to indicate that this dentry should
1549  * not be extracted, unless the appropriate flag
1550  * WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES is set in the extract flags, in
1551  * which case a substitute filename will be created and set instead.
1552  *
1553  * Conflicts with case-insensitive names on Windows are handled similarly; see
1554  * below.
1555  */
1556 static int
1557 dentry_calculate_extraction_path(struct wim_dentry *dentry, void *_args)
1558 {
1559         struct apply_ctx *ctx = _args;
1560         int ret;
1561
1562         if (dentry == ctx->extract_root || dentry->extraction_skipped)
1563                 return 0;
1564
1565         if (!dentry_is_supported(dentry, &ctx->supported_features))
1566                 goto skip_dentry;
1567
1568         if (dentry_is_dot_or_dotdot(dentry)) {
1569                 /* WIM files shouldn't contain . or .. entries.  But if they are
1570                  * there, don't attempt to extract them. */
1571                 WARNING("Skipping extraction of unexpected . or .. file "
1572                         "\"%"TS"\"", dentry_full_path(dentry));
1573                 goto skip_dentry;
1574         }
1575
1576 #ifdef __WIN32__
1577         if (!ctx->ops->supports_case_sensitive_filenames)
1578         {
1579                 struct wim_dentry *other;
1580                 list_for_each_entry(other, &dentry->case_insensitive_conflict_list,
1581                                     case_insensitive_conflict_list)
1582                 {
1583                         if (ctx->extract_flags &
1584                             WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
1585                                 WARNING("\"%"TS"\" has the same "
1586                                         "case-insensitive name as "
1587                                         "\"%"TS"\"; extracting "
1588                                         "dummy name instead",
1589                                         dentry_full_path(dentry),
1590                                         dentry_full_path(other));
1591                                 goto out_replace;
1592                         } else {
1593                                 WARNING("Not extracting \"%"TS"\": "
1594                                         "has same case-insensitive "
1595                                         "name as \"%"TS"\"",
1596                                         dentry_full_path(dentry),
1597                                         dentry_full_path(other));
1598                                 goto skip_dentry;
1599                         }
1600                 }
1601         }
1602 #else   /* __WIN32__ */
1603         wimlib_assert(ctx->ops->supports_case_sensitive_filenames);
1604 #endif  /* !__WIN32__ */
1605
1606         if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
1607 #ifdef __WIN32__
1608                 dentry->extraction_name = dentry->file_name;
1609                 dentry->extraction_name_nchars = dentry->file_name_nbytes / 2;
1610                 return 0;
1611 #else
1612                 return utf16le_to_tstr(dentry->file_name,
1613                                        dentry->file_name_nbytes,
1614                                        &dentry->extraction_name,
1615                                        &dentry->extraction_name_nchars);
1616 #endif
1617         } else {
1618                 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
1619                 {
1620                         WARNING("\"%"TS"\" has an invalid filename "
1621                                 "that is not supported on this platform; "
1622                                 "extracting dummy name instead",
1623                                 dentry_full_path(dentry));
1624                         goto out_replace;
1625                 } else {
1626                         WARNING("Not extracting \"%"TS"\": has an invalid filename "
1627                                 "that is not supported on this platform",
1628                                 dentry_full_path(dentry));
1629                         goto skip_dentry;
1630                 }
1631         }
1632
1633 out_replace:
1634         {
1635                 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
1636
1637                 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
1638                 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
1639
1640                 tchar *tchar_name;
1641                 size_t tchar_nchars;
1642         #ifdef __WIN32__
1643                 tchar_name = utf16_name_copy;
1644                 tchar_nchars = dentry->file_name_nbytes / 2;
1645         #else
1646                 ret = utf16le_to_tstr(utf16_name_copy,
1647                                       dentry->file_name_nbytes,
1648                                       &tchar_name, &tchar_nchars);
1649                 if (ret)
1650                         return ret;
1651         #endif
1652                 size_t fixed_name_num_chars = tchar_nchars;
1653                 tchar fixed_name[tchar_nchars + 50];
1654
1655                 tmemcpy(fixed_name, tchar_name, tchar_nchars);
1656                 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
1657                                                  T(" (invalid filename #%lu)"),
1658                                                  ++ctx->invalid_sequence);
1659         #ifndef __WIN32__
1660                 FREE(tchar_name);
1661         #endif
1662                 dentry->extraction_name = memdup(fixed_name,
1663                                                  2 * fixed_name_num_chars + 2);
1664                 if (!dentry->extraction_name)
1665                         return WIMLIB_ERR_NOMEM;
1666                 dentry->extraction_name_nchars = fixed_name_num_chars;
1667         }
1668         return 0;
1669
1670 skip_dentry:
1671         for_dentry_in_tree(dentry, dentry_mark_skipped, NULL);
1672         return 0;
1673 }
1674
1675 /* Clean up dentry and inode structure after extraction.  */
1676 static int
1677 dentry_reset_needs_extraction(struct wim_dentry *dentry, void *_ignore)
1678 {
1679         struct wim_inode *inode = dentry->d_inode;
1680
1681         dentry->extraction_skipped = 0;
1682         dentry->was_hardlinked = 0;
1683         dentry->skeleton_extracted = 0;
1684         inode->i_visited = 0;
1685         FREE(inode->i_extracted_file);
1686         inode->i_extracted_file = NULL;
1687         inode->i_dos_name_extracted = 0;
1688         if ((void*)dentry->extraction_name != (void*)dentry->file_name)
1689                 FREE(dentry->extraction_name);
1690         dentry->extraction_name = NULL;
1691         return 0;
1692 }
1693
1694 /* Tally features necessary to extract a dentry and the corresponding inode.  */
1695 static int
1696 dentry_tally_features(struct wim_dentry *dentry, void *_features)
1697 {
1698         struct wim_features *features = _features;
1699         struct wim_inode *inode = dentry->d_inode;
1700
1701         if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
1702                 features->archive_files++;
1703         if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
1704                 features->hidden_files++;
1705         if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
1706                 features->system_files++;
1707         if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
1708                 features->compressed_files++;
1709         if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED)
1710                 features->encrypted_files++;
1711         if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
