X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Flookup_table.c;h=f7fb7c3a5dff9ee263e28b5c748798adc437d1f6;hp=e2ca5920745f84b733bdfb794928c4f5a516e821;hb=df1050ef86e54f027912a15fe60bf67c8ab43d2c;hpb=4f8059f2d0a74a9922128b162d9c9343b305999c

diff --git a/src/lookup_table.c b/src/lookup_table.c
index e2ca5920..f7fb7c3a 100644
--- a/src/lookup_table.c
+++ b/src/lookup_table.c
@@ -2,7 +2,7 @@
  * lookup_table.c
  *
  * Lookup table, implemented as a hash table, that maps SHA1 message digests to
- * data streams.
+ * data streams; plus code to read and write the corresponding on-disk data.
  */
 
 /*
@@ -24,14 +24,25 @@
  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  */
 
-#include "wimlib_internal.h"
-#include "lookup_table.h"
-#include "buffer_io.h"
+#ifdef HAVE_CONFIG_H
+#  include "config.h"
+#endif
+
+#include "wimlib/endianness.h"
+#include "wimlib/error.h"
+#include "wimlib/file_io.h"
+#include "wimlib/glob.h"
+#include "wimlib/lookup_table.h"
+#include "wimlib/metadata.h"
+#include "wimlib/paths.h"
+#include "wimlib/resource.h"
+#include "wimlib/util.h"
+#include "wimlib/write.h"
+
 #include <errno.h>
 #include <stdlib.h>
-
 #ifdef WITH_FUSE
-#include <unistd.h>
+#  include <unistd.h> /* for unlink() */
 #endif
 
 struct wim_lookup_table *
@@ -58,7 +69,7 @@ new_lookup_table(size_t capacity)
 }
 
 struct wim_lookup_table_entry *
-new_lookup_table_entry()
+new_lookup_table_entry(void)
 {
 	struct wim_lookup_table_entry *lte;
 
@@ -79,18 +90,15 @@ clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
 {
 	struct wim_lookup_table_entry *new;
 
-	new = MALLOC(sizeof(*new));
+	new = memdup(old, sizeof(struct wim_lookup_table_entry));
 	if (!new)
 		return NULL;
 
-	memcpy(new, old, sizeof(*old));
 	new->extracted_file = NULL;
 	switch (new->resource_location) {
+	case RESOURCE_IN_FILE_ON_DISK:
 #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:
@@ -102,32 +110,30 @@ clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
 			goto out_free;
 		break;
 	case RESOURCE_IN_ATTACHED_BUFFER:
-		new->attached_buffer = MALLOC(wim_resource_size(old));
+		new->attached_buffer = memdup(old->attached_buffer,
+					      wim_resource_size(old));
 		if (!new->attached_buffer)
 			goto out_free;
-		memcpy(new->attached_buffer, old->attached_buffer,
-		       wim_resource_size(old));
 		break;
 #ifdef WITH_NTFS_3G
 	case RESOURCE_IN_NTFS_VOLUME:
 		if (old->ntfs_loc) {
 			struct ntfs_location *loc;
-			loc = MALLOC(sizeof(*loc));
+			loc = memdup(old->ntfs_loc, sizeof(struct ntfs_location));
 			if (!loc)
 				goto out_free;
-			memcpy(loc, old->ntfs_loc, sizeof(*loc));
 			loc->path = NULL;
 			loc->stream_name = NULL;
 			new->ntfs_loc = loc;
 			loc->path = STRDUP(old->ntfs_loc->path);
 			if (!loc->path)
 				goto out_free;
-			loc->stream_name = MALLOC((loc->stream_name_nchars + 1) * 2);
-			if (!loc->stream_name)
-				goto out_free;
-			memcpy(loc->stream_name,
-			       old->ntfs_loc->stream_name,
-			       (loc->stream_name_nchars + 1) * 2);
+			if (loc->stream_name_nchars) {
+				loc->stream_name = memdup(old->ntfs_loc->stream_name,
+							  loc->stream_name_nchars * 2);
+				if (!loc->stream_name)
+					goto out_free;
+			}
 		}
 		break;
 #endif
@@ -145,11 +151,9 @@ free_lookup_table_entry(struct wim_lookup_table_entry *lte)
 {
 	if (lte) {
 		switch (lte->resource_location) {
+		case RESOURCE_IN_FILE_ON_DISK:
 	#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:
@@ -240,7 +244,9 @@ lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
 	wimlib_assert(lte != NULL);
 	wimlib_assert(lte->refcnt != 0);
 	if (--lte->refcnt == 0) {
-		if (!lte->unhashed)
+		if (lte->unhashed)
+			list_del(&lte->unhashed_list);
+		else
 			lookup_table_unlink(table, lte);
 	#ifdef WITH_FUSE
 		if (lte->num_opened_fds == 0)
@@ -283,18 +289,104 @@ for_lookup_table_entry(struct wim_lookup_table *table,
 	return 0;
 }
 
