X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Flookup_table.c;h=e2ca5920745f84b733bdfb794928c4f5a516e821;hp=bfa0c858c4ddea8a84842f630377fb94a3c5ddc7;hb=4f8059f2d0a74a9922128b162d9c9343b305999c;hpb=650997e4865a090b6856c7ca34b02f42994e8e29

diff --git a/src/lookup_table.c b/src/lookup_table.c
index bfa0c858..e2ca5920 100644
--- a/src/lookup_table.c
+++ b/src/lookup_table.c
@@ -28,6 +28,7 @@
 #include "lookup_table.h"
 #include "buffer_io.h"
 #include <errno.h>
+#include <stdlib.h>
 
 #ifdef WITH_FUSE
 #include <unistd.h>
@@ -39,7 +40,7 @@ new_lookup_table(size_t capacity)
 	struct wim_lookup_table *table;
 	struct hlist_head *array;
 
-	table = MALLOC(sizeof(struct wim_lookup_table));
+	table = CALLOC(1, sizeof(struct wim_lookup_table));
 	if (table) {
 		array = CALLOC(capacity, sizeof(array[0]));
 		if (array) {
@@ -49,8 +50,8 @@ new_lookup_table(size_t capacity)
 		} else {
 			FREE(table);
 			table = NULL;
-			ERROR("Failed to allocate memory for lookup table with capacity %zu",
-			      capacity);
+			ERROR("Failed to allocate memory for lookup table "
+			      "with capacity %zu", capacity);
 		}
 	}
 	return table;
@@ -87,13 +88,17 @@ clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
 	switch (new->resource_location) {
 #ifdef __WIN32__
 	case RESOURCE_WIN32:
+	case RESOURCE_WIN32_ENCRYPTED:
+#else
+	case RESOURCE_IN_FILE_ON_DISK:
 #endif
+#ifdef WITH_FUSE
 	case RESOURCE_IN_STAGING_FILE:
-	case RESOURCE_IN_FILE_ON_DISK:
 		BUILD_BUG_ON((void*)&old->file_on_disk !=
 			     (void*)&old->staging_file_name);
-		new->staging_file_name = TSTRDUP(old->staging_file_name);
-		if (!new->staging_file_name)
+#endif
+		new->file_on_disk = TSTRDUP(old->file_on_disk);
+		if (!new->file_on_disk)
 			goto out_free;
 		break;
 	case RESOURCE_IN_ATTACHED_BUFFER:
@@ -140,14 +145,18 @@ free_lookup_table_entry(struct wim_lookup_table_entry *lte)
 {
 	if (lte) {
 		switch (lte->resource_location) {
-		case RESOURCE_IN_STAGING_FILE:
-		case RESOURCE_IN_ATTACHED_BUFFER:
-		case RESOURCE_IN_FILE_ON_DISK:
-#ifdef __WIN32__
+	#ifdef __WIN32__
 		case RESOURCE_WIN32:
-#endif
+		case RESOURCE_WIN32_ENCRYPTED:
+	#else
+		case RESOURCE_IN_FILE_ON_DISK:
+	#endif
+	#ifdef WITH_FUSE
+		case RESOURCE_IN_STAGING_FILE:
 			BUILD_BUG_ON((void*)&lte->file_on_disk !=
 				     (void*)&lte->staging_file_name);
+	#endif
+		case RESOURCE_IN_ATTACHED_BUFFER:
 			BUILD_BUG_ON((void*)&lte->file_on_disk !=
 				     (void*)&lte->attached_buffer);
 			FREE(lte->file_on_disk);
@@ -214,7 +223,7 @@ finalize_lte(struct wim_lookup_table_entry *lte)
 	#ifdef WITH_FUSE
 	if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
 		unlink(lte->staging_file_name);
-		list_del(&lte->staging_list);
+		list_del(&lte->unhashed_list);
 	}
 	#endif
 	free_lookup_table_entry(lte);
@@ -231,7 +240,8 @@ lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
 	wimlib_assert(lte != NULL);
 	wimlib_assert(lte->refcnt != 0);
 	if (--lte->refcnt == 0) {
-		lookup_table_unlink(table, lte);
+		if (!lte->unhashed)
+			lookup_table_unlink(table, lte);
 	#ifdef WITH_FUSE
 		if (lte->num_opened_fds == 0)
 	#endif
@@ -264,73 +274,141 @@ for_lookup_table_entry(struct wim_lookup_table *table,
 		hlist_for_each_entry_safe(lte, pos, tmp, &table->array[i],
 					  hash_list)
 		{
+			wimlib_assert2(!(lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA));
 			ret = visitor(lte, arg);
-			if (ret != 0)
+			if (ret)
 				return ret;
 		}
 	}
 	return 0;
 }
 