1712                 features->not_context_indexed_files++;
1713         if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
1714                 features->sparse_files++;
1715         if (inode_has_named_stream(inode))
1716                 features->named_data_streams++;
1717         if (inode->i_visited)
1718                 features->hard_links++;
1719         if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
1720                 features->reparse_points++;
1721                 if (inode_is_symlink(inode))
1722                         features->symlink_reparse_points++;
1723                 else
1724                         features->other_reparse_points++;
1725         }
1726         if (inode->i_security_id != -1)
1727                 features->security_descriptors++;
1728         if (dentry->short_name_nbytes)
1729                 features->short_names++;
1730         if (inode_has_unix_data(inode))
1731                 features->unix_data++;
1732         inode->i_visited = 1;
1733         return 0;
1734 }
1735
1736 static int
1737 dentry_clear_inode_visited(struct wim_dentry *dentry, void *_ignore)
1738 {
1739         dentry->d_inode->i_visited = 0;
1740         return 0;
1741 }
1742
1743 /* Tally the features necessary to extract a dentry tree.  */
1744 static void
1745 dentry_tree_get_features(struct wim_dentry *root, struct wim_features *features)
1746 {
1747         memset(features, 0, sizeof(struct wim_features));
1748         for_dentry_in_tree(root, dentry_tally_features, features);
1749         for_dentry_in_tree(root, dentry_clear_inode_visited, NULL);
1750 }
1751
1752 static int
1753 do_feature_check(const struct wim_features *required_features,
1754                  const struct wim_features *supported_features,
1755                  int extract_flags,
1756                  const struct apply_operations *ops,
1757                  const tchar *wim_source_path)
1758 {
1759         const tchar *loc;
1760         const tchar *mode = "this extraction mode";
1761
1762         if (wim_source_path[0] == '\0')
1763                 loc = "the WIM image";
1764         else
1765                 loc = wim_source_path;
1766
1767         /* We're an archive program, so theoretically we can do what we want
1768          * with FILE_ATTRIBUTE_ARCHIVE (which is a dumb flag anyway).  Don't
1769          * bother the user about it.  */
1770 #if 0
1771         if (required_features->archive_files && !supported_features->archive_files)
1772         {
1773                 WARNING(
1774           "%lu files in %"TS" are marked as archived, but this attribute\n"
1775 "          is not supported in %"TS".",
1776                         required_features->archive_files, loc, mode);
1777         }
1778 #endif
1779
1780         if (required_features->hidden_files && !supported_features->hidden_files)
1781         {
1782                 WARNING(
1783           "%lu files in %"TS" are marked as hidden, but this\n"
1784 "          attribute is not supported in %"TS".",
1785                         required_features->hidden_files, loc, mode);
1786         }
1787
1788         if (required_features->system_files && !supported_features->system_files)
1789         {
1790                 WARNING(
1791           "%lu files in %"TS" are marked as system files,\n"
1792 "          but this attribute is not supported in %"TS".",
1793                         required_features->system_files, loc, mode);
1794         }
1795
1796         if (required_features->compressed_files && !supported_features->compressed_files)
1797         {
1798                 WARNING(
1799           "%lu files in %"TS" are marked as being transparently\n"
1800 "          compressed, but transparent compression is not supported in\n"
1801 "          %"TS".  These files will be extracted as uncompressed.",
1802                         required_features->compressed_files, loc, mode);
1803         }
1804
1805         if (required_features->encrypted_files && !supported_features->encrypted_files)
1806         {
1807                 WARNING(
1808           "%lu files in %"TS" are marked as being encrypted,\n"
1809 "           but encryption is not supported in %"TS".  These files\n"
1810 "           will be extracted as raw encrypted data instead.",
1811                         required_features->encrypted_files, loc, mode);
1812         }
1813
1814         if (required_features->not_context_indexed_files &&
1815             !supported_features->not_context_indexed_files)
1816         {
1817                 WARNING(
1818           "%lu files in %"TS" are marked as not content indexed,\n"
1819 "          but this attribute is not supported in %"TS".",
1820                         required_features->not_context_indexed_files, loc, mode);
1821         }
1822
1823         if (required_features->sparse_files && !supported_features->sparse_files)
1824         {
1825                 WARNING(
1826           "%lu files in %"TS" are marked as sparse, but creating\n"
1827 "           sparse files is not supported in %"TS".  These files\n"
1828 "           will be extracted as non-sparse.",
1829                         required_features->sparse_files, loc, mode);
1830         }
1831
1832         if (required_features->named_data_streams &&
1833             !supported_features->named_data_streams)
1834         {
1835                 WARNING(
1836           "%lu files in %"TS" contain one or more alternate (named)\n"
1837 "          data streams, which are not supported in %"TS".\n"
1838 "          Alternate data streams will NOT be extracted.",
1839                         required_features->named_data_streams, loc, mode);
1840         }
1841
1842         if (unlikely(extract_flags & (WIMLIB_EXTRACT_FLAG_HARDLINK |
1843                                       WIMLIB_EXTRACT_FLAG_SYMLINK)) &&
1844             required_features->named_data_streams &&
1845             supported_features->named_data_streams)
1846         {
1847                 WARNING(
1848           "%lu files in %"TS" contain one or more alternate (named)\n"
1849 "          data streams, which are not supported in linked extraction mode.\n"
1850 "          Alternate data streams will NOT be extracted.",
1851                         required_features->named_data_streams, loc);
1852         }
1853
1854         if (required_features->hard_links && !supported_features->hard_links)
1855         {
1856                 WARNING(
1857           "%lu files in %"TS" are hard links, but hard links are\n"
1858 "          not supported in %"TS".  Hard links will be extracted as\n"
1859 "          duplicate copies of the linked files.",
1860                         required_features->hard_links, loc, mode);
1861         }
1862
1863         if (required_features->reparse_points && !supported_features->reparse_points)
1864         {
1865                 if (supported_features->symlink_reparse_points) {
1866                         if (required_features->other_reparse_points) {
1867                                 WARNING(
1868           "%lu files in %"TS" are reparse points that are neither\n"
1869 "          symbolic links nor junction points and are not supported in\n"
1870 "          %"TS".  These reparse points will not be extracted.",
1871                                         required_features->other_reparse_points, loc,
1872                                         mode);
1873                         }
1874                 } else {
1875                         WARNING(
1876           "%lu files in %"TS" are reparse points, which are\n"
1877 "          not supported in %"TS" and will not be extracted.",
1878                                 required_features->reparse_points, loc, mode);