-int
-cmp_streams_by_wim_position(const void *p1, const void *p2)
+/* qsort() callback that sorts streams (represented by `struct
+ * wim_lookup_table_entry's) into an order optimized for reading and writing.
+ *
+ * Sorting is done primarily by resource location, then secondarily by a
+ * per-resource location order.  For example, resources in WIM files are sorted
+ * primarily by part number, then secondarily by offset, as to implement optimal
+ * reading of either a standalone or split WIM.  */
+static int
+cmp_streams_by_sequential_order(const void *p1, const void *p2)
 {
 	const struct wim_lookup_table_entry *lte1, *lte2;
+	int v;
+
 	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
+
+	v = (int)lte1->resource_location - (int)lte2->resource_location;
+
+	/* Different resource locations?  */
+	if (v)
+		return v;
+
+	switch (lte1->resource_location) {
+	case RESOURCE_IN_WIM:
+
+		/* Different (possibly split) WIMs?  */
+		if (lte1->wim != lte2->wim) {
+			v = memcmp(lte1->wim->hdr.guid, lte2->wim->hdr.guid,
+				   WIM_GID_LEN);
+			if (v)
+				return v;
+		}
+
+		/* Different part numbers in the same WIM?  */
+		v = (int)lte1->wim->hdr.part_number - (int)lte2->wim->hdr.part_number;
+		if (v)
+			return v;
+
+		/* Compare by offset.  */
+		if (lte1->resource_entry.offset < lte2->resource_entry.offset)
+			return -1;
+		else if (lte1->resource_entry.offset > lte2->resource_entry.offset)
+			return 1;
+		return 0;
+	case RESOURCE_IN_FILE_ON_DISK:
+#ifdef __WIN32__
+	case RESOURCE_WIN32_ENCRYPTED:
+#endif
+		/* Compare files by path: just a heuristic that will place files
+		 * in the same directory next to each other.  */
+		return tstrcmp(lte1->file_on_disk, lte2->file_on_disk);
+#ifdef WITH_NTFS_3G
+	case RESOURCE_IN_NTFS_VOLUME:
+		return tstrcmp(lte1->ntfs_loc->path, lte2->ntfs_loc->path);
+#endif
+	default:
+		/* No additional sorting order defined for this resource
+		 * location (e.g. RESOURCE_IN_ATTACHED_BUFFER); simply compare
+		 * everything equal to each other.  */
 		return 0;
+	}
+}
+
+int
+sort_stream_list_by_sequential_order(struct list_head *stream_list,
+				     size_t list_head_offset)
+{
+	struct list_head *cur;
+	struct wim_lookup_table_entry **array;
+	size_t i;
+	size_t array_size;
+	size_t num_streams = 0;
+
+	list_for_each(cur, stream_list)
+		num_streams++;
+
+	array_size = num_streams * sizeof(array[0]);
+	array = MALLOC(array_size);
+	if (!array)
+		return WIMLIB_ERR_NOMEM;
+	cur = stream_list->next;
+	for (i = 0; i < num_streams; i++) {
+		array[i] = (struct wim_lookup_table_entry*)((u8*)cur -
+							    list_head_offset);
+		cur = cur->next;
+	}
+
+	qsort(array, num_streams, sizeof(array[0]),
+	      cmp_streams_by_sequential_order);
+
+	INIT_LIST_HEAD(stream_list);
+	for (i = 0; i < num_streams; i++) {
+		list_add_tail((struct list_head*)
+			       ((u8*)array[i] + list_head_offset),
+			      stream_list);
+	}
+	FREE(array);
+	return 0;
 }
 
 
@@ -329,7 +421,7 @@ for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 	wimlib_assert(p == lte_array + num_streams);
 
 	qsort(lte_array, num_streams, sizeof(lte_array[0]),
-	      cmp_streams_by_wim_position);
+	      cmp_streams_by_sequential_order);
 	ret = 0;
 	for (size_t i = 0; i < num_streams; i++) {
 		ret = visitor(lte_array[i], arg);
@@ -340,131 +432,169 @@ for_lookup_table_entry_pos_sorted(struct wim_lookup_table *table,
 	return ret;
 }
 
+/* On-disk format of a WIM lookup table entry (stream entry). */
+struct wim_lookup_table_entry_disk {
+	/* Location, offset, compression status, and metadata status of the
+	 * stream. */
+	struct resource_entry_disk resource_entry;
+
+	/* Which part of the split WIM this stream is in; indexed from 1. */
+	le16 part_number;
+
+	/* Reference count of this stream over all WIM images. */
+	le32 refcnt;
+
+	/* SHA1 message digest of the uncompressed data of this stream, or
+	 * optionally all zeroes if this stream is of zero length. */
+	u8 hash[SHA1_HASH_SIZE];
+} _packed_attribute;
+
+#define WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE 50
+
+void
+lte_init_wim(struct wim_lookup_table_entry *lte, WIMStruct *wim)
+{
+	lte->resource_location = RESOURCE_IN_WIM;
+	lte->wim = wim;
+	if (lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
+		lte->compression_type = wim->compression_type;
+	else
+		lte->compression_type = WIMLIB_COMPRESSION_TYPE_NONE;
+
+	if (wim_is_pipable(wim))
+		lte->is_pipable = 1;
+}
+
 /*
  * 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).
+ *
+ * Return values:
+ *	WIMLIB_ERR_SUCCESS (0)
+ *	WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY
+ *	WIMLIB_ERR_RESOURCE_NOT_FOUND
  */
 int
-read_lookup_table(WIMStruct *w)
+read_wim_lookup_table(WIMStruct *wim)
 {
 	int ret;
+	size_t i;
 	size_t num_entries;
 	struct wim_lookup_table *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;
+	void *buf;
+
+	BUILD_BUG_ON(sizeof(struct wim_lookup_table_entry_disk) !=
+		     WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE);
 
