*/
/*
- * Copyright (C) 2012, 2013, 2014 Eric Biggers
+ * Copyright (C) 2012-2016 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/.
*/
/*
* 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,
+ * 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
# include "config.h"
#endif
+#include <errno.h>
+#include <fcntl.h>
+#include <stdlib.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/object_id.h"
#include "wimlib/pathlist.h"
#include "wimlib/paths.h"
+#include "wimlib/pattern.h"
#include "wimlib/reparse.h"
#include "wimlib/resource.h"
#include "wimlib/security.h"
#include "wimlib/unix_data.h"
-#ifdef __WIN32__
-# include "wimlib/win32.h" /* for realpath() equivalent */
-#endif
-#include "wimlib/xml.h"
-#include "wimlib/wildcard.h"
#include "wimlib/wim.h"
-
-#include <errno.h>
-#include <fcntl.h>
-#include <stdlib.h>
-#include <sys/stat.h>
-#include <unistd.h>
+#include "wimlib/win32.h" /* for realpath() equivalent */
+#include "wimlib/xml.h"
#define WIMLIB_EXTRACT_FLAG_FROM_PIPE 0x80000000
#define WIMLIB_EXTRACT_FLAG_IMAGEMODE 0x40000000
WIMLIB_EXTRACT_FLAG_STRICT_GLOB | \
WIMLIB_EXTRACT_FLAG_NO_ATTRIBUTES | \
WIMLIB_EXTRACT_FLAG_NO_PRESERVE_DIR_STRUCTURE | \
- WIMLIB_EXTRACT_FLAG_WIMBOOT)
+ WIMLIB_EXTRACT_FLAG_WIMBOOT | \
+ WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K | \
+ WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K | \
+ WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K | \
+ WIMLIB_EXTRACT_FLAG_COMPACT_LZX \
+ )
/* 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 = 512; /* Arbitrary value to limit calls */
+ ctx->count_until_file_progress = 500; /* Arbitrary value to limit calls */
return extract_progress(ctx, msg);
}
-/* Check whether the extraction of a dentry should be skipped completely. */
-static bool
-dentry_is_supported(struct wim_dentry *dentry,
- const struct wim_features *supported_features)
+static int
+start_file_phase(struct apply_ctx *ctx, u64 end_file_count, enum wimlib_progress_msg msg)
{
- struct wim_inode *inode = dentry->d_inode;
+ ctx->progress.extract.current_file_count = 0;
+ ctx->progress.extract.end_file_count = end_file_count;
+ return do_file_extract_progress(ctx, msg);
+}
- if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
- return supported_features->reparse_points ||
- (inode_is_symlink(inode) &&
- supported_features->symlink_reparse_points);
- }
- if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
- if (inode->i_attributes & FILE_ATTRIBUTE_DIRECTORY)
- return supported_features->encrypted_directories != 0;
- else
- return supported_features->encrypted_files != 0;
- }
- return true;
+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);
}
+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);
+}
-#define PWM_ALLOW_WIM_HDR 0x00001
+static int
+end_file_phase(struct apply_ctx *ctx, enum wimlib_progress_msg msg)
+{
+ ctx->progress.extract.current_file_count = ctx->progress.extract.end_file_count;
+ return do_file_extract_progress(ctx, msg);
+}
-/* Read the header from a stream in a pipable WIM. */
+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);
+}
+
+#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_stream_header(WIMStruct *pwm, struct wim_lookup_table_entry *lte,
- struct wim_resource_spec *rspec,
- int flags, struct wim_header_disk *hdr_ret)
+read_pwm_blob_header(WIMStruct *pwm, u8 hash_ret[SHA1_HASH_SIZE],
+ struct wim_reshdr *reshdr_ret,
+ struct wim_header_disk *pwm_hdr_ret)
{
- union {
- struct pwm_stream_hdr stream_hdr;
- struct wim_header_disk pwm_hdr;
- } buf;
- struct wim_reshdr reshdr;
int ret;
+ struct pwm_blob_hdr blob_hdr;
+ u64 magic;
- ret = full_read(&pwm->in_fd, &buf.stream_hdr, sizeof(buf.stream_hdr));
- if (ret)
+ ret = full_read(&pwm->in_fd, &blob_hdr, sizeof(blob_hdr));
+ if (unlikely(ret))
goto read_error;
- if ((flags & PWM_ALLOW_WIM_HDR) &&
- le64_to_cpu(buf.stream_hdr.magic) == PWM_MAGIC)
- {
- BUILD_BUG_ON(sizeof(buf.pwm_hdr) < sizeof(buf.stream_hdr));
- ret = full_read(&pwm->in_fd, &buf.stream_hdr + 1,
- sizeof(buf.pwm_hdr) - sizeof(buf.stream_hdr));
+ magic = le64_to_cpu(blob_hdr.magic);
- if (ret)
+ 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;
- lte->resource_location = RESOURCE_NONEXISTENT;
- memcpy(hdr_ret, &buf.pwm_hdr, sizeof(buf.pwm_hdr));
- return 0;
+ return PWM_FOUND_WIM_HDR;
}
- if (le64_to_cpu(buf.stream_hdr.magic) != PWM_STREAM_MAGIC) {
- ERROR("Data read on pipe is invalid (expected stream header).");
+ if (unlikely(magic != PWM_BLOB_MAGIC)) {
+ ERROR("Data read on pipe is invalid (expected blob header)");