1879                 }
1880         }
1881
1882         if (required_features->security_descriptors &&
1883             !supported_features->security_descriptors)
1884         {
1885                 WARNING(
1886           "%lu files in %"TS" have Windows NT security descriptors,\n"
1887 "          but extracting security descriptors is not supported in\n"
1888 "          %"TS".  No security descriptors will be extracted.",
1889                         required_features->security_descriptors, loc, mode);
1890         }
1891
1892         if (required_features->short_names && !supported_features->short_names)
1893         {
1894                 WARNING(
1895           "%lu files in %"TS" have short (DOS) names, but\n"
1896 "          extracting short names is not supported in %"TS".\n"
1897 "          Short names will not be extracted.\n",
1898                         required_features->short_names, loc, mode);
1899         }
1900
1901         if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
1902             required_features->unix_data && !supported_features->unix_data)
1903         {
1904                 ERROR("Extracting UNIX data is not supported in %"TS, mode);
1905                 return WIMLIB_ERR_UNSUPPORTED;
1906         }
1907         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) &&
1908             required_features->short_names && !supported_features->short_names)
1909         {
1910                 ERROR("Extracting short names is not supported in %"TS"", mode);
1911                 return WIMLIB_ERR_UNSUPPORTED;
1912         }
1913         if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
1914             !ops->set_timestamps)
1915         {
1916                 ERROR("Extracting timestamps is not supported in %"TS"", mode);
1917                 return WIMLIB_ERR_UNSUPPORTED;
1918         }
1919         if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
1920                                WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1921              == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
1922             required_features->security_descriptors &&
1923             !supported_features->security_descriptors)
1924         {
1925                 ERROR("Extracting security descriptors is not supported in %"TS, mode);
1926                 return WIMLIB_ERR_UNSUPPORTED;
1927         }
1928
1929         if ((extract_flags & WIMLIB_EXTRACT_FLAG_HARDLINK) &&
1930             !supported_features->hard_links)
1931         {
1932                 ERROR("Hard link extraction mode requested, but "
1933                       "%"TS" does not support hard links!", mode);
1934                 return WIMLIB_ERR_UNSUPPORTED;
1935         }
1936
1937         if ((extract_flags & WIMLIB_EXTRACT_FLAG_SYMLINK) &&
1938             !supported_features->symlink_reparse_points)
1939         {
1940                 ERROR("Symbolic link extraction mode requested, but "
1941                       "%"TS" does not support symbolic "
1942                       "links!", mode);
1943                 return WIMLIB_ERR_UNSUPPORTED;
1944         }
1945         return 0;
1946 }
1947
1948 /*
1949  * extract_tree - Extract a file or directory tree from the currently selected
1950  *                WIM image.
1951  *
1952  * @wim:        WIMStruct for the WIM file, with the desired image selected
1953  *              (as wim->current_image).
1954  *
1955  * @wim_source_path:
1956  *              "Canonical" (i.e. no leading or trailing slashes, path
1957  *              separators WIM_PATH_SEPARATOR) path inside the WIM image to
1958  *              extract.  An empty string means the full image.
1959  *
1960  * @target:
1961  *              Filesystem path to extract the file or directory tree to.
1962  *              (Or, with WIMLIB_EXTRACT_FLAG_NTFS: the name of a NTFS volume.)
1963  *
1964  * @extract_flags:
1965  *              WIMLIB_EXTRACT_FLAG_*.  Also, the private flag
1966  *              WIMLIB_EXTRACT_FLAG_MULTI_IMAGE will be set if this is being
1967  *              called through wimlib_extract_image() with WIMLIB_ALL_IMAGES as
1968  *              the image.
1969  *
1970  * @progress_func:
1971  *              If non-NULL, progress function for the extraction.  The messages
1972  *              that may be sent in this function are:
1973  *
1974  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN or
1975  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
1976  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN;
1977  *              WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END;
1978  *              WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY;
1979  *              WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS;
1980  *              WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS;
1981  *              WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END or
1982  *                      WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END.
1983  *
1984  * Returns 0 on success; a positive WIMLIB_ERR_* code on failure.
1985  */
1986 static int
1987 extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target,
1988              int extract_flags, wimlib_progress_func_t progress_func)
1989 {
1990         struct wim_dentry *root;
1991         struct wim_features required_features;
1992         struct apply_ctx ctx;
1993         int ret;
1994         struct wim_lookup_table_entry *lte;
1995
1996         /* Start initializing the apply_ctx.  */
1997         memset(&ctx, 0, sizeof(struct apply_ctx));
1998         ctx.wim = wim;
1999         ctx.extract_flags = extract_flags;
2000         ctx.target = target;
2001         ctx.target_nchars = tstrlen(target);
2002         ctx.progress_func = progress_func;
2003         if (progress_func) {
2004                 ctx.progress.extract.wimfile_name = wim->filename;
2005                 ctx.progress.extract.image = wim->current_image;
2006                 ctx.progress.extract.extract_flags = (extract_flags &
2007                                                       WIMLIB_EXTRACT_MASK_PUBLIC);
2008                 ctx.progress.extract.image_name = wimlib_get_image_name(wim,
2009                                                                         wim->current_image);
2010                 ctx.progress.extract.extract_root_wim_source_path = wim_source_path;
2011                 ctx.progress.extract.target = target;
2012         }
2013         INIT_LIST_HEAD(&ctx.stream_list);
2014
2015         /* Translate the path to extract into the corresponding
2016          * `struct wim_dentry', which will be the root of the
2017          * "dentry tree" to extract.  */
2018         root = get_dentry(wim, wim_source_path);
2019         if (!root) {
2020                 ERROR("Path \"%"TS"\" does not exist in WIM image %d",
2021                       wim_source_path, wim->current_image);
2022                 ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
2023                 goto out;
2024         }
2025
2026         ctx.extract_root = root;
2027
2028         /* Select the appropriate apply_operations based on the
2029          * platform and extract_flags.  */
2030 #ifdef __WIN32__
2031         ctx.ops = &win32_apply_ops;
2032 #else
2033         ctx.ops = &unix_apply_ops;
2034 #endif
2035
2036 #ifdef WITH_NTFS_3G
2037         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
2038                 ctx.ops = &ntfs_3g_apply_ops;
2039 #endif
2040
2041         /* Call the start_extract() callback.  This gives the apply_operations
2042          * implementation a chance to do any setup needed to access the volume.