 	DEBUG("Reading lookup table: offset %"PRIu64", size %"PRIu64"",
-	      w->hdr.lookup_table_res_entry.offset,
-	      w->hdr.lookup_table_res_entry.original_size);
+	      wim->hdr.lookup_table_res_entry.offset,
+	      wim->hdr.lookup_table_res_entry.size);
 
-	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;
-	}
+	/* Calculate number of entries in the lookup table.  */
+	num_entries = wim->hdr.lookup_table_res_entry.size /
+		      sizeof(struct wim_lookup_table_entry_disk);
+
+
+	/* Read the lookup table into a buffer.  */
+	ret = res_entry_to_data(&wim->hdr.lookup_table_res_entry, wim, &buf);
+	if (ret)
+		goto out;
 
-	num_entries = w->hdr.lookup_table_res_entry.size /
-		      WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE;
+	/* Allocate hash table.  */
 	table = new_lookup_table(num_entries * 2 + 1);
-	if (!table)
-		return WIMLIB_ERR_NOMEM;
+	if (!table) {
+		ERROR("Not enough memory to read lookup table.");
+		ret = WIMLIB_ERR_NOMEM;
+		goto out_free_buf;
+	}
+
+	/* Allocate and initalize `struct wim_lookup_table_entry's from the
+	 * on-disk lookup table.  */
+	wim->current_image = 0;
+	for (i = 0; i < num_entries; i++) {
+		const struct wim_lookup_table_entry_disk *disk_entry =
+			&((const struct wim_lookup_table_entry_disk*)buf)[i];
 
-	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;
-			}
-			offset += bytes_to_read;
-			p = table_buf;
-			buf_entries_remaining = entries_to_read;
-		}
 		cur_entry = new_lookup_table_entry();
 		if (!cur_entry) {
+			ERROR("Not enough memory to read lookup table.");
 			ret = WIMLIB_ERR_NOMEM;
 			goto out_free_lookup_table;
 		}
 
-		cur_entry->wim = w;
+		cur_entry->wim = wim;
 		cur_entry->resource_location = RESOURCE_IN_WIM;
-		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);
-
-		if (cur_entry->part_number != w->hdr.part_number) {
-			ERROR("A lookup table entry in part %hu of the WIM "
-			      "points to part %hu",
-			      w->hdr.part_number, cur_entry->part_number);
-			ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
-			goto out_free_cur_entry;
+		get_resource_entry(&disk_entry->resource_entry, &cur_entry->resource_entry);
+		cur_entry->part_number = le16_to_cpu(disk_entry->part_number);
+		cur_entry->refcnt = le32_to_cpu(disk_entry->refcnt);
+		copy_hash(cur_entry->hash, disk_entry->hash);
+		lte_init_wim(cur_entry, wim);
+
+		if (cur_entry->part_number != wim->hdr.part_number) {
+			WARNING("A lookup table entry in part %hu of the WIM "
+				"points to part %hu (ignoring it)",
+				wim->hdr.part_number, cur_entry->part_number);
+			free_lookup_table_entry(cur_entry);
+			continue;
 		}
 
 		if (is_zero_hash(cur_entry->hash)) {
-			ERROR("The WIM lookup table contains an entry with a "
-			      "SHA1 message digest of all 0's");
-			ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
-			goto out_free_cur_entry;
+			WARNING("The WIM lookup table contains an entry with a "
+				"SHA1 message digest of all 0's (ignoring it)");
+			free_lookup_table_entry(cur_entry);
+			continue;
 		}
 
 		if (!(cur_entry->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED)
 		    && (cur_entry->resource_entry.size !=
 		        cur_entry->resource_entry.original_size))
 		{
-		#ifdef ENABLE_ERROR_MESSAGES
-			ERROR("Found uncompressed resource with original size "
-			      "not the same as compressed size");
-			ERROR("The lookup table entry for the resource is as follows:");
-			print_lookup_table_entry(cur_entry, stderr);
-		#endif
-			ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
-			goto out_free_cur_entry;
+			if (wimlib_print_errors) {
+				WARNING("Found uncompressed resource with "
+					"original size (%"PRIu64") not the same "
+					"as compressed size (%"PRIu64")",
+					cur_entry->resource_entry.original_size,
+					cur_entry->resource_entry.size);
+				if (cur_entry->resource_entry.original_size) {
+					WARNING("Overriding compressed size with original size.");
+					cur_entry->resource_entry.size =
+						cur_entry->resource_entry.original_size;
+				} else {
+					WARNING("Overriding original size with compressed size");
+					cur_entry->resource_entry.original_size =
+						cur_entry->resource_entry.size;
+				}
+			}
 		}
 
