X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Flookup_table.c;h=89b57b4e65ee757dc0485c61903750b866319091;hp=61cc5a83ab663132dc87b3d581c9b64fefa2862a;hb=1f2091f5bb860769e20105af2d9eff3a0872c57b;hpb=63a5804943519281f206ca0fefacfe2b99fc9958

diff --git a/src/lookup_table.c b/src/lookup_table.c
index 61cc5a83..89b57b4e 100644
--- a/src/lookup_table.c
+++ b/src/lookup_table.c
@@ -1,12 +1,12 @@
 /*
  * lookup_table.c
  *
- * Lookup table, implemented as a hash table, that maps dentries to file
- * resources.
+ * Lookup table, implemented as a hash table, that maps SHA1 message digests to
+ * data streams.
  */
 
 /*
- * Copyright (C) 2012 Eric Biggers
+ * Copyright (C) 2012, 2013 Eric Biggers
  *
  * This file is part of wimlib, a library for working with WIM files.
  *
@@ -26,75 +26,141 @@
 
 #include "wimlib_internal.h"
 #include "lookup_table.h"
-#include "io.h"
+#include "buffer_io.h"
 #include <errno.h>
+#include <stdlib.h>
 
 #ifdef WITH_FUSE
 #include <unistd.h>
 #endif
 
-struct lookup_table *new_lookup_table(size_t capacity)
+struct wim_lookup_table *
+new_lookup_table(size_t capacity)
 {
-	struct lookup_table *table;
+	struct wim_lookup_table *table;
 	struct hlist_head *array;
 
-	table = MALLOC(sizeof(struct lookup_table));
-	if (!table)
-		goto err;
-	array = CALLOC(capacity, sizeof(array[0]));
-	if (!array) {
-		FREE(table);
-		goto err;
+	table = CALLOC(1, sizeof(struct wim_lookup_table));
+	if (table) {
+		array = CALLOC(capacity, sizeof(array[0]));
+		if (array) {
+			table->num_entries = 0;
+			table->capacity = capacity;
+			table->array = array;
+		} else {
+			FREE(table);
+			table = NULL;
+			ERROR("Failed to allocate memory for lookup table with capacity %zu",
+			      capacity);
+		}
 	}
-	table->num_entries = 0;
-	table->capacity = capacity;
-	table->array = array;
 	return table;
-err:
-	ERROR("Failed to allocate memory for lookup table with capacity %zu",
-	      capacity);
-	return NULL;
 }
 
-struct lookup_table_entry *new_lookup_table_entry()
+struct wim_lookup_table_entry *
+new_lookup_table_entry()
 {
-	struct lookup_table_entry *lte;
-	
-	lte = CALLOC(1, sizeof(struct lookup_table_entry));
-	if (!lte) {
+	struct wim_lookup_table_entry *lte;
+
+	lte = CALLOC(1, sizeof(struct wim_lookup_table_entry));
+	if (lte) {
+		lte->part_number  = 1;
+		lte->refcnt       = 1;
+	} else {
 		ERROR("Out of memory (tried to allocate %zu bytes for "
 		      "lookup table entry)",
-		      sizeof(struct lookup_table_entry));
-		return NULL;
+		      sizeof(struct wim_lookup_table_entry));
 	}
-
-	lte->part_number  = 1;
-	lte->refcnt       = 1;
 	return lte;
 }
 
-void free_lookup_table_entry(struct lookup_table_entry *lte)
+struct wim_lookup_table_entry *
+clone_lookup_table_entry(const struct wim_lookup_table_entry *old)
 {
-	if (lte) {
+	struct wim_lookup_table_entry *new;
+
+	new = MALLOC(sizeof(*new));
+	if (!new)
+		return NULL;
+
+	memcpy(new, old, sizeof(*old));
+	new->extracted_file = NULL;
+	switch (new->resource_location) {
+#ifdef __WIN32__
+	case RESOURCE_WIN32:
+	case RESOURCE_WIN32_ENCRYPTED:
+#endif
 #ifdef WITH_FUSE
-		if (lte->staging_list.next)
-			list_del(&lte->staging_list);
+	case RESOURCE_IN_STAGING_FILE:
+#endif
+	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)
+			goto out_free;
+		break;
+	case RESOURCE_IN_ATTACHED_BUFFER:
+		new->attached_buffer = MALLOC(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));
+			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);
+		}
+		break;
 #endif
+	default:
+		break;
+	}
+	return new;
+out_free:
+	free_lookup_table_entry(new);
+	return NULL;
+}
+
+void
+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:
-			wimlib_assert(((void*)&lte->file_on_disk ==
-				      (void*)&lte->staging_file_name)
-				      && ((void*)&lte->file_on_disk ==
-				      (void*)&lte->attached_buffer));
+#ifdef __WIN32__
+		case RESOURCE_WIN32:
+#endif
+			BUILD_BUG_ON((void*)&lte->file_on_disk !=
+				     (void*)&lte->staging_file_name);
+			BUILD_BUG_ON((void*)&lte->file_on_disk !=
+				     (void*)&lte->attached_buffer);
 			FREE(lte->file_on_disk);
 			break;
 #ifdef WITH_NTFS_3G
 		case RESOURCE_IN_NTFS_VOLUME:
 			if (lte->ntfs_loc) {
-				FREE(lte->ntfs_loc->path_utf8);
-				FREE(lte->ntfs_loc->stream_name_utf16);
+				FREE(lte->ntfs_loc->path);
+				FREE(lte->ntfs_loc->stream_name);
 				FREE(lte->ntfs_loc);
 			}
 			break;
@@ -102,20 +168,20 @@ void free_lookup_table_entry(struct lookup_table_entry *lte)
 		default:
 			break;
 		}
-		FREE(lte->extracted_file);
 		FREE(lte);
 	}
 }
 
