4 * Support for extracting WIM images, or files or directories contained in a WIM
9 * Copyright (C) 2012, 2013, 2014 Eric Biggers
11 * This file is part of wimlib, a library for working with WIM files.
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)
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
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/.
28 * This file provides the API functions wimlib_extract_image(),
29 * wimlib_extract_image_from_pipe(), wimlib_extract_paths(), and
30 * wimlib_extract_pathlist(). Internally, all end up calling
31 * do_wimlib_extract_paths() and extract_trees().
33 * Although wimlib supports multiple extraction modes/backends (NTFS-3g, UNIX,
34 * Win32), this file does not itself have code to extract files or directories
35 * to any specific target; instead, it handles generic functionality and relies
36 * on lower-level callback functions declared in `struct apply_operations' to do
37 * the actual extraction.
44 #include "wimlib/apply.h"
45 #include "wimlib/dentry.h"
46 #include "wimlib/encoding.h"
47 #include "wimlib/endianness.h"
48 #include "wimlib/error.h"
49 #include "wimlib/lookup_table.h"
50 #include "wimlib/metadata.h"
51 #include "wimlib/pathlist.h"
52 #include "wimlib/paths.h"
53 #include "wimlib/reparse.h"
54 #include "wimlib/resource.h"
55 #include "wimlib/security.h"
56 #include "wimlib/unix_data.h"
58 # include "wimlib/win32.h" /* for realpath() equivalent */
60 #include "wimlib/xml.h"
61 #include "wimlib/wildcard.h"
62 #include "wimlib/wim.h"
70 #define WIMLIB_EXTRACT_FLAG_FROM_PIPE 0x80000000
71 #define WIMLIB_EXTRACT_FLAG_IMAGEMODE 0x40000000
73 /* Keep in sync with wimlib.h */
74 #define WIMLIB_EXTRACT_MASK_PUBLIC \
75 (WIMLIB_EXTRACT_FLAG_NTFS | \
76 WIMLIB_EXTRACT_FLAG_UNIX_DATA | \
77 WIMLIB_EXTRACT_FLAG_NO_ACLS | \
78 WIMLIB_EXTRACT_FLAG_STRICT_ACLS | \
79 WIMLIB_EXTRACT_FLAG_RPFIX | \
80 WIMLIB_EXTRACT_FLAG_NORPFIX | \
81 WIMLIB_EXTRACT_FLAG_TO_STDOUT | \
82 WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES | \
83 WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS | \
84 WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS | \
85 WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES | \
86 WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS | \
87 WIMLIB_EXTRACT_FLAG_GLOB_PATHS | \
88 WIMLIB_EXTRACT_FLAG_STRICT_GLOB | \
89 WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES | \
90 WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE | \
91 WIMLIB_EXTRACT_FLAG_WIMBOOT)
93 /* Check whether the extraction of a dentry should be skipped completely. */
95 dentry_is_supported(struct wim_dentry *dentry,
96 const struct wim_features *supported_features)
98 struct wim_inode *inode = dentry->d_inode;
100 if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
101 return supported_features->reparse_points ||
102 (inode_is_symlink(inode) &&
103 supported_features->symlink_reparse_points);
105 if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
106 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
107 return supported_features->encrypted_directories != 0;
109 return supported_features->encrypted_files != 0;
115 #define PWM_ALLOW_WIM_HDR 0x00001
117 /* Read the header from a stream in a pipable WIM. */
119 read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
120 struct wim_resource_spec *rspec,
121 int flags, struct wim_header_disk *hdr_ret)
124 struct pwm_stream_hdr stream_hdr;
125 struct wim_header_disk pwm_hdr;
127 struct wim_reshdr reshdr;
130 ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
134 if ((flags & PWM_ALLOW_WIM_HDR) && buf.stream_hdr.magic == PWM_MAGIC) {
135 BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
136 ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
137 sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
141 lte->resource_location = RESOURCE_NONEXISTENT;
142 memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
146 if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) {
147 ERROR("Data read on pipe is invalid (expected stream header).");
148 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
151 copy_hash(lte->hash, buf.stream_hdr.hash);
153 reshdr.size_in_wim = 0;
154 reshdr.flags = le32_to_cpu(buf.stream_hdr.flags);
155 reshdr.offset_in_wim = pwm->in_fd.offset;
156 reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
157 wim_res_hdr_to_spec(&reshdr, pwm, rspec);
158 lte_bind_wim_resource_spec(lte, rspec);
159 lte->flags = rspec->flags;
160 lte->size = rspec->uncompressed_size;
161 lte->offset_in_res = 0;
165 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
170 load_streams_from_pipe(struct apply_ctx *ctx,
171 const struct read_stream_list_callbacks *cbs)
173 struct wim_lookup_table_entry *found_lte = NULL;
174 struct wim_resource_spec *rspec = NULL;
175 struct wim_lookup_table *lookup_table;
178 ret = WIMLIB_ERR_NOMEM;
179 found_lte = new_lookup_table_entry();
183 rspec = MALLOC(sizeof(struct wim_resource_spec));
187 lookup_table = ctx->wim->lookup_table;
188 memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GUID_LEN);
189 ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
190 ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
191 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
195 while (ctx->num_streams_remaining) {
196 struct wim_header_disk pwm_hdr;
197 struct wim_lookup_table_entry *needed_lte;
199 if (found_lte->resource_location != RESOURCE_NONEXISTENT)
200 lte_unbind_wim_resource_spec(found_lte);
201 ret = read_pwm_stream_header(ctx->wim, found_lte, rspec,
202 PWM_ALLOW_WIM_HDR, &pwm_hdr);
206 if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
207 && !(found_lte->flags & WIM_RESHDR_FLAG_METADATA)
208 && (needed_lte = lookup_stream(lookup_table, found_lte->hash))