 		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
+				if (wimlib_print_errors) {
+					ERROR("Found metadata resource with refcnt != 1:");
+					print_lookup_table_entry(cur_entry, stderr);
+				}
 				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 (wim->hdr.part_number != 1) {
+				WARNING("Ignoring metadata resource found in a "
+					"non-first part of the split WIM");
+				free_lookup_table_entry(cur_entry);
+				continue;
 			}
-			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;
+			if (wim->current_image == wim->hdr.image_count) {
+				WARNING("The WIM header says there are %u images "
+					"in the WIM, but we found more metadata "
+					"resources than this (ignoring the extra)",
+					wim->hdr.image_count);
+				free_lookup_table_entry(cur_entry);
+				continue;
 			}
 
 			/* Notice very carefully:  We are assigning the metadata
@@ -476,153 +606,200 @@ read_lookup_table(WIMStruct *w)
 			 * written in the XML data. */
 			DEBUG("Found metadata resource for image %u at "
 			      "offset %"PRIu64".",
-			      w->current_image + 1,
+			      wim->current_image + 1,
 			      cur_entry->resource_entry.offset);
-			w->image_metadata[
-				w->current_image++]->metadata_lte = cur_entry;
+			wim->image_metadata[
+				wim->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);
+			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;
+				if (wimlib_print_errors) {
+					WARNING("The WIM lookup table contains two entries with the "
+					      "same SHA1 message digest!");
+					WARNING("The first entry is:");
+					print_lookup_table_entry(duplicate_entry, stderr);
+					WARNING("The second entry is:");
+					print_lookup_table_entry(cur_entry, stderr);
+				}
+				free_lookup_table_entry(cur_entry);
+				continue;
+			} else {
+				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;
+	if (wim->hdr.part_number == 1 && wim->current_image != wim->hdr.image_count) {
+		WARNING("The header of \"%"TS"\" says there are %u images in\n"
+			"          the WIM, but we only found %d metadata resources!  Acting as if\n"
+			"          the header specified only %d images instead.",
+			wim->filename, wim->hdr.image_count,
+			wim->current_image, wim->current_image);
+		for (int i = wim->current_image; i < wim->hdr.image_count; i++)
+			put_image_metadata(wim->image_metadata[i], NULL);
+		wim->hdr.image_count = wim->current_image;
 	}
 	DEBUG("Done reading lookup table.");
-	w->lookup_table = table;
+	wim->lookup_table = table;
 	ret = 0;
-	goto out;
+	goto out_free_buf;
 out_free_cur_entry:
 	FREE(cur_entry);
 out_free_lookup_table:
 	free_lookup_table(table);
+out_free_buf:
+	FREE(buf);
 out:
-	w->current_image = 0;
+	wim->current_image = 0;
 	return ret;
 }
 
 
-static u8 *
-write_lookup_table_entry(struct wim_lookup_table_entry *lte, u8 *buf_p)
+static void
+write_wim_lookup_table_entry(const struct wim_lookup_table_entry *lte,
+			     struct wim_lookup_table_entry_disk *disk_entry)
 {
-	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;
+	put_resource_entry(&lte->output_resource_entry, &disk_entry->resource_entry);
+	disk_entry->part_number = cpu_to_le16(lte->part_number);
+	disk_entry->refcnt = cpu_to_le32(lte->out_refcnt);
+	copy_hash(disk_entry->hash, lte->hash);
 }
 
-int
-write_lookup_table_from_stream_list(struct list_head *stream_list,
-				    int out_fd,
-				    struct resource_entry *out_res_entry)
+static int
+write_wim_lookup_table_from_stream_list(struct list_head *stream_list,
+					struct filedes *out_fd,
+					struct resource_entry *out_res_entry,
+					int write_resource_flags)
 {
-	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;
+	struct wim_lookup_table_entry_disk *table_buf;
+	struct wim_lookup_table_entry_disk *table_buf_ptr;
+	int ret;
 