return WIMLIB_ERR_INVALID_PIPABLE_WIM;
}
- copy_hash(lte->hash, buf.stream_hdr.hash);
+ 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);
+
+ 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;
+ }
- reshdr.size_in_wim = 0;
- reshdr.flags = le32_to_cpu(buf.stream_hdr.flags);
- reshdr.offset_in_wim = pwm->in_fd.offset;
- reshdr.uncompressed_size = le64_to_cpu(buf.stream_hdr.uncompressed_size);
- wim_res_hdr_to_spec(&reshdr, pwm, rspec);
- lte_bind_wim_resource_spec(lte, rspec);
- lte->flags = rspec->flags;
- lte->size = rspec->uncompressed_size;
- lte->offset_in_res = 0;
return 0;
read_error:
- ERROR_WITH_ERRNO("Error reading pipable WIM from pipe");
+ 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
-load_streams_from_pipe(struct apply_ctx *ctx,
- const struct read_stream_list_callbacks *cbs)
+read_blobs_from_pipe(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
{
- struct wim_lookup_table_entry *found_lte = NULL;
- struct wim_resource_spec *rspec = NULL;
- struct wim_lookup_table *lookup_table;
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;
- ret = WIMLIB_ERR_NOMEM;
- found_lte = new_lookup_table_entry();
- if (!found_lte)
- goto out;
-
- rspec = MALLOC(sizeof(struct wim_resource_spec));
- if (!rspec)
- goto out;
-
- lookup_table = ctx->wim->lookup_table;
- memcpy(ctx->progress.extract.guid, ctx->wim->hdr.guid, WIM_GUID_LEN);
+ 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)
- goto out;
+ return ret;
- while (ctx->num_streams_remaining) {
- struct wim_header_disk pwm_hdr;
- struct wim_lookup_table_entry *needed_lte;
+ while (ctx->num_blobs_remaining) {
- if (found_lte->resource_location != RESOURCE_NONEXISTENT)
- lte_unbind_wim_resource_spec(found_lte);
- ret = read_pwm_stream_header(ctx->wim, found_lte, rspec,
- PWM_ALLOW_WIM_HDR, &pwm_hdr);
- if (ret)
- goto out;
+ ret = read_pwm_blob_header(ctx->wim, hash, &reshdr, &pwm_hdr);
- if ((found_lte->resource_location != RESOURCE_NONEXISTENT)
- && !(found_lte->flags & WIM_RESHDR_FLAG_METADATA)
- && (needed_lte = lookup_stream(lookup_table, found_lte->hash))
- && (needed_lte->out_refcnt))
- {
- needed_lte->offset_in_res = found_lte->offset_in_res;
- needed_lte->flags = found_lte->flags;
- needed_lte->size = found_lte->size;
+ 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);
- lte_unbind_wim_resource_spec(found_lte);
- lte_bind_wim_resource_spec(needed_lte, rspec);
+ 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;
- ret = (*cbs->begin_stream)(needed_lte,
- cbs->begin_stream_ctx);
- if (ret) {
- lte_unbind_wim_resource_spec(needed_lte);
- goto out;
- }
+ 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;
- ret = extract_stream(needed_lte, needed_lte->size,
- cbs->consume_chunk,
- cbs->consume_chunk_ctx);
+ continue;
+ }
- ret = (*cbs->end_stream)(needed_lte, ret,
- cbs->end_stream_ctx);
- lte_unbind_wim_resource_spec(needed_lte);
- if (ret)
- goto out;
- ctx->num_streams_remaining--;
- } else if (found_lte->resource_location != RESOURCE_NONEXISTENT) {
- ret = skip_wim_stream(found_lte);
+ if (ret)
+ return ret;
+
+ if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)
+ && (blob = lookup_blob(ctx->wim->blob_table, hash))
+ && (blob->out_refcnt))
+ {
+ 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)
- goto out;
+ return ret;
+ ctx->num_blobs_remaining--;
} else {
- 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 ||
- memcmp(pwm_hdr.guid, ctx->progress.extract.guid,
- WIM_GUID_LEN))
- {
- ctx->progress.extract.part_number = part_number;
- ctx->progress.extract.total_parts = total_parts;
- memcpy(ctx->progress.extract.guid,
- pwm_hdr.guid, WIM_GUID_LEN);
- ret = extract_progress(ctx,
- WIMLIB_PROGRESS_MSG_EXTRACT_SPWM_PART_BEGIN);
- if (ret)
- goto out;
- }
+ wim_reshdr_to_desc(&reshdr, ctx->wim, &rdesc);
+ ret = skip_wim_resource(&rdesc);
+ if (ret)
+ return ret;
}
}
- ret = 0;
+
+ return 0;
+}
+
+static int
+handle_pwm_metadata_resource(WIMStruct *pwm, int image, bool is_needed)
+{
+ struct blob_descriptor *blob;
+ struct wim_reshdr reshdr;
+ struct wim_resource_descriptor *rdesc;
+ int ret;
+
+ ret = WIMLIB_ERR_NOMEM;
+ blob = new_blob_descriptor();
+ if (!blob)
+ goto out;
+
+ ret = read_pwm_blob_header(pwm, blob->hash, &reshdr, NULL);
+ if (ret)
+ goto out;
+
+ ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
+ if (!(reshdr.flags & WIM_RESHDR_FLAG_METADATA)) {
+ ERROR("Expected metadata resource, but found non-metadata "
+ "resource");
+ goto out;
+ }
+
+ ret = WIMLIB_ERR_NOMEM;
+ rdesc = MALLOC(sizeof(*rdesc));