209 && (needed_lte->out_refcnt))
211 needed_lte->offset_in_res = found_lte->offset_in_res;
212 needed_lte->flags = found_lte->flags;
213 needed_lte->size = found_lte->size;
215 lte_unbind_wim_resource_spec(found_lte);
216 lte_bind_wim_resource_spec(needed_lte, rspec);
218 ret = (*cbs->begin_stream)(needed_lte, 0,
219 cbs->begin_stream_ctx);
221 lte_unbind_wim_resource_spec(needed_lte);
225 ret = extract_stream(needed_lte, needed_lte->size,
227 cbs->consume_chunk_ctx);
229 ret = (*cbs->end_stream)(needed_lte, ret,
230 cbs->end_stream_ctx);
231 lte_unbind_wim_resource_spec(needed_lte);
234 ctx->num_streams_remaining--;
235 } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
236 ret = skip_wim_stream(found_lte);
240 u16 part_number = le16_to_cpu(pwm_hdr.part_number);
241 u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
243 if (part_number != ctx->progress.extract.part_number ||
244 total_parts != ctx->progress.extract.total_parts ||
245 memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
248 ctx->progress.extract.part_number = part_number;
249 ctx->progress.extract.total_parts = total_parts;
250 memcpy(ctx->progress.extract.guid,
251 pwm_hdr.guid, WIM_GUID_LEN);
252 ret = extract_progress(ctx,
253 WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
261 if (found_lte->resource_location != RESOURCE_IN_WIM)
263 free_lookup_table_entry(found_lte);
268 begin_extract_stream_with_progress(struct wim_lookup_table_entry *lte,
269 u32 flags, void *_ctx)
271 struct apply_ctx *ctx = _ctx;
273 ctx->cur_stream = lte;
275 return (*ctx->saved_cbs->begin_stream)(lte, flags,
276 ctx->saved_cbs->begin_stream_ctx);
280 consume_chunk_with_progress(const void *chunk, size_t size, void *_ctx)
282 struct apply_ctx *ctx = _ctx;
283 union wimlib_progress_info *progress = &ctx->progress;
286 if (likely(ctx->supported_features.hard_links)) {
287 progress->extract.completed_bytes +=
288 (u64)size * ctx->cur_stream->out_refcnt;
290 const struct stream_owner *owners = stream_owners(ctx->cur_stream);
291 for (u32 i = 0; i < ctx->cur_stream->out_refcnt; i++) {
292 const struct wim_inode *inode = owners[i].inode;
293 const struct wim_dentry *dentry;
295 list_for_each_entry(dentry,
296 &inode->i_extraction_aliases,
297 d_extraction_alias_node)
299 progress->extract.completed_bytes += size;
303 if (progress->extract.completed_bytes >= ctx->next_progress) {
305 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
309 if (progress->extract.completed_bytes >=
310 progress->extract.total_bytes)
312 ctx->next_progress = UINT64_MAX;
314 ctx->next_progress += progress->extract.total_bytes / 128;
315 if (ctx->next_progress > progress->extract.total_bytes)
316 ctx->next_progress = progress->extract.total_bytes;
319 return (*ctx->saved_cbs->consume_chunk)(chunk, size,
320 ctx->saved_cbs->consume_chunk_ctx);
324 * Read the list of single-instance streams to extract and feed their data into
325 * the specified callback functions.
327 * This handles checksumming each stream.
329 * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS.
331 * This also works if the WIM is being read from a pipe, whereas attempting to
332 * read streams directly (e.g. with read_full_stream_into_buf()) will not.
335 extract_stream_list(struct apply_ctx *ctx,
336 const struct read_stream_list_callbacks *cbs)
338 struct read_stream_list_callbacks wrapper_cbs = {
339 .begin_stream = begin_extract_stream_with_progress,
340 .begin_stream_ctx = ctx,
341 .consume_chunk = consume_chunk_with_progress,
342 .consume_chunk_ctx = ctx,
343 .end_stream = cbs->end_stream,
344 .end_stream_ctx = cbs->end_stream_ctx,
347 ctx->saved_cbs = cbs;
350 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
351 return load_streams_from_pipe(ctx, cbs);
353 return read_stream_list(&ctx->stream_list,
354 offsetof(struct wim_lookup_table_entry,
356 cbs, VERIFY_STREAM_HASHES);
360 /* Extract a WIM dentry to standard output.
362 * This obviously doesn't make sense in all cases. We return an error if the
363 * dentry does not correspond to a regular file. Otherwise we extract the
364 * unnamed data stream only. */
366 extract_dentry_to_stdout(struct wim_dentry *dentry,
367 const struct wim_lookup_table *lookup_table)
369 struct wim_inode *inode = dentry->d_inode;
370 struct wim_lookup_table_entry *lte;
371 struct filedes _stdout;
373 if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
374 FILE_ATTRIBUTE_DIRECTORY))
376 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
377 "extracted to standard output", dentry_full_path(dentry));
378 return WIMLIB_ERR_NOT_A_REGULAR_FILE;
381 lte = inode_unnamed_lte(inode, lookup_table);
383 const u8 *hash = inode_unnamed_stream_hash(inode);
384 if (!is_zero_hash(hash))
385 return stream_not_found_error(inode, hash);
389 filedes_init(&_stdout, STDOUT_FILENO);
390 return extract_full_stream_to_fd(lte, &_stdout);
394 extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
395 const struct wim_lookup_table *lookup_table)
397 for (size_t i = 0; i < num_dentries; i++) {
398 int ret = extract_dentry_to_stdout(dentries[i], lookup_table);
405 /**********************************************************************/
408 * Removes duplicate dentries from the array.
410 * Returns the new number of dentries, packed at the front of the array.
413 remove_duplicate_trees(struct wim_dentry **trees, size_t num_trees)
416 for (i = 0; i < num_trees; i++) {
417 if (!trees[i]->tmp_flag) {
418 /* Found distinct dentry. */
419 trees[i]->tmp_flag = 1;
420 trees[j++] = trees[i];
423 for (i = 0; i < j; i++)
424 trees[i]->tmp_flag = 0;
429 * Remove dentries that are descendants of other dentries in the array.
431 * Returns the new number of dentries, packed at the front of the array.