-	start_offset = filedes_offset(out_fd);
-	if (start_offset == -1)
-		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;
+	list_for_each_entry(lte, stream_list, lookup_table_list)
+		table_size += sizeof(struct wim_lookup_table_entry_disk);
+
+	DEBUG("Writing WIM lookup table (size=%zu, offset=%"PRIu64")",
+	      table_size, out_fd->offset);
+
+	table_buf = MALLOC(table_size);
+	if (!table_buf) {
+		ERROR("Failed to allocate %zu bytes for temporary lookup table",
+		      table_size);
+		return WIMLIB_ERR_NOMEM;
 	}
-	out_res_entry->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:
+	table_buf_ptr = table_buf;
+	list_for_each_entry(lte, stream_list, lookup_table_list)
+		write_wim_lookup_table_entry(lte, table_buf_ptr++);
+
+	/* Write the lookup table uncompressed.  Although wimlib can handle a
+	 * compressed lookup table, MS software cannot.  */
+	ret = write_wim_resource_from_buffer(table_buf,
+					     table_size,
+					     WIM_RESHDR_FLAG_METADATA,
+					     out_fd,
+					     WIMLIB_COMPRESSION_TYPE_NONE,
+					     out_res_entry,
+					     NULL,
+					     write_resource_flags);
+	FREE(table_buf);
 	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)
+	/* Lookup table entries with 'out_refcnt' == 0 correspond to streams not
+	 * written and not present in the resulting WIM file, and should not be
+	 * included in the lookup table.
+	 *
+	 * Lookup table entries marked as filtered (EXTERNAL_WIM) with
+	 * 'out_refcnt != 0' were referenced as part of the logical write but
+	 * correspond to streams that were not in fact written, and should not
+	 * be included in the lookup table.
+	 *
+	 * Lookup table entries marked as filtered (SAME_WIM) with 'out_refcnt
+	 * != 0' were referenced as part of the logical write but correspond to
+	 * streams that were not in fact written, but nevertheless were already
+	 * present in the WIM being overwritten in-place.  These entries must be
+	 * included in the lookup table, and the resource information to write
+	 * needs to be copied from the resource information read originally.
+	 */
+	if (lte->out_refcnt != 0 && !(lte->filtered & FILTERED_EXTERNAL_WIM)) {
+		if (lte->filtered & FILTERED_SAME_WIM) {
+			copy_resource_entry(&lte->output_resource_entry,
+					    &lte->resource_entry);
+		}
 		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)
+write_wim_lookup_table(WIMStruct *wim, int image, int write_flags,
+		       struct resource_entry *out_res_entry,
+		       struct list_head *stream_list_override)
 {
-	LIST_HEAD(stream_list);
-	int start_image;
-	int end_image;
+	int write_resource_flags;
+	struct list_head _stream_list;
+	struct list_head *stream_list;
 
-	if (image == WIMLIB_ALL_IMAGES) {
-		start_image = 1;
-		end_image = w->hdr.image_count;
+	if (stream_list_override) {
+		stream_list = stream_list_override;
 	} else {
-		start_image = image;
-		end_image = image;
+		stream_list = &_stream_list;
+		INIT_LIST_HEAD(stream_list);
 	}
 
-	for (int i = start_image; i <= end_image; i++) {
-		struct wim_lookup_table_entry *metadata_lte;
+	if (!(write_flags & WIMLIB_WRITE_FLAG_NO_METADATA)) {
+		int start_image;
+		int end_image;
 
-		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);
+		if (image == WIMLIB_ALL_IMAGES) {
+			start_image = 1;
+			end_image = wim->hdr.image_count;
+		} else {
+			start_image = image;
+			end_image = image;
+		}
+
+		/* Push metadata resource lookup table entries onto the front of
+		 * the list in reverse order, so that they're written in order.
+		 */
+		for (int i = end_image; i >= start_image; i--) {
+			struct wim_lookup_table_entry *metadata_lte;
+
+			metadata_lte = wim->image_metadata[i - 1]->metadata_lte;
+			metadata_lte->out_refcnt = 1;
+			metadata_lte->part_number = wim->hdr.part_number;
+			metadata_lte->output_resource_entry.flags |= WIM_RESHDR_FLAG_METADATA;
+
+			list_add(&metadata_lte->lookup_table_list, 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);
+
+	/* Append additional lookup table entries that need to be written, with
+	 * some special handling for streams that have been marked as filtered.
+	 */
+	if (!stream_list_override) {
+		for_lookup_table_entry(wim->lookup_table,
+				       append_lookup_table_entry, stream_list);
+	}
+
+	write_resource_flags = 0;
+	if (write_flags & WIMLIB_WRITE_FLAG_PIPABLE)
+		write_resource_flags |= WIMLIB_WRITE_RESOURCE_FLAG_PIPABLE;
+	return write_wim_lookup_table_from_stream_list(stream_list,
+						       &wim->out_fd,
+						       out_res_entry,
+						       write_resource_flags);
 }
 
+
 int
 lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 {
@@ -694,11 +871,9 @@ print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 		}
 		break;
 #ifdef __WIN32__
-	case RESOURCE_WIN32:
 	case RESOURCE_WIN32_ENCRYPTED:
-#else
-	case RESOURCE_IN_FILE_ON_DISK:
 #endif
+	case RESOURCE_IN_FILE_ON_DISK:
 		tfprintf(out, T("File on Disk      = `%"TS"'\n"),
 			 lte->file_on_disk);
 		break;
@@ -714,28 +889,63 @@ print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 	tputc(T('\n'), out);
 }
 