+ if (!rdesc)
+ goto out;
+
+ wim_reshdr_to_desc_and_blob(&reshdr, pwm, rdesc, blob);
+ pwm->refcnt++;
+
+ ret = WIMLIB_ERR_NOMEM;
+ pwm->image_metadata[image - 1] = new_unloaded_image_metadata(blob);
+ if (!pwm->image_metadata[image - 1])
+ goto out;
+ blob = NULL;
+
+ /* If the metadata resource is for the image being extracted, then parse
+ * it and save the metadata in memory. Otherwise, skip over it. */
+ if (is_needed)
+ ret = select_wim_image(pwm, image);
+ else
+ ret = skip_wim_resource(rdesc);
out:
- if (found_lte && found_lte->resource_location != RESOURCE_IN_WIM)
- FREE(rspec);
- free_lookup_table_entry(found_lte);
+ free_blob_descriptor(blob);
return ret;
}
create_temporary_file(struct filedes *fd_ret, tchar **name_ret)
{
tchar *name;
- int open_flags;
int raw_fd;
+#ifdef __WIN32__
retry:
- name = ttempnam(NULL, T("wimlib"));
+ name = _wtempnam(NULL, L"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__
- open_flags |= _O_SHORT_LIVED;
-#endif
- raw_fd = topen(name, open_flags, 0600);
+ raw_fd = _wopen(name, O_WRONLY | O_CREAT | O_EXCL | O_BINARY |
+ _O_SHORT_LIVED, 0600);
+ if (raw_fd < 0 && errno == EEXIST) {
+ FREE(name);
+ goto retry;
+ }
+#else /* __WIN32__ */
+ const char *tmpdir = getenv("TMPDIR");
+ if (!tmpdir)
+ tmpdir = P_tmpdir;
+ name = MALLOC(strlen(tmpdir) + 1 + 6 + 6 + 1);
+ if (!name)
+ return WIMLIB_ERR_NOMEM;
+ sprintf(name, "%s/wimlibXXXXXX", tmpdir);
+ raw_fd = mkstemp(name);
+#endif /* !__WIN32__ */
if (raw_fd < 0) {
- if (errno == EEXIST) {
- FREE(name);
- goto retry;
- }
ERROR_WITH_ERRNO("Failed to create temporary file "
"\"%"TS"\"", name);
FREE(name);
}
static int
-begin_extract_stream_wrapper(struct wim_lookup_table_entry *lte, void *_ctx)
+begin_extract_blob_wrapper(struct blob_descriptor *blob, void *_ctx)
{
struct apply_ctx *ctx = _ctx;
- ctx->cur_stream = lte;
- ctx->cur_stream_offset = 0;
+ ctx->cur_blob = blob;
+ ctx->cur_blob_offset = 0;
- if (unlikely(lte->out_refcnt > MAX_OPEN_STREAMS))
+ if (unlikely(blob->out_refcnt > MAX_OPEN_FILES))
return create_temporary_file(&ctx->tmpfile_fd, &ctx->tmpfile_name);
- else
- return (*ctx->saved_cbs->begin_stream)(lte, ctx->saved_cbs->begin_stream_ctx);
+
+ return call_begin_blob(blob, ctx->saved_cbs);
}
static int
union wimlib_progress_info *progress = &ctx->progress;
int ret;
- ctx->cur_stream_offset += size;
+ ctx->cur_blob_offset += size;
if (likely(ctx->supported_features.hard_links)) {
progress->extract.completed_bytes +=
- (u64)size * ctx->cur_stream->out_refcnt;
- if (ctx->cur_stream_offset == ctx->cur_stream->size)
- progress->extract.completed_streams += ctx->cur_stream->out_refcnt;
+ (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 stream_owner *owners = stream_owners(ctx->cur_stream);
- for (u32 i = 0; i < ctx->cur_stream->out_refcnt; i++) {
- const struct wim_inode *inode = owners[i].inode;
+ 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)
- {
+ inode_for_each_extraction_alias(dentry, inode) {
progress->extract.completed_bytes += size;
- if (ctx->cur_stream_offset == ctx->cur_stream->size)
+ if (ctx->cur_blob_offset == ctx->cur_blob->size)
progress->extract.completed_streams++;
}
}
if (ret)
return ret;
- if (progress->extract.completed_bytes >=
- progress->extract.total_bytes)
- {
- ctx->next_progress = UINT64_MAX;
- } else {
- /* 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;
- }
+ set_next_progress(progress->extract.completed_bytes,
+ progress->extract.total_bytes,
+ &ctx->next_progress);
}
if (unlikely(filedes_valid(&ctx->tmpfile_fd))) {
ctx->tmpfile_name);
}
return ret;
- } else {
- return (*ctx->saved_cbs->consume_chunk)(chunk, size,
- ctx->saved_cbs->consume_chunk_ctx);
}
+
+ return call_consume_chunk(chunk, size, ctx->saved_cbs);
}
+/* 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, struct apply_ctx *ctx)
+extract_from_tmpfile(const tchar *tmpfile_name,
+ const struct blob_descriptor *orig_blob,
+ const struct read_blob_callbacks *cbs)
{
- struct wim_lookup_table_entry tmpfile_lte;
- struct wim_lookup_table_entry *orig_lte = ctx->cur_stream;
- const struct read_stream_list_callbacks *cbs = ctx->saved_cbs;
+ struct blob_descriptor tmpfile_blob;
+ const struct blob_extraction_target *targets = blob_extraction_targets(orig_blob);
int ret;
- const u32 orig_refcnt = orig_lte->out_refcnt;
-
- BUILD_BUG_ON(MAX_OPEN_STREAMS < ARRAY_LEN(orig_lte->inline_stream_owners));
-
- struct stream_owner *owners = orig_lte->stream_owners;
-
- /* Copy the stream's data from the temporary file to each of its
- * destinations.
