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"
57 # include "wimlib/win32.h" /* for realpath() equivalent */
59 #include "wimlib/xml.h"
60 #include "wimlib/wildcard.h"
61 #include "wimlib/wim.h"
69 #define WIMLIB_EXTRACT_FLAG_MULTI_IMAGE 0x80000000
70 #define WIMLIB_EXTRACT_FLAG_FROM_PIPE 0x40000000
71 #define WIMLIB_EXTRACT_FLAG_IMAGEMODE 0x20000000
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
116 #define PWM_SILENT_EOF 0x00002
118 /* Read the header from a stream in a pipable WIM. */
120 read_pwm_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
121 struct wim_resource_spec *rspec,
122 int flags, struct wim_header_disk *hdr_ret)
125 struct pwm_stream_hdr stream_hdr;
126 struct wim_header_disk pwm_hdr;
128 struct wim_reshdr reshdr;
131 ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
135 if ((flags & PWM_ALLOW_WIM_HDR) && buf.stream_hdr.magic == PWM_MAGIC) {
136 BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
137 ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
138 sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
142 lte->resource_location = RESOURCE_NONEXISTENT;
143 memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
147 if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) {
148 ERROR("Data read on pipe is invalid (expected stream header).");
149 return WIMLIB_ERR_INVALID_PIPABLE_WIM;
152 copy_hash(lte->hash, buf.stream_hdr.hash);
154 reshdr.size_in_wim = 0;
155 reshdr.flags = le32_to_cpu(buf.stream_hdr.flags);
156 reshdr.offset_in_wim = pwm->in_fd.offset;
157 reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
158 wim_res_hdr_to_spec(&reshdr, pwm, rspec);
159 lte_bind_wim_resource_spec(lte, rspec);
160 lte->flags = rspec->flags;
161 lte->size = rspec->uncompressed_size;
162 lte->offset_in_res = 0;
166 if (ret != WIMLIB_ERR_UNEXPECTED_END_OF_FILE || !(flags & PWM_SILENT_EOF))
167 ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
172 load_streams_from_pipe(struct apply_ctx *ctx,
173 const struct read_stream_list_callbacks *cbs)
175 struct wim_lookup_table_entry *found_lte = NULL;
176 struct wim_resource_spec *rspec = NULL;
177 struct wim_lookup_table *lookup_table;
180 ret = WIMLIB_ERR_NOMEM;
181 found_lte = new_lookup_table_entry();
185 rspec = MALLOC(sizeof(struct wim_resource_spec));
189 lookup_table = ctx->wim->lookup_table;
190 memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GUID_LEN);
191 ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
192 ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
193 if (ctx->progress_func) {
194 ctx->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
197 while (ctx->num_streams_remaining) {
198 struct wim_header_disk pwm_hdr;
199 struct wim_lookup_table_entry *needed_lte;
201 if (found_lte->resource_location != RESOURCE_NONEXISTENT)
202 lte_unbind_wim_resource_spec(found_lte);
203 ret = read_pwm_stream_header(ctx->wim, found_lte, rspec,
204 PWM_ALLOW_WIM_HDR, &pwm_hdr);
208 if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
209 && !(found_lte->flags & WIM_RESHDR_FLAG_METADATA)
210 && (needed_lte = lookup_stream(lookup_table, found_lte->hash))
211 && (needed_lte->out_refcnt))
213 needed_lte->offset_in_res = found_lte->offset_in_res;
214 needed_lte->flags = found_lte->flags;
215 needed_lte->size = found_lte->size;
217 lte_unbind_wim_resource_spec(found_lte);
218 lte_bind_wim_resource_spec(needed_lte, rspec);
220 ret = (*cbs->begin_stream)(needed_lte, 0,
221 cbs->begin_stream_ctx);
223 lte_unbind_wim_resource_spec(needed_lte);
227 ret = extract_stream(needed_lte, needed_lte->size,
229 cbs->consume_chunk_ctx);
231 ret = (*cbs->end_stream)(needed_lte, ret,
232 cbs->end_stream_ctx);
233 lte_unbind_wim_resource_spec(needed_lte);
236 ctx->num_streams_remaining--;
237 } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
238 ret = skip_wim_stream(found_lte);
242 u16 part_number = le16_to_cpu(pwm_hdr.part_number);
243 u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
245 if (part_number != ctx->progress.extract.part_number ||
246 total_parts != ctx->progress.extract.total_parts ||
247 memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
250 ctx->progress.extract.part_number = part_number;
251 ctx->progress.extract.total_parts = total_parts;
252 memcpy(ctx->progress.extract.guid,
253 pwm_hdr.guid, WIM_GUID_LEN);
254 if (ctx->progress_func) {
256 WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN,
264 if (found_lte->resource_location != RESOURCE_IN_WIM)
266 free_lookup_table_entry(found_lte);
271 begin_extract_stream_with_progress(struct wim_lookup_table_entry *lte,
272 u32 flags, void *_ctx)
274 struct apply_ctx *ctx = _ctx;
276 ctx->cur_stream = lte;
278 return (*ctx->saved_cbs->begin_stream)(lte, flags,
279 ctx->saved_cbs->begin_stream_ctx);
283 consume_chunk_with_progress(const void *chunk, size_t size, void *_ctx)
285 struct apply_ctx *ctx = _ctx;
286 wimlib_progress_func_t progress_func = ctx->progress_func;
287 union wimlib_progress_info *progress = &ctx->progress;
289 if (likely(ctx->supported_features.hard_links)) {
290 progress->extract.completed_bytes +=
291 (u64)size * ctx->cur_stream->out_refcnt;
293 const struct stream_owner *owners = stream_owners(ctx->cur_stream);
294 for (u32 i = 0; i < ctx->cur_stream->out_refcnt; i++) {
295 const struct wim_inode *inode = owners[i].inode;
296 const struct wim_dentry *dentry;
298 list_for_each_entry(dentry,
299 &inode->i_extraction_aliases,
300 d_extraction_alias_node)
302 progress->extract.completed_bytes += size;
306 if (progress->extract.completed_bytes >= ctx->next_progress) {
307 progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS, progress);
308 if (progress->extract.completed_bytes >=
309 progress->extract.total_bytes)
311 ctx->next_progress = UINT64_MAX;
313 ctx->next_progress += progress->extract.total_bytes / 128;
314 if (ctx->next_progress > progress->extract.total_bytes)
315 ctx->next_progress = progress->extract.total_bytes;
318 return (*ctx->saved_cbs->consume_chunk)(chunk, size,
319 ctx->saved_cbs->consume_chunk_ctx);
323 * Read the list of single-instance streams to extract and feed their data into
324 * the specified callback functions.
