+ tmemcpy(fixed_name, tchar_name, tchar_nchars);
+ fixed_name_num_chars += tsprintf(fixed_name + tchar_nchars,
+ T(" (invalid filename #%lu)"),
+ ++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->d_extraction_name_nchars = fixed_name_num_chars;
+ }
+ return 0;
+
+skip_dentry:
+ 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_list_calculate_extraction_names(struct list_head *dentry_list,
+ struct apply_ctx *ctx)
+{
+ 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;
+}
+
+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;
+}
+
+/*
+ * For each dentry to be extracted, resolve all streams in the corresponding
+ * inode and set 'out_refcnt' in all referenced blob_descriptors to 0.
+ *
+ * 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 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;
+ }
+ 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.
+ *
+ * This also initializes the extract progress info with byte and blob
+ * information.
+ *
+ * ctx->supported_features must be filled in.
+ */
+static int
+dentry_list_ref_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_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;
+
+ 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) {
+ dentry->d_next_extraction_alias = dentry->d_inode->i_first_extraction_alias;
+ dentry->d_inode->i_first_extraction_alias = dentry;
+ }
+}
+
+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)
+ {
+ 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);
+ }
+ }
+
+ /* 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)
+ {
+ 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;
+ }
+
+ if (required_features->unix_data &&
+ !(extract_flags & WIMLIB_EXTRACT_FLAG_UNIX_DATA))
+ {
+ WARNING("Ignoring UNIX metadata of %lu files",
+ required_features->unix_data);
+ }
+
+ /* 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;
+ }
+ WARNING("Ignoring DOS names of %lu files",
+ required_features->short_names);