- *
- * This is executed only in the very uncommon case that a
- * single-instance stream is being extracted to more than
- * MAX_OPEN_STREAMS locations! */
-
- memcpy(&tmpfile_lte, orig_lte, sizeof(struct wim_lookup_table_entry));
- tmpfile_lte.resource_location = RESOURCE_IN_FILE_ON_DISK;
- tmpfile_lte.file_on_disk = ctx->tmpfile_name;
- ret = 0;
- for (u32 i = 0; i < orig_refcnt; i++) {
-
- /* Note: it usually doesn't matter whether we pass the original
- * stream entry to callbacks provided by the extraction backend
- * as opposed to the tmpfile stream entry, since they shouldn't
- * actually read data from the stream other than through the
- * read_stream_prefix() call below. But for
- * WIMLIB_EXTRACT_FLAG_WIMBOOT mode on Windows it does matter
- * because it needs the original stream location in order to
- * create the external backing reference. */
-
- orig_lte->out_refcnt = 1;
- orig_lte->inline_stream_owners[0] = owners[i];
-
- ret = (*cbs->begin_stream)(orig_lte, cbs->begin_stream_ctx);
- if (ret)
- break;
- /* Extra SHA-1 isn't necessary here, but it shouldn't hurt as
- * this case is very rare anyway. */
- ret = extract_stream(&tmpfile_lte, tmpfile_lte.size,
- cbs->consume_chunk,
- cbs->consume_chunk_ctx);
+ 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;
- ret = (*cbs->end_stream)(orig_lte, ret, cbs->end_stream_ctx);
+ 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)
- break;
+ return ret;
}
- FREE(owners);
- orig_lte->out_refcnt = 0;
- return ret;
+ return 0;
}
static int
-end_extract_stream_wrapper(struct wim_lookup_table_entry *stream,
- int status, void *_ctx)
+end_extract_blob_wrapper(struct blob_descriptor *blob, int status, void *_ctx)
{
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, ctx);
+ 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;
- } else {
- return (*ctx->saved_cbs->end_stream)(stream, status,
- ctx->saved_cbs->end_stream_ctx);
}
+
+ return call_end_blob(blob, status, ctx->saved_cbs);
}
/*
- * Read the list of single-instance streams to extract and feed their data into
- * the specified callback functions.
+ * Read the list of blobs to extract and feed their data into the specified
+ * callback functions.
*
- * This handles checksumming each stream.
+ * 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, whereas attempting to
- * read streams directly (e.g. with read_full_stream_into_buf()) will not.
+ * This also works if the WIM is being read from a pipe.
*
- * This also will split up streams that will need to be extracted to more than
- * MAX_OPEN_STREAMS locations, as measured by the 'out_refcnt' of each stream.
+ * 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 nominal destination (e.g. Win32 currently might because the destination
- * file system might not support hard links).
+ * per 'blob_extraction_target' (e.g. Win32 currently might because the
+ * destination file system might not support hard links).
*/
int
-extract_stream_list(struct apply_ctx *ctx,
- const struct read_stream_list_callbacks *cbs)
+extract_blob_list(struct apply_ctx *ctx, const struct read_blob_callbacks *cbs)
{
- struct read_stream_list_callbacks wrapper_cbs = {
- .begin_stream = begin_extract_stream_wrapper,
- .begin_stream_ctx = ctx,
- .consume_chunk = extract_chunk_wrapper,
- .consume_chunk_ctx = ctx,
- .end_stream = end_extract_stream_wrapper,
- .end_stream_ctx = ctx,
+ 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 load_streams_from_pipe(ctx, &wrapper_cbs);
+ return read_blobs_from_pipe(ctx, &wrapper_cbs);
} else {
- return read_stream_list(&ctx->stream_list,
- offsetof(struct wim_lookup_table_entry,
- extraction_list),
- &wrapper_cbs, VERIFY_STREAM_HASHES);
+ return read_blob_list(&ctx->blob_list,
+ offsetof(struct blob_descriptor,
+ extraction_list),
+ &wrapper_cbs, VERIFY_BLOB_HASHES);
}
}
* unnamed data stream only. */
static int
extract_dentry_to_stdout(struct wim_dentry *dentry,
- const struct wim_lookup_table *lookup_table)
+ const struct blob_table *blob_table)
{
struct wim_inode *inode = dentry->d_inode;
- struct wim_lookup_table_entry *lte;
+ struct blob_descriptor *blob;
struct filedes _stdout;
if (inode->i_attributes & (FILE_ATTRIBUTE_REPARSE_POINT |
- FILE_ATTRIBUTE_DIRECTORY))
+ 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;
}
- lte = inode_unnamed_lte(inode, lookup_table);
- if (!lte) {
- const u8 *hash = inode_unnamed_stream_hash(inode);
+ 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 stream_not_found_error(inode, hash);
+ return blob_not_found_error(inode, hash);
return 0;
}
filedes_init(&_stdout, STDOUT_FILENO);
- return extract_full_stream_to_fd(lte, &_stdout);
+ return extract_blob_to_fd(blob, &_stdout);
}
static int
extract_dentries_to_stdout(struct wim_dentry **dentries, size_t num_dentries,
- const struct wim_lookup_table *lookup_table)
+ const struct blob_table *blob_table)
{
for (size_t i = 0; i < num_dentries; i++) {
- int ret = extract_dentry_to_stdout(dentries[i], lookup_table);
+ int ret = extract_dentry_to_stdout(dentries[i], blob_table);
if (ret)
return ret;
}
{
size_t i, j = 0;
for (i = 0; i < num_trees; i++) {
- if (!trees[i]->tmp_flag) {
+ if (!trees[i]->d_tmp_flag) {
/* Found distinct dentry. */