+int
+cmp_streams_by_wim_position(const void *p1, const void *p2)
+{
+	const struct wim_lookup_table_entry *lte1, *lte2;
+	lte1 = *(const struct wim_lookup_table_entry**)p1;
+	lte2 = *(const struct wim_lookup_table_entry**)p2;
+	if (lte1->resource_entry.offset < lte2->resource_entry.offset)
+		return -1;
+	else if (lte1->resource_entry.offset > lte2->resource_entry.offset)
+		return 1;
+	else
+		return 0;
+}
+
+
+static int
+add_lte_to_array(struct wim_lookup_table_entry *lte,
+		 void *_pp)
+{
+	struct wim_lookup_table_entry ***pp = _pp;
+	*(*pp)++ = lte;
+	return 0;
+}
+
+/* Iterate through the lookup table entries, but first sort them by stream
+ * offset in the WIM.  Caution: this is intended to be used when the stream
+ * offset field has actually been set. */
+int
+for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
+				  int (*visitor)(struct wim_lookup_table_entry *,
+						 void *),
+				  void *arg)
+{
+	struct wim_lookup_table_entry **lte_array, **p;
+	size_t num_streams = table->num_entries;
+	int ret;
+
+	lte_array = MALLOC(num_streams * sizeof(lte_array[0]));
+	if (!lte_array)
+		return WIMLIB_ERR_NOMEM;
+	p = lte_array;
+	for_lookup_table_entry(table, add_lte_to_array, &p);
+
+	wimlib_assert(p == lte_array + num_streams);
+
+	qsort(lte_array, num_streams, sizeof(lte_array[0]),
+	      cmp_streams_by_wim_position);
+	ret = 0;
+	for (size_t i = 0; i < num_streams; i++) {
+		ret = visitor(lte_array[i], arg);
+		if (ret)
+			break;
+	}
+	FREE(lte_array);
+	return ret;
+}
 
 /*
  * Reads the lookup table from a WIM file.
+ *
+ * Saves lookup table entries for non-metadata streams in a hash table, and
+ * saves the metadata entry for each image in a special per-image location (the
+ * image_metadata array).
  */
 int
 read_lookup_table(WIMStruct *w)
 {
-	u64 num_entries;
-	u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 	int ret;
+	size_t num_entries;
 	struct wim_lookup_table *table;
-	struct wim_lookup_table_entry *cur_entry = NULL, *duplicate_entry;
-
-	if (resource_is_compressed(&w->hdr.lookup_table_res_entry)) {
-		ERROR("Didn't expect a compressed lookup table!");
-		ERROR("Ask the author to implement support for this.");
-		return WIMLIB_ERR_COMPRESSED_LOOKUP_TABLE;
-	}
+	struct wim_lookup_table_entry *cur_entry, *duplicate_entry;
+	u8 table_buf[(BUFFER_SIZE / WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE) *
+			WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
+	const u8 *p;
+	off_t offset;
+	size_t buf_entries_remaining;
 
 	DEBUG("Reading lookup table: offset %"PRIu64", size %"PRIu64"",
 	      w->hdr.lookup_table_res_entry.offset,
 	      w->hdr.lookup_table_res_entry.original_size);
 
