*/
/*
- * Copyright (C) 2012, 2013 Eric Biggers
+ * Copyright (C) 2012, 2013, 2014, 2015 Eric Biggers
*
- * This file is part of wimlib, a library for working with WIM files.
+ * This file is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU Lesser General Public License as published by the Free
+ * Software Foundation; either version 3 of the License, or (at your option) any
+ * later version.
*
- * wimlib is free software; you can redistribute it and/or modify it under the
- * terms of the GNU General Public License as published by the Free
- * Software Foundation; either version 3 of the License, or (at your option)
- * any later version.
- *
- * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
- * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
- * A PARTICULAR PURPOSE. See the GNU General Public License for more
+ * This file is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+ * FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
* details.
*
- * You should have received a copy of the GNU General Public License
- * along with wimlib; if not, see http://www.gnu.org/licenses/.
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this file; if not, see http://www.gnu.org/licenses/.
+ */
+
+/*
+ * This file provides the API functions wimlib_extract_image(),
+ * wimlib_extract_image_from_pipe(), wimlib_extract_paths(), and
+ * wimlib_extract_pathlist(). Internally, all end up calling
+ * do_wimlib_extract_paths() and extract_trees().
+ *
+ * Although wimlib supports multiple extraction modes/backends (NTFS-3g, UNIX,
+ * Win32), this file does not itself have code to extract files or directories
+ * to any specific target; instead, it handles generic functionality and relies
+ * on lower-level callback functions declared in `struct apply_operations' to do
+ * the actual extraction.
*/
#ifdef HAVE_CONFIG_H
# include "config.h"
#endif
-#ifdef __WIN32__
-# include "wimlib/win32_common.h" /* For GetFullPathName() */
-#endif
+#include <errno.h>
+#include <fcntl.h>
+#include <sys/stat.h>
+#include <unistd.h>
#include "wimlib/apply.h"
+#include "wimlib/assert.h"
+#include "wimlib/blob_table.h"
#include "wimlib/dentry.h"
#include "wimlib/encoding.h"
#include "wimlib/endianness.h"
#include "wimlib/error.h"
-#include "wimlib/lookup_table.h"
+#include "wimlib/metadata.h"
+#include "wimlib/pathlist.h"
#include "wimlib/paths.h"
+#include "wimlib/reparse.h"
#include "wimlib/resource.h"
-#include "wimlib/swm.h"
-#ifdef __WIN32__
-# include "wimlib/win32.h" /* for realpath() equivalent */
-#endif
+#include "wimlib/security.h"
+#include "wimlib/unix_data.h"
+#include "wimlib/wildcard.h"
+#include "wimlib/wim.h"
+#include "wimlib/win32.h" /* for realpath() equivalent */
#include "wimlib/xml.h"
-#include <errno.h>
-#include <limits.h>
-#ifdef WITH_NTFS_3G
-# include <ntfs-3g/volume.h> /* for ntfs_mount(), ntfs_umount() */
-#endif
-#include <stdlib.h>
-#include <sys/stat.h>
-#include <unistd.h>
+#define WIMLIB_EXTRACT_FLAG_FROM_PIPE 0x80000000
+#define WIMLIB_EXTRACT_FLAG_IMAGEMODE 0x40000000
+
+/* Keep in sync with wimlib.h */
+#define WIMLIB_EXTRACT_MASK_PUBLIC \
+ (WIMLIB_EXTRACT_FLAG_NTFS | \
+ WIMLIB_EXTRACT_FLAG_UNIX_DATA | \
+ WIMLIB_EXTRACT_FLAG_NO_ACLS | \
+ WIMLIB_EXTRACT_FLAG_STRICT_ACLS | \
+ WIMLIB_EXTRACT_FLAG_RPFIX | \
+ WIMLIB_EXTRACT_FLAG_NORPFIX | \
+ WIMLIB_EXTRACT_FLAG_TO_STDOUT | \
+ WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES | \
+ WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS | \
+ WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS | \
+ WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES | \
+ WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS | \
+ WIMLIB_EXTRACT_FLAG_GLOB_PATHS | \
+ WIMLIB_EXTRACT_FLAG_STRICT_GLOB | \
+ WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES | \
+ WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE | \
+ WIMLIB_EXTRACT_FLAG_WIMBOOT)
+
+/* Send WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE or
+ * WIMLIB_PROGRESS_MSG_EXTRACT_METADATA. */
+int
+do_file_extract_progress(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
+{
+ ctx->count_until_file_progress = 500; /* Arbitrary value to limit calls */
+ return extract_progress(ctx, msg);
+}
+
+static int
+start_file_phase(struct apply_ctx *ctx, u64 end_file_count, enum wimlib_progress_msg msg)
+{
+ ctx->progress.extract.current_file_count = 0;
+ ctx->progress.extract.end_file_count = end_file_count;
+ return do_file_extract_progress(ctx, msg);
+}
+
+int
+start_file_structure_phase(struct apply_ctx *ctx, u64 end_file_count)
+{
+ return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
+}
-#define MAX_EXTRACT_LONG_PATH_WARNINGS 5
+int
+start_file_metadata_phase(struct apply_ctx *ctx, u64 end_file_count)
+{
+ return start_file_phase(ctx, end_file_count, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
+}
static int
-do_apply_op(struct wim_dentry *dentry, struct apply_args *args,
- int (*apply_dentry_func)(const tchar *, size_t,
- struct wim_dentry *, struct apply_args *))
+end_file_phase(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
{
- tchar *p;
- size_t extraction_path_nchars;
- struct wim_dentry *d;
- LIST_HEAD(ancestor_list);
- const tchar *target;
- size_t target_nchars;
+ ctx->progress.extract.current_file_count = ctx->progress.extract.end_file_count;
+ return do_file_extract_progress(ctx, msg);
+}
-#ifdef __WIN32__
- if (args->target_lowlevel_path) {
- target = args->target_lowlevel_path;
- target_nchars = args->target_lowlevel_path_nchars;
- } else
-#endif
- {
- target = args->target;
- target_nchars = args->target_nchars;
- }
+int
+end_file_structure_phase(struct apply_ctx *ctx)
+{
+ return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_FILE_STRUCTURE);
+}
+
+int
+end_file_metadata_phase(struct apply_ctx *ctx)
+{
+ return end_file_phase(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_METADATA);
+}
- extraction_path_nchars = target_nchars;
+#define PWM_FOUND_WIM_HDR (-1)
+
+/* Read the header for a blob in a pipable WIM. If @pwm_hdr_ret is not NULL,
+ * also look for a pipable WIM header and return PWM_FOUND_WIM_HDR if found. */
+static int
+read_pwm_blob_header(WIMStruct *pwm, u8 hash_ret[SHA1_HASH_SIZE],
+ struct wim_reshdr *reshdr_ret,
+ struct wim_header_disk *pwm_hdr_ret)
+{
+ int ret;
+ struct pwm_blob_hdr blob_hdr;
+ u64 magic;
+
+ ret = full_read(&pwm->in_fd, &blob_hdr, sizeof(blob_hdr));
+ if (unlikely(ret))
+ goto read_error;
+
+ magic = le64_to_cpu(blob_hdr.magic);
+
+ if (magic == PWM_MAGIC && pwm_hdr_ret != NULL) {
+ memcpy(pwm_hdr_ret, &blob_hdr, sizeof(blob_hdr));
+ ret = full_read(&pwm->in_fd,
+ (u8 *)pwm_hdr_ret + sizeof(blob_hdr),
+ sizeof(*pwm_hdr_ret) - sizeof(blob_hdr));
+ if (unlikely(ret))
+ goto read_error;
+ return PWM_FOUND_WIM_HDR;
+ }
- for (d = dentry; d != args->extract_root; d = d->parent) {
- if (d->not_extracted)
- return 0;
- extraction_path_nchars += d->extraction_name_nchars + 1;
- list_add(&d->tmp_list, &ancestor_list);
+ if (unlikely(magic != PWM_BLOB_MAGIC)) {
+ ERROR("Data read on pipe is invalid (expected blob header)");
+ return WIMLIB_ERR_INVALID_PIPABLE_WIM;
}
- tchar extraction_path[extraction_path_nchars + 1];
- p = tmempcpy(extraction_path, target, target_nchars);
+ copy_hash(hash_ret, blob_hdr.hash);
+ reshdr_ret->size_in_wim = 0; /* Not available */
+ reshdr_ret->flags = le32_to_cpu(blob_hdr.flags);
+ reshdr_ret->offset_in_wim = pwm->in_fd.offset;
+ reshdr_ret->uncompressed_size = le64_to_cpu(blob_hdr.uncompressed_size);
- list_for_each_entry(d, &ancestor_list, tmp_list) {
- *p++ = OS_PREFERRED_PATH_SEPARATOR;
- p = tmempcpy(p, d->extraction_name, d->extraction_name_nchars);
+ if (unlikely(reshdr_ret->uncompressed_size == 0)) {
+ ERROR("Data read on pipe is invalid (resource is of 0 size)");
+ return WIMLIB_ERR_INVALID_PIPABLE_WIM;
}
- *p = T('\0');
-#ifdef __WIN32__
- /* Warn the user if the path exceeds MAX_PATH */
+ return 0;
+
+read_error:
+ if (ret == WIMLIB_ERR_UNEXPECTED_END_OF_FILE)
+ ERROR("The pipe ended before all needed data was sent!");
+ else
+ ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
+ return ret;
+}
+
+static int
+read_blobs_from_pipe(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
+{
+ int ret;
+ u8 hash[SHA1_HASH_SIZE];
+ struct wim_reshdr reshdr;
+ struct wim_header_disk pwm_hdr;
+ struct wim_resource_descriptor rdesc;
+ struct blob_descriptor *blob;
+
+ copy_guid(ctx->progress.extract.guid, ctx->wim->hdr.guid);
+ ctx->progress.extract.part_number = ctx->wim->hdr.part_number;
+ ctx->progress.extract.total_parts = ctx->wim->hdr.total_parts;
+ ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
+ if (ret)
+ return ret;
+
+ while (ctx->num_blobs_remaining) {
+
+ ret = read_pwm_blob_header(ctx->wim, hash, &reshdr, &pwm_hdr);
+
+ if (ret == PWM_FOUND_WIM_HDR) {
+ u16 part_number = le16_to_cpu(pwm_hdr.part_number);
+ u16 total_parts = le16_to_cpu(pwm_hdr.total_parts);
+
+ if (part_number == ctx->progress.extract.part_number &&
+ total_parts == ctx->progress.extract.total_parts &&
+ guids_equal(pwm_hdr.guid, ctx->progress.extract.guid))
+ continue;
+
+ copy_guid(ctx->progress.extract.guid, pwm_hdr.guid);
+ ctx->progress.extract.part_number = part_number;
+ ctx->progress.extract.total_parts = total_parts;
+ ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
+ if (ret)
+ return ret;
+
+ continue;
+ }
+
+ if (ret)
+ return ret;
- /* + 1 for '\0', -4 for \\?\. */
- if (extraction_path_nchars + 1 - 4 > MAX_PATH) {
- if (dentry->needs_extraction &&
- args->num_long_paths < MAX_EXTRACT_LONG_PATH_WARNINGS)
+ if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)
+ && (blob = lookup_blob(ctx->wim->blob_table, hash))
+ && (blob->out_refcnt))
{
- WARNING("Path \"%ls\" exceeds MAX_PATH and will not be accessible "
- "to most Windows software", extraction_path);
- if (++args->num_long_paths == MAX_EXTRACT_LONG_PATH_WARNINGS)
- WARNING("Suppressing further warnings about long paths");
+ wim_reshdr_to_desc_and_blob(&reshdr, ctx->wim, &rdesc, blob);
+ ret = read_blob_with_sha1(blob, cbs);
+ blob_unset_is_located_in_wim_resource(blob);
+ if (ret)
+ return ret;
+ ctx->num_blobs_remaining--;
+ } else {
+ wim_reshdr_to_desc(&reshdr, ctx->wim, &rdesc);
+ ret = skip_wim_resource(&rdesc);
+ if (ret)
+ return ret;
}
}
-#endif
- return (*apply_dentry_func)(extraction_path, extraction_path_nchars,
- dentry, args);
-}
+ return 0;
+}
-/* Extracts a file, directory, or symbolic link from the WIM archive. */
+/* Creates a temporary file opened for writing. The open file descriptor is