-static int do_free_lookup_table_entry(struct lookup_table_entry *entry,
-				      void *ignore)
+static int
+do_free_lookup_table_entry(struct wim_lookup_table_entry *entry, void *ignore)
 {
 	free_lookup_table_entry(entry);
 	return 0;
 }
 
 
-void free_lookup_table(struct lookup_table *table)
+void
+free_lookup_table(struct wim_lookup_table *table)
 {
 	DEBUG2("Freeing lookup table");
 	if (table) {
@@ -133,10 +199,11 @@ void free_lookup_table(struct lookup_table *table)
  * Inserts an entry into the lookup table.
  *
  * @table:	A pointer to the lookup table.
- * @entry:	A pointer to the entry to insert.
+ * @lte:	A pointer to the entry to insert.
  */
-void lookup_table_insert(struct lookup_table *table, 
-			 struct lookup_table_entry *lte)
+void
+lookup_table_insert(struct wim_lookup_table *table,
+		    struct wim_lookup_table_entry *lte)
 {
 	size_t i = lte->hash_short % table->capacity;
 	hlist_add_head(&lte->hash_list, &table->array[i]);
@@ -145,11 +212,14 @@ void lookup_table_insert(struct lookup_table *table,
 	table->num_entries++;
 }
 
-static void finalize_lte(struct lookup_table_entry *lte)
+static void
+finalize_lte(struct wim_lookup_table_entry *lte)
 {
 	#ifdef WITH_FUSE
-	if (lte->resource_location == RESOURCE_IN_STAGING_FILE)
+	if (lte->resource_location == RESOURCE_IN_STAGING_FILE) {
 		unlink(lte->staging_file_name);
+		list_del(&lte->unhashed_list);
+	}
 	#endif
 	free_lookup_table_entry(lte);
 }
@@ -158,51 +228,40 @@ static void finalize_lte(struct lookup_table_entry *lte)
  * reference count reaches 0, it is unlinked from the lookup table.  If,
  * furthermore, the entry has no opened file descriptors associated with it, the
  * entry is freed.  */
-struct lookup_table_entry *
-lte_decrement_refcnt(struct lookup_table_entry *lte, struct lookup_table *table)
+void
+lte_decrement_refcnt(struct wim_lookup_table_entry *lte,
+		     struct wim_lookup_table *table)
 {
-	if (lte) {
-		wimlib_assert(lte->refcnt);
-		if (--lte->refcnt == 0) {
+	wimlib_assert(lte != NULL);
+	wimlib_assert(lte->refcnt != 0);
+	if (--lte->refcnt == 0) {
+		if (!lte->unhashed)
 			lookup_table_unlink(table, lte);
-		#ifdef WITH_FUSE
-			if (lte->num_opened_fds == 0)
-		#endif
-			{
-				finalize_lte(lte);
-				lte = NULL;
-			}
-		}
+	#ifdef WITH_FUSE
+		if (lte->num_opened_fds == 0)
+	#endif
+			finalize_lte(lte);
 	}
-	return lte;
 }
 