- trees[i]->tmp_flag = 1;
+ trees[i]->d_tmp_flag = 1;
trees[j++] = trees[i];
}
}
for (i = 0; i < j; i++)
- trees[i]->tmp_flag = 0;
+ trees[i]->d_tmp_flag = 0;
return j;
}
{
size_t i, j = 0;
for (i = 0; i < num_trees; i++)
- trees[i]->tmp_flag = 1;
+ trees[i]->d_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)
+ if (d->d_tmp_flag)
goto tree_contained;
}
trees[j++] = trees[i];
continue;
tree_contained:
- trees[i]->tmp_flag = 0;
+ trees[i]->d_tmp_flag = 0;
}
for (i = 0; i < j; i++)
- trees[i]->tmp_flag = 0;
+ trees[i]->d_tmp_flag = 0;
return j;
}
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);
+ if (will_extract_dentry(dentry)) {
+ list_del(&dentry->d_extraction_list_node);
+ dentry_reset_extraction_list_node(dentry);
+ }
return 0;
}
inode = dentry->d_inode;
dentry_reset_extraction_list_node(dentry);
inode->i_visited = 0;
- if ((void *)dentry->d_extraction_name != (void *)dentry->file_name)
+ inode->i_can_externally_back = 0;
+ if ((void *)dentry->d_extraction_name != (void *)dentry->d_name)
FREE(dentry->d_extraction_name);
dentry->d_extraction_name = NULL;
dentry->d_extraction_name_nchars = 0;
}
static void
-destroy_stream_list(struct list_head *stream_list)
+destroy_blob_list(struct list_head *blob_list)
{
- struct wim_lookup_table_entry *lte;
+ struct blob_descriptor *blob;
- list_for_each_entry(lte, stream_list, extraction_list)
- if (lte->out_refcnt > ARRAY_LEN(lte->inline_stream_owners))
- FREE(lte->stream_owners);
+ 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__
if (num_chars == 0)
return true;
for (i = 0; i < num_chars; i++) {
- switch (name[i]) {
+ switch (le16_to_cpu(name[i])) {
#ifdef __WIN32__
- case cpu_to_le16('\\'):
- case cpu_to_le16(':'):
- case cpu_to_le16('*'):
- case cpu_to_le16('?'):
- case cpu_to_le16('"'):
- case cpu_to_le16('<'):
- case cpu_to_le16('>'):
- case cpu_to_le16('|'):
+ case '\x01'...'\x1F':
+ case '\\':
+ case ':':
+ case '*':
+ case '?':
+ case '"':
+ case '<':
+ case '>':
+ case '|':
#endif
- case cpu_to_le16('/'):
- case cpu_to_le16('\0'):
+ case '/':
+ case '\0':
if (fix)
name[i] = replacement_char;
else
}
}
-#ifdef __WIN32__
- if (name[num_chars - 1] == cpu_to_le16(' ') ||
- name[num_chars - 1] == cpu_to_le16('.'))
- {
- if (fix)
- name[num_chars - 1] = replacement_char;
- else
- return false;
- }
-#endif
return true;
}
{
int ret;
- if (!dentry_is_supported(dentry, &ctx->supported_features))
- goto skip_dentry;
-
if (dentry_is_root(dentry))
return 0;
#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 /
+ dentry->d_extraction_name = dentry->d_name;
+ dentry->d_extraction_name_nchars = dentry->d_name_nbytes /
sizeof(utf16lechar);
return 0;
}
}
}
- if (file_name_valid(dentry->file_name, dentry->file_name_nbytes / 2, false)) {
- ret = utf16le_get_tstr(dentry->file_name,
- dentry->file_name_nbytes,
+ if (file_name_valid(dentry->d_name, dentry->d_name_nbytes / 2, false)) {
+ size_t nbytes = 0;
+ ret = utf16le_get_tstr(dentry->d_name,
+ dentry->d_name_nbytes,
(const tchar **)&dentry->d_extraction_name,
- &dentry->d_extraction_name_nchars);
- dentry->d_extraction_name_nchars /= sizeof(tchar);
+ &nbytes);
+ dentry->d_extraction_name_nchars = nbytes / sizeof(tchar);
return ret;
} else {
if (ctx->extract_flags & WIMLIB_EXTRACT_FLAG_REPLACE_INVALID_FILENAMES)
out_replace:
{
- utf16lechar utf16_name_copy[dentry->file_name_nbytes / 2];
+ utf16lechar utf16_name_copy[dentry->d_name_nbytes / 2];
- memcpy(utf16_name_copy, dentry->file_name, dentry->file_name_nbytes);
- file_name_valid(utf16_name_copy, dentry->file_name_nbytes / 2, true);
+ memcpy(utf16_name_copy, dentry->d_name, dentry->d_name_nbytes);
+ file_name_valid(utf16_name_copy, dentry->d_name_nbytes / 2, true);
const tchar *tchar_name;
size_t tchar_nchars;
ret = utf16le_get_tstr(utf16_name_copy,
- dentry->file_name_nbytes,
+ dentry->d_name_nbytes,
&tchar_name, &tchar_nchars);
if (ret)
return ret;
utf16le_put_tstr(tchar_name);
- dentry->d_extraction_name = memdup(fixed_name,
- 2 * fixed_name_num_chars + 2);
+ dentry->d_extraction_name = TSTRDUP(fixed_name);
if (!dentry->d_extraction_name)
return WIMLIB_ERR_NOMEM;
dentry->d_extraction_name_nchars = fixed_name_num_chars;
static int
dentry_resolve_streams(struct wim_dentry *dentry, int extract_flags,
- struct wim_lookup_table *lookup_table)
+ struct blob_table *blob_table)
{
struct wim_inode *inode = dentry->d_inode;
- struct wim_lookup_table_entry *lte;
+ struct blob_descriptor *blob;
int ret;
bool force = false;
- /* Special case: when extracting from a pipe, the WIM lookup table is
+ /* 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 streams,
- * identified by SHA1 message digests, need to be extracted, so we
- * "resolve" the inode's streams anyway by allocating new entries. */
+ * 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, lookup_table, force);
+ ret = inode_resolve_streams(inode, blob_table, force);
if (ret)
return ret;
- for (u32 i = 0; i <= inode->i_num_ads; i++) {
- lte = inode_stream_lte_resolved(inode, i);
- if (lte)
- lte->out_refcnt = 0;
+ 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;
}
/*
* For each dentry to be extracted, resolve all streams in the corresponding
- * inode and set 'out_refcnt' in each to 0.