+ * returned in @fd_ret and its name is returned in @name_ret (dynamically
+ * allocated). */
static int
-apply_dentry_normal(struct wim_dentry *dentry, void *arg)
+create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
{
+ tchar *name;
+ int open_flags;
+ int raw_fd;
+
+retry:
+ name = ttempnam(NULL, T("wimlib"));
+ if (!name) {
+ ERROR_WITH_ERRNO("Failed to create temporary filename");
+ return WIMLIB_ERR_NOMEM;
+ }
+
+ open_flags = O_WRONLY | O_CREAT | O_EXCL | O_BINARY;
#ifdef __WIN32__
- return do_apply_op(dentry, arg, win32_do_apply_dentry);
-#else
- return do_apply_op(dentry, arg, unix_do_apply_dentry);
+ open_flags |= _O_SHORT_LIVED;
#endif
-}
+ raw_fd = topen(name, open_flags, 0600);
+ if (raw_fd < 0) {
+ if (errno == EEXIST) {
+ FREE(name);
+ goto retry;
+ }
+ ERROR_WITH_ERRNO("Failed to create temporary file "
+ "\"%"TS"\"", name);
+ FREE(name);
+ return WIMLIB_ERR_OPEN;
+ }
-/* Apply timestamps to an extracted file or directory */
-static int
-apply_dentry_timestamps_normal(struct wim_dentry *dentry, void *arg)
-{
-#ifdef __WIN32__
- return do_apply_op(dentry, arg, win32_do_apply_dentry_timestamps);
-#else
- return do_apply_op(dentry, arg, unix_do_apply_dentry_timestamps);
-#endif
+ filedes_init(fd_ret, raw_fd);
+ *name_ret = name;
+ return 0;
}
-static bool
-dentry_is_dot_or_dotdot(const struct wim_dentry *dentry)
+static int
+begin_extract_blob_wrapper(struct blob_descriptor *blob, void *_ctx)
{
- const utf16lechar *file_name = dentry->file_name;
- return file_name != NULL &&
- file_name[0] == cpu_to_le16('.') &&
- (file_name[1] == cpu_to_le16('\0') ||
- (file_name[1] == cpu_to_le16('.') &&
- file_name[2] == cpu_to_le16('\0')));
+ struct apply_ctx *ctx = _ctx;
+
+ ctx->cur_blob = blob;
+ ctx->cur_blob_offset = 0;
+
+ if (unlikely(blob->out_refcnt > MAX_OPEN_FILES))
+ return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
+
+ return call_begin_blob(blob, ctx->saved_cbs);
}
-/* Extract a dentry if it hasn't already been extracted and either
- * WIMLIB_EXTRACT_FLAG_NO_STREAMS is not specified, or the dentry is a directory
- * and/or has no unnamed stream. */
static int
-maybe_apply_dentry(struct wim_dentry *dentry, void *arg)
+extract_chunk_wrapper(const void *chunk, size_t size, void *_ctx)
{
- struct apply_args *args = arg;
+ struct apply_ctx *ctx = _ctx;
+ union wimlib_progress_info *progress = &ctx->progress;
int ret;
- if (!dentry->needs_extraction)
- return 0;
+ ctx->cur_blob_offset += size;
- if (args->extract_flags & WIMLIB_EXTRACT_FLAG_NO_STREAMS &&
- !dentry_is_directory(dentry) &&
- inode_unnamed_lte_resolved(dentry->d_inode) != NULL)
- return 0;
+ if (likely(ctx->supported_features.hard_links)) {
+ progress->extract.completed_bytes +=
+ (u64)size * ctx->cur_blob->out_refcnt;
+ if (ctx->cur_blob_offset == ctx->cur_blob->size)
+ progress->extract.completed_streams += ctx->cur_blob->out_refcnt;
+ } else {
+ const struct blob_extraction_target *targets =
+ blob_extraction_targets(ctx->cur_blob);
+ for (u32 i = 0; i < ctx->cur_blob->out_refcnt; i++) {
+ const struct wim_inode *inode = targets[i].inode;
+ const struct wim_dentry *dentry;
+
+ list_for_each_entry(dentry,
+ &inode->i_extraction_aliases,
+ d_extraction_alias_node)
+ {
+ progress->extract.completed_bytes += size;
+ if (ctx->cur_blob_offset == ctx->cur_blob->size)
+ progress->extract.completed_streams++;
+ }
+ }
+ }
+ if (progress->extract.completed_bytes >= ctx->next_progress) {
- if ((args->extract_flags & WIMLIB_EXTRACT_FLAG_VERBOSE) &&
- args->progress_func) {
- ret = calculate_dentry_full_path(dentry);
+ ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
if (ret)
return ret;
- args->progress.extract.cur_path = dentry->_full_path;
- args->progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY,
- &args->progress);
- }
- ret = args->apply_dentry(dentry, args);
- if (ret == 0)
- dentry->needs_extraction = 0;
- return ret;
-}
-static void
-calculate_bytes_to_extract(struct list_head *stream_list,
- int extract_flags,
- union wimlib_progress_info *progress)
-{
- struct wim_lookup_table_entry *lte;
- u64 total_bytes = 0;
- u64 num_streams = 0;
-
- /* For each stream to be extracted... */
- list_for_each_entry(lte, stream_list, extraction_list) {
- if (extract_flags &
- (WIMLIB_EXTRACT_FLAG_SYMLINK | WIMLIB_EXTRACT_FLAG_HARDLINK))
+ if (progress->extract.completed_bytes >=
+ progress->extract.total_bytes)
{
- /* In the symlink or hard link extraction mode, each
- * stream will be extracted one time regardless of how
- * many dentries share the stream. */
- wimlib_assert(!(extract_flags & WIMLIB_EXTRACT_FLAG_NTFS));
- if (!lte->extracted_file) {
- num_streams++;
- total_bytes += wim_resource_size(lte);
- }
+ ctx->next_progress = UINT64_MAX;
} else {
- num_streams += lte->out_refcnt;
- total_bytes += lte->out_refcnt * wim_resource_size(lte);
+ /* Send new message as soon as another 1/128 of the
+ * total has been extracted. (Arbitrary number.) */
+ ctx->next_progress =
+ progress->extract.completed_bytes +
+ progress->extract.total_bytes / 128;
+
+ /* ... Unless that would be more than 5000000 bytes, in
+ * which case send the next after the next 5000000
+ * bytes. (Another arbitrary number.) */
+ if (progress->extract.completed_bytes + 5000000 <
+ ctx->next_progress)
+ ctx->next_progress =
+ progress->extract.completed_bytes + 5000000;
+
+ /* ... But always send a message as soon as we're
+ * completely done. */
+ if (progress->extract.total_bytes < ctx->next_progress)
+ ctx->next_progress = progress->extract.total_bytes;
+ }
+ }
+
+ if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
+ /* Just extracting to temporary file for now. */
+ ret = full_write(&ctx->tmpfile_fd, chunk, size);
+ if (ret) {
+ ERROR_WITH_ERRNO("Error writing data to "
+ "temporary file \"%"TS"\"",
+ ctx->tmpfile_name);
}
+ return ret;
}
- progress->extract.num_streams = num_streams;
- progress->extract.total_bytes = total_bytes;
- progress->extract.completed_bytes = 0;
+
+ return call_consume_chunk(chunk, size, ctx->saved_cbs);
}
-static void
-maybe_add_stream_for_extraction(struct wim_lookup_table_entry *lte,
- struct list_head *stream_list)
+/* Copy the blob's data from the temporary file to each of its targets.
+ *
+ * This is executed only in the very uncommon case that a blob is being
+ * extracted to more than MAX_OPEN_FILES targets! */
+static int
+extract_from_tmpfile(const tchar *tmpfile_name,
+ const struct blob_descriptor *orig_blob,
+ const struct read_blob_callbacks *cbs)
{
- if (++lte->out_refcnt == 1) {
- INIT_LIST_HEAD(<e->lte_dentry_list);
- list_add_tail(<e->extraction_list, stream_list);
+ struct blob_descriptor tmpfile_blob;
+ const struct blob_extraction_target *targets = blob_extraction_targets(orig_blob);
+ int ret;
+
+ memcpy(&tmpfile_blob, orig_blob, sizeof(struct blob_descriptor));
+ tmpfile_blob.blob_location = BLOB_IN_FILE_ON_DISK;
+ tmpfile_blob.file_on_disk = (tchar *)tmpfile_name;
+ tmpfile_blob.out_refcnt = 1;
+
+ for (u32 i = 0; i < orig_blob->out_refcnt; i++) {
+ tmpfile_blob.inline_blob_extraction_targets[0] = targets[i];
+ ret = read_blob_with_cbs(&tmpfile_blob, cbs);
+ if (ret)
+ return ret;
}
+ return 0;
}
-struct find_streams_ctx {
- struct list_head stream_list;
- int extract_flags;
-};
-
static int
-dentry_find_streams_to_extract(struct wim_dentry *dentry, void *_ctx)
+end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx)
{
- struct find_streams_ctx *ctx = _ctx;
- struct wim_inode *inode = dentry->d_inode;
- struct wim_lookup_table_entry *lte;
- bool dentry_added = false;
- struct list_head *stream_list = &ctx->stream_list;
- int extract_flags = ctx->extract_flags;
+ struct apply_ctx *ctx = _ctx;
+
+ if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
+ filedes_close(&ctx->tmpfile_fd);
+ if (!status)
+ status = extract_from_tmpfile(ctx->tmpfile_name, blob,
+ ctx->saved_cbs);
+ filedes_invalidate(&ctx->tmpfile_fd);
+ tunlink(ctx->tmpfile_name);
+ FREE(ctx->tmpfile_name);
+ return status;
+ }
- if (!dentry->needs_extraction)
- return 0;
+ return call_end_blob(blob, status, ctx->saved_cbs);
+}
- lte = inode_unnamed_lte_resolved(inode);
- if (lte) {
- if (!inode->i_visited)
- maybe_add_stream_for_extraction(lte, stream_list);
- list_add_tail(&dentry->extraction_stream_list, <e->lte_dentry_list);
- dentry_added = true;
- }
-
- /* Determine whether to include alternate data stream entries or not.
- *
- * UNIX: Include them if extracting using NTFS-3g.
- *
- * Windows: Include them undconditionally, although if the filesystem is
- * not NTFS we won't actually be able to extract them. */
-#if defined(WITH_NTFS_3G)
- if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
-#elif defined(__WIN32__)
- if (1)
-#else
- if (0)
-#endif
- {
- for (unsigned i = 0; i < inode->i_num_ads; i++) {
- if (inode->i_ads_entries[i].stream_name_nbytes != 0) {
- lte = inode->i_ads_entries[i].lte;
- if (lte) {
- if (!inode->i_visited) {
- maybe_add_stream_for_extraction(lte,
- stream_list);
- }
- if (!dentry_added) {
- list_add_tail(&dentry->extraction_stream_list,
- <e->lte_dentry_list);
- dentry_added = true;
- }
- }
- }
- }
+/*
+ * Read the list of blobs to extract and feed their data into the specified
+ * callback functions.
+ *
+ * This handles checksumming each blob.
+ *
+ * This also handles sending WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS.
+ *
+ * This also works if the WIM is being read from a pipe.
+ *
+ * This also will split up blobs that will need to be extracted to more than
+ * MAX_OPEN_FILES locations, as measured by the 'out_refcnt' of each blob.
+ * Therefore, the apply_operations implementation need not worry about running
+ * out of file descriptors, unless it might open more than one file descriptor
+ * per 'blob_extraction_target' (e.g. Win32 currently might because the
+ * destination file system might not support hard links).
+ */
+int
+extract_blob_list(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
+{
+ struct read_blob_callbacks wrapper_cbs = {
+ .begin_blob = begin_extract_blob_wrapper,
+ .consume_chunk = extract_chunk_wrapper,
+ .end_blob = end_extract_blob_wrapper,
+ .ctx = ctx,
+ };
+ ctx->saved_cbs = cbs;
+ if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
+ return read_blobs_from_pipe(ctx, &wrapper_cbs);
+ } else {
+ return read_blob_list(&ctx->blob_list,
+ offsetof(struct blob_descriptor,
+ extraction_list),
+ &wrapper_cbs, VERIFY_BLOB_HASHES);
}
- inode->i_visited = 1;
- return 0;
}
+/* Extract a WIM dentry to standard output.