+void
+lte_to_wimlib_resource_entry(const struct wim_lookup_table_entry *lte,
+			     struct wimlib_resource_entry *wentry)
+{
+	wentry->uncompressed_size = lte->resource_entry.original_size;
+	wentry->compressed_size = lte->resource_entry.size;
+	wentry->offset = lte->resource_entry.offset;
+	copy_hash(wentry->sha1_hash, lte->hash);
+	wentry->part_number = lte->part_number;
+	wentry->reference_count = lte->refcnt;
+	wentry->is_compressed = (lte->resource_entry.flags & WIM_RESHDR_FLAG_COMPRESSED) != 0;
+	wentry->is_metadata = (lte->resource_entry.flags & WIM_RESHDR_FLAG_METADATA) != 0;
+	wentry->is_free = (lte->resource_entry.flags & WIM_RESHDR_FLAG_FREE) != 0;
+	wentry->is_spanned = (lte->resource_entry.flags & WIM_RESHDR_FLAG_SPANNED) != 0;
+}
+
+struct iterate_lte_context {
+	wimlib_iterate_lookup_table_callback_t cb;
+	void *user_ctx;
+};
+
 static int
-do_print_lookup_table_entry(struct wim_lookup_table_entry *lte, void *fp)
+do_iterate_lte(struct wim_lookup_table_entry *lte, void *_ctx)
 {
-	print_lookup_table_entry(lte, (FILE*)fp);
-	return 0;
+	struct iterate_lte_context *ctx = _ctx;
+	struct wimlib_resource_entry entry;
+
+	lte_to_wimlib_resource_entry(lte, &entry);
+	return (*ctx->cb)(&entry, ctx->user_ctx);
 }
 
-/*
- * Prints the lookup table of a WIM file.
- */
-WIMLIBAPI void
-wimlib_print_lookup_table(WIMStruct *w)
+/* API function documented in wimlib.h  */
+WIMLIBAPI int
+wimlib_iterate_lookup_table(WIMStruct *wim, int flags,
+			    wimlib_iterate_lookup_table_callback_t cb,
+			    void *user_ctx)
 {
-	for_lookup_table_entry(w->lookup_table,
-			       do_print_lookup_table_entry,
-			       stdout);
+	struct iterate_lte_context ctx = {
+		.cb = cb,
+		.user_ctx = user_ctx,
+	};
+	if (wim->hdr.part_number == 1) {
+		int ret;
+		for (int i = 0; i < wim->hdr.image_count; i++) {
+			ret = do_iterate_lte(wim->image_metadata[i]->metadata_lte,
+					     &ctx);
+			if (ret)
+				return ret;
+		}
+	}
+	return for_lookup_table_entry(wim->lookup_table, do_iterate_lte, &ctx);
 }
 
 /* Given a SHA1 message digest, return the corresponding entry in the WIM's
  * lookup table, or NULL if there is none.  */
 struct wim_lookup_table_entry *
-__lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
+lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
 {
 	size_t i;
 	struct wim_lookup_table_entry *lte;
@@ -759,12 +969,12 @@ __lookup_resource(const struct wim_lookup_table *table, const u8 hash[])
  * This is only for pre-resolved inodes.
  */
 int
-lookup_resource(WIMStruct *w,
-		const tchar *path,
-		int lookup_flags,
-		struct wim_dentry **dentry_ret,
-		struct wim_lookup_table_entry **lte_ret,
-		u16 *stream_idx_ret)
+wim_pathname_to_stream(WIMStruct *wim,
+		       const tchar *path,
+		       int lookup_flags,
+		       struct wim_dentry **dentry_ret,
+		       struct wim_lookup_table_entry **lte_ret,
+		       u16 *stream_idx_ret)
 {
 	struct wim_dentry *dentry;
 	struct wim_lookup_table_entry *lte;
@@ -781,7 +991,7 @@ lookup_resource(WIMStruct *w,
 		}
 	}
 
-	dentry = get_dentry(w, path);
+	dentry = get_dentry(wim, path);
 	if (p)
 		*p = T(':');
 	if (!dentry)
@@ -789,7 +999,9 @@ lookup_resource(WIMStruct *w,
 
 	inode = dentry->d_inode;
 
-	wimlib_assert(inode->i_resolved);
+	if (!inode->i_resolved)
+		if (inode_resolve_ltes(inode, wim->lookup_table, false))
+			return -EIO;
 
 	if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
 	      && inode_is_directory(inode))
@@ -822,33 +1034,92 @@ out:
 }
 #endif
 