+ * inode and set 'out_refcnt' in all referenced blob_descriptors to 0.
*
* Possible error codes: WIMLIB_ERR_RESOURCE_NOT_FOUND, WIMLIB_ERR_NOMEM.
*/
list_for_each_entry(dentry, dentry_list, d_extraction_list_node) {
ret = dentry_resolve_streams(dentry,
ctx->extract_flags,
- ctx->wim->lookup_table);
+ ctx->wim->blob_table);
if (ret)
return ret;
}
}
static int
-ref_stream(struct wim_lookup_table_entry *lte, u32 stream_idx,
- struct wim_dentry *dentry, struct apply_ctx *ctx)
+ref_stream(struct wim_inode_stream *strm, struct wim_dentry *dentry,
+ struct apply_ctx *ctx)
{
struct wim_inode *inode = dentry->d_inode;
- struct stream_owner *stream_owners;
+ struct blob_descriptor *blob = stream_blob_resolved(strm);
+ struct blob_extraction_target *targets;
- if (!lte)
+ if (!blob)
return 0;
- /* Tally the size only for each extraction of the stream (not hard
- * links). */
+ /* 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 += lte->size;
+ ctx->progress.extract.total_bytes += blob->size;
ctx->progress.extract.total_streams++;
if (inode->i_visited)
return 0;
- /* Add stream to the dentry_list only one time, even if it's going
- * to be extracted to multiple inodes. */
- if (lte->out_refcnt == 0) {
- list_add_tail(<e->extraction_list, &ctx->stream_list);
- ctx->num_streams_remaining++;
+ /* 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++;
}
- /* If inode not yet been visited, append it to the stream_owners array. */
- if (lte->out_refcnt < ARRAY_LEN(lte->inline_stream_owners)) {
- stream_owners = lte->inline_stream_owners;
+ /* 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 stream_owner *prev_stream_owners;
- size_t alloc_stream_owners;
+ struct blob_extraction_target *prev_targets;
+ size_t alloc_blob_extraction_targets;
- if (lte->out_refcnt == ARRAY_LEN(lte->inline_stream_owners)) {
- prev_stream_owners = NULL;
- alloc_stream_owners = ARRAY_LEN(lte->inline_stream_owners);
+ 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_stream_owners = lte->stream_owners;
- alloc_stream_owners = lte->alloc_stream_owners;
+ prev_targets = blob->blob_extraction_targets;
+ alloc_blob_extraction_targets = blob->alloc_blob_extraction_targets;
}
- if (lte->out_refcnt == alloc_stream_owners) {
- alloc_stream_owners *= 2;
- stream_owners = REALLOC(prev_stream_owners,
- alloc_stream_owners *
- sizeof(stream_owners[0]));
- if (!stream_owners)
+ 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_stream_owners) {
- memcpy(stream_owners,
- lte->inline_stream_owners,
- sizeof(lte->inline_stream_owners));
+ if (!prev_targets) {
+ memcpy(targets,
+ blob->inline_blob_extraction_targets,
+ sizeof(blob->inline_blob_extraction_targets));
}
- lte->stream_owners = stream_owners;
- lte->alloc_stream_owners = alloc_stream_owners;
+ blob->blob_extraction_targets = targets;
+ blob->alloc_blob_extraction_targets = alloc_blob_extraction_targets;
}
- stream_owners = lte->stream_owners;
+ targets = blob->blob_extraction_targets;
}
- stream_owners[lte->out_refcnt].inode = inode;
- if (stream_idx == 0) {
- stream_owners[lte->out_refcnt].stream_name = NULL;
- } else {
- stream_owners[lte->out_refcnt].stream_name =
- inode->i_ads_entries[stream_idx - 1].stream_name;
+ 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 from the WIM archive itself
+ */
+ if (ctx->apply_ops->will_back_from_wim) {
+ int ret = (*ctx->apply_ops->will_back_from_wim)(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;
}
- lte->out_refcnt++;
+ if (need_stream)
+ return ref_stream(strm, dentry, ctx);
return 0;
}
dentry_ref_streams(struct wim_dentry *dentry, struct apply_ctx *ctx)
{
struct wim_inode *inode = dentry->d_inode;
- int ret;
-
- /* The unnamed data stream will always be extracted, except in an
- * unlikely case. */
- if (!inode_is_encrypted_directory(inode)) {
- u16 stream_idx;
- struct wim_lookup_table_entry *stream;
-
- stream = inode_unnamed_stream_resolved(inode, &stream_idx);
- ret = ref_stream(stream, stream_idx, dentry, ctx);
+ 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;
}
-
- /* Named data streams will be extracted only if supported in the current
- * extraction mode and volume, and to avoid complications, if not doing
- * a linked extraction. */
- if (ctx->supported_features.named_data_streams) {
- for (u16 i = 0; i < inode->i_num_ads; i++) {
- if (!ads_entry_is_named_stream(&inode->i_ads_entries[i]))
- continue;
- ret = ref_stream(inode->i_ads_entries[i].lte, i + 1,
- dentry, ctx);
- if (ret)
- return ret;
- }
- }
inode->i_visited = 1;
return 0;
}
/*
- * For each dentry to be extracted, iterate through the data streams of the
- * corresponding inode. For each such stream that is not to be ignored due to
- * the supported features or extraction flags, add it to the list of streams to
- * be extracted (ctx->stream_list) if not already done so.
- *
- * Also builds a mapping from each stream to the inodes referencing it.
+ * 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.
*
- * This also initializes the extract progress info with byte and stream
+ * This also initializes the extract progress info with byte and blob
* information.
*
* ctx->supported_features must be filled in.
- *
- * Possible error codes: WIMLIB_ERR_NOMEM.