2043          * Furthermore, it's expected to set the supported features of this
2044          * extraction mode (ctx.supported_features), which are determined at
2045          * runtime as they may vary depending on the actual volume.  These
2046          * features are then compared with the actual features extracting this
2047          * dentry tree requires.  Some mismatches will merely produce warnings
2048          * and the unsupported data will be ignored; others will produce errors.
2049          */
2050         ret = ctx.ops->start_extract(target, &ctx);
2051         if (ret)
2052                 goto out;
2053
2054         dentry_tree_get_features(root, &required_features);
2055         ret = do_feature_check(&required_features, &ctx.supported_features,
2056                                extract_flags, ctx.ops, wim_source_path);
2057         if (ret)
2058                 goto out_finish_or_abort_extract;
2059
2060         /* Figure out whether the root dentry is being extracted to the root of
2061          * a volume and therefore needs to be treated "specially", for example
2062          * not being explicitly created and not having attributes set.  */
2063         if (ctx.ops->target_is_root && ctx.ops->root_directory_is_special)
2064                 ctx.root_dentry_is_special = ctx.ops->target_is_root(target);
2065
2066         /* Calculate the actual filename component of each extracted dentry.  In
2067          * the process, set the dentry->extraction_skipped flag on dentries that
2068          * are being skipped for some reason (e.g. invalid filename).  */
2069         ret = for_dentry_in_tree(root, dentry_calculate_extraction_path, &ctx);
2070         if (ret)
2071                 goto out_dentry_reset_needs_extraction;
2072
2073         /* Build the list of the streams that need to be extracted and
2074          * initialize ctx.progress.extract with stream information.  */
2075         ret = for_dentry_in_tree(ctx.extract_root,
2076                                  dentry_resolve_and_zero_lte_refcnt, &ctx);
2077         if (ret)
2078                 goto out_dentry_reset_needs_extraction;
2079
2080         ret = for_dentry_in_tree(ctx.extract_root,
2081                                  dentry_add_streams_to_extract, &ctx);
2082         if (ret)
2083                 goto out_teardown_stream_list;
2084
2085         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2086                 /* When extracting from a pipe, the number of bytes of data to
2087                  * extract can't be determined in the normal way (examining the
2088                  * lookup table), since at this point all we have is a set of
2089                  * SHA1 message digests of streams that need to be extracted.
2090                  * However, we can get a reasonably accurate estimate by taking
2091                  * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
2092                  * data.  This does assume that a full image is being extracted,
2093                  * but currently there is no API for doing otherwise.  (Also,
2094                  * subtract <HARDLINKBYTES> from this if hard links are
2095                  * supported by the extraction mode.)  */
2096                 ctx.progress.extract.total_bytes =
2097                         wim_info_get_image_total_bytes(wim->wim_info,
2098                                                        wim->current_image);
2099                 if (ctx.supported_features.hard_links) {
2100                         ctx.progress.extract.total_bytes -=
2101                                 wim_info_get_image_hard_link_bytes(wim->wim_info,
2102                                                                    wim->current_image);
2103                 }
2104         }
2105
2106         /* Handle the special case of extracting a file to standard
2107          * output.  In that case, "root" should be a single file, not a
2108          * directory tree.  (If not, extract_dentry_to_stdout() will
2109          * return an error.)  */
2110         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
2111                 ret = extract_dentry_to_stdout(root);
2112                 goto out_teardown_stream_list;
2113         }
2114
2115         /* If a sequential extraction was specified, sort the streams to be
2116          * extracted by their position in the WIM file so that the WIM file can
2117          * be read sequentially.  */
2118         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SEQUENTIAL |
2119                               WIMLIB_EXTRACT_FLAG_FROM_PIPE))
2120                                         == WIMLIB_EXTRACT_FLAG_SEQUENTIAL)
2121         {
2122                 ret = sort_stream_list_by_sequential_order(
2123                                 &ctx.stream_list,
2124                                 offsetof(struct wim_lookup_table_entry,
2125                                          extraction_list));
2126                 if (ret)
2127                         goto out_teardown_stream_list;
2128         }
2129
2130         if (ctx.ops->realpath_works_on_nonexisting_files &&
2131             ((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) ||
2132              ctx.ops->requires_realtarget_in_paths))
2133         {
2134                 ctx.realtarget = realpath(target, NULL);
2135                 if (!ctx.realtarget) {
2136                         ret = WIMLIB_ERR_NOMEM;
2137                         goto out_teardown_stream_list;
2138                 }
2139                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2140         }
2141
2142         if (progress_func) {
2143                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN :
2144                                                  WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
2145                               &ctx.progress);
2146         }
2147
2148         if (!ctx.root_dentry_is_special)
2149         {
2150                 tchar path[ctx.ops->path_max];
2151                 if (build_extraction_path(path, root, &ctx))
2152                 {
2153                         ret = extract_inode(path, &ctx, root->d_inode);
2154                         if (ret)
2155                                 goto out_free_realtarget;
2156                 }
2157         }
2158
2159         /* If we need to fix up the targets of absolute symbolic links
2160          * (WIMLIB_EXTRACT_FLAG_RPFIX) or the extraction mode requires paths to
2161          * be absolute, use realpath() (or its replacement on Windows) to get
2162          * the absolute path to the extraction target.  Note that this requires
2163          * the target directory to exist, unless
2164          * realpath_works_on_nonexisting_files is set in the apply_operations.