+ *
+ * This obviously doesn't make sense in all cases. We return an error if the
+ * dentry does not correspond to a regular file. Otherwise we extract the
+ * unnamed data stream only. */
static int
-dentry_resolve_and_zero_lte_refcnt(struct wim_dentry *dentry, void *_lookup_table)
+extract_dentry_to_stdout(struct wim_dentry *dentry,
+ const struct blob_table *blob_table)
{
struct wim_inode *inode = dentry->d_inode;
- struct wim_lookup_table *lookup_table = _lookup_table;
- struct wim_lookup_table_entry *lte;
- int ret;
+ struct blob_descriptor *blob;
+ struct filedes _stdout;
- ret = inode_resolve_ltes(inode, lookup_table);
- if (ret)
- return ret;
- for (unsigned i = 0; i <= inode->i_num_ads; i++) {
- lte = inode_stream_lte_resolved(inode, i);
- if (lte)
- lte->out_refcnt = 0;
+ if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
+ FILE_ATTRIBUTE_DIRECTORY |
+ FILE_ATTRIBUTE_ENCRYPTED))
+ {
+ ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
+ "extracted to standard output", dentry_full_path(dentry));
+ return WIMLIB_ERR_NOT_A_REGULAR_FILE;
}
- return 0;
+
+ blob = inode_get_blob_for_unnamed_data_stream(inode, blob_table);
+ if (!blob) {
+ const u8 *hash = inode_get_hash_of_unnamed_data_stream(inode);
+ if (!is_zero_hash(hash))
+ return blob_not_found_error(inode, hash);
+ return 0;
+ }
+
+ filedes_init(&_stdout, STDOUT_FILENO);
+ return extract_blob_to_fd(blob, &_stdout);
}
static int
-find_streams_for_extraction(struct wim_dentry *root,
- struct list_head *stream_list,
- struct wim_lookup_table *lookup_table,
- int extract_flags)
+extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
+ const struct blob_table *blob_table)
{
- struct find_streams_ctx ctx;
- int ret;
-
- INIT_LIST_HEAD(&ctx.stream_list);
- ctx.extract_flags = extract_flags;
- ret = for_dentry_in_tree(root, dentry_resolve_and_zero_lte_refcnt, lookup_table);
- if (ret)
- return ret;
- for_dentry_in_tree(root, dentry_find_streams_to_extract, &ctx);
- list_transfer(&ctx.stream_list, stream_list);
+ for (size_t i = 0; i < num_dentries; i++) {
+ int ret = extract_dentry_to_stdout(dentries[i], blob_table);
+ if (ret)
+ return ret;
+ }
return 0;
}
-struct apply_operations {
- int (*apply_dentry)(struct wim_dentry *dentry, void *arg);
- int (*apply_dentry_timestamps)(struct wim_dentry *dentry, void *arg);
-};
-
-static const struct apply_operations normal_apply_operations = {
- .apply_dentry = apply_dentry_normal,
- .apply_dentry_timestamps = apply_dentry_timestamps_normal,
-};
-
-#ifdef WITH_NTFS_3G
-static const struct apply_operations ntfs_apply_operations = {
- .apply_dentry = apply_dentry_ntfs,
- .apply_dentry_timestamps = apply_dentry_timestamps_ntfs,
-};
-#endif
+/**********************************************************************/
-static int
-apply_stream_list(struct list_head *stream_list,
- struct apply_args *args,
- const struct apply_operations *ops,
- wimlib_progress_func_t progress_func)
+/*
+ * Removes duplicate dentries from the array.
+ *
+ * Returns the new number of dentries, packed at the front of the array.
+ */
+static size_t
+remove_duplicate_trees(struct wim_dentry **trees, size_t num_trees)
{
- uint64_t bytes_per_progress = args->progress.extract.total_bytes / 100;
- uint64_t next_progress = bytes_per_progress;
- struct wim_lookup_table_entry *lte;
- struct wim_dentry *dentry;
- int ret;
+ size_t i, j = 0;
+ for (i = 0; i < num_trees; i++) {
+ if (!trees[i]->tmp_flag) {
+ /* Found distinct dentry. */
+ trees[i]->tmp_flag = 1;
+ trees[j++] = trees[i];
+ }
+ }
+ for (i = 0; i < j; i++)
+ trees[i]->tmp_flag = 0;
+ return j;
+}
- /* This complicated loop is essentially looping through the dentries,
- * although dentries may be visited more than once (if a dentry contains
- * two different nonempty streams) or not at all (if a dentry contains
- * no non-empty streams).
- *
- * The outer loop is over the distinct streams to be extracted so that
- * sequential reading of the WIM can be implemented. */
-
- /* For each distinct stream to be extracted */
- list_for_each_entry(lte, stream_list, extraction_list) {
- /* For each dentry to be extracted that is a name for an inode
- * containing the stream */
- list_for_each_entry(dentry, <e->lte_dentry_list, extraction_stream_list) {
- /* Extract the dentry if it was not already
- * extracted */
- ret = maybe_apply_dentry(dentry, args);
- if (ret)
- return ret;
- if (progress_func &&
- args->progress.extract.completed_bytes >= next_progress)
- {
- progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS,
- &args->progress);
- if (args->progress.extract.completed_bytes >=
- args->progress.extract.total_bytes)
- {
- next_progress = ~0ULL;
- } else {
- next_progress =
- min (args->progress.extract.completed_bytes +
- bytes_per_progress,
- args->progress.extract.total_bytes);
- }
- }
+/*
+ * Remove dentries that are descendants of other dentries in the array.
+ *
+ * Returns the new number of dentries, packed at the front of the array.
+ */
+static size_t
+remove_contained_trees(struct wim_dentry **trees, size_t num_trees)
+{
+ size_t i, j = 0;
+ for (i = 0; i < num_trees; i++)
+ trees[i]->tmp_flag = 1;
+ for (i = 0; i < num_trees; i++) {
+ struct wim_dentry *d = trees[i];
+ while (!dentry_is_root(d)) {
+ d = d->d_parent;
+ if (d->tmp_flag)
+ goto tree_contained;
}
+ trees[j++] = trees[i];
+ continue;
+
+ tree_contained:
+ trees[i]->tmp_flag = 0;
}
- return 0;
+
+ for (i = 0; i < j; i++)
+ trees[i]->tmp_flag = 0;
+ return j;
}
static int
-sort_stream_list_by_wim_position(struct list_head *stream_list)
+dentry_append_to_list(struct wim_dentry *dentry, void *_dentry_list)
{
- struct list_head *cur;
- size_t num_streams;
- struct wim_lookup_table_entry **array;
- size_t i;
- size_t array_size;
-
- num_streams = 0;
- list_for_each(cur, stream_list)
- num_streams++;
- array_size = num_streams * sizeof(array[0]);
- array = MALLOC(array_size);
- if (!array) {
- ERROR("Failed to allocate %zu bytes to sort stream entries",
- array_size);
- return WIMLIB_ERR_NOMEM;
- }
- cur = stream_list->next;
- for (i = 0; i < num_streams; i++) {
- array[i] = container_of(cur, struct wim_lookup_table_entry, extraction_list);
- cur = cur->next;
- }
+ struct list_head *dentry_list = _dentry_list;
+ list_add_tail(&dentry->d_extraction_list_node, dentry_list);
+ return 0;
+}
- qsort(array, num_streams, sizeof(array[0]), cmp_streams_by_wim_position);
+static void
+dentry_reset_extraction_list_node(struct wim_dentry *dentry)
+{
+ dentry->d_extraction_list_node = (struct list_head){NULL, NULL};
+}
- INIT_LIST_HEAD(stream_list);
- for (i = 0; i < num_streams; i++)
- list_add_tail(&array[i]->extraction_list, stream_list);
- FREE(array);
+static int
+dentry_delete_from_list(struct wim_dentry *dentry, void *_ignore)
+{
+ list_del(&dentry->d_extraction_list_node);
+ dentry_reset_extraction_list_node(dentry);
return 0;
}
/*
- * Extract a dentry to standard output.
+ * Build the preliminary list of dentries to be extracted.
*
- * This obviously doesn't make sense in all cases. We return an error if the
- * dentry does not correspond to a regular file. Otherwise we extract the
- * unnamed data stream only.
+ * The list maintains the invariant that if d1 and d2 are in the list and d1 is
+ * an ancestor of d2, then d1 appears before d2 in the list.
*/
-static int
-extract_dentry_to_stdout(struct wim_dentry *dentry)
+static void
+build_dentry_list(struct list_head *dentry_list, struct wim_dentry **trees,
+ size_t num_trees, bool add_ancestors)
{
- int ret = 0;
- if (!dentry_is_regular_file(dentry)) {
- ERROR("\"%"TS"\" is not a regular file and therefore cannot be "
- "extracted to standard output", dentry->_full_path);
- ret = WIMLIB_ERR_NOT_A_REGULAR_FILE;
- } else {
- struct wim_lookup_table_entry *lte;
-
- lte = inode_unnamed_lte_resolved(dentry->d_inode);
- if (lte) {
- ret = extract_wim_resource_to_fd(lte, STDOUT_FILENO,
- wim_resource_size(lte));
+ INIT_LIST_HEAD(dentry_list);
+
+ /* Add the trees recursively. */
+ for (size_t i = 0; i < num_trees; i++)
+ for_dentry_in_tree(trees[i], dentry_append_to_list, dentry_list);
+
+ /* If requested, add ancestors of the trees. */
+ if (add_ancestors) {
+ for (size_t i = 0; i < num_trees; i++) {
+ struct wim_dentry *dentry = trees[i];
+ struct wim_dentry *ancestor;
+ struct list_head *place_after;
+
+ if (dentry_is_root(dentry))
+ continue;
+
+ place_after = dentry_list;
+ ancestor = dentry;
+ do {
+ ancestor = ancestor->d_parent;
+ if (will_extract_dentry(ancestor)) {
+ place_after = &ancestor->d_extraction_list_node;
+ break;
+ }
+ } while (!dentry_is_root(ancestor));
+
+ ancestor = dentry;
+ do {
+ ancestor = ancestor->d_parent;
+ if (will_extract_dentry(ancestor))
+ break;
+ list_add(&ancestor->d_extraction_list_node, place_after);
+ } while (!dentry_is_root(ancestor));
}
}
- return ret;
+}
+
+static void
+destroy_dentry_list(struct list_head *dentry_list)
+{
+ struct wim_dentry *dentry, *tmp;
+ struct wim_inode *inode;
+
+ list_for_each_entry_safe(dentry, tmp, dentry_list, d_extraction_list_node) {
+ inode = dentry->d_inode;
+ dentry_reset_extraction_list_node(dentry);
+ inode->i_visited = 0;
+ inode->i_can_externally_back = 0;
+ if ((void *)dentry->d_extraction_name != (void *)dentry->file_name)
+ FREE(dentry->d_extraction_name);
+ dentry->d_extraction_name = NULL;
+ dentry->d_extraction_name_nchars = 0;
+ }
+}
+
+static void
+destroy_blob_list(struct list_head *blob_list)
+{
+ struct blob_descriptor *blob;
+
+ list_for_each_entry(blob, blob_list, extraction_list)
+ if (blob->out_refcnt > ARRAY_LEN(blob->inline_blob_extraction_targets))
+ FREE(blob->blob_extraction_targets);
}
#ifdef __WIN32__
return true;
}
-/*
- * dentry_calculate_extraction_path-
- *
- * Calculate the actual filename component at which a WIM dentry will be
- * extracted, handling invalid filenames "properly".
- *
- * dentry->extraction_name usually will be set the same as dentry->file_name (on
- * UNIX, converted into the platform's multibyte encoding). However, if the
- * file name contains characters that are not valid on the current platform or
- * has some other format that is not valid, leave dentry->extraction_name as
- * NULL and clear dentry->needs_extraction to indicate that this dentry should
- * not be extracted, unless the appropriate flag
- * WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES is set in the extract flags, in
- * which case a substitute filename will be created and set instead.