326 * This handles checksumming each stream.
328 * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS.
330 * This also works if the WIM is being read from a pipe, whereas attempting to
331 * read streams directly (e.g. with read_full_stream_into_buf()) will not.
334 extract_stream_list(struct apply_ctx *ctx,
335 const struct read_stream_list_callbacks *cbs)
337 struct read_stream_list_callbacks wrapper_cbs = {
338 .begin_stream = begin_extract_stream_with_progress,
339 .begin_stream_ctx = ctx,
340 .consume_chunk = consume_chunk_with_progress,
341 .consume_chunk_ctx = ctx,
342 .end_stream = cbs->end_stream,
343 .end_stream_ctx = cbs->end_stream_ctx,
345 if (ctx->progress_func) {
346 ctx->saved_cbs = cbs;
349 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
350 return load_streams_from_pipe(ctx, cbs);
352 return read_stream_list(&ctx->stream_list,
353 offsetof(struct wim_lookup_table_entry,
355 cbs, VERIFY_STREAM_HASHES);
359 /* Extract a WIM dentry to standard output.
361 * This obviously doesn't make sense in all cases. We return an error if the
362 * dentry does not correspond to a regular file. Otherwise we extract the
363 * unnamed data stream only. */
365 extract_dentry_to_stdout(struct wim_dentry *dentry,
366 const struct wim_lookup_table *lookup_table)
368 struct wim_inode *inode = dentry->d_inode;
369 struct wim_lookup_table_entry *lte;
370 struct filedes _stdout;
372 if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
373 FILE_ATTRIBUTE_DIRECTORY))
375 ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
376 "extracted to standard output", dentry_full_path(dentry));
377 return WIMLIB_ERR_NOT_A_REGULAR_FILE;
380 lte = inode_unnamed_lte(inode, lookup_table);
382 const u8 *hash = inode_unnamed_stream_hash(inode);
383 if (!is_zero_hash(hash))
384 return stream_not_found_error(inode, hash);
388 filedes_init(&_stdout, STDOUT_FILENO);
389 return extract_full_stream_to_fd(lte, &_stdout);
393 extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
394 const struct wim_lookup_table *lookup_table)
396 for (size_t i = 0; i < num_dentries; i++) {
397 int ret = extract_dentry_to_stdout(dentries[i], lookup_table);
404 /**********************************************************************/
407 * Removes duplicate dentries from the array.
409 * Returns the new number of dentries, packed at the front of the array.
412 remove_duplicate_trees(struct wim_dentry **trees, size_t num_trees)
415 for (i = 0; i < num_trees; i++) {
416 if (!trees[i]->tmp_flag) {
417 /* Found distinct dentry. */
418 trees[i]->tmp_flag = 1;
419 trees[j++] = trees[i];
422 for (i = 0; i < j; i++)
423 trees[i]->tmp_flag = 0;
428 * Remove dentries that are descendants of other dentries in the array.
430 * Returns the new number of dentries, packed at the front of the array.
433 remove_contained_trees(struct wim_dentry **trees, size_t num_trees)
436 for (i = 0; i < num_trees; i++)
437 trees[i]->tmp_flag = 1;
438 for (i = 0; i < num_trees; i++) {
439 struct wim_dentry *d = trees[i];
440 while (!dentry_is_root(d)) {
445 trees[j++] = trees[i];
449 trees[i]->tmp_flag = 0;
452 for (i = 0; i < j; i++)
453 trees[i]->tmp_flag = 0;
458 dentry_append_to_list(struct wim_dentry *dentry, void *_dentry_list)
460 struct list_head *dentry_list = _dentry_list;
461 list_add_tail(&dentry->d_extraction_list_node, dentry_list);
466 dentry_reset_extraction_list_node(struct wim_dentry *dentry)
468 dentry->d_extraction_list_node = (struct list_head){NULL, NULL};
472 dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore)
474 list_del(&dentry->d_extraction_list_node);
475 dentry_reset_extraction_list_node(dentry);
480 * Build the preliminary list of dentries to be extracted.
482 * The list maintains the invariant that if d1 and d2 are in the list and d1 is
483 * an ancestor of d2, then d1 appears before d2 in the list.