2165          * */
2166         if (!ctx.realtarget &&
2167             (((extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) &&
2168               required_features.symlink_reparse_points) ||
2169              ctx.ops->requires_realtarget_in_paths))
2170         {
2171                 ctx.realtarget = realpath(target, NULL);
2172                 if (!ctx.realtarget) {
2173                         ret = WIMLIB_ERR_NOMEM;
2174                         goto out_free_realtarget;
2175                 }
2176                 ctx.realtarget_nchars = tstrlen(ctx.realtarget);
2177         }
2178
2179         /* Finally, the important part: extract the tree of files.  */
2180         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SEQUENTIAL |
2181                              WIMLIB_EXTRACT_FLAG_FROM_PIPE)) {
2182                 /* Sequential extraction requested, so two passes are needed
2183                  * (one for directory structure, one for streams.)  */
2184                 if (progress_func)
2185                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2186                                       &ctx.progress);
2187
2188                 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_RESUME)) {
2189                         ret = for_dentry_in_tree(root, dentry_extract_skeleton, &ctx);
2190                         if (ret)
2191                                 goto out_free_realtarget;
2192                 }
2193                 if (progress_func)
2194                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2195                                       &ctx.progress);
2196                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
2197                         ret = extract_streams_from_pipe(&ctx);
2198                 else
2199                         ret = extract_stream_list(&ctx);
2200                 if (ret)
2201                         goto out_free_realtarget;
2202         } else {
2203                 /* Sequential extraction was not requested, so we can make do
2204                  * with one pass where we both create the files and extract
2205                  * streams.   */
2206                 if (progress_func)
2207                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
2208                                       &ctx.progress);
2209                 ret = for_dentry_in_tree(root, dentry_extract, &ctx);
2210                 if (ret)
2211                         goto out_free_realtarget;
2212                 if (progress_func)
2213                         progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
2214                                       &ctx.progress);
2215         }
2216
2217         /* If the total number of bytes to extract was miscalculated, just jump
2218          * to the calculated number in order to avoid confusing the progress
2219          * function.  This should only occur when extracting from a pipe.  */
2220         if (ctx.progress.extract.completed_bytes != ctx.progress.extract.total_bytes)
2221         {
2222                 DEBUG("Calculated %"PRIu64" bytes to extract, but actually "
2223                       "extracted %"PRIu64,
2224                       ctx.progress.extract.total_bytes,
2225                       ctx.progress.extract.completed_bytes);
2226         }
2227         if (progress_func &&
2228             ctx.progress.extract.completed_bytes < ctx.progress.extract.total_bytes)
2229         {
2230                 ctx.progress.extract.completed_bytes = ctx.progress.extract.total_bytes;
2231                 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, &ctx.progress);
2232         }
2233
2234         /* Apply security descriptors and timestamps.  This is done at the end,
2235          * and in a depth-first manner, to prevent timestamps from getting
2236          * changed by subsequent extract operations and to minimize the chance
2237          * of the restored security descriptors getting in our way.  */
2238         if (progress_func)
2239                 progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
2240                               &ctx.progress);
2241         ret = for_dentry_in_tree_depth(root, dentry_extract_final, &ctx);
2242         if (ret)
2243                 goto out_free_realtarget;
2244
2245         if (progress_func) {
2246                 progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END :
2247                               WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
2248                               &ctx.progress);
2249         }
2250
2251         ret = 0;
2252 out_free_realtarget:
2253         FREE(ctx.realtarget);
2254 out_teardown_stream_list:
2255         /* Free memory allocated as part of the mapping from each
2256          * wim_lookup_table_entry to the dentries that reference it.  */
2257         if (ctx.extract_flags & WIMLIB_EXTRACT_FLAG_SEQUENTIAL)
2258                 list_for_each_entry(lte, &ctx.stream_list, extraction_list)
2259                         if (lte->out_refcnt > ARRAY_LEN(lte->inline_lte_dentries))
2260                                 FREE(lte->lte_dentries);
2261 out_dentry_reset_needs_extraction:
2262         for_dentry_in_tree(root, dentry_reset_needs_extraction, NULL);
2263 out_finish_or_abort_extract:
2264         if (ret) {
2265                 if (ctx.ops->abort_extract)
2266                         ctx.ops->abort_extract(&ctx);
2267         } else {
2268                 if (ctx.ops->finish_extract)
2269                         ret = ctx.ops->finish_extract(&ctx);
2270         }
2271 out:
2272         return ret;
2273 }
2274
2275 /* Validates a single wimlib_extract_command, mostly checking to make sure the
2276  * extract flags make sense. */
2277 static int
2278 check_extract_command(struct wimlib_extract_command *cmd, int wim_header_flags)
2279 {
2280         int extract_flags;
2281
2282         /* Empty destination path? */
2283         if (cmd->fs_dest_path[0] == T('\0'))
2284                 return WIMLIB_ERR_INVALID_PARAM;
2285
2286         extract_flags = cmd->extract_flags;
2287
2288         /* Check for invalid flag combinations  */
2289         if ((extract_flags &
2290              (WIMLIB_EXTRACT_FLAG_SYMLINK |
2291               WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
2292                                                  WIMLIB_EXTRACT_FLAG_HARDLINK))
2293                 return WIMLIB_ERR_INVALID_PARAM;
2294
2295         if ((extract_flags &
2296              (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2297               WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
2298                                                     WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
2299                 return WIMLIB_ERR_INVALID_PARAM;
2300
2301         if ((extract_flags &
2302              (WIMLIB_EXTRACT_FLAG_RPFIX |
2303               WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
2304                                                 WIMLIB_EXTRACT_FLAG_NORPFIX))
2305                 return WIMLIB_ERR_INVALID_PARAM;
2306
2307         if ((extract_flags &
2308              (WIMLIB_EXTRACT_FLAG_RESUME |
2309               WIMLIB_EXTRACT_FLAG_FROM_PIPE)) == WIMLIB_EXTRACT_FLAG_RESUME)
2310                 return WIMLIB_ERR_INVALID_PARAM;
2311
2312         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2313 #ifndef WITH_NTFS_3G
2314                 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