- *
- * Conflicts with case-insensitive names on Windows are handled similarly; see
- * below.
- */
static int
-dentry_calculate_extraction_path(struct wim_dentry *dentry, void *_args)
+dentry_calculate_extraction_name(struct wim_dentry *dentry,
+ struct apply_ctx *ctx)
{
- struct apply_args *args = _args;
int ret;
- dentry->needs_extraction = 1;
-
- if (dentry == args->extract_root)
+ if (dentry_is_root(dentry))
return 0;
- if (dentry_is_dot_or_dotdot(dentry)) {
- /* WIM files shouldn't contain . or .. entries. But if they are
- * there, don't attempt to extract them. */
- WARNING("Skipping extraction of unexpected . or .. file \"%"TS"\"",
- dentry_full_path(dentry));
- goto skip_dentry;
+#ifdef WITH_NTFS_3G
+ if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
+ dentry->d_extraction_name = dentry->file_name;
+ dentry->d_extraction_name_nchars = dentry->file_name_nbytes /
+ sizeof(utf16lechar);
+ return 0;
}
+#endif
-#ifdef __WIN32__
- struct wim_dentry *other;
- list_for_each_entry(other, &dentry->case_insensitive_conflict_list,
- case_insensitive_conflict_list)
- {
- if (other->needs_extraction) {
- if (args->extract_flags & WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS)
- {
- WARNING("\"%"TS"\" has the same case-insensitive "
- "name as \"%"TS"\"; extracting dummy name instead",
- dentry_full_path(dentry),
- dentry_full_path(other));
- goto out_replace;
- } else {
- WARNING("Not extracting \"%"TS"\": has same case-insensitive "
- "name as \"%"TS"\"",
- dentry_full_path(dentry),
- dentry_full_path(other));
- goto skip_dentry;
+ if (!ctx->supported_features.case_sensitive_filenames) {
+ struct wim_dentry *other;
+ list_for_each_entry(other, &dentry->d_ci_conflict_list,
+ d_ci_conflict_list)
+ {
+ if (will_extract_dentry(other)) {
+ if (ctx->extract_flags &
+ WIMLIB_EXTRACT_FLAG_ALL_CASE_CONFLICTS) {
+ WARNING("\"%"TS"\" has the same "
+ "case-insensitive name as "
+ "\"%"TS"\"; extracting "
+ "dummy name instead",
+ dentry_full_path(dentry),
+ dentry_full_path(other));
+ goto out_replace;
+ } else {
+ WARNING("Not extracting \"%"TS"\": "
+ "has same case-insensitive "
+ "name as \"%"TS"\"",
+ dentry_full_path(dentry),
+ dentry_full_path(other));
+ goto skip_dentry;
+ }
}
}
}
-#endif
if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
-#ifdef __WIN32__
- dentry->extraction_name = dentry->file_name;
- dentry->extraction_name_nchars = dentry->file_name_nbytes / 2;
- return 0;
-#else
- return utf16le_to_tstr(dentry->file_name,
+ ret = utf16le_get_tstr(dentry->file_name,
dentry->file_name_nbytes,
- &dentry->extraction_name,
- &dentry->extraction_name_nchars);
-#endif
+ (const tchar **)&dentry->d_extraction_name,
+ &dentry->d_extraction_name_nchars);
+ dentry->d_extraction_name_nchars /= sizeof(tchar);
+ return ret;
} else {
- if (args->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
+ if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
{
WARNING("\"%"TS"\" has an invalid filename "
"that is not supported on this platform; "
memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
- tchar *tchar_name;
+ const tchar *tchar_name;
size_t tchar_nchars;
- #ifdef __WIN32__
- tchar_name = utf16_name_copy;
- tchar_nchars = dentry->file_name_nbytes / 2;
- #else
- ret = utf16le_to_tstr(utf16_name_copy,
- dentry->file_name_nbytes,
- &tchar_name, &tchar_nchars);
+
+ ret = utf16le_get_tstr(utf16_name_copy,
+ dentry->file_name_nbytes,
+ &tchar_name, &tchar_nchars);
if (ret)
return ret;
- #endif
+
+ tchar_nchars /= sizeof(tchar);
+
size_t fixed_name_num_chars = tchar_nchars;
tchar fixed_name[tchar_nchars + 50];
tmemcpy(fixed_name, tchar_name, tchar_nchars);
fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
T(" (invalid filename #%lu)"),
- ++args->invalid_sequence);
- #ifndef __WIN32__
- FREE(tchar_name);
- #endif
- dentry->extraction_name = memdup(fixed_name, 2 * fixed_name_num_chars + 2);
- if (!dentry->extraction_name)
+ ++ctx->invalid_sequence);
+
+ utf16le_put_tstr(tchar_name);
+
+ dentry->d_extraction_name = TSTRDUP(fixed_name);
+ if (!dentry->d_extraction_name)
return WIMLIB_ERR_NOMEM;
- dentry->extraction_name_nchars = fixed_name_num_chars;
+ dentry->d_extraction_name_nchars = fixed_name_num_chars;
}
return 0;
+
skip_dentry:
- dentry->needs_extraction = 0;
- dentry->not_extracted = 1;
+ for_dentry_in_tree(dentry, dentry_delete_from_list, NULL);
return 0;
}
+/*
+ * Calculate the actual filename component at which each WIM dentry will be
+ * extracted, with special handling for dentries that are unsupported by the
+ * extraction backend or have invalid names.
+ *
+ * ctx->supported_features must be filled in.
+ *
+ * Possible error codes: WIMLIB_ERR_NOMEM, WIMLIB_ERR_INVALID_UTF16_STRING
+ */
static int
-dentry_reset_needs_extraction(struct wim_dentry *dentry, void *_ignore)
+dentry_list_calculate_extraction_names(struct list_head *dentry_list,
+ struct apply_ctx *ctx)
{
- dentry->needs_extraction = 0;
- dentry->not_extracted = 0;
- dentry->is_win32_name = 0;
- dentry->d_inode->i_visited = 0;
- dentry->d_inode->i_dos_name_extracted = 0;
- FREE(dentry->d_inode->i_extracted_file);
- dentry->d_inode->i_extracted_file = NULL;
- if ((void*)dentry->extraction_name != (void*)dentry->file_name)
- FREE(dentry->extraction_name);
- dentry->extraction_name = NULL;
+ struct list_head *prev, *cur;
+
+ /* Can't use list_for_each_entry() because a call to
+ * dentry_calculate_extraction_name() may delete the current dentry and
+ * its children from the list. */
+
+ prev = dentry_list;
+ for (;;) {
+ struct wim_dentry *dentry;
+ int ret;
+
+ cur = prev->next;
+ if (cur == dentry_list)
+ break;
+
+ dentry = list_entry(cur, struct wim_dentry, d_extraction_list_node);
+
+ ret = dentry_calculate_extraction_name(dentry, ctx);
+ if (ret)
+ return ret;
+
+ if (prev->next == cur)
+ prev = cur;
+ else
+ ; /* Current dentry and its children (which follow in
+ the list) were deleted. prev stays the same. */
+ }
return 0;
}
-#define WINDOWS_NT_MAX_PATH 32768
+static int
+dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
+ struct blob_table *blob_table)
+{
+ struct wim_inode *inode = dentry->d_inode;
+ struct blob_descriptor *blob;
+ int ret;
+ bool force = false;
+
+ /* Special case: when extracting from a pipe, the WIM blob table is
+ * initially empty, so "resolving" an inode's streams is initially not
+ * possible. However, we still need to keep track of which blobs,
+ * identified by SHA-1 message digests, need to be extracted, so we
+ * "resolve" the inode's streams anyway by allocating a 'struct
+ * blob_descriptor' for each one. */
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE)
+ force = true;
+ ret = inode_resolve_streams(inode, blob_table, force);
+ if (ret)
+ return ret;
+ for (unsigned i = 0; i < inode->i_num_streams; i++) {
+ blob = stream_blob_resolved(&inode->i_streams[i]);
+ if (blob)
+ blob->out_refcnt = 0;
+ }
+ return 0;
+}
/*
- * extract_tree - Extract a file or directory tree from the currently selected
- * WIM image.
- *
- * @wim: WIMStruct for the WIM file, with the desired image selected
- * (as wim->current_image).
- * @wim_source_path:
- * "Canonical" (i.e. no leading or trailing slashes, path
- * separators forwald slashes) path inside the WIM image to
- * extract. An empty string means the full image.
- * @target:
- * Filesystem path to extract the file or directory tree to.
+ * For each dentry to be extracted, resolve all streams in the corresponding
+ * inode and set 'out_refcnt' in all referenced blob_descriptors to 0.
*
- * @extract_flags:
- * WIMLIB_EXTRACT_FLAG_*. Also, the private flag
- * WIMLIB_EXTRACT_FLAG_MULTI_IMAGE will be set if this is being
- * called through wimlib_extract_image() with WIMLIB_ALL_IMAGES as
- * the image.
- *
- * @progress_func:
- * If non-NULL, progress function for the extraction. The messages
- * we may in this function are:
+ * Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
+ */
+static int
+dentry_list_resolve_streams(struct list_head *dentry_list,
+ struct apply_ctx *ctx)
+{
+ struct wim_dentry *dentry;
+ int ret;
+
+ list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
+ ret = dentry_resolve_streams(dentry,
+ ctx->extract_flags,
+ ctx->wim->blob_table);
+ if (ret)
+ return ret;
+ }
+ return 0;
+}
+
+static int
+ref_stream(struct wim_inode_stream *strm, struct wim_dentry *dentry,
+ struct apply_ctx *ctx)
+{
+ struct wim_inode *inode = dentry->d_inode;
+ struct blob_descriptor *blob = stream_blob_resolved(strm);
+ struct blob_extraction_target *targets;
+
+ if (!blob)
+ return 0;
+
+ /* Tally the size only for each actual extraction of the stream (not
+ * additional hard links to the inode). */
+ if (inode->i_visited && ctx->supported_features.hard_links)
+ return 0;
+
+ ctx->progress.extract.total_bytes += blob->size;
+ ctx->progress.extract.total_streams++;
+
+ if (inode->i_visited)
+ return 0;
+
+ /* Add each blob to 'ctx->blob_list' only one time, regardless of how
+ * many extraction targets it will have. */
+ if (blob->out_refcnt == 0) {
+ list_add_tail(&blob->extraction_list, &ctx->blob_list);
+ ctx->num_blobs_remaining++;
+ }
+
+ /* Set this stream as an extraction target of 'blob'. */
+
+ if (blob->out_refcnt < ARRAY_LEN(blob->inline_blob_extraction_targets)) {
+ targets = blob->inline_blob_extraction_targets;
+ } else {
+ struct blob_extraction_target *prev_targets;
+ size_t alloc_blob_extraction_targets;
+
+ if (blob->out_refcnt == ARRAY_LEN(blob->inline_blob_extraction_targets)) {
+ prev_targets = NULL;
+ alloc_blob_extraction_targets = ARRAY_LEN(blob->inline_blob_extraction_targets);
+ } else {
+ prev_targets = blob->blob_extraction_targets;
+ alloc_blob_extraction_targets = blob->alloc_blob_extraction_targets;
+ }
+
+ if (blob->out_refcnt == alloc_blob_extraction_targets) {
+ alloc_blob_extraction_targets *= 2;
+ targets = REALLOC(prev_targets,
+ alloc_blob_extraction_targets *
+ sizeof(targets[0]));
+ if (!targets)
+ return WIMLIB_ERR_NOMEM;
+ if (!prev_targets) {
+ memcpy(targets,
+ blob->inline_blob_extraction_targets,
+ sizeof(blob->inline_blob_extraction_targets));
+ }
+ blob->blob_extraction_targets = targets;
+ blob->alloc_blob_extraction_targets = alloc_blob_extraction_targets;
+ }
+ targets = blob->blob_extraction_targets;
+ }
+ targets[blob->out_refcnt].inode = inode;
+ targets[blob->out_refcnt].stream = strm;
+ blob->out_refcnt++;
+ return 0;
+}
+
+static int
+ref_stream_if_needed(struct wim_dentry *dentry, struct wim_inode *inode,
+ struct wim_inode_stream *strm, struct apply_ctx *ctx)
+{
+ bool need_stream = false;
+ switch (strm->stream_type) {
+ case STREAM_TYPE_DATA:
+ if (stream_is_named(strm)) {
+ /* Named data stream */
+ if (ctx->supported_features.named_data_streams)
+ need_stream = true;
+ } else if (!(inode->i_attributes & (FILE_ATTRIBUTE_DIRECTORY |
+ FILE_ATTRIBUTE_ENCRYPTED))
+ && !(inode_is_symlink(inode)
+ && !ctx->supported_features.reparse_points
+ && ctx->supported_features.symlink_reparse_points))
+ {
+ /*
+ * Unnamed data stream. Skip if any of the following is true:
+ *
+ * - file is a directory
+ * - file is encrypted
+ * - backend needs to create the file as UNIX symlink
+ * - backend will extract the stream as externally backed
+ */
+ if (ctx->apply_ops->will_externally_back) {
+ int ret = (*ctx->apply_ops->will_externally_back)(dentry, ctx);
+ if (ret > 0) /* Error? */
+ return ret;
+ if (ret < 0) /* Won't externally back? */
+ need_stream = true;
+ } else {
+ need_stream = true;
+ }
+ }
+ break;
+ case STREAM_TYPE_REPARSE_POINT:
+ wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT);
+ if (ctx->supported_features.reparse_points ||
+ (inode_is_symlink(inode) &&
+ ctx->supported_features.symlink_reparse_points))
+ need_stream = true;
+ break;
+ case STREAM_TYPE_EFSRPC_RAW_DATA:
+ wimlib_assert(inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED);
+ if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY) {
+ if (ctx->supported_features.encrypted_directories)
+ need_stream = true;
+ } else {
+ if (ctx->supported_features.encrypted_files)
+ need_stream = true;
+ }
+ break;
+ }
+ if (need_stream)
+ return ref_stream(strm, dentry, ctx);
+ return 0;
+}
+
+static int
+dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
+{
+ struct wim_inode *inode = dentry->d_inode;
+ for (unsigned i = 0; i < inode->i_num_streams; i++) {
+ int ret = ref_stream_if_needed(dentry, inode,
+ &inode->i_streams[i], ctx);
+ if (ret)
+ return ret;
+ }
+ inode->i_visited = 1;
+ return 0;
+}
+
+/*
+ * Given a list of dentries to be extracted, build the list of blobs that need
+ * to be extracted, and for each blob determine the streams to which that blob
+ * will be extracted.