434 remove_contained_trees(struct wim_dentry **trees, size_t num_trees)
437 for (i = 0; i < num_trees; i++)
438 trees[i]->tmp_flag = 1;
439 for (i = 0; i < num_trees; i++) {
440 struct wim_dentry *d = trees[i];
441 while (!dentry_is_root(d)) {
446 trees[j++] = trees[i];
450 trees[i]->tmp_flag = 0;
453 for (i = 0; i < j; i++)
454 trees[i]->tmp_flag = 0;
459 dentry_append_to_list(struct wim_dentry *dentry, void *_dentry_list)
461 struct list_head *dentry_list = _dentry_list;
462 list_add_tail(&dentry->d_extraction_list_node, dentry_list);
467 dentry_reset_extraction_list_node(struct wim_dentry *dentry)
469 dentry->d_extraction_list_node = (struct list_head){NULL, NULL};
473 dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore)
475 list_del(&dentry->d_extraction_list_node);
476 dentry_reset_extraction_list_node(dentry);
481 * Build the preliminary list of dentries to be extracted.
483 * The list maintains the invariant that if d1 and d2 are in the list and d1 is
484 * an ancestor of d2, then d1 appears before d2 in the list.
487 build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees,
488 size_t num_trees, bool add_ancestors)
490 INIT_LIST_HEAD(dentry_list);
492 /* Add the trees recursively. */
493 for (size_t i = 0; i < num_trees; i++)
494 for_dentry_in_tree(trees[i], dentry_append_to_list, dentry_list);
496 /* If requested, add ancestors of the trees. */
498 for (size_t i = 0; i < num_trees; i++) {
499 struct wim_dentry *dentry = trees[i];
500 struct wim_dentry *ancestor;
501 struct list_head *place_after;
503 if (dentry_is_root(dentry))
506 place_after = dentry_list;
509 ancestor = ancestor->d_parent;
510 if (will_extract_dentry(ancestor)) {
511 place_after = &ancestor->d_extraction_list_node;
514 } while (!dentry_is_root(ancestor));
518 ancestor = ancestor->d_parent;
519 if (will_extract_dentry(ancestor))
521 list_add(&ancestor->d_extraction_list_node, place_after);
522 } while (!dentry_is_root(ancestor));
528 destroy_dentry_list(struct list_head *dentry_list)
530 struct wim_dentry *dentry, *tmp;
531 struct wim_inode *inode;
533 list_for_each_entry_safe(dentry, tmp, dentry_list, d_extraction_list_node) {
534 inode = dentry->d_inode;
535 dentry_reset_extraction_list_node(dentry);
536 inode->i_visited = 0;
537 if ((void *)dentry->d_extraction_name != (void *)dentry->file_name)
538 FREE(dentry->d_extraction_name);
539 dentry->d_extraction_name = NULL;
540 dentry->d_extraction_name_nchars = 0;
545 destroy_stream_list(struct list_head *stream_list)
547 struct wim_lookup_table_entry *lte;
549 list_for_each_entry(lte, stream_list, extraction_list)
550 if (lte->out_refcnt > ARRAY_LEN(lte->inline_stream_owners))
551 FREE(lte->stream_owners);
555 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
557 static const utf16lechar replacement_char = cpu_to_le16('?');
561 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
567 for (i = 0; i < num_chars; i++) {
570 case cpu_to_le16('\\'):
571 case cpu_to_le16(':'):
572 case cpu_to_le16('*'):
573 case cpu_to_le16('?'):
574 case cpu_to_le16('"'):
575 case cpu_to_le16('<'):
576 case cpu_to_le16('>'):
577 case cpu_to_le16('|'):
579 case cpu_to_le16('/'):
580 case cpu_to_le16('\0'):
582 name[i] = replacement_char;
589 if (name[num_chars - 1] == cpu_to_le16(' ') ||
590 name[num_chars - 1] == cpu_to_le16('.'))
593 name[num_chars - 1] = replacement_char;
602 dentry_calculate_extraction_name(struct wim_dentry *dentry,
603 struct apply_ctx *ctx)
607 if (!dentry_is_supported(dentry, &ctx->supported_features))
610 if (dentry_is_root(dentry))
613 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
614 dentry->d_extraction_name = dentry->file_name;
615 dentry->d_extraction_name_nchars = dentry->file_name_nbytes /
620 if (!ctx->supported_features.case_sensitive_filenames) {
621 struct wim_dentry *other;
622 list_for_each_entry(other, &dentry->d_ci_conflict_list,
625 if (will_extract_dentry(other)) {
626 if (ctx->extract_flags &
627 WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
628 WARNING("\"%"TS"\" has the same "
629 "case-insensitive name as "
630 "\"%"TS"\"; extracting "
631 "dummy name instead",
632 dentry_full_path(dentry),
633 dentry_full_path(other));
636 WARNING("Not extracting \"%"TS"\": "
637 "has same case-insensitive "
639 dentry_full_path(dentry),
640 dentry_full_path(other));
647 if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
648 ret = utf16le_get_tstr(dentry->file_name,
649 dentry->file_name_nbytes,
650 (const tchar **)&dentry->d_extraction_name,
651 &dentry->d_extraction_name_nchars);
652 dentry->d_extraction_name_nchars /= sizeof(tchar);
655 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
657 WARNING("\"%"TS"\" has an invalid filename "
658 "that is not supported on this platform; "
659 "extracting dummy name instead",
660 dentry_full_path(dentry));
663 WARNING("Not extracting \"%"TS"\": has an invalid filename "
664 "that is not supported on this platform",
665 dentry_full_path(dentry));
672 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
674 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
675 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
677 const tchar *tchar_name;
680 ret = utf16le_get_tstr(utf16_name_copy,
681 dentry->file_name_nbytes,
682 &tchar_name, &tchar_nchars);
686 tchar_nchars /= sizeof(tchar);
688 size_t fixed_name_num_chars = tchar_nchars;
689 tchar fixed_name[tchar_nchars + 50];
691 tmemcpy(fixed_name, tchar_name, tchar_nchars);
692 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
693 T(" (invalid filename #%lu)"),
694 ++ctx->invalid_sequence);
696 utf16le_put_tstr(tchar_name);
698 dentry->d_extraction_name = memdup(fixed_name,
699 2 * fixed_name_num_chars + 2);
700 if (!dentry->d_extraction_name)
701 return WIMLIB_ERR_NOMEM;
702 dentry->d_extraction_name_nchars = fixed_name_num_chars;
707 for_dentry_in_tree(dentry, dentry_delete_from_list, NULL);
712 * Calculate the actual filename component at which each WIM dentry will be
713 * extracted, with special handling for dentries that are unsupported by the
714 * extraction backend or have invalid names.