2315                       "        we cannot apply a WIM image directly to a NTFS volume.");
2316                 return WIMLIB_ERR_UNSUPPORTED;
2317 #endif
2318         }
2319
2320         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
2321                               WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
2322         {
2323                 /* Do reparse point fixups by default if the WIM header says
2324                  * they are enabled and we are extracting a full image. */
2325                 if (wim_header_flags & WIM_HDR_FLAG_RP_FIX)
2326                         extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
2327         }
2328
2329         /* TODO: Since UNIX data entries are stored in the file resources, in a
2330          * completely sequential extraction they may come up before the
2331          * corresponding file or symbolic link data.  This needs to be handled
2332          * better.  */
2333         if ((extract_flags & (WIMLIB_EXTRACT_FLAG_UNIX_DATA |
2334                               WIMLIB_EXTRACT_FLAG_SEQUENTIAL))
2335                                     == (WIMLIB_EXTRACT_FLAG_UNIX_DATA |
2336                                         WIMLIB_EXTRACT_FLAG_SEQUENTIAL))
2337         {
2338                 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2339                         WARNING("Setting UNIX file/owner group may "
2340                                 "be impossible on some\n"
2341                                 "          symbolic links "
2342                                 "when applying from a pipe.");
2343                 } else {
2344                         extract_flags &= ~WIMLIB_EXTRACT_FLAG_SEQUENTIAL;
2345                         WARNING("Disabling sequential extraction for "
2346                                 "UNIX data mode");
2347                 }
2348         }
2349
2350         cmd->extract_flags = extract_flags;
2351         return 0;
2352 }
2353
2354
2355 /* Internal function to execute extraction commands for a WIM image.  The paths
2356  * in the extract commands are expected to be already "canonicalized".  */
2357 static int
2358 do_wimlib_extract_files(WIMStruct *wim,
2359                         int image,
2360                         struct wimlib_extract_command *cmds,
2361                         size_t num_cmds,
2362                         wimlib_progress_func_t progress_func)
2363 {
2364         int ret;
2365         bool found_link_cmd = false;
2366         bool found_nolink_cmd = false;
2367
2368         /* Select the image from which we are extracting files */
2369         ret = select_wim_image(wim, image);
2370         if (ret)
2371                 return ret;
2372
2373         /* Make sure there are no streams in the WIM that have not been
2374          * checksummed yet.  */
2375         ret = wim_checksum_unhashed_streams(wim);
2376         if (ret)
2377                 return ret;
2378
2379         /* Check for problems with the extraction commands */
2380         for (size_t i = 0; i < num_cmds; i++) {
2381                 ret = check_extract_command(&cmds[i], wim->hdr.flags);
2382                 if (ret)
2383                         return ret;
2384                 if (cmds[i].extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2385                                              WIMLIB_EXTRACT_FLAG_HARDLINK)) {
2386                         found_link_cmd = true;
2387                 } else {
2388                         found_nolink_cmd = true;
2389                 }
2390                 if (found_link_cmd && found_nolink_cmd) {
2391                         ERROR("Symlink or hardlink extraction mode must "
2392                               "be set on all extraction commands");
2393                         return WIMLIB_ERR_INVALID_PARAM;
2394                 }
2395         }
2396
2397         /* Execute the extraction commands */
2398         for (size_t i = 0; i < num_cmds; i++) {
2399                 ret = extract_tree(wim,
2400                                    cmds[i].wim_source_path,
2401                                    cmds[i].fs_dest_path,
2402                                    cmds[i].extract_flags,
2403                                    progress_func);
2404                 if (ret)
2405                         return ret;
2406         }
2407         return 0;
2408 }
2409
2410 /* API function documented in wimlib.h  */
2411 WIMLIBAPI int
2412 wimlib_extract_files(WIMStruct *wim,
2413                      int image,
2414                      const struct wimlib_extract_command *cmds,
2415                      size_t num_cmds,
2416                      int default_extract_flags,
2417                      WIMStruct **additional_swms,
2418                      unsigned num_additional_swms,
2419                      wimlib_progress_func_t progress_func)
2420 {
2421         int ret;
2422         struct wimlib_extract_command *cmds_copy;
2423         int all_flags = 0;
2424
2425         default_extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2426
2427         ret = verify_swm_set(wim, additional_swms, num_additional_swms);
2428         if (ret)
2429                 goto out;
2430
2431         if (num_cmds == 0)
2432                 goto out;
2433
2434         if (num_additional_swms)
2435                 merge_lookup_tables(wim, additional_swms, num_additional_swms);
2436
2437         cmds_copy = CALLOC(num_cmds, sizeof(cmds[0]));
2438         if (!cmds_copy) {
2439                 ret = WIMLIB_ERR_NOMEM;
2440                 goto out_restore_lookup_table;
2441         }
2442
2443         for (size_t i = 0; i < num_cmds; i++) {
2444                 cmds_copy[i].extract_flags = (default_extract_flags |
2445                                                  cmds[i].extract_flags)
2446                                                 & WIMLIB_EXTRACT_MASK_PUBLIC;
2447                 all_flags |= cmds_copy[i].extract_flags;
2448
2449                 cmds_copy[i].wim_source_path = canonicalize_wim_path(cmds[i].wim_source_path);
2450                 if (!cmds_copy[i].wim_source_path) {
2451                         ret = WIMLIB_ERR_NOMEM;
2452                         goto out_free_cmds_copy;
2453                 }
2454
2455                 cmds_copy[i].fs_dest_path = canonicalize_fs_path(cmds[i].fs_dest_path);
2456                 if (!cmds_copy[i].fs_dest_path) {
2457                         ret = WIMLIB_ERR_NOMEM;
2458                         goto out_free_cmds_copy;
2459                 }
2460
2461         }
2462         ret = do_wimlib_extract_files(wim, image,
2463                                       cmds_copy, num_cmds,
2464                                       progress_func);
2465
2466         if (all_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2467                          WIMLIB_EXTRACT_FLAG_HARDLINK))
2468         {
2469                 for_lookup_table_entry(wim->lookup_table,
2470                                        lte_free_extracted_file, NULL);
2471         }
2472 out_free_cmds_copy:
2473         for (size_t i = 0; i < num_cmds; i++) {
2474                 FREE(cmds_copy[i].wim_source_path);
2475                 FREE(cmds_copy[i].fs_dest_path);
2476         }
2477         FREE(cmds_copy);
2478 out_restore_lookup_table:
2479         if (num_additional_swms)
2480                 unmerge_lookup_table(wim);
2481 out:
2482         return ret;
2483 }
2484
2485 /*
2486  * Extracts an image from a WIM file.