-/* Resolve an inode's lookup table entries
+int
+resource_not_found_error(struct wim_inode *inode, const u8 *hash)
+{
+	if (wimlib_print_errors) {
+		ERROR("\"%"TS"\": resource not found", inode_first_full_path(inode));
+		tfprintf(stderr, T("        SHA-1 message digest of missing resource:\n        "));
+		print_hash(hash, stderr);
+		tputc(T('\n'), stderr);
+	}
+	return WIMLIB_ERR_RESOURCE_NOT_FOUND;
+}
+
+/*
+ * Resolve an inode's lookup table entries.
  *
  * This replaces the SHA1 hash fields (which are used to lookup an entry in the
- * lookup table) with pointers directly to the lookup table entries.  A circular
- * linked list of streams sharing the same lookup table entry is created.
+ * lookup table) with pointers directly to the lookup table entries.
  *
- * This function always succeeds; unresolved lookup table entries are given a
- * NULL pointer.
+ * If @force is %false:
+ *	If any needed SHA1 message digests are not found in the lookup table,
+ *	WIMLIB_ERR_RESOURCE_NOT_FOUND is returned and the inode is left
+ *	unmodified.
+ * If @force is %true:
+ *	If any needed SHA1 message digests are not found in the lookup table,
+ *	new entries are allocated and inserted into the lookup table.
  */
-void
-inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table)
+int
+inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table,
+		   bool force)
 {
+	const u8 *hash;
 
 	if (!inode->i_resolved) {
-		struct wim_lookup_table_entry *lte;
+		struct wim_lookup_table_entry *lte, *ads_lte;
+
 		/* Resolve the default file stream */
-		lte = __lookup_resource(table, inode->i_hash);
-		inode->i_lte = lte;
-		inode->i_resolved = 1;
+		lte = NULL;
+		hash = inode->i_hash;
+		if (!is_zero_hash(hash)) {
+			lte = lookup_resource(table, hash);
+			if (!lte) {
+				if (force) {
+					lte = new_lookup_table_entry();
+					if (!lte)
+						return WIMLIB_ERR_NOMEM;
+					copy_hash(lte->hash, hash);
+					lookup_table_insert(table, lte);
+				} else {
+					goto resource_not_found;
+				}
+			}
+		}
 
 		/* Resolve the alternate data streams */
+		struct wim_lookup_table_entry *ads_ltes[inode->i_num_ads];
 		for (u16 i = 0; i < inode->i_num_ads; i++) {
-			struct wim_ads_entry *cur_entry = &inode->i_ads_entries[i];
-			lte = __lookup_resource(table, cur_entry->hash);
-			cur_entry->lte = lte;
+			struct wim_ads_entry *cur_entry;
+
+			ads_lte = NULL;
+			cur_entry = &inode->i_ads_entries[i];
+			hash = cur_entry->hash;
+			if (!is_zero_hash(hash)) {
+				ads_lte = lookup_resource(table, hash);
+				if (!ads_lte) {
+					if (force) {
+						ads_lte = new_lookup_table_entry();
+						if (!ads_lte)
+							return WIMLIB_ERR_NOMEM;
+						copy_hash(ads_lte->hash, hash);
+						lookup_table_insert(table, ads_lte);
+					} else {
+						goto resource_not_found;
+					}
+				}
+			}
+			ads_ltes[i] = ads_lte;
 		}
+		inode->i_lte = lte;
+		for (u16 i = 0; i < inode->i_num_ads; i++)
+			inode->i_ads_entries[i].lte = ads_ltes[i];
+		inode->i_resolved = 1;
 	}
+	return 0;
+
+resource_not_found:
+	return resource_not_found_error(inode, hash);
 }
 
 void
@@ -1008,7 +1279,7 @@ hash_unhashed_stream(struct wim_lookup_table_entry *lte,
 		return ret;
 
 	/* Look for a duplicate stream */
-	duplicate_lte = __lookup_resource(lookup_table, lte->hash);
+	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
@@ -1017,7 +1288,7 @@ hash_unhashed_stream(struct wim_lookup_table_entry *lte,
 		 * duplicate, then free this stream. */
 		wimlib_assert(!(duplicate_lte->unhashed));
 		duplicate_lte->refcnt += lte->refcnt;
-		duplicate_lte->out_refcnt += lte->refcnt;
+		duplicate_lte->out_refcnt += lte->out_refcnt;
 		*back_ptr = duplicate_lte;
 		free_lookup_table_entry(lte);
 		lte = duplicate_lte;
@@ -1033,3 +1304,182 @@ hash_unhashed_stream(struct wim_lookup_table_entry *lte,
 	return 0;
 }
 