716 * ctx->supported_features must be filled in.
718 * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING
721 dentry_list_calculate_extraction_names(struct list_head *dentry_list,
722 struct apply_ctx *ctx)
724 struct list_head *prev, *cur;
726 /* Can't use list_for_each_entry() because a call to
727 * dentry_calculate_extraction_name() may delete the current dentry and
728 * its children from the list. */
732 struct wim_dentry *dentry;
736 if (cur == dentry_list)
739 dentry = list_entry(cur, struct wim_dentry, d_extraction_list_node);
741 ret = dentry_calculate_extraction_name(dentry, ctx);
745 if (prev->next == cur)
748 ; /* Current dentry and its children (which follow in
749 the list) were deleted. prev stays the same. */
755 dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
756 struct wim_lookup_table *lookup_table)
758 struct wim_inode *inode = dentry->d_inode;
759 struct wim_lookup_table_entry *lte;
763 /* Special case: when extracting from a pipe, the WIM lookup table is
764 * initially empty, so "resolving" an inode's streams is initially not
765 * possible. However, we still need to keep track of which streams,
766 * identified by SHA1 message digests, need to be extracted, so we
767 * "resolve" the inode's streams anyway by allocating new entries. */
768 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
770 ret = inode_resolve_streams(inode, lookup_table, force);
773 for (u32 i = 0; i <= inode->i_num_ads; i++) {
774 lte = inode_stream_lte_resolved(inode, i);
782 * For each dentry to be extracted, resolve all streams in the corresponding
783 * inode and set 'out_refcnt' in each to 0.
785 * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
788 dentry_list_resolve_streams(struct list_head *dentry_list,
789 struct apply_ctx *ctx)
791 struct wim_dentry *dentry;
794 list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
795 ret = dentry_resolve_streams(dentry,
797 ctx->wim->lookup_table);
805 ref_stream(struct wim_lookup_table_entry *lte, u32 stream_idx,
806 struct wim_dentry *dentry, struct apply_ctx *ctx)
808 struct wim_inode *inode = dentry->d_inode;
809 struct stream_owner *stream_owners;
814 /* Tally the size only for each extraction of the stream (not hard
816 if (inode->i_visited && ctx->supported_features.hard_links)
819 ctx->progress.extract.total_bytes += lte->size;
820 ctx->progress.extract.num_streams++;
822 if (inode->i_visited)
825 /* Add stream to the dentry_list only one time, even if it's going
826 * to be extracted to multiple inodes. */
827 if (lte->out_refcnt == 0) {
828 list_add_tail(<e->extraction_list, &ctx->stream_list);
829 ctx->num_streams_remaining++;
832 /* If inode not yet been visited, append it to the stream_owners array. */
833 if (lte->out_refcnt < ARRAY_LEN(lte->inline_stream_owners)) {
834 stream_owners = lte->inline_stream_owners;
836 struct stream_owner *prev_stream_owners;
837 size_t alloc_stream_owners;
839 if (lte->out_refcnt == ARRAY_LEN(lte->inline_stream_owners)) {
840 prev_stream_owners = NULL;
841 alloc_stream_owners = ARRAY_LEN(lte->inline_stream_owners);
843 prev_stream_owners = lte->stream_owners;
844 alloc_stream_owners = lte->alloc_stream_owners;
847 if (lte->out_refcnt == alloc_stream_owners) {
848 alloc_stream_owners *= 2;
849 stream_owners = REALLOC(prev_stream_owners,
850 alloc_stream_owners *
851 sizeof(stream_owners[0]));
853 return WIMLIB_ERR_NOMEM;
854 if (!prev_stream_owners) {
855 memcpy(stream_owners,
856 lte->inline_stream_owners,
857 sizeof(lte->inline_stream_owners));
859 lte->stream_owners = stream_owners;
860 lte->alloc_stream_owners = alloc_stream_owners;
862 stream_owners = lte->stream_owners;
864 stream_owners[lte->out_refcnt].inode = inode;
865 if (stream_idx == 0) {
866 stream_owners[lte->out_refcnt].stream_name = NULL;
868 stream_owners[lte->out_refcnt].stream_name =
869 inode->i_ads_entries[stream_idx - 1].stream_name;
876 dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
878 struct wim_inode *inode = dentry->d_inode;
881 /* The unnamed data stream will always be extracted, except in an
883 if (!inode_is_encrypted_directory(inode)) {
885 struct wim_lookup_table_entry *stream;
887 stream = inode_unnamed_stream_resolved(inode, &stream_idx);
888 ret = ref_stream(stream, stream_idx, dentry, ctx);
893 /* Named data streams will be extracted only if supported in the current
894 * extraction mode and volume, and to avoid complications, if not doing
895 * a linked extraction. */
896 if (ctx->supported_features.named_data_streams) {
897 for (u16 i = 0; i < inode->i_num_ads; i++) {
898 if (!ads_entry_is_named_stream(&inode->i_ads_entries[i]))
900 ret = ref_stream(inode->i_ads_entries[i].lte, i + 1,
906 inode->i_visited = 1;
911 * For each dentry to be extracted, iterate through the data streams of the
912 * corresponding inode. For each such stream that is not to be ignored due to
913 * the supported features or extraction flags, add it to the list of streams to
914 * be extracted (ctx->stream_list) if not already done so.