2487  *
2488  * @wim:                WIMStruct for the WIM file.
2489  *
2490  * @image:              Number of the single image to extract.
2491  *
2492  * @target:             Directory or NTFS volume to extract the image to.
2493  *
2494  * @extract_flags:      Bitwise or of WIMLIB_EXTRACT_FLAG_*.
2495  *
2496  * @progress_func:      If non-NULL, a progress function to be called
2497  *                      periodically.
2498  *
2499  * Returns 0 on success; nonzero on failure.
2500  */
2501 static int
2502 extract_single_image(WIMStruct *wim, int image,
2503                      const tchar *target, int extract_flags,
2504                      wimlib_progress_func_t progress_func)
2505 {
2506         int ret;
2507         tchar *target_copy = canonicalize_fs_path(target);
2508         if (!target_copy)
2509                 return WIMLIB_ERR_NOMEM;
2510         struct wimlib_extract_command cmd = {
2511                 .wim_source_path = T(""),
2512                 .fs_dest_path = target_copy,
2513                 .extract_flags = extract_flags,
2514         };
2515         ret = do_wimlib_extract_files(wim, image, &cmd, 1, progress_func);
2516         FREE(target_copy);
2517         return ret;
2518 }
2519
2520 static const tchar * const filename_forbidden_chars =
2521 T(
2522 #ifdef __WIN32__
2523 "<>:\"/\\|?*"
2524 #else
2525 "/"
2526 #endif
2527 );
2528
2529 /* This function checks if it is okay to use a WIM image's name as a directory
2530  * name.  */
2531 static bool
2532 image_name_ok_as_dir(const tchar *image_name)
2533 {
2534         return image_name && *image_name &&
2535                 !tstrpbrk(image_name, filename_forbidden_chars) &&
2536                 tstrcmp(image_name, T(".")) &&
2537                 tstrcmp(image_name, T(".."));
2538 }
2539
2540 /* Extracts all images from the WIM to the directory @target, with the images
2541  * placed in subdirectories named by their image names. */
2542 static int
2543 extract_all_images(WIMStruct *wim,
2544                    const tchar *target,
2545                    int extract_flags,
2546                    wimlib_progress_func_t progress_func)
2547 {
2548         size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
2549         size_t output_path_len = tstrlen(target);
2550         tchar buf[output_path_len + 1 + image_name_max_len + 1];
2551         int ret;
2552         int image;
2553         const tchar *image_name;
2554         struct stat stbuf;
2555
2556         extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
2557
2558         if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
2559                 ERROR("Cannot extract multiple images in NTFS extraction mode.");
2560                 return WIMLIB_ERR_INVALID_PARAM;
2561         }
2562
2563         if (tstat(target, &stbuf)) {
2564                 if (errno == ENOENT) {
2565                         if (tmkdir(target, 0755)) {
2566                                 ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
2567                                 return WIMLIB_ERR_MKDIR;
2568                         }
2569                 } else {
2570                         ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
2571                         return WIMLIB_ERR_STAT;
2572                 }
2573         } else if (!S_ISDIR(stbuf.st_mode)) {
2574                 ERROR("\"%"TS"\" is not a directory", target);
2575                 return WIMLIB_ERR_NOTDIR;
2576         }
2577
2578         tmemcpy(buf, target, output_path_len);
2579         buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
2580         for (image = 1; image <= wim->hdr.image_count; image++) {
2581                 image_name = wimlib_get_image_name(wim, image);
2582                 if (image_name_ok_as_dir(image_name)) {
2583                         tstrcpy(buf + output_path_len + 1, image_name);
2584                 } else {
2585                         /* Image name is empty or contains forbidden characters.
2586                          * Use image number instead. */
2587                         tsprintf(buf + output_path_len + 1, T("%d"), image);
2588                 }
2589                 ret = extract_single_image(wim, image, buf, extract_flags,
2590                                            progress_func);
2591                 if (ret)
2592                         return ret;
2593         }
2594         return 0;
2595 }
2596
2597 static int
2598 do_wimlib_extract_image(WIMStruct *wim,
2599                         int image,
2600                         const tchar *target,
2601                         int extract_flags,
2602                         WIMStruct **additional_swms,
2603                         unsigned num_additional_swms,
2604                         wimlib_progress_func_t progress_func)
2605 {
2606         int ret;
2607
2608         if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
2609                 wimlib_assert(wim->hdr.part_number == 1);
2610                 wimlib_assert(num_additional_swms == 0);
2611         } else {
2612                 ret = verify_swm_set(wim, additional_swms, num_additional_swms);
2613                 if (ret)
2614                         return ret;
2615
2616                 if (num_additional_swms)
2617                         merge_lookup_tables(wim, additional_swms, num_additional_swms);
2618         }
2619
2620         if (image == WIMLIB_ALL_IMAGES) {
2621                 ret = extract_all_images(wim, target, extract_flags,
2622                                          progress_func);
2623         } else {
2624                 ret = extract_single_image(wim, image, target, extract_flags,
2625                                            progress_func);
2626         }
2627
2628         if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
2629                              WIMLIB_EXTRACT_FLAG_HARDLINK))
2630         {
2631                 for_lookup_table_entry(wim->lookup_table,
2632                                        lte_free_extracted_file,
2633                                        NULL);
2634         }
2635         if (num_additional_swms)
2636                 unmerge_lookup_table(wim);
2637         return ret;
2638 }
2639
2640 /* API function documented in wimlib.h  */
2641 WIMLIBAPI int
2642 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
2643                                const tchar *target, int extract_flags,
2644                                wimlib_progress_func_t progress_func)
2645 {
2646         int ret;
2647         WIMStruct *pwm;
2648         struct filedes *in_fd;
2649         int image;
2650         unsigned i;
2651
2652         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2653
2654         if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT)
2655                 return WIMLIB_ERR_INVALID_PARAM;
2656
2657         extract_flags |= WIMLIB_EXTRACT_FLAG_SEQUENTIAL;
2658
2659         /* Read the WIM header from the pipe and get a WIMStruct to represent
2660          * the pipable WIM.  Caveats:  Unlike getting a WIMStruct with
2661          * wimlib_open_wim(), getting a WIMStruct in this way will result in