486 build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees,
487 size_t num_trees, bool add_ancestors)
489 INIT_LIST_HEAD(dentry_list);
491 /* Add the trees recursively. */
492 for (size_t i = 0; i < num_trees; i++)
493 for_dentry_in_tree(trees[i], dentry_append_to_list, dentry_list);
495 /* If requested, add ancestors of the trees. */
497 for (size_t i = 0; i < num_trees; i++) {
498 struct wim_dentry *dentry = trees[i];
499 struct wim_dentry *ancestor;
500 struct list_head *place_after;
502 if (dentry_is_root(dentry))
505 place_after = dentry_list;
508 ancestor = ancestor->parent;
509 if (will_extract_dentry(ancestor)) {
510 place_after = &ancestor->d_extraction_list_node;
513 } while (!dentry_is_root(ancestor));
517 ancestor = ancestor->parent;
518 if (will_extract_dentry(ancestor))
520 list_add(&ancestor->d_extraction_list_node, place_after);
521 } while (!dentry_is_root(ancestor));
527 destroy_dentry_list(struct list_head *dentry_list)
529 struct wim_dentry *dentry, *tmp;
530 struct wim_inode *inode;
532 list_for_each_entry_safe(dentry, tmp, dentry_list, d_extraction_list_node) {
533 inode = dentry->d_inode;
534 dentry_reset_extraction_list_node(dentry);
535 inode->i_visited = 0;
536 if ((void *)dentry->d_extraction_name != (void *)dentry->file_name)
537 FREE(dentry->d_extraction_name);
538 dentry->d_extraction_name = NULL;
539 dentry->d_extraction_name_nchars = 0;
544 destroy_stream_list(struct list_head *stream_list)
546 struct wim_lookup_table_entry *lte;
548 list_for_each_entry(lte, stream_list, extraction_list)
549 if (lte->out_refcnt > ARRAY_LEN(lte->inline_stream_owners))
550 FREE(lte->stream_owners);
554 static const utf16lechar replacement_char = cpu_to_le16(0xfffd);
556 static const utf16lechar replacement_char = cpu_to_le16('?');
560 file_name_valid(utf16lechar *name, size_t num_chars, bool fix)
566 for (i = 0; i < num_chars; i++) {
569 case cpu_to_le16('\\'):
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('|'):
578 case cpu_to_le16('/'):
579 case cpu_to_le16('\0'):
581 name[i] = replacement_char;
588 if (name[num_chars - 1] == cpu_to_le16(' ') ||
589 name[num_chars - 1] == cpu_to_le16('.'))
592 name[num_chars - 1] = replacement_char;
601 dentry_calculate_extraction_name(struct wim_dentry *dentry,
602 struct apply_ctx *ctx)
606 if (!dentry_is_supported(dentry, &ctx->supported_features))
609 if (dentry_is_root(dentry))
612 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
613 dentry->d_extraction_name = dentry->file_name;
614 dentry->d_extraction_name_nchars = dentry->file_name_nbytes /
619 if (!ctx->supported_features.case_sensitive_filenames) {
620 struct wim_dentry *other;
621 list_for_each_entry(other, &dentry->d_ci_conflict_list,
624 if (will_extract_dentry(other)) {
625 if (ctx->extract_flags &
626 WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
627 WARNING("\"%"TS"\" has the same "
628 "case-insensitive name as "
629 "\"%"TS"\"; extracting "
630 "dummy name instead",
631 dentry_full_path(dentry),
632 dentry_full_path(other));
635 WARNING("Not extracting \"%"TS"\": "
636 "has same case-insensitive "
638 dentry_full_path(dentry),
639 dentry_full_path(other));
646 if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
647 ret = utf16le_get_tstr(dentry->file_name,
648 dentry->file_name_nbytes,
649 (const tchar **)&dentry->d_extraction_name,
650 &dentry->d_extraction_name_nchars);
651 dentry->d_extraction_name_nchars /= sizeof(tchar);
654 if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
656 WARNING("\"%"TS"\" has an invalid filename "
657 "that is not supported on this platform; "
658 "extracting dummy name instead",
659 dentry_full_path(dentry));
662 WARNING("Not extracting \"%"TS"\": has an invalid filename "
663 "that is not supported on this platform",
664 dentry_full_path(dentry));
671 utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
673 memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
674 file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
676 const tchar *tchar_name;
679 ret = utf16le_get_tstr(utf16_name_copy,
680 dentry->file_name_nbytes,
681 &tchar_name, &tchar_nchars);
685 tchar_nchars /= sizeof(tchar);
687 size_t fixed_name_num_chars = tchar_nchars;
688 tchar fixed_name[tchar_nchars + 50];
690 tmemcpy(fixed_name, tchar_name, tchar_nchars);
691 fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
692 T(" (invalid filename #%lu)"),
693 ++ctx->invalid_sequence);
695 utf16le_put_tstr(tchar_name);
697 dentry->d_extraction_name = memdup(fixed_name,
698 2 * fixed_name_num_chars + 2);
699 if (!dentry->d_extraction_name)
700 return WIMLIB_ERR_NOMEM;
701 dentry->d_extraction_name_nchars = fixed_name_num_chars;
706 for_dentry_in_tree(dentry, dentry_delete_from_list, NULL);
711 * Calculate the actual filename component at which each WIM dentry will be
712 * extracted, with special handling for dentries that are unsupported by the
713 * extraction backend or have invalid names.
715 * ctx->supported_features must be filled in.
717 * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING
720 dentry_list_calculate_extraction_names(struct list_head *dentry_list,
721 struct apply_ctx *ctx)
723 struct list_head *prev, *cur;
725 /* Can't use list_for_each_entry() because a call to
726 * dentry_calculate_extraction_name() may delete the current dentry and
727 * its children from the list. */
731 struct wim_dentry *dentry;
735 if (cur == dentry_list)
738 dentry = list_entry(cur, struct wim_dentry, d_extraction_list_node);
740 ret = dentry_calculate_extraction_name(dentry, ctx);
744 if (prev->next == cur)
747 ; /* Current dentry and its children (which follow in
748 the list) were deleted. prev stays the same. */
754 dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
755 struct wim_lookup_table *lookup_table)
757 struct wim_inode *inode = dentry->d_inode;
758 struct wim_lookup_table_entry *lte;
762 /* Special case: when extracting from a pipe, the WIM lookup table is
763 * initially empty, so "resolving" an inode's streams is initially not
764 * possible. However, we still need to keep track of which streams,
765 * identified by SHA1 message digests, need to be extracted, so we
766 * "resolve" the inode's streams anyway by allocating new entries. */
767 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
769 ret = inode_resolve_streams(inode, lookup_table, force);
772 for (u32 i = 0; i <= inode->i_num_ads; i++) {
773 lte = inode_stream_lte_resolved(inode, i);
781 * For each dentry to be extracted, resolve all streams in the corresponding
782 * inode and set 'out_refcnt' in each to 0.