+static int
+lte_clone_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
+{
+	struct wim_lookup_table *lookup_table = _lookup_table;
+
+	if (lookup_resource(lookup_table, lte->hash))
+		return 0;  /*  Resource already present.  */
+
+	lte = clone_lookup_table_entry(lte);
+	if (!lte)
+		return WIMLIB_ERR_NOMEM;
+	lte->out_refcnt = 1;
+	lookup_table_insert(lookup_table, lte);
+	return 0;
+}
+
+static int
+lte_delete_if_new(struct wim_lookup_table_entry *lte, void *_lookup_table)
+{
+	struct wim_lookup_table *lookup_table = _lookup_table;
+
+	if (lte->out_refcnt) {
+		lookup_table_unlink(lookup_table, lte);
+		free_lookup_table_entry(lte);
+	}
+	return 0;
+}
+
+/* API function documented in wimlib.h  */
+WIMLIBAPI int
+wimlib_reference_resources(WIMStruct *wim,
+			   WIMStruct **resource_wims, unsigned num_resource_wims,
+			   int ref_flags)
+{
+	int ret;
+	unsigned i;
+
+	if (wim == NULL)
+		return WIMLIB_ERR_INVALID_PARAM;
+
+	if (num_resource_wims != 0 && resource_wims == NULL)
+		return WIMLIB_ERR_INVALID_PARAM;
+
+	for (i = 0; i < num_resource_wims; i++)
+		if (resource_wims[i] == NULL)
+			return WIMLIB_ERR_INVALID_PARAM;
+
+	for_lookup_table_entry(wim->lookup_table, lte_zero_out_refcnt, NULL);
+
+	for (i = 0; i < num_resource_wims; i++) {
+		ret = for_lookup_table_entry(resource_wims[i]->lookup_table,
+					     lte_clone_if_new,
+					     wim->lookup_table);
+		if (ret)
+			goto out_rollback;
+	}
+	return 0;
+
+out_rollback:
+	for_lookup_table_entry(wim->lookup_table, lte_delete_if_new,
+			       wim->lookup_table);
+	return ret;
+}
+
+static int
+reference_resource_paths(WIMStruct *wim,
+			 const tchar * const *resource_wimfiles,
+			 unsigned num_resource_wimfiles,
+			 int ref_flags,
+			 int open_flags,
+			 wimlib_progress_func_t progress_func)
+{
+	WIMStruct **resource_wims;
+	unsigned i;
+	int ret;
+
+	resource_wims = CALLOC(num_resource_wimfiles, sizeof(resource_wims[0]));
+	if (!resource_wims)
+		return WIMLIB_ERR_NOMEM;
+
+	for (i = 0; i < num_resource_wimfiles; i++) {
+		DEBUG("Referencing resources from path \"%"TS"\"",
+		      resource_wimfiles[i]);
+		ret = wimlib_open_wim(resource_wimfiles[i], open_flags,
+				      &resource_wims[i], progress_func);
+		if (ret)
+			goto out_free_resource_wims;
+	}
+
+	ret = wimlib_reference_resources(wim, resource_wims,
+					 num_resource_wimfiles, ref_flags);
+	if (ret)
+		goto out_free_resource_wims;
+
+	for (i = 0; i < num_resource_wimfiles; i++)
+		list_add_tail(&resource_wims[i]->subwim_node, &wim->subwims);
+
+	ret = 0;
+	goto out_free_array;
+
+out_free_resource_wims:
+	for (i = 0; i < num_resource_wimfiles; i++)
+		wimlib_free(resource_wims[i]);
+out_free_array:
+	FREE(resource_wims);
+	return ret;
+}
+
+static int
+reference_resource_glob(WIMStruct *wim, const tchar *refglob,
+			int ref_flags, int open_flags,
+			wimlib_progress_func_t progress_func)
+{
+	glob_t globbuf;
+	int ret;
+
+	/* Note: glob() is replaced in Windows native builds.  */
+	ret = tglob(refglob, GLOB_ERR | GLOB_NOSORT, NULL, &globbuf);
+	if (ret) {
+		if (ret == GLOB_NOMATCH) {
+			if (ref_flags & WIMLIB_REF_FLAG_GLOB_ERR_ON_NOMATCH) {
+				ERROR("Found no files for glob \"%"TS"\"", refglob);
+				return WIMLIB_ERR_GLOB_HAD_NO_MATCHES;
+			} else {
+				return reference_resource_paths(wim,
+								&refglob,
+								1,
+								ref_flags,
+								open_flags,
+								progress_func);
+			}
+		} else {
+			ERROR_WITH_ERRNO("Failed to process glob \"%"TS"\"", refglob);
+			if (ret == GLOB_NOSPACE)
+				return WIMLIB_ERR_NOMEM;
+			else
+				return WIMLIB_ERR_READ;
+		}
+	}
+
+	ret = reference_resource_paths(wim,
+				       (const tchar * const *)globbuf.gl_pathv,
+				       globbuf.gl_pathc,
+				       ref_flags,
+				       open_flags,
+				       progress_func);
+	globfree(&globbuf);
+	return ret;
+}
+
+/* API function documented in wimlib.h  */
+WIMLIBAPI int
+wimlib_reference_resource_files(WIMStruct *wim,
+				const tchar * const * resource_wimfiles_or_globs,
+				unsigned count,
+				int ref_flags,
+				int open_flags,
+				wimlib_progress_func_t progress_func)
+{
+	unsigned i;
+	int ret;
+
+	if (ref_flags & WIMLIB_REF_FLAG_GLOB_ENABLE) {
+		for (i = 0; i < count; i++) {
+			ret = reference_resource_glob(wim,
+						      resource_wimfiles_or_globs[i],
+						      ref_flags,
+						      open_flags,
+						      progress_func);
+			if (ret)
+				return ret;
+		}
+		return 0;
+	} else {
+		return reference_resource_paths(wim, resource_wimfiles_or_globs,
+						count, ref_flags,
+						open_flags, progress_func);
+	}
+}