-	if (fseeko(w->fp, w->hdr.lookup_table_res_entry.offset, SEEK_SET) != 0)
-	{
-		ERROR_WITH_ERRNO("Failed to seek to byte %"PRIu64" to read "
-				 "lookup table",
-				 w->hdr.lookup_table_res_entry.offset);
-		return WIMLIB_ERR_READ;
+	if (resource_is_compressed(&w->hdr.lookup_table_res_entry)) {
+		ERROR("Didn't expect a compressed lookup table!");
+		ERROR("Ask the author to implement support for this.");
+		return WIMLIB_ERR_COMPRESSED_LOOKUP_TABLE;
 	}
 
-	num_entries = w->hdr.lookup_table_res_entry.original_size /
+	num_entries = w->hdr.lookup_table_res_entry.size /
 		      WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE;
 	table = new_lookup_table(num_entries * 2 + 1);
 	if (!table)
 		return WIMLIB_ERR_NOMEM;
 
-	while (num_entries--) {
-		const u8 *p;
-
-		if (fread(buf, 1, sizeof(buf), w->fp) != sizeof(buf)) {
-			if (feof(w->fp)) {
-				ERROR("Unexpected EOF in WIM lookup table!");
-			} else {
-				ERROR_WITH_ERRNO("Error reading WIM lookup "
-						 "table");
+	w->current_image = 0;
+	offset = w->hdr.lookup_table_res_entry.offset;
+	buf_entries_remaining = 0;
+	for (; num_entries != 0; num_entries--, buf_entries_remaining--) {
+		if (buf_entries_remaining == 0) {
+			size_t entries_to_read, bytes_to_read;
+
+			entries_to_read = min(sizeof(table_buf) /
+						WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE,
+					      num_entries);
+			bytes_to_read = entries_to_read *
+						WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE;
+			if (full_pread(w->in_fd, table_buf,
+				       bytes_to_read, offset) != bytes_to_read)
+			{
+				ERROR_WITH_ERRNO("Error reading lookup table "
+						 "(offset=%"PRIu64")", offset);
+				ret = WIMLIB_ERR_READ;
+				goto out_free_lookup_table;
 			}
-			ret = WIMLIB_ERR_READ;
-			goto out;
+			offset += bytes_to_read;
+			p = table_buf;
+			buf_entries_remaining = entries_to_read;
 		}
 		cur_entry = new_lookup_table_entry();
 		if (!cur_entry) {
 			ret = WIMLIB_ERR_NOMEM;
-			goto out;
+			goto out_free_lookup_table;
 		}
+
 		cur_entry->wim = w;
 		cur_entry->resource_location = RESOURCE_IN_WIM;
-
-		p = get_resource_entry(buf, &cur_entry->resource_entry);
+		p = get_resource_entry(p, &cur_entry->resource_entry);
 		p = get_u16(p, &cur_entry->part_number);
 		p = get_u32(p, &cur_entry->refcnt);
 		p = get_bytes(p, SHA1_HASH_SIZE, cur_entry->hash);
@@ -350,27 +428,6 @@ read_lookup_table(WIMStruct *w)
 			goto out_free_cur_entry;
 		}
 