916 * Also builds a mapping from each stream to the inodes referencing it.
918 * This also initializes the extract progress info with byte and stream
921 * ctx->supported_features must be filled in.
923 * Possible error codes: WIMLIB_ERR_NOMEM.
926 dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
928 struct wim_dentry *dentry;
931 list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
932 ret = dentry_ref_streams(dentry, ctx);
936 list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
937 dentry->d_inode->i_visited = 0;
942 dentry_list_build_inode_alias_lists(struct list_head *dentry_list)
944 struct wim_dentry *dentry;
945 struct wim_inode *inode;
947 list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
948 inode = dentry->d_inode;
949 if (!inode->i_visited)
950 INIT_LIST_HEAD(&inode->i_extraction_aliases);
951 list_add_tail(&dentry->d_extraction_alias_node,
952 &inode->i_extraction_aliases);
953 inode->i_visited = 1;
955 list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
956 dentry->d_inode->i_visited = 0;
960 inode_tally_features(const struct wim_inode *inode,
961 struct wim_features *features)
963 if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
964 features->archive_files++;
965 if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
966 features->hidden_files++;
967 if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
968 features->system_files++;
969 if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
970 features->compressed_files++;
971 if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
972 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
973 features->encrypted_directories++;
975 features->encrypted_files++;
977 if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
978 features->not_context_indexed_files++;
979 if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
980 features->sparse_files++;
981 if (inode_has_named_stream(inode))
982 features->named_data_streams++;
983 if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
984 features->reparse_points++;
985 if (inode_is_symlink(inode))
986 features->symlink_reparse_points++;
988 features->other_reparse_points++;
990 if (inode->i_security_id != -1)
991 features->security_descriptors++;
992 if (inode_has_unix_data(inode))
993 features->unix_data++;
996 /* Tally features necessary to extract a dentry and the corresponding inode. */
998 dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features)
1000 struct wim_inode *inode = dentry->d_inode;
1002 if (dentry_has_short_name(dentry))
1003 features->short_names++;
1005 if (inode->i_visited) {
1006 features->hard_links++;
1008 inode_tally_features(inode, features);
1009 inode->i_visited = 1;
1013 /* Tally the features necessary to extract the specified dentries. */
1015 dentry_list_get_features(struct list_head *dentry_list,
1016 struct wim_features *features)
1018 struct wim_dentry *dentry;
1020 list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1021 dentry_tally_features(dentry, features);
1023 list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1024 dentry->d_inode->i_visited = 0;
1028 do_feature_check(const struct wim_features *required_features,
1029 const struct wim_features *supported_features,
1032 /* File attributes. */
1033 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
1034 /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
1035 * We're an archive program, so theoretically we can do what we
1038 if (required_features->hidden_files &&
1039 !supported_features->hidden_files)
1040 WARNING("Ignoring FILE_ATTRIBUTE_HIDDEN of %lu files",
1041 required_features->hidden_files);
1043 if (required_features->system_files &&
1044 !supported_features->system_files)
1045 WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
1046 required_features->system_files);
1048 if (required_features->compressed_files &&
1049 !supported_features->compressed_files)
1050 WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
1051 required_features->compressed_files);
1053 if (required_features->not_context_indexed_files &&
1054 !supported_features->not_context_indexed_files)
1055 WARNING("Ignoring FILE_ATTRIBUTE_NOT_CONTENT_INDEXED of %lu files",
1056 required_features->not_context_indexed_files);
1058 if (required_features->sparse_files &&
1059 !supported_features->sparse_files)
1060 WARNING("Ignoring FILE_ATTRIBUTE_SPARSE_FILE of %lu files",
1061 required_features->sparse_files);
1063 if (required_features->encrypted_directories &&
1064 !supported_features->encrypted_directories)
1065 WARNING("Ignoring FILE_ATTRIBUTE_ENCRYPTED of %lu directories",
1066 required_features->encrypted_directories);
1069 /* Encrypted files. */
1070 if (required_features->encrypted_files &&
1071 !supported_features->encrypted_files)
1072 WARNING("Ignoring %lu encrypted files",
1073 required_features->encrypted_files);
1075 /* Named data streams. */
1076 if (required_features->named_data_streams &&
1077 (!supported_features->named_data_streams))
1078 WARNING("Ignoring named data streams of %lu files",
1079 required_features->named_data_streams);
1082 if (required_features->hard_links && !supported_features->hard_links)
1083 WARNING("Extracting %lu hard links as independent files",
1084 required_features->hard_links);
1086 /* Symbolic links and reparse points. */
1087 if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
1088 required_features->symlink_reparse_points &&
1089 !supported_features->symlink_reparse_points &&
1090 !supported_features->reparse_points)
1092 ERROR("Extraction backend does not support symbolic links!");
1093 return WIMLIB_ERR_UNSUPPORTED;
1095 if (required_features->reparse_points &&
1096 !supported_features->reparse_points)
1098 if (supported_features->symlink_reparse_points) {
1099 if (required_features->other_reparse_points) {
1100 WARNING("Ignoring %lu non-symlink/junction "
1101 "reparse point files",
1102 required_features->other_reparse_points);
1105 WARNING("Ignoring %lu reparse point files",
1106 required_features->reparse_points);
1110 /* Security descriptors. */
1111 if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