 #ifdef WITH_FUSE
-struct lookup_table_entry *
-lte_decrement_num_opened_fds(struct lookup_table_entry *lte,
-			     struct lookup_table *table)
+void
+lte_decrement_num_opened_fds(struct wim_lookup_table_entry *lte)
 {
-	if (lte) {
-		wimlib_assert(lte->num_opened_fds);
-		if (--lte->num_opened_fds == 0 && lte->refcnt == 0) {
-			lookup_table_unlink(table, lte);
+	if (lte->num_opened_fds != 0)
+		if (--lte->num_opened_fds == 0 && lte->refcnt == 0)
 			finalize_lte(lte);
-			lte = NULL;
-		}
-	}
-	return lte;
 }
 #endif
 
-/* 
- * Calls a function on all the entries in the lookup table.  Stop early and
- * return nonzero if any call to the function returns nonzero.
- */
-int for_lookup_table_entry(struct lookup_table *table, 
-			   int (*visitor)(struct lookup_table_entry *, void *),
-			   void *arg)
+/* Calls a function on all the entries in the WIM lookup table.  Stop early and
+ * return nonzero if any call to the function returns nonzero. */
+int
+for_lookup_table_entry(struct wim_lookup_table *table,
+		       int (*visitor)(struct wim_lookup_table_entry *, void *),
+		       void *arg)
 {
-	struct lookup_table_entry *lte;
+	struct wim_lookup_table_entry *lte;
 	struct hlist_node *pos, *tmp;
 	int ret;
 
@@ -210,31 +269,100 @@ int for_lookup_table_entry(struct 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)
+int
+read_lookup_table(WIMStruct *w)
 {
-	u64    num_entries;
-	u8     buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
-	int    ret;
-	struct lookup_table *table;
-	struct lookup_table_entry *cur_entry = NULL, *duplicate_entry;
+	u64 num_entries;
+	u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
+	int ret;
+	struct wim_lookup_table *table;
+	struct wim_lookup_table_entry *cur_entry, *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;
+	}
 
 	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) {
+	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);
@@ -247,6 +375,7 @@ int read_lookup_table(WIMStruct *w)
 	if (!table)
 		return WIMLIB_ERR_NOMEM;
 
+	w->current_image = 0;
 	while (num_entries--) {
 		const u8 *p;
 
@@ -258,16 +387,16 @@ int read_lookup_table(WIMStruct *w)
 						 "table");
 			}
 			ret = WIMLIB_ERR_READ;
-			goto out;
+			goto out_free_lookup_table;
 		}
 		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_u16(p, &cur_entry->part_number);
 		p = get_u32(p, &cur_entry->refcnt);
@@ -279,7 +408,6 @@ int read_lookup_table(WIMStruct *w)
 			      w->hdr.part_number, cur_entry->part_number);
 			ret = WIMLIB_ERR_INVALID_LOOKUP_TABLE_ENTRY;
 			goto out_free_cur_entry;
-			
 		}
 
 		if (is_zero_hash(cur_entry->hash)) {
@@ -289,62 +417,123 @@ int read_lookup_table(WIMStruct *w)
 			goto out_free_cur_entry;
 		}
 
-		duplicate_entry = __lookup_resource(table, cur_entry->hash);
-		if (duplicate_entry) {
-			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);
-			ERROR("The second entry is:");
-			print_lookup_table_entry(cur_entry);
-			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))
+		        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);
+			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);
 
+		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);
+		}
+	}
+
+	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 lookup_table_entry *lte, void *__out)
+int
+write_lookup_table_entry(struct wim_lookup_table_entry *lte, void *_out)
 {
 	FILE *out;
 	u8 buf[WIM_LOOKUP_TABLE_ENTRY_DISK_SIZE];
 	u8 *p;
 
-	out = __out;
+	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 %lu (orig size = %zu)",
+	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);
@@ -357,100 +546,173 @@ int write_lookup_table_entry(struct lookup_table_entry *lte, void *__out)
 	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)
+{
+	FILE *out = w->out_fp;
+	off_t start_offset, end_offset;
+	int ret;
+	int start_image, end_image;
+
+	start_offset = ftello(out);
+	if (start_offset == -1)
+		return WIMLIB_ERR_WRITE;
+
+	/* Write lookup table entries for metadata resources */
+	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;
+		ret = write_lookup_table_entry(metadata_lte, out);
+		if (ret)
+			return ret;
+	}
+
+	/* Write lookup table entries for other resources */
+	ret = for_lookup_table_entry(w->lookup_table, write_lookup_table_entry, out);
+	if (ret)
+		return ret;
+
+	/* Fill in the resource entry for the lookup table itself */
+	end_offset = ftello(out);
+	if (end_offset == -1)
+		return WIMLIB_ERR_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->flags         = WIM_RESHDR_FLAG_METADATA;
+	return 0;
+}
 