-		/* Ordinarily, no two streams should share the same SHA1 message
-		 * digest.  However, this constraint can be broken for metadata
-		 * resources--- two identical images will have the same metadata
-		 * resource, but their lookup table entries are not shared. */
-		duplicate_entry = __lookup_resource(table, cur_entry->hash);
-		if (duplicate_entry
-		    && !((duplicate_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA)
-			  && cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA))
-		{
-		#ifdef ENABLE_ERROR_MESSAGES
-			ERROR("The WIM lookup table contains two entries with the "
-			      "same SHA1 message digest!");
-			ERROR("The first entry is:");
-			print_lookup_table_entry(duplicate_entry, stderr);
-			ERROR("The second entry is:");
-			print_lookup_table_entry(cur_entry, stderr);
-		#endif
-			ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
-			goto out_free_cur_entry;
-		}
-
 		if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 		    && (cur_entry->resource_entry.size !=
 		        cur_entry->resource_entry.original_size))
@@ -384,109 +441,204 @@ read_lookup_table(WIMStruct *w)
 			ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 			goto out_free_cur_entry;
 		}
-		if ((cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA)
-		    && cur_entry->refcnt != 1)
-		{
-		#ifdef ENABLE_ERROR_MESSAGES
-			ERROR("Found metadata resource with refcnt != 1:");
-			print_lookup_table_entry(cur_entry, stderr);
-		#endif
-			ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
-			goto out_free_cur_entry;
+
+		if (cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_METADATA) {
+			/* Lookup table entry for a metadata resource */
+			if (cur_entry->refcnt != 1) {
+			#ifdef ENABLE_ERROR_MESSAGES
+				ERROR("Found metadata resource with refcnt != 1:");
+				print_lookup_table_entry(cur_entry, stderr);
+			#endif
+				ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
+				goto out_free_cur_entry;
+			}
+
+			if (w->hdr.part_number != 1) {
+				ERROR("Found a metadata resource in a "
+				      "non-first part of the split WIM!");
+				ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
+				goto out_free_cur_entry;
+			}
+			if (w->current_image == w->hdr.image_count) {
+				ERROR("The WIM header says there are %u images "
+				      "in the WIM, but we found more metadata "
+				      "resources than this", w->hdr.image_count);
+				ret = WIMLIB_ERR_IMAGE_COUNT;
+				goto out_free_cur_entry;
+			}
+
+			/* Notice very carefully:  We are assigning the metadata
+			 * resources in the exact order mirrored by their lookup
+			 * table entries on disk, which is the behavior of
+			 * Microsoft's software.  In particular, this overrides
+			 * the actual locations of the metadata resources
+			 * themselves in the WIM file as well as any information
+			 * written in the XML data. */
+			DEBUG("Found metadata resource for image %u at "
+			      "offset %"PRIu64".",
+			      w->current_image + 1,
+			      cur_entry->resource_entry.offset);
+			w->image_metadata[
+				w->current_image++]->metadata_lte = cur_entry;
+		} else {
+			/* Lookup table entry for a stream that is not a
+			 * metadata resource */
+			duplicate_entry = __lookup_resource(table, cur_entry->hash);
+			if (duplicate_entry) {
+			#ifdef ENABLE_ERROR_MESSAGES
+				ERROR("The WIM lookup table contains two entries with the "
+				      "same SHA1 message digest!");
+				ERROR("The first entry is:");
+				print_lookup_table_entry(duplicate_entry, stderr);
+				ERROR("The second entry is:");
+				print_lookup_table_entry(cur_entry, stderr);
+			#endif
+				ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
+				goto out_free_cur_entry;
+			}
+			lookup_table_insert(table, cur_entry);
 		}
-		lookup_table_insert(table, cur_entry);
+	}
 
+	if (w->hdr.part_number == 1 && w->current_image != w->hdr.image_count)
+	{
+		ERROR("The WIM header says there are %u images "
+		      "in the WIM, but we only found %d metadata "
+		      "resources!", w->hdr.image_count, w->current_image);
+		ret = WIMLIB_ERR_IMAGE_COUNT;
+		goto out_free_lookup_table;
 	}
 	DEBUG("Done reading lookup table.");
 	w->lookup_table = table;
-	return 0;
+	ret = 0;
+	goto out;
 out_free_cur_entry:
 	FREE(cur_entry);
-out:
+out_free_lookup_table:
 	free_lookup_table(table);
+out:
+	w->current_image = 0;
 	return ret;
 }
 