1112 WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1113 == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
1114 required_features->security_descriptors &&
1115 !supported_features->security_descriptors)
1117 ERROR("Extraction backend does not support security descriptors!");
1118 return WIMLIB_ERR_UNSUPPORTED;
1120 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) &&
1121 required_features->security_descriptors &&
1122 !supported_features->security_descriptors)
1123 WARNING("Ignoring Windows NT security descriptors of %lu files",
1124 required_features->security_descriptors);
1127 if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
1128 required_features->unix_data && !supported_features->unix_data)
1130 ERROR("Extraction backend does not support UNIX data!");
1131 return WIMLIB_ERR_UNSUPPORTED;
1134 if (required_features->unix_data &&
1135 !(extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1137 WARNING("Ignoring UNIX metadata of %lu files",
1138 required_features->unix_data);
1142 if (required_features->short_names &&
1143 !supported_features->short_names)
1145 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) {
1146 ERROR("Extraction backend does not support DOS names!");
1147 return WIMLIB_ERR_UNSUPPORTED;
1149 WARNING("Ignoring DOS names of %lu files",
1150 required_features->short_names);
1154 if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
1155 !supported_features->timestamps)
1157 ERROR("Extraction backend does not support timestamps!");
1158 return WIMLIB_ERR_UNSUPPORTED;
1164 static const struct apply_operations *
1165 select_apply_operations(int extract_flags)
1168 if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
1169 return &ntfs_3g_apply_ops;
1172 return &win32_apply_ops;
1174 return &unix_apply_ops;
1179 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
1180 const tchar *target, int extract_flags)
1182 const struct apply_operations *ops;
1183 struct apply_ctx *ctx;
1185 LIST_HEAD(dentry_list);
1187 if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
1188 ret = extract_dentries_to_stdout(trees, num_trees,
1193 num_trees = remove_duplicate_trees(trees, num_trees);
1194 num_trees = remove_contained_trees(trees, num_trees);
1196 ops = select_apply_operations(extract_flags);
1198 if (num_trees > 1 && ops->single_tree_only) {
1199 ERROR("Extracting multiple directory trees "
1200 "at once is not supported in %s extraction mode!",
1202 ret = WIMLIB_ERR_UNSUPPORTED;
1206 ctx = CALLOC(1, ops->context_size);
1208 ret = WIMLIB_ERR_NOMEM;
1213 ctx->target = target;
1214 ctx->target_nchars = tstrlen(target);
1215 ctx->extract_flags = extract_flags;
1216 if (ctx->wim->progfunc) {
1217 ctx->progfunc = ctx->wim->progfunc;
1218 ctx->progctx = ctx->wim->progctx;
1219 ctx->progress.extract.image = wim->current_image;
1220 ctx->progress.extract.extract_flags = (extract_flags &
1221 WIMLIB_EXTRACT_MASK_PUBLIC);
1222 ctx->progress.extract.wimfile_name = wim->filename;
1223 ctx->progress.extract.image_name = wimlib_get_image_name(wim,
1224 wim->current_image);
1225 ctx->progress.extract.target = target;
1227 INIT_LIST_HEAD(&ctx->stream_list);
1229 ret = (*ops->get_supported_features)(target, &ctx->supported_features);
1233 build_dentry_list(&dentry_list, trees, num_trees,
1235 WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE));
1237 dentry_list_get_features(&dentry_list, &ctx->required_features);
1239 ret = do_feature_check(&ctx->required_features, &ctx->supported_features,
1240 ctx->extract_flags);
1244 ret = dentry_list_calculate_extraction_names(&dentry_list, ctx);
1248 ret = dentry_list_resolve_streams(&dentry_list, ctx);
1252 ret = dentry_list_ref_streams(&dentry_list, ctx);
1256 dentry_list_build_inode_alias_lists(&dentry_list);
1258 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
1259 /* When extracting from a pipe, the number of bytes of data to
1260 * extract can't be determined in the normal way (examining the
1261 * lookup table), since at this point all we have is a set of
1262 * SHA1 message digests of streams that need to be extracted.
1263 * However, we can get a reasonably accurate estimate by taking
1264 * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
1265 * data. This does assume that a full image is being extracted,
1266 * but currently there is no API for doing otherwise. (Also,
1267 * subtract <HARDLINKBYTES> from this if hard links are
1268 * supported by the extraction mode.) */
1269 ctx->progress.extract.total_bytes =
1270 wim_info_get_image_total_bytes(wim->wim_info,
1271 wim->current_image);
1272 if (ctx->supported_features.hard_links) {
1273 ctx->progress.extract.total_bytes -=
1274 wim_info_get_image_hard_link_bytes(wim->wim_info,
1275 wim->current_image);
1279 ret = extract_progress(ctx,
1280 ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1281 WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN :
1282 WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN));
1286 ret = (*ops->extract)(&dentry_list, ctx);
1290 if (ctx->progress.extract.completed_bytes <
1291 ctx->progress.extract.total_bytes)
1293 ctx->progress.extract.completed_bytes =
1294 ctx->progress.extract.total_bytes;
1295 ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
1300 ret = extract_progress(ctx,
1301 ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
1302 WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END :
1303 WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END));
1305 destroy_stream_list(&ctx->stream_list);
1306 destroy_dentry_list(&dentry_list);
1313 mkdir_if_needed(const tchar *target)
1316 if (tstat(target, &stbuf)) {
1317 if (errno == ENOENT) {
1318 if (tmkdir(target, 0755)) {
1319 ERROR_WITH_ERRNO("Failed to create directory "
1320 "\"%"TS"\"", target);
1321 return WIMLIB_ERR_MKDIR;
1324 ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
1325 return WIMLIB_ERR_STAT;
1327 } else if (!S_ISDIR(stbuf.st_mode)) {
1328 ERROR("\"%"TS"\" is not a directory", target);
1329 return WIMLIB_ERR_NOTDIR;
1334 /* Make sure the extraction flags make sense, and update them if needed. */