*
- * WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN or
- * WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN;
- * WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN;
- * WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END;
- * WIMLIB_PROGRESS_MSG_EXTRACT_DENTRY;
- * WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS;
- * WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS;
- * WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END or
- * WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END.
+ * This also initializes the extract progress info with byte and blob
+ * information.
*
- * Returns 0 on success; nonzero on failure.
+ * ctx->supported_features must be filled in.
*/
static int
-extract_tree(WIMStruct *wim, const tchar *wim_source_path, const tchar *target,
- int extract_flags, wimlib_progress_func_t progress_func)
+dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
{
+ struct wim_dentry *dentry;
int ret;
- struct list_head stream_list;
- struct apply_args args;
- const struct apply_operations *ops;
- struct wim_dentry *root;
- memset(&args, 0, sizeof(args));
+ list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
+ ret = dentry_ref_streams(dentry, ctx);
+ if (ret)
+ return ret;
+ }
+ list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
+ dentry->d_inode->i_visited = 0;
+ return 0;
+}
+static void
+dentry_list_build_inode_alias_lists(struct list_head *dentry_list)
+{
+ struct wim_dentry *dentry;
+ struct wim_inode *inode;
- args.w = wim;
- args.target = target;
- args.target_nchars = tstrlen(target);
- args.extract_flags = extract_flags;
- args.progress_func = progress_func;
+ list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
+ inode = dentry->d_inode;
+ if (!inode->i_visited)
+ INIT_LIST_HEAD(&inode->i_extraction_aliases);
+ list_add_tail(&dentry->d_extraction_alias_node,
+ &inode->i_extraction_aliases);
+ inode->i_visited = 1;
+ }
+ list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
+ dentry->d_inode->i_visited = 0;
+}
-#ifdef __WIN32__
- /* Work around defective behavior in Windows where paths longer than 260
- * characters are not supported by default; instead they need to be
- * turned into absolute paths and prefixed with "\\?\". */
- args.target_lowlevel_path = MALLOC(WINDOWS_NT_MAX_PATH * sizeof(wchar_t));
- if (!args.target_lowlevel_path)
+static void
+inode_tally_features(const struct wim_inode *inode,
+ struct wim_features *features)
+{
+ if (inode->i_attributes & FILE_ATTRIBUTE_ARCHIVE)
+ features->archive_files++;
+ if (inode->i_attributes & FILE_ATTRIBUTE_HIDDEN)
+ features->hidden_files++;
+ if (inode->i_attributes & FILE_ATTRIBUTE_SYSTEM)
+ features->system_files++;
+ if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
+ features->compressed_files++;
+ if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
+ if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
+ features->encrypted_directories++;
+ else
+ features->encrypted_files++;
+ }
+ if (inode->i_attributes & FILE_ATTRIBUTE_NOT_CONTENT_INDEXED)
+ features->not_context_indexed_files++;
+ if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
+ features->sparse_files++;
+ if (inode_has_named_data_stream(inode))
+ features->named_data_streams++;
+ if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
+ features->reparse_points++;
+ if (inode_is_symlink(inode))
+ features->symlink_reparse_points++;
+ else
+ features->other_reparse_points++;
+ }
+ if (inode_has_security_descriptor(inode))
+ features->security_descriptors++;
+ if (inode_has_unix_data(inode))
+ features->unix_data++;
+}
+
+/* Tally features necessary to extract a dentry and the corresponding inode. */
+static void
+dentry_tally_features(struct wim_dentry *dentry, struct wim_features *features)
+{
+ struct wim_inode *inode = dentry->d_inode;
+
+ if (dentry_has_short_name(dentry))
+ features->short_names++;
+
+ if (inode->i_visited) {
+ features->hard_links++;
+ } else {
+ inode_tally_features(inode, features);
+ inode->i_visited = 1;
+ }
+}
+
+/* Tally the features necessary to extract the specified dentries. */
+static void
+dentry_list_get_features(struct list_head *dentry_list,
+ struct wim_features *features)
+{
+ struct wim_dentry *dentry;
+
+ list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
+ dentry_tally_features(dentry, features);
+
+ list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
+ dentry->d_inode->i_visited = 0;
+}
+
+static int
+do_feature_check(const struct wim_features *required_features,
+ const struct wim_features *supported_features,
+ int extract_flags)
+{
+ /* Encrypted files. */
+ if (required_features->encrypted_files &&
+ !supported_features->encrypted_files)
+ WARNING("Ignoring EFS-encrypted data of %lu files",
+ required_features->encrypted_files);
+
+ /* Named data streams. */
+ if (required_features->named_data_streams &&
+ !supported_features->named_data_streams)
+ WARNING("Ignoring named data streams of %lu files",
+ required_features->named_data_streams);
+
+ /* File attributes. */
+ if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES)) {
+ /* Note: Don't bother the user about FILE_ATTRIBUTE_ARCHIVE.
+ * We're an archive program, so theoretically we can do what we
+ * want with it. */
+
+ if (required_features->hidden_files &&
+ !supported_features->hidden_files)
+ WARNING("Ignoring FILE_ATTRIBUTE_HIDDEN of %lu files",
+ required_features->hidden_files);
+
+ if (required_features->system_files &&
+ !supported_features->system_files)
+ WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
+ required_features->system_files);
+
+ if (required_features->compressed_files &&
+ !supported_features->compressed_files)
+ WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
+ required_features->compressed_files);
+
+ if (required_features->not_context_indexed_files &&
+ !supported_features->not_context_indexed_files)
+ WARNING("Ignoring FILE_ATTRIBUTE_NOT_CONTENT_INDEXED of %lu files",
+ required_features->not_context_indexed_files);
+
+ if (required_features->sparse_files &&
+ !supported_features->sparse_files)
+ WARNING("Ignoring FILE_ATTRIBUTE_SPARSE_FILE of %lu files",
+ required_features->sparse_files);
+
+ if (required_features->encrypted_directories &&
+ !supported_features->encrypted_directories)
+ WARNING("Ignoring FILE_ATTRIBUTE_ENCRYPTED of %lu directories",
+ required_features->encrypted_directories);
+ }
+
+ /* Hard links. */
+ if (required_features->hard_links && !supported_features->hard_links)
+ WARNING("Extracting %lu hard links as independent files",
+ required_features->hard_links);
+
+ /* Symbolic links and reparse points. */
+ if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SYMLINKS) &&
+ required_features->symlink_reparse_points &&
+ !supported_features->symlink_reparse_points &&
+ !supported_features->reparse_points)
{
- ret = WIMLIB_ERR_NOMEM;
- goto out;
+ ERROR("Extraction backend does not support symbolic links!");
+ return WIMLIB_ERR_UNSUPPORTED;
+ }
+ if (required_features->reparse_points &&
+ !supported_features->reparse_points)
+ {
+ if (supported_features->symlink_reparse_points) {
+ if (required_features->other_reparse_points) {
+ WARNING("Ignoring reparse data of %lu non-symlink/junction files",
+ required_features->other_reparse_points);
+ }
+ } else {
+ WARNING("Ignoring reparse data of %lu files",
+ required_features->reparse_points);
+ }
}
- args.target_lowlevel_path_nchars =
- GetFullPathName(args.target, WINDOWS_NT_MAX_PATH - 4,
- &args.target_lowlevel_path[4], NULL);
- if (args.target_lowlevel_path_nchars == 0 ||
- args.target_lowlevel_path_nchars >= WINDOWS_NT_MAX_PATH - 4)
+ /* Security descriptors. */
+ if (((extract_flags & (WIMLIB_EXTRACT_FLAG_STRICT_ACLS |
+ WIMLIB_EXTRACT_FLAG_UNIX_DATA))
+ == WIMLIB_EXTRACT_FLAG_STRICT_ACLS) &&
+ required_features->security_descriptors &&
+ !supported_features->security_descriptors)
{
- WARNING("Can't get full path name for \"%ls\"", args.target);
- FREE(args.target_lowlevel_path);
- args.target_lowlevel_path = NULL;
- } else {
- wmemcpy(args.target_lowlevel_path, L"\\\\?\\", 4);
- args.target_lowlevel_path_nchars += 4;
+ ERROR("Extraction backend does not support security descriptors!");
+ return WIMLIB_ERR_UNSUPPORTED;
+ }
+ if (!(extract_flags & WIMLIB_EXTRACT_FLAG_NO_ACLS) &&
+ required_features->security_descriptors &&
+ !supported_features->security_descriptors)
+ WARNING("Ignoring Windows NT security descriptors of %lu files",
+ required_features->security_descriptors);
+
+ /* UNIX data. */
+ if ((extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) &&
+ required_features->unix_data && !supported_features->unix_data)
+ {
+ ERROR("Extraction backend does not support UNIX data!");
+ return WIMLIB_ERR_UNSUPPORTED;
}
-#endif
- if (progress_func) {
- args.progress.extract.wimfile_name = wim->filename;
- args.progress.extract.image = wim->current_image;
- args.progress.extract.extract_flags = (extract_flags &
- WIMLIB_EXTRACT_MASK_PUBLIC);
- args.progress.extract.image_name = wimlib_get_image_name(wim,