*/
static int
dentry_list_ref_streams(struct list_head *dentry_list, struct apply_ctx *ctx)
dentry_list_build_inode_alias_lists(struct list_head *dentry_list)
{
struct wim_dentry *dentry;
- struct wim_inode *inode;
+
+ list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
+ dentry->d_inode->i_first_extraction_alias = NULL;
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;
+ dentry->d_next_extraction_alias = dentry->d_inode->i_first_extraction_alias;
+ dentry->d_inode->i_first_extraction_alias = dentry;
}
- list_for_each_entry(dentry, dentry_list, d_extraction_list_node)
- dentry->d_inode->i_visited = 0;
}
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_READONLY)
+ features->readonly_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_ARCHIVE)
+ features->archive_files++;
if (inode->i_attributes & FILE_ATTRIBUTE_COMPRESSED)
features->compressed_files++;
if (inode->i_attributes & FILE_ATTRIBUTE_ENCRYPTED) {
features->not_context_indexed_files++;
if (inode->i_attributes & FILE_ATTRIBUTE_SPARSE_FILE)
features->sparse_files++;
- if (inode_has_named_stream(inode))
+ if (inode_has_named_data_stream(inode))
features->named_data_streams++;
if (inode->i_attributes & FILE_ATTRIBUTE_REPARSE_POINT) {
features->reparse_points++;
else
features->other_reparse_points++;
}
- if (inode->i_security_id != -1)
+ if (inode_has_security_descriptor(inode))
features->security_descriptors++;
if (inode_has_unix_data(inode))
features->unix_data++;
+ if (inode_has_object_id(inode))
+ features->object_ids++;
}
/* Tally features necessary to extract a dentry and the corresponding inode. */
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->readonly_files &&
+ !supported_features->readonly_files)
+ WARNING("Ignoring FILE_ATTRIBUTE_READONLY of %lu files",
+ required_features->readonly_files);
if (required_features->hidden_files &&
!supported_features->hidden_files)
WARNING("Ignoring FILE_ATTRIBUTE_SYSTEM of %lu files",
required_features->system_files);
+ /* 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->compressed_files &&
!supported_features->compressed_files)
WARNING("Ignoring FILE_ATTRIBUTE_COMPRESSED of %lu files",
required_features->encrypted_directories);
}
- /* Encrypted files. */
- if (required_features->encrypted_files &&
- !supported_features->encrypted_files)
- WARNING("Ignoring %lu encrypted 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);
-
/* Hard links. */
if (required_features->hard_links && !supported_features->hard_links)
WARNING("Extracting %lu hard links as independent files",
{
if (supported_features->symlink_reparse_points) {
if (required_features->other_reparse_points) {
- WARNING("Ignoring %lu non-symlink/junction "
- "reparse point files",
+ WARNING("Ignoring reparse data of %lu non-symlink/junction files",
required_features->other_reparse_points);
}
} else {
- WARNING("Ignoring %lu reparse point files",
+ WARNING("Ignoring reparse data of %lu files",
required_features->reparse_points);
}
}
required_features->unix_data);
}
+ /* Object IDs. */
+ if (required_features->object_ids && !supported_features->object_ids) {
+ WARNING("Ignoring object IDs of %lu files",
+ required_features->object_ids);
+ }
+
/* DOS Names. */
if (required_features->short_names &&
!supported_features->short_names)
if (extract_flags & WIMLIB_EXTRACT_FLAG_TO_STDOUT) {
ret = extract_dentries_to_stdout(trees, num_trees,
- wim->lookup_table);
+ wim->blob_table);
goto out;
}
wim->current_image);
ctx->progress.extract.target = target;
}
- INIT_LIST_HEAD(&ctx->stream_list);
+ INIT_LIST_HEAD(&ctx->blob_list);
filedes_invalidate(&ctx->tmpfile_fd);
+ ctx->apply_ops = ops;
ret = (*ops->get_supported_features)(target, &ctx->supported_features);
if (ret)
if (ret)
goto out_cleanup;
+ if (unlikely(list_empty(&dentry_list))) {
+ WARNING("There is nothing to extract!");
+ goto out_cleanup;
+ }
+
ret = dentry_list_resolve_streams(&dentry_list, ctx);
if (ret)
goto out_cleanup;
+ dentry_list_build_inode_alias_lists(&dentry_list);
+
ret = dentry_list_ref_streams(&dentry_list, ctx);
if (ret)
goto out_cleanup;
- dentry_list_build_inode_alias_lists(&dentry_list);
-
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
- * lookup table), since at this point all we have is a set of
- * SHA1 message digests of streams that need to be extracted.