 
-/*
- * Writes a lookup table entry to the output file.
- */
-int
-write_lookup_table_entry(struct wim_lookup_table_entry *lte, void *__out)
+static u8 *
+write_lookup_table_entry(struct wim_lookup_table_entry *lte, u8 *buf_p)
 {
-	FILE *out;
-	u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
-	u8 *p;
-
-	out = __out;
-
-	/* Don't write entries that have not had file resources or metadata
-	 * resources written for them. */
-	if (lte->out_refcnt == 0)
-		return 0;
-
-	if (lte->output_resource_entry.flags & WIM_RESHDR_FLAG_METADATA) {
-		DEBUG("Writing metadata entry at %"PRIu64" "
-		      "(orig size = %"PRIu64")",
-		      ftello(out), lte->output_resource_entry.original_size);
-	}
-
-	p = put_resource_entry(buf, &lte->output_resource_entry);
-	p = put_u16(p, lte->part_number);
-	p = put_u32(p, lte->out_refcnt);
-	p = put_bytes(p, SHA1_HASH_SIZE, lte->hash);
-	if (fwrite(buf, 1, sizeof(buf), out) != sizeof(buf)) {
-		ERROR_WITH_ERRNO("Failed to write lookup table entry");
-		return WIMLIB_ERR_WRITE;
-	}
-	return 0;
+	buf_p = put_resource_entry(buf_p, &lte->output_resource_entry);
+	buf_p = put_u16(buf_p, lte->part_number);
+	buf_p = put_u32(buf_p, lte->out_refcnt);
+	buf_p = put_bytes(buf_p, SHA1_HASH_SIZE, lte->hash);
+	return buf_p;
 }
 
-/* Writes the lookup table to the output file. */
 int
-write_lookup_table(struct wim_lookup_table *table, FILE *out,
-		   struct resource_entry *out_res_entry)
+write_lookup_table_from_stream_list(struct list_head *stream_list,
+				    int out_fd,
+				    struct resource_entry *out_res_entry)
 {
-	off_t start_offset, end_offset;
 	int ret;
+	off_t start_offset;
+	u8 table_buf[(BUFFER_SIZE / WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE) *
+			WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
+	u8 *buf_p;
+	size_t table_size;
+	size_t bytes_to_write;
+	struct wim_lookup_table_entry *lte;
 
-	start_offset = ftello(out);
+	start_offset = filedes_offset(out_fd);
 	if (start_offset == -1)
-		return WIMLIB_ERR_WRITE;
-
-	ret = for_lookup_table_entry(table, write_lookup_table_entry, out);
-	if (ret != 0)
-		return ret;
-
-	end_offset = ftello(out);
-	if (end_offset == -1)
-		return WIMLIB_ERR_WRITE;
-
+		goto write_error;
+
+	buf_p = table_buf;
+	table_size = 0;
+	list_for_each_entry(lte, stream_list, lookup_table_list) {
+		if (buf_p == table_buf + sizeof(table_buf)) {
+			bytes_to_write = sizeof(table_buf);
+			if (full_write(out_fd, table_buf,
+				       bytes_to_write) != bytes_to_write)
+				goto write_error;
+			table_size += bytes_to_write;
+			buf_p = table_buf;
+		}
+		buf_p = write_lookup_table_entry(lte, buf_p);
+	}
+	bytes_to_write = buf_p - table_buf;
+	if (bytes_to_write != 0) {
+		if (full_write(out_fd, table_buf,
+			       bytes_to_write) != bytes_to_write)
+			goto write_error;
+		table_size += bytes_to_write;
+	}
 	out_res_entry->offset        = start_offset;
-	out_res_entry->size          = end_offset - start_offset;
-	out_res_entry->original_size = end_offset - start_offset;
+	out_res_entry->size          = table_size;
+	out_res_entry->original_size = table_size;
 	out_res_entry->flags         = WIM_RESHDR_FLAG_METADATA;
+	ret = 0;
+out:
+	return ret;
+write_error:
+	ERROR_WITH_ERRNO("Failed to write lookup table");
+	ret = WIMLIB_ERR_WRITE;
+	goto out;
+}
 