- wim->current_image);
- args.progress.extract.extract_root_wim_source_path = wim_source_path;
- args.progress.extract.target = target;
+ if (required_features->unix_data &&
+ !(extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA))
+ {
+ WARNING("Ignoring UNIX metadata of %lu files",
+ required_features->unix_data);
}
-#ifdef WITH_NTFS_3G
- if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
- args.vol = ntfs_mount(target, 0);
- if (!args.vol) {
- ERROR_WITH_ERRNO("Failed to mount NTFS volume `%"TS"'",
- target);
- ret = WIMLIB_ERR_NTFS_3G;
- goto out_free_target_lowlevel_path;
+ /* DOS Names. */
+ if (required_features->short_names &&
+ !supported_features->short_names)
+ {
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_SHORT_NAMES) {
+ ERROR("Extraction backend does not support DOS names!");
+ return WIMLIB_ERR_UNSUPPORTED;
}
- ops = &ntfs_apply_operations;
- } else
-#endif
- ops = &normal_apply_operations;
+ WARNING("Ignoring DOS names of %lu files",
+ required_features->short_names);
+ }
- root = get_dentry(wim, wim_source_path);
- if (!root) {
- ERROR("Path \"%"TS"\" does not exist in WIM image %d",
- wim_source_path, wim->current_image);
- ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
- goto out_ntfs_umount;
+ /* Timestamps. */
+ if ((extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_TIMESTAMPS) &&
+ !supported_features->timestamps)
+ {
+ ERROR("Extraction backend does not support timestamps!");
+ return WIMLIB_ERR_UNSUPPORTED;
}
- args.extract_root = root;
- /* Calculate the actual filename component of each extracted dentry, and
- * in the process set the dentry->needs_extraction flag on dentries that
- * will be extracted. */
- ret = for_dentry_in_tree(root, dentry_calculate_extraction_path, &args);
- if (ret)
- goto out_dentry_reset_needs_extraction;
+ return 0;
+}
- /* Build a list of the streams that need to be extracted */
- ret = find_streams_for_extraction(root,
- &stream_list,
- wim->lookup_table, extract_flags);
- if (ret)
- goto out_dentry_reset_needs_extraction;
+static const struct apply_operations *
+select_apply_operations(int extract_flags)
+{
+#ifdef WITH_NTFS_3G
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS)
+ return &ntfs_3g_apply_ops;
+#endif
+#ifdef __WIN32__
+ return &win32_apply_ops;
+#else
+ return &unix_apply_ops;
+#endif
+}
- /* Calculate the number of bytes of data that will be extracted */
- calculate_bytes_to_extract(&stream_list, extract_flags,
- &args.progress);
+static int
+extract_trees(WIMStruct *wim, struct wim_dentry **trees, size_t num_trees,
+ const tchar *target, int extract_flags)
+{
+ const struct apply_operations *ops;
+ struct apply_ctx *ctx;
+ int ret;
+ LIST_HEAD(dentry_list);
if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
- ret = extract_dentry_to_stdout(root);
- goto out_dentry_reset_needs_extraction;
+ ret = extract_dentries_to_stdout(trees, num_trees,
+ wim->blob_table);
+ goto out;
}
- if (progress_func) {
- progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN :
- WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN,
- &args.progress);
+ num_trees = remove_duplicate_trees(trees, num_trees);
+ num_trees = remove_contained_trees(trees, num_trees);
+
+ ops = select_apply_operations(extract_flags);
+
+ if (num_trees > 1 && ops->single_tree_only) {
+ ERROR("Extracting multiple directory trees "
+ "at once is not supported in %s extraction mode!",
+ ops->name);
+ ret = WIMLIB_ERR_UNSUPPORTED;
+ goto out;
}
- /* If a sequential extraction was specified, sort the streams to be
- * extracted by their position in the WIM file, so that the WIM file can
- * be read sequentially. */
- if (extract_flags & WIMLIB_EXTRACT_FLAG_SEQUENTIAL) {
- ret = sort_stream_list_by_wim_position(&stream_list);
- if (ret != 0) {
- WARNING("Falling back to non-sequential extraction");
- extract_flags &= ~WIMLIB_EXTRACT_FLAG_SEQUENTIAL;
- }
+ ctx = CALLOC(1, ops->context_size);
+ if (!ctx) {
+ ret = WIMLIB_ERR_NOMEM;
+ goto out;
}
- if (progress_func) {
- progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_BEGIN,
- &args.progress);
+ ctx->wim = wim;
+ ctx->target = target;
+ ctx->target_nchars = tstrlen(target);
+ ctx->extract_flags = extract_flags;
+ if (ctx->wim->progfunc) {
+ ctx->progfunc = ctx->wim->progfunc;
+ ctx->progctx = ctx->wim->progctx;
+ ctx->progress.extract.image = wim->current_image;
+ ctx->progress.extract.extract_flags = (extract_flags &
+ WIMLIB_EXTRACT_MASK_PUBLIC);
+ ctx->progress.extract.wimfile_name = wim->filename;
+ ctx->progress.extract.image_name = wimlib_get_image_name(wim,
+ wim->current_image);
+ ctx->progress.extract.target = target;
}
+ INIT_LIST_HEAD(&ctx->blob_list);
+ filedes_invalidate(&ctx->tmpfile_fd);
+ ctx->apply_ops = ops;
- /* Make the directory structure and extract empty files */
- args.extract_flags |= WIMLIB_EXTRACT_FLAG_NO_STREAMS;
- args.apply_dentry = ops->apply_dentry;
- ret = for_dentry_in_tree(root, maybe_apply_dentry, &args);
- args.extract_flags &= ~WIMLIB_EXTRACT_FLAG_NO_STREAMS;
+ ret = (*ops->get_supported_features)(target, &ctx->supported_features);
if (ret)
- goto out_dentry_reset_needs_extraction;
+ goto out_cleanup;
- if (progress_func) {
- progress_func(WIMLIB_PROGRESS_MSG_EXTRACT_DIR_STRUCTURE_END,
- &args.progress);
- }
+ build_dentry_list(&dentry_list, trees, num_trees,
+ !(extract_flags &
+ WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE));
- if (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX) {
- args.target_realpath = realpath(target, NULL);
- if (!args.target_realpath) {
- ret = WIMLIB_ERR_NOMEM;
- goto out_dentry_reset_needs_extraction;
- }
- args.target_realpath_len = tstrlen(args.target_realpath);
- }
+ dentry_list_get_features(&dentry_list, &ctx->required_features);
+
+ ret = do_feature_check(&ctx->required_features, &ctx->supported_features,
+ ctx->extract_flags);
+ if (ret)
+ goto out_cleanup;
- /* Extract non-empty files */
- ret = apply_stream_list(&stream_list, &args, ops, progress_func);
+ ret = dentry_list_calculate_extraction_names(&dentry_list, ctx);
if (ret)
- goto out_free_target_realpath;
+ goto out_cleanup;
- if (progress_func) {
- progress_func(WIMLIB_PROGRESS_MSG_APPLY_TIMESTAMPS,
- &args.progress);
+ if (unlikely(list_empty(&dentry_list))) {
+ WARNING("There is nothing to extract!");
+ goto out_cleanup;
}
- /* Apply timestamps */
- ret = for_dentry_in_tree_depth(root,
- ops->apply_dentry_timestamps, &args);
+ ret = dentry_list_resolve_streams(&dentry_list, ctx);
if (ret)
- goto out_free_target_realpath;
-
- if (progress_func) {
- progress_func(*wim_source_path ? WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END :
- WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END,
- &args.progress);
- }
-out_free_target_realpath:
- FREE(args.target_realpath);
-out_dentry_reset_needs_extraction:
- for_dentry_in_tree(root, dentry_reset_needs_extraction, NULL);
-out_ntfs_umount:
-#ifdef WITH_NTFS_3G
- /* Unmount the NTFS volume */
- if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
- if (ntfs_umount(args.vol, FALSE) != 0) {
- ERROR_WITH_ERRNO("Failed to unmount NTFS volume `%"TS"'",
- args.target);
- if (ret == 0)
- ret = WIMLIB_ERR_NTFS_3G;
+ goto out_cleanup;
+
+ dentry_list_build_inode_alias_lists(&dentry_list);
+
+ ret = dentry_list_ref_streams(&dentry_list, ctx);
+ if (ret)
+ goto out_cleanup;
+
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_FROM_PIPE) {
+ /* When extracting from a pipe, the number of bytes of data to
+ * extract can't be determined in the normal way (examining the
+ * blob table), since at this point all we have is a set of
+ * SHA-1 message digests of blobs that need to be extracted.
+ * However, we can get a reasonably accurate estimate by taking
+ * <TOTALBYTES> from the corresponding <IMAGE> in the WIM XML
+ * data. This does assume that a full image is being extracted,
+ * but currently there is no API for doing otherwise. (Also,
+ * subtract <HARDLINKBYTES> from this if hard links are
+ * supported by the extraction mode.) */
+ ctx->progress.extract.total_bytes =
+ wim_info_get_image_total_bytes(wim->wim_info,
+ wim->current_image);
+ if (ctx->supported_features.hard_links) {
+ ctx->progress.extract.total_bytes -=
+ wim_info_get_image_hard_link_bytes(wim->wim_info,
+ wim->current_image);
}
}
-#endif
-out_free_target_lowlevel_path:
-#ifdef __WIN32__
- FREE(args.target_lowlevel_path);
-#endif
+
+ ret = extract_progress(ctx,
+ ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
+ WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_BEGIN :
+ WIMLIB_PROGRESS_MSG_EXTRACT_TREE_BEGIN));
+ if (ret)
+ goto out_cleanup;
+
+ ret = (*ops->extract)(&dentry_list, ctx);
+ if (ret)
+ goto out_cleanup;
+
+ if (ctx->progress.extract.completed_bytes <
+ ctx->progress.extract.total_bytes)
+ {
+ ctx->progress.extract.completed_bytes =
+ ctx->progress.extract.total_bytes;
+ ret = extract_progress(ctx, WIMLIB_PROGRESS_MSG_EXTRACT_STREAMS);
+ if (ret)
+ goto out_cleanup;
+ }
+
+ ret = extract_progress(ctx,
+ ((extract_flags & WIMLIB_EXTRACT_FLAG_IMAGEMODE) ?
+ WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END :
+ WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END));
+out_cleanup:
+ destroy_blob_list(&ctx->blob_list);
+ destroy_dentry_list(&dentry_list);
+ FREE(ctx);
out:
return ret;
}
-/* Validates a single wimlib_extract_command, mostly checking to make sure the
- * extract flags make sense. */
static int
-check_extract_command(struct wimlib_extract_command *cmd, int wim_header_flags)
+mkdir_if_needed(const tchar *target)
{
- int extract_flags;
- bool is_entire_image = (cmd->wim_source_path[0] == T('\0'));
+ if (!tmkdir(target, 0755))
+ return 0;
- /* Empty destination path? */
- if (cmd->fs_dest_path[0] == T('\0'))
- return WIMLIB_ERR_INVALID_PARAM;
+ if (errno == EEXIST)
+ return 0;
+
+#ifdef __WIN32__
+ /* _wmkdir() fails with EACCES if called on a drive root directory. */
+ if (errno == EACCES)
+ return 0;
+#endif
+
+ ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
+ return WIMLIB_ERR_MKDIR;
+}
+
+/* Make sure the extraction flags make sense, and update them if needed. */
+static int
+check_extract_flags(const WIMStruct *wim, int *extract_flags_p)
+{
+ int extract_flags = *extract_flags_p;
- extract_flags = cmd->extract_flags;
+ /* Check for invalid flag combinations */
- /* Specified both symlink and hardlink modes? */
if ((extract_flags &
- (WIMLIB_EXTRACT_FLAG_SYMLINK |
- WIMLIB_EXTRACT_FLAG_HARDLINK)) == (WIMLIB_EXTRACT_FLAG_SYMLINK |
- WIMLIB_EXTRACT_FLAG_HARDLINK))
+ (WIMLIB_EXTRACT_FLAG_NO_ACLS |
+ WIMLIB_EXTRACT_FLAG_STRICT_ACLS)) == (WIMLIB_EXTRACT_FLAG_NO_ACLS |
+ WIMLIB_EXTRACT_FLAG_STRICT_ACLS))
return WIMLIB_ERR_INVALID_PARAM;
-#ifdef __WIN32__
- /* Wanted UNIX data on Windows? */
- if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
- ERROR("Extracting UNIX data is not supported on Windows");
- return WIMLIB_ERR_INVALID_PARAM;
- }
- /* Wanted linked extraction on Windows? (XXX This is possible, just not
- * implemented yet.) */
- if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
- WIMLIB_EXTRACT_FLAG_HARDLINK))
- {
- ERROR("Linked extraction modes are not supported on Windows");
+ if ((extract_flags &
+ (WIMLIB_EXTRACT_FLAG_RPFIX |
+ WIMLIB_EXTRACT_FLAG_NORPFIX)) == (WIMLIB_EXTRACT_FLAG_RPFIX |
+ WIMLIB_EXTRACT_FLAG_NORPFIX))
return WIMLIB_ERR_INVALID_PARAM;
+
+#ifndef WITH_NTFS_3G
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
+ ERROR("wimlib was compiled without support for NTFS-3g, so\n"
+ " it cannot apply a WIM image directly to an NTFS volume.");
+ return WIMLIB_ERR_UNSUPPORTED;
}
#endif
- if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
- /* NTFS-3g extraction mode requested */
-#ifdef WITH_NTFS_3G
- if ((extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
- WIMLIB_EXTRACT_FLAG_HARDLINK))) {
- ERROR("Cannot specify symlink or hardlink flags when applying\n"
- " directly to a NTFS volume");
- return WIMLIB_ERR_INVALID_PARAM;
- }
- if (!is_entire_image &&
- (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS))
- {
- ERROR("When applying directly to a NTFS volume you can "
- "only extract a full image, not part of one");
- return WIMLIB_ERR_INVALID_PARAM;
- }
- if (extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA) {
- ERROR("Cannot restore UNIX-specific data in "
- "the NTFS extraction mode");
- return WIMLIB_ERR_INVALID_PARAM;
- }
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
+#ifdef __WIN32__
+ if (!wim->filename)
+ return WIMLIB_ERR_NO_FILENAME;
#else
- ERROR("wimlib was compiled without support for NTFS-3g, so");
- ERROR("we cannot apply a WIM image directly to a NTFS volume");
+ ERROR("WIMBoot extraction is only supported on Windows!");
return WIMLIB_ERR_UNSUPPORTED;
#endif
}
- if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
- WIMLIB_EXTRACT_FLAG_NORPFIX)) ==
- (WIMLIB_EXTRACT_FLAG_RPFIX | WIMLIB_EXTRACT_FLAG_NORPFIX))
- {
- ERROR("Cannot specify RPFIX and NORPFIX flags at the same time!");
- return WIMLIB_ERR_INVALID_PARAM;
- }
if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
- WIMLIB_EXTRACT_FLAG_NORPFIX)) == 0)
+ WIMLIB_EXTRACT_FLAG_NORPFIX |
+ WIMLIB_EXTRACT_FLAG_IMAGEMODE)) ==
+ WIMLIB_EXTRACT_FLAG_IMAGEMODE)
{
- /* Do reparse point fixups by default if the WIM header says
- * they are enabled and we are extracting a full image. */
- if ((wim_header_flags & WIM_HDR_FLAG_RP_FIX) && is_entire_image)
+ /* For full-image extraction, do reparse point fixups by default
+ * if the WIM header says they are enabled. */
+ if (wim->hdr.flags & WIM_HDR_FLAG_RP_FIX)
extract_flags |= WIMLIB_EXTRACT_FLAG_RPFIX;
}
- if (!is_entire_image && (extract_flags & WIMLIB_EXTRACT_FLAG_RPFIX)) {
- ERROR("Cannot specify --rpfix when not extracting entire image");
- return WIMLIB_ERR_INVALID_PARAM;
- }
-
- cmd->extract_flags = extract_flags;
+ *extract_flags_p = extract_flags;
return 0;
}
-
-/* Internal function to execute extraction commands for a WIM image. */
-static int
-do_wimlib_extract_files(WIMStruct *wim,
- int image,
- struct wimlib_extract_command *cmds,
- size_t num_cmds,
- wimlib_progress_func_t progress_func)
+static u32
+get_wildcard_flags(int extract_flags)
{
- int ret;
- bool found_link_cmd = false;
- bool found_nolink_cmd = false;
+ u32 wildcard_flags = 0;
- /* Select the image from which we are extracting files */
- ret = select_wim_image(wim, image);
- if (ret)
- return ret;
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB)
+ wildcard_flags |= WILDCARD_FLAG_ERROR_IF_NO_MATCH;
+ else
+ wildcard_flags |= WILDCARD_FLAG_WARN_IF_NO_MATCH;
- /* Make sure there are no streams in the WIM that have not been
- * checksummed yet. */
- ret = wim_checksum_unhashed_streams(wim);
- if (ret)
- return ret;
+ if (default_ignore_case)
+ wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
- /* Check for problems with the extraction commands */
- for (size_t i = 0; i < num_cmds; i++) {
- ret = check_extract_command(&cmds[i], wim->hdr.flags);
- if (ret)
- return ret;
- if (cmds[i].extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
- WIMLIB_EXTRACT_FLAG_HARDLINK)) {
- found_link_cmd = true;
- } else {
- found_nolink_cmd = true;
- }
- if (found_link_cmd && found_nolink_cmd) {
- ERROR("Symlink or hardlink extraction mode must "
- "be set on all extraction commands");
- return WIMLIB_ERR_INVALID_PARAM;
- }
- }
+ return wildcard_flags;
+}
- /* Execute the extraction commands */
- for (size_t i = 0; i < num_cmds; i++) {
- ret = extract_tree(wim,
- cmds[i].wim_source_path,
- cmds[i].fs_dest_path,
- cmds[i].extract_flags,
- progress_func);
- if (ret)
- return ret;
+struct append_dentry_ctx {
+ struct wim_dentry **dentries;
+ size_t num_dentries;
+ size_t num_alloc_dentries;
+};
+
+static int
+append_dentry_cb(struct wim_dentry *dentry, void *_ctx)
+{
+ struct append_dentry_ctx *ctx = _ctx;
+
+ if (ctx->num_dentries == ctx->num_alloc_dentries) {
+ struct wim_dentry **new_dentries;
+ size_t new_length;
+
+ new_length = max(ctx->num_alloc_dentries + 8,
+ ctx->num_alloc_dentries * 3 / 2);
+ new_dentries = REALLOC(ctx->dentries,
+ new_length * sizeof(ctx->dentries[0]));
+ if (new_dentries == NULL)
+ return WIMLIB_ERR_NOMEM;
+ ctx->dentries = new_dentries;
+ ctx->num_alloc_dentries = new_length;
}
+ ctx->dentries[ctx->num_dentries++] = dentry;
return 0;
}
-/* Extract files or directories from a WIM image. */
-WIMLIBAPI int
-wimlib_extract_files(WIMStruct *wim,
- int image,
- const struct wimlib_extract_command *cmds,
- size_t num_cmds,
- int default_extract_flags,
- WIMStruct **additional_swms,
- unsigned num_additional_swms,
- wimlib_progress_func_t progress_func)
+static int
+do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
+ const tchar * const *paths, size_t num_paths,
+ int extract_flags)
{
int ret;
- struct wimlib_extract_command *cmds_copy;
- int all_flags = 0;
+ struct wim_dentry **trees;
+ size_t num_trees;
- default_extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
+ if (wim == NULL || target == NULL || target[0] == T('\0') ||
+ (num_paths != 0 && paths == NULL))
+ return WIMLIB_ERR_INVALID_PARAM;
- ret = verify_swm_set(wim, additional_swms, num_additional_swms);
+ ret = check_extract_flags(wim, &extract_flags);
if (ret)
- goto out;
+ return ret;
- if (num_cmds == 0)
- goto out;
+ ret = select_wim_image(wim, image);
+ if (ret)
+ return ret;
- if (num_additional_swms)
- merge_lookup_tables(wim, additional_swms, num_additional_swms);
+ ret = wim_checksum_unhashed_blobs(wim);
+ if (ret)
+ return ret;
- cmds_copy = CALLOC(num_cmds, sizeof(cmds[0]));
- if (!cmds_copy) {
- ret = WIMLIB_ERR_NOMEM;
- goto out_restore_lookup_table;
+ if ((extract_flags & (WIMLIB_EXTRACT_FLAG_NTFS |
+ WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE)) ==
+ (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE))
+ {
+ ret = mkdir_if_needed(target);
+ if (ret)
+ return ret;
}
- for (size_t i = 0; i < num_cmds; i++) {
- cmds_copy[i].extract_flags = (default_extract_flags |
- cmds[i].extract_flags)
- & WIMLIB_EXTRACT_MASK_PUBLIC;
- all_flags |= cmds_copy[i].extract_flags;
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_GLOB_PATHS) {
- cmds_copy[i].wim_source_path = canonicalize_wim_path(cmds[i].wim_source_path);
- if (!cmds_copy[i].wim_source_path) {
- ret = WIMLIB_ERR_NOMEM;
- goto out_free_cmds_copy;
+ struct append_dentry_ctx append_dentry_ctx = {
+ .dentries = NULL,
+ .num_dentries = 0,
+ .num_alloc_dentries = 0,
+ };
+
+ u32 wildcard_flags = get_wildcard_flags(extract_flags);
+
+ for (size_t i = 0; i < num_paths; i++) {
+ tchar *path = canonicalize_wim_path(paths[i]);
+ if (path == NULL) {
+ ret = WIMLIB_ERR_NOMEM;
+ trees = append_dentry_ctx.dentries;
+ goto out_free_trees;
+ }
+ ret = expand_wildcard(wim, path,
+ append_dentry_cb,
+ &append_dentry_ctx,
+ wildcard_flags);
+ FREE(path);
+ if (ret) {
+ trees = append_dentry_ctx.dentries;
+ goto out_free_trees;
+ }
}
+ trees = append_dentry_ctx.dentries;
+ num_trees = append_dentry_ctx.num_dentries;
+ } else {
+ trees = MALLOC(num_paths * sizeof(trees[0]));
+ if (trees == NULL)
+ return WIMLIB_ERR_NOMEM;
+
+ for (size_t i = 0; i < num_paths; i++) {
+
+ tchar *path = canonicalize_wim_path(paths[i]);
+ if (path == NULL) {
+ ret = WIMLIB_ERR_NOMEM;
+ goto out_free_trees;
+ }
- cmds_copy[i].fs_dest_path = canonicalize_fs_path(cmds[i].fs_dest_path);
- if (!cmds_copy[i].fs_dest_path) {
- ret = WIMLIB_ERR_NOMEM;
- goto out_free_cmds_copy;
+ trees[i] = get_dentry(wim, path,
+ WIMLIB_CASE_PLATFORM_DEFAULT);
+ FREE(path);
+ if (trees[i] == NULL) {
+ ERROR("Path \"%"TS"\" does not exist "
+ "in WIM image %d",
+ paths[i], wim->current_image);
+ ret = WIMLIB_ERR_PATH_DOES_NOT_EXIST;
+ goto out_free_trees;
+ }
}
+ num_trees = num_paths;
+ }
+ if (num_trees == 0) {
+ ret = 0;
+ goto out_free_trees;
}
- ret = do_wimlib_extract_files(wim, image,
- cmds_copy, num_cmds,
- progress_func);
- if (all_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
- WIMLIB_EXTRACT_FLAG_HARDLINK))
- {
- for_lookup_table_entry(wim->lookup_table,
- lte_free_extracted_file, NULL);
- }
-out_free_cmds_copy:
- for (size_t i = 0; i < num_cmds; i++) {
- FREE(cmds_copy[i].wim_source_path);
- FREE(cmds_copy[i].fs_dest_path);
- }
- FREE(cmds_copy);
-out_restore_lookup_table:
- if (num_additional_swms)
- unmerge_lookup_table(wim);
-out:
+ ret = extract_trees(wim, trees, num_trees, target, extract_flags);
+out_free_trees:
+ FREE(trees);
return ret;
}
-/*
- * Extracts an image from a WIM file.
- *
- * @wim: WIMStruct for the WIM file.
- *
- * @image: Number of the single image to extract.
- *
- * @target: Directory or NTFS volume to extract the image to.
- *
- * @extract_flags: Bitwise or of WIMLIB_EXTRACT_FLAG_*.
- *
- * @progress_func: If non-NULL, a progress function to be called
- * periodically.
- *
- * Returns 0 on success; nonzero on failure.