784 * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
787 dentry_list_resolve_streams(struct list_head *dentry_list,
788 struct apply_ctx *ctx)
790 struct wim_dentry *dentry;
793 list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
794 ret = dentry_resolve_streams(dentry,
796 ctx->wim->lookup_table);
804 ref_stream(struct wim_lookup_table_entry *lte, u32 stream_idx,
805 struct wim_dentry *dentry, struct apply_ctx *ctx)
807 struct wim_inode *inode = dentry->d_inode;
808 struct stream_owner *stream_owners;
813 /* Tally the size only for each extraction of the stream (not hard
815 if (inode->i_visited && ctx->supported_features.hard_links)
818 ctx->progress.extract.total_bytes += lte->size;
819 ctx->progress.extract.num_streams++;
821 if (inode->i_visited)
824 /* Add stream to the dentry_list only one time, even if it's going
825 * to be extracted to multiple inodes. */
826 if (lte->out_refcnt == 0) {
827 list_add_tail(<e->extraction_list, &ctx->stream_list);
828 ctx->num_streams_remaining++;
831 /* If inode not yet been visited, append it to the stream_owners array. */
832 if (lte->out_refcnt < ARRAY_LEN(lte->inline_stream_owners)) {
833 stream_owners = lte->inline_stream_owners;
835 struct stream_owner *prev_stream_owners;
836 size_t alloc_stream_owners;
838 if (lte->out_refcnt == ARRAY_LEN(lte->inline_stream_owners)) {
839 prev_stream_owners = NULL;
840 alloc_stream_owners = ARRAY_LEN(lte->inline_stream_owners);
842 prev_stream_owners = lte->stream_owners;
843 alloc_stream_owners = lte->alloc_stream_owners;
846 if (lte->out_refcnt == alloc_stream_owners) {
847 alloc_stream_owners *= 2;
848 stream_owners = REALLOC(prev_stream_owners,
849 alloc_stream_owners *
850 sizeof(stream_owners[0]));
852 return WIMLIB_ERR_NOMEM;
853 if (!prev_stream_owners) {
854 memcpy(stream_owners,
855 lte->inline_stream_owners,
856 sizeof(lte->inline_stream_owners));
858 lte->stream_owners = stream_owners;
859 lte->alloc_stream_owners = alloc_stream_owners;
861 stream_owners = lte->stream_owners;
863 stream_owners[lte->out_refcnt].inode = inode;
864 if (stream_idx == 0) {
865 stream_owners[lte->out_refcnt].stream_name = NULL;
867 stream_owners[lte->out_refcnt].stream_name =
868 inode->i_ads_entries[stream_idx - 1].stream_name;
875 dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
877 struct wim_inode *inode = dentry->d_inode;
880 /* The unnamed data stream will always be extracted, except in an
882 if (!inode_is_encrypted_directory(inode)) {
884 struct wim_lookup_table_entry *stream;
886 stream = inode_unnamed_stream_resolved(inode, &stream_idx);
887 ret = ref_stream(stream, stream_idx, dentry, ctx);
892 /* Named data streams will be extracted only if supported in the current
893 * extraction mode and volume, and to avoid complications, if not doing
894 * a linked extraction. */
895 if (ctx->supported_features.named_data_streams) {
896 for (u16 i = 0; i < inode->i_num_ads; i++) {
897 if (!ads_entry_is_named_stream(&inode->i_ads_entries[i]))
899 ret = ref_stream(inode->i_ads_entries[i].lte, i + 1,
905 inode->i_visited = 1;
910 * For each dentry to be extracted, iterate through the data streams of the
911 * corresponding inode. For each such stream that is not to be ignored due to
912 * the supported features or extraction flags, add it to the list of streams to
913 * be extracted (ctx->stream_list) if not already done so.
915 * Also builds a mapping from each stream to the inodes referencing it.
917 * This also initializes the extract progress info with byte and stream
920 * ctx->supported_features must be filled in.
922 * Possible error codes: WIMLIB_ERR_NOMEM.