1336 check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
1338 int extract_flags = *extract_flags_p;
1340 /* Check for invalid flag combinations */
1342 if ((extract_flags &
1343 (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1344 WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1345 WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
1346 return WIMLIB_ERR_INVALID_PARAM;
1348 if ((extract_flags &
1349 (WIMLIB_EXTRACT_FLAG_RPFIX |
1350 WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
1351 WIMLIB_EXTRACT_FLAG_NORPFIX))
1352 return WIMLIB_ERR_INVALID_PARAM;
1354 #ifndef WITH_NTFS_3G
1355 if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1356 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1357 " it cannot apply a WIM image directly to an NTFS volume.");
1358 return WIMLIB_ERR_UNSUPPORTED;
1363 if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
1364 ERROR("WIMBoot extraction is only supported on Windows!");
1365 return WIMLIB_ERR_UNSUPPORTED;
1369 if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1370 WIMLIB_EXTRACT_FLAG_NORPFIX |
1371 WIMLIB_EXTRACT_FLAG_IMAGEMODE)) ==
1372 WIMLIB_EXTRACT_FLAG_IMAGEMODE)
1374 /* For full-image extraction, do reparse point fixups by default
1375 * if the WIM header says they are enabled. */
1376 if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
1377 extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
1380 *extract_flags_p = extract_flags;
1385 get_wildcard_flags(int extract_flags)
1387 u32 wildcard_flags = 0;
1389 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB)
1390 wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH;
1392 wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH;
1394 if (default_ignore_case)
1395 wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
1397 return wildcard_flags;
1400 struct append_dentry_ctx {
1401 struct wim_dentry **dentries;
1402 size_t num_dentries;
1403 size_t num_alloc_dentries;
1407 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
1409 struct append_dentry_ctx *ctx = _ctx;
1411 if (ctx->num_dentries == ctx->num_alloc_dentries) {
1412 struct wim_dentry **new_dentries;
1415 new_length = max(ctx->num_alloc_dentries + 8,
1416 ctx->num_alloc_dentries * 3 / 2);
1417 new_dentries = REALLOC(ctx->dentries,
1418 new_length * sizeof(ctx->dentries[0]));
1419 if (new_dentries == NULL)
1420 return WIMLIB_ERR_NOMEM;
1421 ctx->dentries = new_dentries;
1422 ctx->num_alloc_dentries = new_length;
1424 ctx->dentries[ctx->num_dentries++] = dentry;
1429 do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1430 const tchar * const *paths, size_t num_paths,
1434 struct wim_dentry **trees;
1437 if (wim == NULL || target == NULL || target[0] == T('\0') ||
1438 (num_paths != 0 && paths == NULL))
1439 return WIMLIB_ERR_INVALID_PARAM;
1441 ret = check_extract_flags(wim, &extract_flags);
1445 ret = select_wim_image(wim, image);
1449 ret = wim_checksum_unhashed_streams(wim);
1453 if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NTFS |
1454 WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) ==
1455 (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE))
1457 ret = mkdir_if_needed(target);
1462 if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
1464 struct append_dentry_ctx append_dentry_ctx = {
1467 .num_alloc_dentries = 0,
1470 u32 wildcard_flags = get_wildcard_flags(extract_flags);
1472 for (size_t i = 0; i < num_paths; i++) {
1473 tchar *path = canonicalize_wim_path(paths[i]);
1475 ret = WIMLIB_ERR_NOMEM;
1476 trees = append_dentry_ctx.dentries;
1477 goto out_free_trees;
1479 ret = expand_wildcard(wim, path,
1485 trees = append_dentry_ctx.dentries;
1486 goto out_free_trees;
1489 trees = append_dentry_ctx.dentries;
1490 num_trees = append_dentry_ctx.num_dentries;
1492 trees = MALLOC(num_paths * sizeof(trees[0]));
1494 return WIMLIB_ERR_NOMEM;
1496 for (size_t i = 0; i < num_paths; i++) {
1498 tchar *path = canonicalize_wim_path(paths[i]);
1500 ret = WIMLIB_ERR_NOMEM;
1501 goto out_free_trees;
1504 trees[i] = get_dentry(wim, path,
1505 WIMLIB_CASE_PLATFORM_DEFAULT);
1507 if (trees[i] == NULL) {
1508 ERROR("Path \"%"TS"\" does not exist "
1510 paths[i], wim->current_image);
1511 ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1512 goto out_free_trees;
1515 num_trees = num_paths;
1518 if (num_trees == 0) {
1520 goto out_free_trees;
1523 ret = extract_trees(wim, trees, num_trees, target, extract_flags);
1530 extract_single_image(WIMStruct *wim, int image,
1531 const tchar *target, int extract_flags)
1533 const tchar *path = WIMLIB_WIM_ROOT_PATH;
1534 extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE;
1535 return do_wimlib_extract_paths(wim, image, target, &path, 1, extract_flags);
1538 static const tchar * const filename_forbidden_chars =
1547 /* This function checks if it is okay to use a WIM image's name as a directory
1550 image_name_ok_as_dir(const tchar *image_name)
1552 return image_name && *image_name &&
1553 !tstrpbrk(image_name, filename_forbidden_chars) &&
1554 tstrcmp(image_name, T(".")) &&
1555 tstrcmp(image_name, T(".."));
1558 /* Extracts all images from the WIM to the directory @target, with the images
1559 * placed in subdirectories named by their image names. */
1561 extract_all_images(WIMStruct *wim, const tchar *target, int extract_flags)
1563 size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
1564 size_t output_path_len = tstrlen(target);
1565 tchar buf[output_path_len + 1 + image_name_max_len + 1];
1568 const tchar *image_name;
1570 if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1571 ERROR("Cannot extract multiple images in NTFS extraction mode.");
1572 return WIMLIB_ERR_INVALID_PARAM;
1575 ret = mkdir_if_needed(target);
1578 tmemcpy(buf, target, output_path_len);
1579 buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
1580 for (image = 1; image <= wim->hdr.image_count; image++) {
1581 image_name = wimlib_get_image_name(wim, image);
1582 if (image_name_ok_as_dir(image_name)) {
1583 tstrcpy(buf + output_path_len + 1, image_name);
1585 /* Image name is empty or contains forbidden characters.