+ * 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,
* 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);
+ xml_get_image_total_bytes(wim->xml_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);
+ xml_get_image_hard_link_bytes(wim->xml_info,
+ wim->current_image);
}
}
WIMLIB_PROGRESS_MSG_EXTRACT_IMAGE_END :
WIMLIB_PROGRESS_MSG_EXTRACT_TREE_END));
out_cleanup:
- destroy_stream_list(&ctx->stream_list);
+ destroy_blob_list(&ctx->blob_list);
destroy_dentry_list(&dentry_list);
FREE(ctx);
out:
#ifndef WITH_NTFS_3G
if (extract_flags & WIMLIB_EXTRACT_FLAG_NTFS) {
- ERROR("wimlib was compiled without support for NTFS-3g, so\n"
+ 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_COMPACT_XPRESS4K |
+ WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K |
+ WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K |
+ WIMLIB_EXTRACT_FLAG_COMPACT_LZX))
+ {
+ #ifdef __WIN32__
+ int count = 0;
+ count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS4K) != 0);
+ count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS8K) != 0);
+ count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_XPRESS16K) != 0);
+ count += ((extract_flags & WIMLIB_EXTRACT_FLAG_COMPACT_LZX) != 0);
+ if (count != 1) {
+ ERROR("Only one compression format can be specified "
+ "for compact-mode extraction!");
+ return WIMLIB_ERR_INVALID_PARAM;
+ }
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_WIMBOOT) {
+ ERROR("Compact-mode extraction and WIMBoot-mode "
+ "extraction are mutually exclusive!");
+ return WIMLIB_ERR_INVALID_PARAM;
+ }
+ #else
+ ERROR("Compact-mode extraction (System Compression) "
+ "is only supported on Windows!");
+ return WIMLIB_ERR_UNSUPPORTED;
+ #endif
+ }
+
if ((extract_flags & (WIMLIB_EXTRACT_FLAG_RPFIX |
WIMLIB_EXTRACT_FLAG_NORPFIX |
return 0;
}
-static u32
-get_wildcard_flags(int extract_flags)
-{
- u32 wildcard_flags = 0;
-
- 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;
-
- if (default_ignore_case)
- wildcard_flags |= WILDCARD_FLAG_CASE_INSENSITIVE;
-
- return wildcard_flags;
-}
-
struct append_dentry_ctx {
struct wim_dentry **dentries;
size_t num_dentries;
return 0;
}
+/* Append dentries matched by a path which can contain wildcard characters. */
+static int
+append_matched_dentries(WIMStruct *wim, const tchar *orig_pattern,
+ int extract_flags, struct append_dentry_ctx *ctx)
+{
+ const size_t count_before = ctx->num_dentries;
+ tchar *pattern;
+ int ret;
+
+ pattern = canonicalize_wim_path(orig_pattern);
+ if (!pattern)
+ return WIMLIB_ERR_NOMEM;
+ ret = expand_path_pattern(wim_get_current_root_dentry(wim), pattern,
+ append_dentry_cb, ctx);
+ FREE(pattern);
+ if (ret || ctx->num_dentries > count_before)
+ return ret;
+ if (extract_flags & WIMLIB_EXTRACT_FLAG_STRICT_GLOB) {
+ ERROR("No matches for path pattern \"%"TS"\"", orig_pattern);
+ return WIMLIB_ERR_PATH_DOES_NOT_EXIST;
+ }
+ WARNING("No matches for path pattern \"%"TS"\"", orig_pattern);
+ return 0;
+}
+
static int
do_wimlib_extract_paths(WIMStruct *wim, int image, const tchar *target,
const tchar * const *paths, size_t num_paths,
if (ret)
return ret;
- ret = wim_checksum_unhashed_streams(wim);
+ ret = wim_checksum_unhashed_blobs(wim);
if (ret)
return ret;
.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);
+ ret = append_matched_dentries(wim, paths[i],
+ extract_flags,
+ &append_dentry_ctx);
if (ret) {
trees = append_dentry_ctx.dentries;
goto out_free_trees;
return image_name && *image_name &&
!tstrpbrk(image_name, filename_forbidden_chars) &&
tstrcmp(image_name, T(".")) &&
- tstrcmp(image_name, T(".."));
+ tstrcmp(image_name, T("..")) &&
+ tstrlen(image_name) <= 128;
}
/* Extracts all images from the WIM to the directory @target, with the images
static int
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);
- tchar buf[output_path_len + 1 + image_name_max_len + 1];
+ tchar buf[output_path_len + 1 + 128 + 1];
int ret;
int image;
const tchar *image_name;
/* 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 lookup table, no XML data read, and no filename set. */
+ * 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)
* write_pipable_wim() for more details about the format of pipable
* WIMs.) */
{
- struct wim_lookup_table_entry xml_lte;
- struct wim_resource_spec xml_rspec;
- ret = read_pwm_stream_header(pwm, &xml_lte, &xml_rspec, 0, 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 (!(xml_lte.flags & WIM_RESHDR_FLAG_METADATA))
- {
- ERROR("Expected XML data, but found non-metadata "
- "stream.");
+ 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;
}
- wim_res_spec_to_hdr(&xml_rspec, &pwm->hdr.xml_data_reshdr);
-
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) {
+ if (xml_get_image_count(pwm->xml_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;
/* Load the needed metadata resource. */
for (i = 1; i <= pwm->hdr.image_count; i++) {
- struct wim_lookup_table_entry *metadata_lte;
- struct wim_image_metadata *imd;
- struct wim_resource_spec *metadata_rspec;
-
- metadata_lte = new_lookup_table_entry();
- if (metadata_lte == NULL) {
- ret = WIMLIB_ERR_NOMEM;
- goto out_wimlib_free;
- }
- metadata_rspec = MALLOC(sizeof(struct wim_resource_spec));
- if (metadata_rspec == NULL) {
- ret = WIMLIB_ERR_NOMEM;
- free_lookup_table_entry(metadata_lte);
- goto out_wimlib_free;
- }
-
- ret = read_pwm_stream_header(pwm, metadata_lte, metadata_rspec, 0, NULL);
- imd = pwm->image_metadata[i - 1];
- imd->metadata_lte = metadata_lte;
- if (ret) {
- FREE(metadata_rspec);
- goto out_wimlib_free;
- }
-
- if (!(metadata_lte->flags & WIM_RESHDR_FLAG_METADATA)) {
- ERROR("Expected metadata resource, but found "
- "non-metadata stream.");
- ret = WIMLIB_ERR_INVALID_PIPABLE_WIM;
+ ret = handle_pwm_metadata_resource(pwm, i, i == image);
+ if (ret)
goto out_wimlib_free;
- }
-
- if (i == image) {
- /* Metadata resource is for the image being extracted.
- * Parse it and save the metadata in memory. */
- ret = read_metadata_resource(pwm, 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_stream(metadata_lte);
- if (ret)
- goto out_wimlib_free;
- }
}
/* Extract the image. */
extract_flags |= WIMLIB_EXTRACT_FLAG_FROM_PIPE;
git://wimlib.net / wimlib / blobdiff