925 dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
927 struct wim_dentry *dentry;
930 list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
931 ret = dentry_ref_streams(dentry, ctx);
935 list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
936 dentry->d_inode->i_visited = 0;
941 dentry_list_build_inode_alias_lists(struct list_head *dentry_list)
943 struct wim_dentry *dentry;
944 struct wim_inode *inode;
946 list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
947 inode = dentry->d_inode;
948 if (!inode->i_visited)
949 INIT_LIST_HEAD(&inode->i_extraction_aliases);
950 list_add_tail(&dentry->d_extraction_alias_node,
951 &inode->i_extraction_aliases);
952 inode->i_visited = 1;
954 list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
955 dentry->d_inode->i_visited = 0;
959 inode_tally_features(const struct wim_inode *inode,
960 struct wim_features *features)
962 if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
963 features->archive_files++;
964 if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
965 features->hidden_files++;
966 if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
967 features->system_files++;
968 if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
969 features->compressed_files++;
970 if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
971 if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
972 features->encrypted_directories++;
974 features->encrypted_files++;
976 if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
977 features->not_context_indexed_files++;
978 if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
979 features->sparse_files++;
980 if (inode_has_named_stream(inode))
981 features->named_data_streams++;
982 if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
983 features->reparse_points++;
984 if (inode_is_symlink(inode))
985 features->symlink_reparse_points++;
987 features->other_reparse_points++;
989 if (inode->i_security_id != -1)
990 features->security_descriptors++;
991 if (inode_has_unix_data(inode))
992 features->unix_data++;
995 /* Tally features necessary to extract a dentry and the corresponding inode. */
997 dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features)
999 struct wim_inode *inode = dentry->d_inode;
1001 if (dentry_has_short_name(dentry))
1002 features->short_names++;
1004 if (inode->i_visited) {
1005 features->hard_links++;
1007 inode_tally_features(inode, features);
1008 inode->i_visited = 1;
1012 /* Tally the features necessary to extract the specified dentries. */
1014 dentry_list_get_features(struct list_head *dentry_list,
1015 struct wim_features *features)
1017 struct wim_dentry *dentry;
1019 list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1020 dentry_tally_features(dentry, features);
1022 list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
1023 dentry->d_inode->i_visited = 0;
1027 do_feature_check(const struct wim_features *required_features,
1028 const struct wim_features *supported_features,
1031 /* File attributes. */
1032 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
1033 /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
1034 * We're an archive program, so theoretically we can do what we
1037 if (required_features->hidden_files &&
1038 !supported_features->hidden_files)
1039 WARNING("Ignoring FILE_ATTRIBUTE_HIDDEN of %lu files",
1040 required_features->hidden_files);
1042 if (required_features->system_files &&
1043 !supported_features->system_files)
1044 WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
1045 required_features->system_files);
1047 if (required_features->compressed_files &&
1048 !supported_features->compressed_files)
1049 WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
1050 required_features->compressed_files);
1052 if (required_features->not_context_indexed_files &&
1053 !supported_features->not_context_indexed_files)
1054 WARNING("Ignoring FILE_ATTRIBUTE_NOT_CONTENT_INDEXED of %lu files",
1055 required_features->not_context_indexed_files);
1057 if (required_features->sparse_files &&
1058 !supported_features->sparse_files)
1059 WARNING("Ignoring FILE_ATTRIBUTE_SPARSE_FILE of %lu files",
1060 required_features->sparse_files);
1062 if (required_features->encrypted_directories &&
1063 !supported_features->encrypted_directories)
1064 WARNING("Ignoring FILE_ATTRIBUTE_ENCRYPTED of %lu directories",
1065 required_features->encrypted_directories);
1068 /* Encrypted files. */
1069 if (required_features->encrypted_files &&
1070 !supported_features->encrypted_files)
1071 WARNING("Ignoring %lu encrypted files",
1072 required_features->encrypted_files);
1074 /* Named data streams. */
1075 if (required_features->named_data_streams &&
1076 (!supported_features->named_data_streams))
1077 WARNING("Ignoring named data streams of %lu files",
1078 required_features->named_data_streams);
1081 if (required_features->hard_links && !supported_features->hard_links)
1082 WARNING("Extracting %lu hard links as independent files",
1083 required_features->hard_links);
1085 /* Symbolic links and reparse points. */
1086 if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
1087 required_features->symlink_reparse_points &&
1088 !supported_features->symlink_reparse_points &&
1089 !supported_features->reparse_points)
1091 ERROR("Extraction backend does not support symbolic links!");
1092 return WIMLIB_ERR_UNSUPPORTED;
1094 if (required_features->reparse_points &&
1095 !supported_features->reparse_points)
1097 if (supported_features->symlink_reparse_points) {
1098 if (required_features->other_reparse_points) {
1099 WARNING("Ignoring %lu non-symlink/junction "
1100 "reparse point files",
1101 required_features->other_reparse_points);
1104 WARNING("Ignoring %lu reparse point files",
1105 required_features->reparse_points);
1109 /* Security descriptors. */
1110 if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
1111 WIMLIB_EXTRACT_FLAG_UNIX_DATA))
1112 == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
1113 required_features->security_descriptors &&
1114 !supported_features->security_descriptors)