1586 * Use image number instead. */
1587 tsprintf(buf + output_path_len + 1, T("%d"), image);
1589 ret = extract_single_image(wim, image, buf, extract_flags);
1597 do_wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1600 if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
1601 WIMLIB_EXTRACT_FLAG_TO_STDOUT |
1602 WIMLIB_EXTRACT_FLAG_GLOB_PATHS))
1603 return WIMLIB_ERR_INVALID_PARAM;
1605 if (image == WIMLIB_ALL_IMAGES)
1606 return extract_all_images(wim, target, extract_flags);
1608 return extract_single_image(wim, image, target, extract_flags);
1612 /****************************************************************************
1614 ****************************************************************************/
1617 wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1618 const tchar * const *paths, size_t num_paths,
1621 if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1622 return WIMLIB_ERR_INVALID_PARAM;
1624 return do_wimlib_extract_paths(wim, image, target, paths, num_paths,
1629 wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target,
1630 const tchar *path_list_file, int extract_flags)
1637 ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
1639 ERROR("Failed to read path list file \"%"TS"\"",
1644 ret = wimlib_extract_paths(wim, image, target,
1645 (const tchar * const *)paths, num_paths,
1653 wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
1654 const tchar *image_num_or_name,
1655 const tchar *target,
1657 wimlib_progress_func_t progfunc,
1662 struct filedes *in_fd;
1666 if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1667 return WIMLIB_ERR_INVALID_PARAM;
1669 /* Read the WIM header from the pipe and get a WIMStruct to represent
1670 * the pipable WIM. Caveats: Unlike getting a WIMStruct with
1671 * wimlib_open_wim(), getting a WIMStruct in this way will result in
1672 * an empty lookup table, no XML data read, and no filename set. */
1673 ret = open_wim_as_WIMStruct(&pipe_fd, WIMLIB_OPEN_FLAG_FROM_PIPE, &pwm,
1678 /* Sanity check to make sure this is a pipable WIM. */
1679 if (pwm->hdr.magic != PWM_MAGIC) {
1680 ERROR("The WIM being read from file descriptor %d "
1681 "is not pipable!", pipe_fd);
1682 ret = WIMLIB_ERR_NOT_PIPABLE;
1683 goto out_wimlib_free;
1686 /* Sanity check to make sure the first part of a pipable split WIM is
1687 * sent over the pipe first. */
1688 if (pwm->hdr.part_number != 1) {
1689 ERROR("The first part of the split WIM must be "
1690 "sent over the pipe first.");
1691 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1692 goto out_wimlib_free;
1695 in_fd = &pwm->in_fd;
1696 wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
1698 /* As mentioned, the WIMStruct we created from the pipe does not have
1699 * XML data yet. Fix this by reading the extra copy of the XML data
1700 * that directly follows the header in pipable WIMs. (Note: see
1701 * write_pipable_wim() for more details about the format of pipable
1704 struct wim_lookup_table_entry xml_lte;
1705 struct wim_resource_spec xml_rspec;
1706 ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 0, NULL);
1708 goto out_wimlib_free;
1710 if (!(xml_lte.flags & WIM_RESHDR_FLAG_METADATA))
1712 ERROR("Expected XML data, but found non-metadata "
1714 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1715 goto out_wimlib_free;
1718 wim_res_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr);
1720 ret = read_wim_xml_data(pwm);
1722 goto out_wimlib_free;
1724 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
1725 ERROR("Image count in XML data is not the same as in WIM header.");
1726 ret = WIMLIB_ERR_IMAGE_COUNT;
1727 goto out_wimlib_free;
1731 /* Get image index (this may use the XML data that was just read to
1732 * resolve an image name). */
1733 if (image_num_or_name) {
1734 image = wimlib_resolve_image(pwm, image_num_or_name);
1735 if (image == WIMLIB_NO_IMAGE) {
1736 ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
1738 ret = WIMLIB_ERR_INVALID_IMAGE;
1739 goto out_wimlib_free;
1740 } else if (image == WIMLIB_ALL_IMAGES) {
1741 ERROR("Applying all images from a pipe is not supported!");
1742 ret = WIMLIB_ERR_INVALID_IMAGE;
1743 goto out_wimlib_free;
1746 if (pwm->hdr.image_count != 1) {
1747 ERROR("No image was specified, but the pipable WIM "
1748 "did not contain exactly 1 image");
1749 ret = WIMLIB_ERR_INVALID_IMAGE;
1750 goto out_wimlib_free;
1755 /* Load the needed metadata resource. */
1756 for (i = 1; i <= pwm->hdr.image_count; i++) {
1757 struct wim_lookup_table_entry *metadata_lte;
1758 struct wim_image_metadata *imd;
1759 struct wim_resource_spec *metadata_rspec;
1761 metadata_lte = new_lookup_table_entry();
1762 if (metadata_lte == NULL) {
1763 ret = WIMLIB_ERR_NOMEM;
1764 goto out_wimlib_free;
1766 metadata_rspec = MALLOC(sizeof(struct wim_resource_spec));
1767 if (metadata_rspec == NULL) {
1768 ret = WIMLIB_ERR_NOMEM;
1769 free_lookup_table_entry(metadata_lte);
1770 goto out_wimlib_free;
1773 ret = read_pwm_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL);
1774 imd = pwm->image_metadata[i - 1];
1775 imd->metadata_lte = metadata_lte;
1777 FREE(metadata_rspec);
1778 goto out_wimlib_free;
1781 if (!(metadata_lte->flags & WIM_RESHDR_FLAG_METADATA)) {
1782 ERROR("Expected metadata resource, but found "
1783 "non-metadata stream.");
1784 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1785 goto out_wimlib_free;
1789 /* Metadata resource is for the image being extracted.
1790 * Parse it and save the metadata in memory. */
1791 ret = read_metadata_resource(pwm, imd);
1793 goto out_wimlib_free;
1796 /* Metadata resource is not for the image being
1797 * extracted. Skip over it. */
1798 ret = skip_wim_stream(metadata_lte);
1800 goto out_wimlib_free;
1803 /* Extract the image. */
1804 extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
1805 ret = do_wimlib_extract_image(pwm, image, target, extract_flags);
1806 /* Clean up and return. */
1814 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
1815 const tchar *target, int extract_flags)
1817 return wimlib_extract_image_from_pipe_with_progress(pipe_fd,
1826 wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1829 if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1830 return WIMLIB_ERR_INVALID_PARAM;
1831 return do_wimlib_extract_image(wim, image, target, extract_flags);