2662          * an empty lookup table, no XML data read, and no filename set.  */
2663         ret = open_wim_as_WIMStruct(&pipe_fd,
2664                                     WIMLIB_OPEN_FLAG_FROM_PIPE |
2665                                                 WIMLIB_OPEN_FLAG_SPLIT_OK,
2666                                     &pwm, progress_func);
2667         if (ret)
2668                 return ret;
2669
2670         /* Sanity check to make sure this is a pipable WIM.  */
2671         if (pwm->hdr.magic != PWM_MAGIC) {
2672                 ERROR("The WIM being read from file descriptor %d "
2673                       "is not pipable!", pipe_fd);
2674                 ret = WIMLIB_ERR_NOT_PIPABLE;
2675                 goto out_wimlib_free;
2676         }
2677
2678         /* Sanity check to make sure the first part of a pipable split WIM is
2679          * sent over the pipe first.  */
2680         if (pwm->hdr.part_number != 1) {
2681                 ERROR("The first part of the split WIM must be "
2682                       "sent over the pipe first.");
2683                 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2684                 goto out_wimlib_free;
2685         }
2686
2687         in_fd = &pwm->in_fd;
2688         wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
2689
2690         /* As mentioned, the WIMStruct we created from the pipe does not have
2691          * XML data yet.  Fix this by reading the extra copy of the XML data
2692          * that directly follows the header in pipable WIMs.  (Note: see
2693          * write_pipable_wim() for more details about the format of pipable
2694          * WIMs.)  */
2695         {
2696                 struct wim_lookup_table_entry xml_lte;
2697                 ret = read_pwm_stream_header(pwm, &xml_lte, 0, NULL);
2698                 if (ret)
2699                         goto out_wimlib_free;
2700
2701                 if (!(xml_lte.resource_entry.flags & WIM_RESHDR_FLAG_METADATA))
2702                 {
2703                         ERROR("Expected XML data, but found non-metadata "
2704                               "stream.");
2705                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2706                         goto out_wimlib_free;
2707                 }
2708
2709                 copy_resource_entry(&pwm->hdr.xml_res_entry,
2710                                     &xml_lte.resource_entry);
2711
2712                 ret = read_wim_xml_data(pwm);
2713                 if (ret)
2714                         goto out_wimlib_free;
2715                 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
2716                         ERROR("Image count in XML data is not the same as in WIM header.");
2717                         ret = WIMLIB_ERR_XML;
2718                         goto out_wimlib_free;
2719                 }
2720         }
2721
2722         /* Get image index (this may use the XML data that was just read to
2723          * resolve an image name).  */
2724         if (image_num_or_name) {
2725                 image = wimlib_resolve_image(pwm, image_num_or_name);
2726                 if (image == WIMLIB_NO_IMAGE) {
2727                         ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
2728                               image_num_or_name);
2729                         ret = WIMLIB_ERR_INVALID_IMAGE;
2730                         goto out_wimlib_free;
2731                 } else if (image == WIMLIB_ALL_IMAGES) {
2732                         ERROR("Applying all images from a pipe is not supported.");
2733                         ret = WIMLIB_ERR_INVALID_IMAGE;
2734                         goto out_wimlib_free;
2735                 }
2736         } else {
2737                 if (pwm->hdr.image_count != 1) {
2738                         ERROR("No image was specified, but the pipable WIM "
2739                               "did not contain exactly 1 image");
2740                         ret = WIMLIB_ERR_INVALID_IMAGE;
2741                         goto out_wimlib_free;
2742                 }
2743                 image = 1;
2744         }
2745
2746         /* Load the needed metadata resource.  */
2747         for (i = 1; i <= pwm->hdr.image_count; i++) {
2748                 struct wim_lookup_table_entry *metadata_lte;
2749                 struct wim_image_metadata *imd;
2750
2751                 metadata_lte = new_lookup_table_entry();
2752                 if (!metadata_lte) {
2753                         ret = WIMLIB_ERR_NOMEM;
2754                         goto out_wimlib_free;
2755                 }
2756
2757                 ret = read_pwm_stream_header(pwm, metadata_lte, 0, NULL);
2758                 imd = pwm->image_metadata[i - 1];
2759                 imd->metadata_lte = metadata_lte;
2760                 if (ret)
2761                         goto out_wimlib_free;
2762
2763                 if (!(metadata_lte->resource_entry.flags &
2764                       WIM_RESHDR_FLAG_METADATA))
2765                 {
2766                         ERROR("Expected metadata resource, but found "
2767                               "non-metadata stream.");
2768                         ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
2769                         goto out_wimlib_free;
2770                 }
2771
2772                 if (i == image) {
2773                         /* Metadata resource is for the images being extracted.
2774                          * Parse it and save the metadata in memory.  */
2775                         ret = read_metadata_resource(pwm, imd);
2776                         if (ret)
2777                                 goto out_wimlib_free;
2778                         imd->modified = 1;
2779                 } else {
2780                         /* Metadata resource is not for the image being
2781                          * extracted.  Skip over it.  */
2782                         ret = skip_pwm_stream(metadata_lte);
2783                         if (ret)
2784                                 goto out_wimlib_free;
2785                 }
2786         }
2787         /* Extract the image.  */
2788         extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
2789         ret = do_wimlib_extract_image(pwm, image, target,
2790                                       extract_flags, NULL, 0, progress_func);
2791         /* Clean up and return.  */
2792 out_wimlib_free:
2793         wimlib_free(pwm);
2794         return ret;
2795 }
2796
2797 /* API function documented in wimlib.h  */
2798 WIMLIBAPI int
2799 wimlib_extract_image(WIMStruct *wim,
2800                      int image,
2801                      const tchar *target,
2802                      int extract_flags,
2803                      WIMStruct **additional_swms,
2804                      unsigned num_additional_swms,
2805                      wimlib_progress_func_t progress_func)
2806 {
2807         extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
2808         return do_wimlib_extract_image(wim, image, target, extract_flags,
2809                                        additional_swms, num_additional_swms,
2810                                        progress_func);
2811 }