1116 ERROR("Extraction backend does not support security descriptors!");
1117 return WIMLIB_ERR_UNSUPPORTED;
1119 if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) &&
1120 required_features->security_descriptors &&
1121 !supported_features->security_descriptors)
1122 WARNING("Ignoring Windows NT security descriptors of %lu files",
1123 required_features->security_descriptors);
1126 if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
1127 required_features->unix_data && !supported_features->unix_data)
1129 ERROR("Extraction backend does not support UNIX data!");
1130 return WIMLIB_ERR_UNSUPPORTED;
1134 if (required_features->short_names &&
1135 !supported_features->short_names)
1137 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) {
1138 ERROR("Extraction backend does not support DOS names!");
1139 return WIMLIB_ERR_UNSUPPORTED;
1141 WARNING("Ignoring DOS names of %lu files",
1142 required_features->short_names);
1146 if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
1147 !supported_features->timestamps)
1149 ERROR("Extraction backend does not support timestamps!");
1150 return WIMLIB_ERR_UNSUPPORTED;
1156 static const struct apply_operations *
1157 select_apply_operations(int extract_flags)
1160 if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
1161 return &ntfs_3g_apply_ops;
1164 return &win32_apply_ops;
1166 return &unix_apply_ops;
1171 extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
1172 const tchar *target, int extract_flags,
1173 wimlib_progress_func_t progress_func)
1175 const struct apply_operations *ops;
1176 struct apply_ctx *ctx;
1178 LIST_HEAD(dentry_list);
1180 if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
1181 ret = extract_dentries_to_stdout(trees, num_trees,
1186 num_trees = remove_duplicate_trees(trees, num_trees);
1187 num_trees = remove_contained_trees(trees, num_trees);
1189 ops = select_apply_operations(extract_flags);
1191 if (num_trees > 1 && ops->single_tree_only) {
1192 ERROR("Extracting multiple directory trees "
1193 "at once is not supported in %s extraction mode!",
1195 ret = WIMLIB_ERR_UNSUPPORTED;
1199 ctx = CALLOC(1, ops->context_size);
1201 ret = WIMLIB_ERR_NOMEM;
1206 ctx->target = target;
1207 ctx->target_nchars = tstrlen(target);
1208 ctx->extract_flags = extract_flags;
1209 if (progress_func) {
1210 ctx->progress_func = progress_func;
1211 ctx->progress.extract.image = wim->current_image;
1212 ctx->progress.extract.extract_flags = (extract_flags &
1213 WIMLIB_EXTRACT_MASK_PUBLIC);
1214 ctx->progress.extract.wimfile_name = wim->filename;
1215 ctx->progress.extract.image_name = wimlib_get_image_name(wim,
1216 wim->current_image);
1217 ctx->progress.extract.target = target;
1219 INIT_LIST_HEAD(&ctx->stream_list);
1221 ret = (*ops->get_supported_features)(target, &ctx->supported_features);
1225 build_dentry_list(&dentry_list, trees, num_trees,
1227 WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE));
1229 dentry_list_get_features(&dentry_list, &ctx->required_features);
1231 ret = do_feature_check(&ctx->required_features, &ctx->supported_features,
1232 ctx->extract_flags);
1236 ret = dentry_list_calculate_extraction_names(&dentry_list, ctx);
1240 ret = dentry_list_resolve_streams(&dentry_list, ctx);
1244 ret = dentry_list_ref_streams(&dentry_list, ctx);
1248 dentry_list_build_inode_alias_lists(&dentry_list);
1250 if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
1251 /* When extracting from a pipe, the number of bytes of data to
1252 * extract can't be determined in the normal way (examining the
1253 * lookup table), since at this point all we have is a set of
1254 * SHA1 message digests of streams that need to be extracted.
1255 * However, we can get a reasonably accurate estimate by taking
1256 * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
1257 * data. This does assume that a full image is being extracted,
1258 * but currently there is no API for doing otherwise. (Also,
1259 * subtract <HARDLINKBYTES> from this if hard links are
1260 * supported by the extraction mode.) */
1261 ctx->progress.extract.total_bytes =
1262 wim_info_get_image_total_bytes(wim->wim_info,
1263 wim->current_image);
1264 if (ctx->supported_features.hard_links) {
1265 ctx->progress.extract.total_bytes -=
1266 wim_info_get_image_hard_link_bytes(wim->wim_info,
1267 wim->current_image);
1271 if (ctx->progress_func) {
1273 if (extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE)
1274 msg = WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
1276 msg = WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN;
1277 (*ctx->progress_func)(msg, &ctx->progress);
1280 ret = (*ops->extract)(&dentry_list, ctx);
1284 if (ctx->progress_func &&
1285 ctx->progress.extract.completed_bytes <
1286 ctx->progress.extract.total_bytes)
1288 ctx->progress.extract.completed_bytes =
1289 ctx->progress.extract.total_bytes;
1290 (*ctx->progress_func)(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS,
1294 if (ctx->progress_func) {
1296 if (extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE)
1297 msg = WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END;
1299 msg = WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END;
1300 (*ctx->progress_func)(msg, &ctx->progress);
1304 destroy_stream_list(&ctx->stream_list);
1305 destroy_dentry_list(&dentry_list);
1312 mkdir_if_needed(const tchar *target)
1315 if (tstat(target, &stbuf)) {
1316 if (errno == ENOENT) {
1317 if (tmkdir(target, 0755)) {
1318 ERROR_WITH_ERRNO("Failed to create directory "
1319 "\"%"TS"\"", target);
1320 return WIMLIB_ERR_MKDIR;
1323 ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
1324 return WIMLIB_ERR_STAT;
1326 } else if (!S_ISDIR(stbuf.st_mode)) {
1327 ERROR("\"%"TS"\" is not a directory", target);
1328 return WIMLIB_ERR_NOTDIR;
1333 /* Make sure the extraction flags make sense, and update them if needed. */
1335 check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
1337 int extract_flags = *extract_flags_p;
1339 /* Check for invalid flag combinations */
1341 if ((extract_flags &
1342 (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1343 WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