- */
static int
extract_single_image(WIMStruct *wim, int image,
- const tchar *target, int extract_flags,
- wimlib_progress_func_t progress_func)
+ const tchar *target, int extract_flags)
{
- int ret;
- tchar *target_copy = canonicalize_fs_path(target);
- if (!target_copy)
- return WIMLIB_ERR_NOMEM;
- struct wimlib_extract_command cmd = {
- .wim_source_path = T(""),
- .fs_dest_path = target_copy,
- .extract_flags = extract_flags,
- };
- ret = do_wimlib_extract_files(wim, image, &cmd, 1, progress_func);
- FREE(target_copy);
- return ret;
+ const tchar *path = WIMLIB_WIM_ROOT_PATH;
+ extract_flags |= WIMLIB_EXTRACT_FLAG_IMAGEMODE;
+ return do_wimlib_extract_paths(wim, image, target, &path, 1, extract_flags);
}
static const tchar * const filename_forbidden_chars =
/* Extracts all images from the WIM to the directory @target, with the images
* placed in subdirectories named by their image names. */
static int
-extract_all_images(WIMStruct *wim,
- const tchar *target,
- int extract_flags,
- wimlib_progress_func_t progress_func)
+extract_all_images(WIMStruct *wim, const tchar *target, int extract_flags)
{
size_t image_name_max_len = max(xml_get_max_image_name_len(wim), 20);
size_t output_path_len = tstrlen(target);
int ret;
int image;
const tchar *image_name;
- struct stat stbuf;
- if (tstat(target, &stbuf)) {
- if (errno == ENOENT)
- {
- if (tmkdir(target, S_IRWXU | S_IRGRP | S_IXGRP |
- S_IROTH | S_IXOTH))
- {
- ERROR_WITH_ERRNO("Failed to create directory \"%"TS"\"", target);
- return WIMLIB_ERR_MKDIR;
- }
- } else {
- ERROR_WITH_ERRNO("Failed to stat \"%"TS"\"", target);
- return WIMLIB_ERR_STAT;
- }
- } else if (!S_ISDIR(stbuf.st_mode)) {
- ERROR("\"%"TS"\" is not a directory", target);
- return WIMLIB_ERR_NOTDIR;
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
+ ERROR("Cannot extract multiple images in NTFS extraction mode.");
+ return WIMLIB_ERR_INVALID_PARAM;
}
+ ret = mkdir_if_needed(target);
+ if (ret)
+ return ret;
tmemcpy(buf, target, output_path_len);
buf[output_path_len] = OS_PREFERRED_PATH_SEPARATOR;
for (image = 1; image <= wim->hdr.image_count; image++) {
* Use image number instead. */
tsprintf(buf + output_path_len + 1, T("%d"), image);
}
- ret = extract_single_image(wim, image, buf, extract_flags,
- progress_func);
+ ret = extract_single_image(wim, image, buf, extract_flags);
if (ret)
return ret;
}
return 0;
}
-/* Extracts a single image or all images from a WIM file to a directory or NTFS
- * volume. */
+static int
+do_wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
+ int extract_flags)
+{
+ if (extract_flags & (WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE |
+ WIMLIB_EXTRACT_FLAG_TO_STDOUT |
+ WIMLIB_EXTRACT_FLAG_GLOB_PATHS))
+ return WIMLIB_ERR_INVALID_PARAM;
+
+ if (image == WIMLIB_ALL_IMAGES)
+ return extract_all_images(wim, target, extract_flags);
+ else
+ return extract_single_image(wim, image, target, extract_flags);
+}
+
+
+/****************************************************************************
+ * Extraction API *
+ ****************************************************************************/
+
WIMLIBAPI int
-wimlib_extract_image(WIMStruct *wim,
- int image,
- const tchar *target,
- int extract_flags,
- WIMStruct **additional_swms,
- unsigned num_additional_swms,
- wimlib_progress_func_t progress_func)
+wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
+ const tchar * const *paths, size_t num_paths,
+ int extract_flags)
+{
+ if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
+ return WIMLIB_ERR_INVALID_PARAM;
+
+ return do_wimlib_extract_paths(wim, image, target, paths, num_paths,
+ extract_flags);
+}
+
+WIMLIBAPI int
+wimlib_extract_pathlist(WIMStruct *wim, int image, const tchar *target,
+ const tchar *path_list_file, int extract_flags)
{
int ret;
+ tchar **paths;
+ size_t num_paths;
+ void *mem;
+
+ ret = read_path_list_file(path_list_file, &paths, &num_paths, &mem);
+ if (ret) {
+ ERROR("Failed to read path list file \"%"TS"\"",
+ path_list_file);
+ return ret;
+ }
- extract_flags &= WIMLIB_EXTRACT_MASK_PUBLIC;
+ ret = wimlib_extract_paths(wim, image, target,
+ (const tchar * const *)paths, num_paths,
+ extract_flags);
+ FREE(paths);
+ FREE(mem);
+ return ret;
+}
- ret = verify_swm_set(wim, additional_swms, num_additional_swms);
+WIMLIBAPI int
+wimlib_extract_image_from_pipe_with_progress(int pipe_fd,
+ const tchar *image_num_or_name,
+ const tchar *target,
+ int extract_flags,
+ wimlib_progress_func_t progfunc,
+ void *progctx)
+{
+ int ret;
+ WIMStruct *pwm;
+ struct filedes *in_fd;
+ int image;
+ unsigned i;
+
+ if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
+ return WIMLIB_ERR_INVALID_PARAM;
+
+ /* Read the WIM header from the pipe and get a WIMStruct to represent
+ * the pipable WIM. Caveats: Unlike getting a WIMStruct with
+ * wimlib_open_wim(), getting a WIMStruct in this way will result in an
+ * empty blob table, no XML data read, and no filename set. */
+ ret = open_wim_as_WIMStruct(&pipe_fd, WIMLIB_OPEN_FLAG_FROM_PIPE, &pwm,
+ progfunc, progctx);
if (ret)
return ret;
- if (num_additional_swms)
- merge_lookup_tables(wim, additional_swms, num_additional_swms);
+ /* Sanity check to make sure this is a pipable WIM. */
+ if (pwm->hdr.magic != PWM_MAGIC) {
+ ERROR("The WIM being read from file descriptor %d "
+ "is not pipable!", pipe_fd);
+ ret = WIMLIB_ERR_NOT_PIPABLE;
+ goto out_wimlib_free;
+ }
- if (image == WIMLIB_ALL_IMAGES) {
- ret = extract_all_images(wim, target,
- extract_flags | WIMLIB_EXTRACT_FLAG_MULTI_IMAGE,
- progress_func);
- } else {
- ret = extract_single_image(wim, image, target, extract_flags,
- progress_func);
+ /* Sanity check to make sure the first part of a pipable split WIM is
+ * sent over the pipe first. */
+ if (pwm->hdr.part_number != 1) {
+ ERROR("The first part of the split WIM must be "
+ "sent over the pipe first.");
+ ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
+ goto out_wimlib_free;
}
- if (extract_flags & (WIMLIB_EXTRACT_FLAG_SYMLINK |
- WIMLIB_EXTRACT_FLAG_HARDLINK))
+ in_fd = &pwm->in_fd;
+ wimlib_assert(in_fd->offset == WIM_HEADER_DISK_SIZE);
+
+ /* As mentioned, the WIMStruct we created from the pipe does not have
+ * XML data yet. Fix this by reading the extra copy of the XML data
+ * that directly follows the header in pipable WIMs. (Note: see
+ * write_pipable_wim() for more details about the format of pipable
+ * WIMs.) */
{
- for_lookup_table_entry(wim->lookup_table,
- lte_free_extracted_file,
- NULL);
+ u8 hash[SHA1_HASH_SIZE];
+
+ ret = read_pwm_blob_header(pwm, hash,
+ &pwm->hdr.xml_data_reshdr, NULL);
+ if (ret)
+ goto out_wimlib_free;
+
+ if (!(pwm->hdr.xml_data_reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
+ ERROR("Expected XML data, but found non-metadata resource.");
+ ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
+ goto out_wimlib_free;
+ }
+
+ ret = read_wim_xml_data(pwm);
+ if (ret)
+ goto out_wimlib_free;
+
+ if (wim_info_get_num_images(pwm->wim_info) != pwm->hdr.image_count) {
+ ERROR("Image count in XML data is not the same as in WIM header.");
+ ret = WIMLIB_ERR_IMAGE_COUNT;
+ goto out_wimlib_free;
+ }
+ }
+
+ /* Get image index (this may use the XML data that was just read to
+ * resolve an image name). */
+ if (image_num_or_name) {
+ image = wimlib_resolve_image(pwm, image_num_or_name);
+ if (image == WIMLIB_NO_IMAGE) {
+ ERROR("\"%"TS"\" is not a valid image in the pipable WIM!",
+ image_num_or_name);
+ ret = WIMLIB_ERR_INVALID_IMAGE;
+ goto out_wimlib_free;
+ } else if (image == WIMLIB_ALL_IMAGES) {
+ ERROR("Applying all images from a pipe is not supported!");
+ ret = WIMLIB_ERR_INVALID_IMAGE;
+ goto out_wimlib_free;
+ }
+ } else {
+ if (pwm->hdr.image_count != 1) {
+ ERROR("No image was specified, but the pipable WIM "
+ "did not contain exactly 1 image");
+ ret = WIMLIB_ERR_INVALID_IMAGE;
+ goto out_wimlib_free;
+ }
+ image = 1;
+ }
+
+ /* Load the needed metadata resource. */
+ for (i = 1; i <= pwm->hdr.image_count; i++) {
+ struct wim_image_metadata *imd;
+ struct wim_reshdr reshdr;
+ struct wim_resource_descriptor *metadata_rdesc;
+
+ imd = pwm->image_metadata[i - 1];
+
+ ret = WIMLIB_ERR_NOMEM;
+ imd->metadata_blob = new_blob_descriptor();
+ if (!imd->metadata_blob)
+ goto out_wimlib_free;
+
+ imd->metadata_blob->is_metadata = 1;
+
+ ret = read_pwm_blob_header(pwm, imd->metadata_blob->hash,
+ &reshdr, NULL);
+ if (ret)
+ goto out_wimlib_free;
+
+ if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
+ ERROR("Expected metadata resource, but found "
+ "non-metadata resource");
+ ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
+ goto out_wimlib_free;
+ }
+
+ ret = WIMLIB_ERR_NOMEM;
+ metadata_rdesc = MALLOC(sizeof(struct wim_resource_descriptor));
+ if (!metadata_rdesc)
+ goto out_wimlib_free;
+ wim_reshdr_to_desc_and_blob(&reshdr, pwm, metadata_rdesc,
+ imd->metadata_blob);
+
+ if (i == image) {
+ /* Metadata resource is for the image being extracted.
+ * Parse it and save the metadata in memory. */
+ ret = read_metadata_resource(imd);
+ if (ret)
+ goto out_wimlib_free;
+ imd->modified = 1;
+ } else {
+ /* Metadata resource is not for the image being
+ * extracted. Skip over it. */
+ ret = skip_wim_resource(metadata_rdesc);
+ if (ret)
+ goto out_wimlib_free;
+ }
}
- if (num_additional_swms)
- unmerge_lookup_table(wim);
+ /* Extract the image. */
+ extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
+ ret = do_wimlib_extract_image(pwm, image, target, extract_flags);
+ /* Clean up and return. */
+out_wimlib_free:
+ wimlib_free(pwm);
return ret;
}
+
+
+WIMLIBAPI int
+wimlib_extract_image_from_pipe(int pipe_fd, const tchar *image_num_or_name,
+ const tchar *target, int extract_flags)
+{
+ return wimlib_extract_image_from_pipe_with_progress(pipe_fd,
+ image_num_or_name,
+ target,
+ extract_flags,
+ NULL,
+ NULL);
+}
+
+WIMLIBAPI int
+wimlib_extract_image(WIMStruct *wim, int image, const tchar *target,
+ int extract_flags)
+{
+ if (extract_flags & ~WIMLIB_EXTRACT_MASK_PUBLIC)
+ return WIMLIB_ERR_INVALID_PARAM;
+ return do_wimlib_extract_image(wim, image, target, extract_flags);
+}