+static int
+append_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_list)
+{
+	if (lte->out_refcnt != 0)
+		list_add_tail(&lte->lookup_table_list, (struct list_head*)_list);
 	return 0;
 }
 
+/* Writes the WIM lookup table to the output file. */
+int
+write_lookup_table(WIMStruct *w, int image, struct resource_entry *out_res_entry)
+{
+	LIST_HEAD(stream_list);
+	int start_image;
+	int end_image;
+
+	if (image == WIMLIB_ALL_IMAGES) {
+		start_image = 1;
+		end_image = w->hdr.image_count;
+	} else {
+		start_image = image;
+		end_image = image;
+	}
+
+	for (int i = start_image; i <= end_image; i++) {
+		struct wim_lookup_table_entry *metadata_lte;
+
+		metadata_lte = w->image_metadata[i - 1]->metadata_lte;
+		metadata_lte->out_refcnt = 1;
+		metadata_lte->output_resource_entry.flags |= WIM_RESHDR_FLAG_METADATA;
+		append_lookup_table_entry(metadata_lte, &stream_list);
+	}
+	for_lookup_table_entry(w->lookup_table,
+			       append_lookup_table_entry,
+			       &stream_list);
+	return write_lookup_table_from_stream_list(&stream_list,
+						   w->out_fd,
+						   out_res_entry);
+}
 
 int
-lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *ignore)
+lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 {
 	lte->real_refcnt = 0;
 	return 0;
 }
 
 int
-lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *ignore)
+lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 {
 	lte->out_refcnt = 0;
 	return 0;
 }
 
 int
-lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *ignore)
+lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 {
 	if (lte->extracted_file != NULL) {
 		FREE(lte->extracted_file);
@@ -514,9 +666,14 @@ print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 	tfprintf(out, T("Part Number       = %hu\n"), lte->part_number);
 	tfprintf(out, T("Reference Count   = %u\n"), lte->refcnt);
 
-	tfprintf(out, T("Hash              = 0x"));
-	print_hash(lte->hash);
-	tputc(T('\n'), out);
+	if (lte->unhashed) {
+		tfprintf(out, T("(Unhashed: inode %p, stream_id = %u)\n"),
+			 lte->back_inode, lte->back_stream_id);
+	} else {
+		tfprintf(out, T("Hash              = 0x"));
+		print_hash(lte->hash, out);
+		tputc(T('\n'), out);
+	}
 
 	tfprintf(out, T("Flags             = "));
 	u8 flags = lte->resource_entry.flags;
@@ -538,15 +695,19 @@ print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 		break;
 #ifdef __WIN32__
 	case RESOURCE_WIN32:
-#endif
+	case RESOURCE_WIN32_ENCRYPTED:
+#else
 	case RESOURCE_IN_FILE_ON_DISK:
+#endif
 		tfprintf(out, T("File on Disk      = `%"TS"'\n"),
 			 lte->file_on_disk);
 		break;
+#ifdef WITH_FUSE
 	case RESOURCE_IN_STAGING_FILE:
 		tfprintf(out, T("Staging File      = `%"TS"'\n"),
 				lte->staging_file_name);
 		break;
+#endif
 	default:
 		break;
 	}
@@ -727,6 +888,34 @@ inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
 		return inode_stream_lte_unresolved(inode, stream_idx, table);
 }
 