1344 WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
1345 return WIMLIB_ERR_INVALID_PARAM;
1347 if ((extract_flags &
1348 (WIMLIB_EXTRACT_FLAG_RPFIX |
1349 WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
1350 WIMLIB_EXTRACT_FLAG_NORPFIX))
1351 return WIMLIB_ERR_INVALID_PARAM;
1353 #ifndef WITH_NTFS_3G
1354 if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1355 ERROR("wimlib was compiled without support for NTFS-3g, so\n"
1356 " it cannot apply a WIM image directly to an NTFS volume.");
1357 return WIMLIB_ERR_UNSUPPORTED;
1362 if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
1363 ERROR("WIMBoot extraction is only supported on Windows!");
1364 return WIMLIB_ERR_UNSUPPORTED;
1368 if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
1369 WIMLIB_EXTRACT_FLAG_NORPFIX |
1370 WIMLIB_EXTRACT_FLAG_IMAGEMODE)) ==
1371 WIMLIB_EXTRACT_FLAG_IMAGEMODE)
1373 /* For full-image extraction, do reparse point fixups by default
1374 * if the WIM header says they are enabled. */
1375 if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
1376 extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
1379 *extract_flags_p = extract_flags;
1384 get_wildcard_flags(int extract_flags)
1386 u32 wildcard_flags = 0;
1388 if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB)
1389 wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH;
1391 wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH;
1393 if (default_ignore_case)
1394 wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
1396 return wildcard_flags;
1399 struct append_dentry_ctx {
1400 struct wim_dentry **dentries;
1401 size_t num_dentries;
1402 size_t num_alloc_dentries;
1406 append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
1408 struct append_dentry_ctx *ctx = _ctx;
1410 if (ctx->num_dentries == ctx->num_alloc_dentries) {
1411 struct wim_dentry **new_dentries;
1414 new_length = max(ctx->num_alloc_dentries + 8,
1415 ctx->num_alloc_dentries * 3 / 2);
1416 new_dentries = REALLOC(ctx->dentries,
1417 new_length * sizeof(ctx->dentries[0]));
1418 if (new_dentries == NULL)
1419 return WIMLIB_ERR_NOMEM;
1420 ctx->dentries = new_dentries;
1421 ctx->num_alloc_dentries = new_length;
1423 ctx->dentries[ctx->num_dentries++] = dentry;
1428 do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1429 const tchar * const *paths, size_t num_paths,
1430 int extract_flags, wimlib_progress_func_t progress_func)
1433 struct wim_dentry **trees;
1436 if (wim == NULL || target == NULL || target[0] == T('\0') ||
1437 (num_paths != 0 && paths == NULL))
1438 return WIMLIB_ERR_INVALID_PARAM;
1440 ret = check_extract_flags(wim, &extract_flags);
1444 ret = select_wim_image(wim, image);
1448 ret = wim_checksum_unhashed_streams(wim);
1452 if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NTFS |
1453 WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) ==
1454 (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE))
1456 ret = mkdir_if_needed(target);
1461 if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
1463 struct append_dentry_ctx append_dentry_ctx = {
1466 .num_alloc_dentries = 0,
1469 u32 wildcard_flags = get_wildcard_flags(extract_flags);
1471 for (size_t i = 0; i < num_paths; i++) {
1472 tchar *path = canonicalize_wim_path(paths[i]);
1474 ret = WIMLIB_ERR_NOMEM;
1475 trees = append_dentry_ctx.dentries;
1476 goto out_free_trees;
1478 ret = expand_wildcard(wim, path,
1484 trees = append_dentry_ctx.dentries;
1485 goto out_free_trees;
1488 trees = append_dentry_ctx.dentries;
1489 num_trees = append_dentry_ctx.num_dentries;
1491 trees = MALLOC(num_paths * sizeof(trees[0]));
1493 return WIMLIB_ERR_NOMEM;
1495 for (size_t i = 0; i < num_paths; i++) {
1497 tchar *path = canonicalize_wim_path(paths[i]);
1499 ret = WIMLIB_ERR_NOMEM;
1500 goto out_free_trees;
1503 trees[i] = get_dentry(wim, path,
1504 WIMLIB_CASE_PLATFORM_DEFAULT);
1506 if (trees[i] == NULL) {
1507 ERROR("Path \"%"TS"\" does not exist "
1509 paths[i], wim->current_image);
1510 ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
1511 goto out_free_trees;
1514 num_trees = num_paths;
1517 if (num_trees == 0) {
1519 goto out_free_trees;
1522 ret = extract_trees(wim, trees, num_trees,
1523 target, extract_flags, progress_func);
1530 extract_single_image(WIMStruct *wim, int image,
1531 const tchar *target, int extract_flags,
1532 wimlib_progress_func_t progress_func)
1534 const tchar *path = WIMLIB_WIM_ROOT_PATH;
1535 extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE;
1536 return do_wimlib_extract_paths(wim, image, target, &path, 1,
1537 extract_flags, progress_func);
1540 static const tchar * const filename_forbidden_chars =
1549 /* This function checks if it is okay to use a WIM image's name as a directory
1552 image_name_ok_as_dir(const tchar *image_name)
1554 return image_name && *image_name &&
1555 !tstrpbrk(image_name, filename_forbidden_chars) &&
1556 tstrcmp(image_name, T(".")) &&
1557 tstrcmp(image_name, T(".."));
1560 /* Extracts all images from the WIM to the directory @target, with the images
1561 * placed in subdirectories named by their image names. */
1563 extract_all_images(WIMStruct *wim,
1564 const tchar *target,
1566 wimlib_progress_func_t progress_func)
1568 size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
1569 size_t output_path_len = tstrlen(target);
1570 tchar buf[output_path_len + 1 + image_name_max_len + 1];
1573 const tchar *image_name;
1575 extract_flags |= WIMLIB_EXTRACT_FLAG_MULTI_IMAGE;
1577 if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
1578 ERROR("Cannot extract multiple images in NTFS extraction mode.");
1579 return WIMLIB_ERR_INVALID_PARAM;
1582 ret = mkdir_if_needed(target);
1585 tmemcpy(buf, target, output_path_len);
1586 buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
1587 for (image = 1; image <= wim->hdr.image_count; image++) {
1588 image_name = wimlib_get_image_name(wim, image);
1589 if (image_name_ok_as_dir(image_name)) {
1590 tstrcpy(buf + output_path_len + 1, image_name);
1592 /* Image name is empty or contains forbidden characters.
1593 * Use image number instead. */
1594 tsprintf(buf + output_path_len + 1, T("%d"), image);
1596 ret = extract_single_image(wim, image, buf, extract_flags,
1605 do_wimlib_extract_image(WIMStruct *wim,
1607 const tchar *target,
1609 wimlib_progress_func_t progress_func)
1611 if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
1612 WIMLIB_EXTRACT_FLAG_TO_STDOUT |
1613 WIMLIB_EXTRACT_FLAG_GLOB_PATHS))
1614 return WIMLIB_ERR_INVALID_PARAM;
1616 if (image == WIMLIB_ALL_IMAGES)
1617 return extract_all_images(wim, target, extract_flags,
1620 return extract_single_image(wim, image, target, extract_flags,
1625 /****************************************************************************
1627 ****************************************************************************/
1630 wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
1631 const tchar * const *paths, size_t num_paths,
1632 int extract_flags, wimlib_progress_func_t progress_func)
1634 if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1635 return WIMLIB_ERR_INVALID_PARAM;
1637 return do_wimlib_extract_paths(wim, image, target, paths, num_paths,
1638 extract_flags, progress_func);
1642 wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target,
1643 const tchar *path_list_file, int extract_flags,
1644 wimlib_progress_func_t progress_func)
1651 ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
1653 ERROR("Failed to read path list file \"%"TS"\"",
1658 ret = wimlib_extract_paths(wim, image, target,
1659 (const tchar * const *)paths, num_paths,
1660 extract_flags, progress_func);
1667 wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
1668 const tchar *target, int extract_flags,
1669 wimlib_progress_func_t progress_func)
1673 struct filedes *in_fd;
1677 if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
1678 return WIMLIB_ERR_INVALID_PARAM;
1680 /* Read the WIM header from the pipe and get a WIMStruct to represent
1681 * the pipable WIM. Caveats: Unlike getting a WIMStruct with
1682 * wimlib_open_wim(), getting a WIMStruct in this way will result in
1683 * an empty lookup table, no XML data read, and no filename set. */
1684 ret = open_wim_as_WIMStruct(&pipe_fd,
1685 WIMLIB_OPEN_FLAG_FROM_PIPE,
1686 &pwm, progress_func);
1690 /* Sanity check to make sure this is a pipable WIM. */
1691 if (pwm->hdr.magic != PWM_MAGIC) {
1692 ERROR("The WIM being read from file descriptor %d "
1693 "is not pipable!", pipe_fd);
1694 ret = WIMLIB_ERR_NOT_PIPABLE;
1695 goto out_wimlib_free;
1698 /* Sanity check to make sure the first part of a pipable split WIM is
1699 * sent over the pipe first. */
1700 if (pwm->hdr.part_number != 1) {
1701 ERROR("The first part of the split WIM must be "
1702 "sent over the pipe first.");
1703 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1704 goto out_wimlib_free;
1707 in_fd = &pwm->in_fd;
1708 wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
1710 /* As mentioned, the WIMStruct we created from the pipe does not have
1711 * XML data yet. Fix this by reading the extra copy of the XML data
1712 * that directly follows the header in pipable WIMs. (Note: see
1713 * write_pipable_wim() for more details about the format of pipable
1716 struct wim_lookup_table_entry xml_lte;
1717 struct wim_resource_spec xml_rspec;
1718 ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 0, NULL);
1720 goto out_wimlib_free;
1722 if (!(xml_lte.flags & WIM_RESHDR_FLAG_METADATA))
1724 ERROR("Expected XML data, but found non-metadata "
1726 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1727 goto out_wimlib_free;
1730 wim_res_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr);
1732 ret = read_wim_xml_data(pwm);
1734 goto out_wimlib_free;
1736 if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
1737 ERROR("Image count in XML data is not the same as in WIM header.");
1738 ret = WIMLIB_ERR_IMAGE_COUNT;
1739 goto out_wimlib_free;
1743 /* Get image index (this may use the XML data that was just read to
1744 * resolve an image name). */
1745 if (image_num_or_name) {
1746 image = wimlib_resolve_image(pwm, image_num_or_name);
1747 if (image == WIMLIB_NO_IMAGE) {
1748 ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
1750 ret = WIMLIB_ERR_INVALID_IMAGE;
1751 goto out_wimlib_free;
1752 } else if (image == WIMLIB_ALL_IMAGES) {
1753 ERROR("Applying all images from a pipe is not supported!");
1754 ret = WIMLIB_ERR_INVALID_IMAGE;
1755 goto out_wimlib_free;
1758 if (pwm->hdr.image_count != 1) {
1759 ERROR("No image was specified, but the pipable WIM "
1760 "did not contain exactly 1 image");
1761 ret = WIMLIB_ERR_INVALID_IMAGE;
1762 goto out_wimlib_free;
1767 /* Load the needed metadata resource. */
1768 for (i = 1; i <= pwm->hdr.image_count; i++) {
1769 struct wim_lookup_table_entry *metadata_lte;
1770 struct wim_image_metadata *imd;
1771 struct wim_resource_spec *metadata_rspec;
1773 metadata_lte = new_lookup_table_entry();
1774 if (metadata_lte == NULL) {
1775 ret = WIMLIB_ERR_NOMEM;
1776 goto out_wimlib_free;
1778 metadata_rspec = MALLOC(sizeof(struct wim_resource_spec));
1779 if (metadata_rspec == NULL) {
1780 ret = WIMLIB_ERR_NOMEM;
1781 free_lookup_table_entry(metadata_lte);
1782 goto out_wimlib_free;
1785 ret = read_pwm_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL);
1786 imd = pwm->image_metadata[i - 1];
1787 imd->metadata_lte = metadata_lte;
1789 FREE(metadata_rspec);
1790 goto out_wimlib_free;
1793 if (!(metadata_lte->flags & WIM_RESHDR_FLAG_METADATA)) {
1794 ERROR("Expected metadata resource, but found "
1795 "non-metadata stream.");
1796 ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
1797 goto out_wimlib_free;
1801 /* Metadata resource is for the image being extracted.
1802 * Parse it and save the metadata in memory. */
1803 ret = read_metadata_resource(pwm, imd);
1805 goto out_wimlib_free;
1808 /* Metadata resource is not for the image being
1809 * extracted. Skip over it. */
1810 ret = skip_wim_stream(metadata_lte);
1812 goto out_wimlib_free;
1815 /* Extract the image. */
1816 extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
1817 ret = do_wimlib_extract_image(pwm, image, target,
1818 extract_flags, progress_func);
1819 /* Clean up and return. */
1826 wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
1827 int extract_flags, wimlib_progress_func_t progress_func)
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,