+struct wim_lookup_table_entry *
+inode_unnamed_lte_resolved(const struct wim_inode *inode)
+{
+	wimlib_assert(inode->i_resolved);
+	for (unsigned i = 0; i <= inode->i_num_ads; i++) {
+		if (inode_stream_name_nbytes(inode, i) == 0 &&
+		    !is_zero_hash(inode_stream_hash_resolved(inode, i)))
+		{
+			return inode_stream_lte_resolved(inode, i);
+		}
+	}
+	return NULL;
+}
+
+struct wim_lookup_table_entry *
+inode_unnamed_lte_unresolved(const struct wim_inode *inode,
+			     const struct wim_lookup_table *table)
+{
+	wimlib_assert(!inode->i_resolved);
+	for (unsigned i = 0; i <= inode->i_num_ads; i++) {
+		if (inode_stream_name_nbytes(inode, i) == 0 &&
+		    !is_zero_hash(inode_stream_hash_unresolved(inode, i)))
+		{
+			return inode_stream_lte_unresolved(inode, i, table);
+		}
+	}
+	return NULL;
+}
 
 /* Return the lookup table entry for the unnamed data stream of an inode, or
  * NULL if there is none.
@@ -768,3 +957,79 @@ lookup_table_total_stream_size(struct wim_lookup_table *table)
 	for_lookup_table_entry(table, lte_add_stream_size, &total_size);
 	return total_size;
 }
+
+struct wim_lookup_table_entry **
+retrieve_lte_pointer(struct wim_lookup_table_entry *lte)
+{
+	wimlib_assert(lte->unhashed);
+	struct wim_inode *inode = lte->back_inode;
+	u32 stream_id = lte->back_stream_id;
+	if (stream_id == 0)
+		return &inode->i_lte;
+	else
+		for (u16 i = 0; i < inode->i_num_ads; i++)
+			if (inode->i_ads_entries[i].stream_id == stream_id)
+				return &inode->i_ads_entries[i].lte;
+	wimlib_assert(0);
+	return NULL;
+}
+
+/* Calculate the SHA1 message digest of a stream and move it from the list of
+ * unhashed streams to the stream lookup table, possibly joining it with an
+ * existing lookup table entry for an identical stream.
+ *
+ * @lte:  An unhashed lookup table entry.
+ * @lookup_table:  Lookup table for the WIM.
+ * @lte_ret:  On success, write a pointer to the resulting lookup table
+ *            entry to this location.  This will be the same as @lte
+ *            if it was inserted into the lookup table, or different if
+ *            a duplicate stream was found.
+ *
+ * Returns 0 on success; nonzero if there is an error reading the stream.
+ */
+int
+hash_unhashed_stream(struct wim_lookup_table_entry *lte,
+		     struct wim_lookup_table *lookup_table,
+		     struct wim_lookup_table_entry **lte_ret)
+{
+	int ret;
+	struct wim_lookup_table_entry *duplicate_lte;
+	struct wim_lookup_table_entry **back_ptr;
+
+	wimlib_assert(lte->unhashed);
+
+	/* back_ptr must be saved because @back_inode and @back_stream_id are in
+	 * union with the SHA1 message digest and will no longer be valid once
+	 * the SHA1 has been calculated. */
+	back_ptr = retrieve_lte_pointer(lte);
+
+	ret = sha1_resource(lte);
+	if (ret)
+		return ret;
+
+	/* Look for a duplicate stream */
+	duplicate_lte = __lookup_resource(lookup_table, lte->hash);
+	list_del(&lte->unhashed_list);
+	if (duplicate_lte) {
+		/* We have a duplicate stream.  Transfer the reference counts
+		 * from this stream to the duplicate, update the reference to
+		 * this stream (in an inode or ads_entry) to point to the
+		 * duplicate, then free this stream. */
+		wimlib_assert(!(duplicate_lte->unhashed));
+		duplicate_lte->refcnt += lte->refcnt;
+		duplicate_lte->out_refcnt += lte->refcnt;
+		*back_ptr = duplicate_lte;
+		free_lookup_table_entry(lte);
+		lte = duplicate_lte;
+	} else {
+		/* No duplicate stream, so we need to insert
+		 * this stream into the lookup table and treat
+		 * it as a hashed stream. */
+		lookup_table_insert(lookup_table, lte);
+		lte->unhashed = 0;
+	}
+	if (lte_ret)
+		*lte_ret = lte;
+	return 0;
+}
+