-int lte_zero_real_refcnt(struct lookup_table_entry *lte, void *ignore)
+int
+lte_zero_real_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 {
 	lte->real_refcnt = 0;
 	return 0;
 }
 
-int lte_zero_out_refcnt(struct lookup_table_entry *lte, void *ignore)
+int
+lte_zero_out_refcnt(struct wim_lookup_table_entry *lte, void *_ignore)
 {
 	lte->out_refcnt = 0;
 	return 0;
 }
 
-int lte_free_extracted_file(struct lookup_table_entry *lte, void *ignone)
+int
+lte_free_extracted_file(struct wim_lookup_table_entry *lte, void *_ignore)
 {
-	FREE(lte->extracted_file);
-	lte->extracted_file = NULL;
+	if (lte->extracted_file != NULL) {
+		FREE(lte->extracted_file);
+		lte->extracted_file = NULL;
+	}
 	return 0;
 }
 
-void print_lookup_table_entry(const struct lookup_table_entry *lte)
+void
+print_lookup_table_entry(const struct wim_lookup_table_entry *lte, FILE *out)
 {
 	if (!lte) {
-		putchar('\n');
+		tputc(T('\n'), out);
 		return;
 	}
-	printf("Offset            = %"PRIu64" bytes\n", 
-	       lte->resource_entry.offset);
-	printf("Size              = %"PRIu64" bytes\n", 
-	       (u64)lte->resource_entry.size);
-	printf("Original size     = %"PRIu64" bytes\n", 
-	       lte->resource_entry.original_size);
-	printf("Part Number       = %hu\n", lte->part_number);
-	printf("Reference Count   = %u\n", lte->refcnt);
-	printf("Hash              = 0x");
-	print_hash(lte->hash);
-	putchar('\n');
-	printf("Flags             = ");
+	tfprintf(out, T("Offset            = %"PRIu64" bytes\n"),
+		 lte->resource_entry.offset);
+
+	tfprintf(out, T("Size              = %"PRIu64" bytes\n"),
+		 (u64)lte->resource_entry.size);
+
+	tfprintf(out, T("Original size     = %"PRIu64" bytes\n"),
+		 lte->resource_entry.original_size);
+
+	tfprintf(out, T("Part Number       = %hu\n"), lte->part_number);
+	tfprintf(out, T("Reference Count   = %u\n"), lte->refcnt);
+
+	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;
 	if (flags & WIM_RESHDR_FLAG_COMPRESSED)
-		fputs("WIM_RESHDR_FLAG_COMPRESSED, ", stdout);
+		tfputs(T("WIM_RESHDR_FLAG_COMPRESSED, "), out);
 	if (flags & WIM_RESHDR_FLAG_FREE)
-		fputs("WIM_RESHDR_FLAG_FREE, ", stdout);
+		tfputs(T("WIM_RESHDR_FLAG_FREE, "), out);
 	if (flags & WIM_RESHDR_FLAG_METADATA)
-		fputs("WIM_RESHDR_FLAG_METADATA, ", stdout);
+		tfputs(T("WIM_RESHDR_FLAG_METADATA, "), out);
 	if (flags & WIM_RESHDR_FLAG_SPANNED)
-		fputs("WIM_RESHDR_FLAG_SPANNED, ", stdout);
-	putchar('\n');
+		tfputs(T("WIM_RESHDR_FLAG_SPANNED, "), out);
+	tputc(T('\n'), out);
 	switch (lte->resource_location) {
 	case RESOURCE_IN_WIM:
 		if (lte->wim->filename) {
-			printf("WIM file          = `%s'\n",
-			       lte->wim->filename);
+			tfprintf(out, T("WIM file          = `%"TS"'\n"),
+				 lte->wim->filename);
 		}
 		break;
+#ifdef __WIN32__
+	case RESOURCE_WIN32:
+	case RESOURCE_WIN32_ENCRYPTED:
+#endif
 	case RESOURCE_IN_FILE_ON_DISK:
-		printf("File on Disk      = `%s'\n", lte->file_on_disk);
+		tfprintf(out, T("File on Disk      = `%"TS"'\n"),
+			 lte->file_on_disk);
 		break;
 	case RESOURCE_IN_STAGING_FILE:
-		printf("Staging File      = `%s'\n", lte->staging_file_name);
+		tfprintf(out, T("Staging File      = `%"TS"'\n"),
+				lte->staging_file_name);
 		break;
 	default:
 		break;
 	}
-	putchar('\n');
+	tputc(T('\n'), out);
 }
 
-static int do_print_lookup_table_entry(struct lookup_table_entry *lte,
-				       void *ignore)
+static int
+do_print_lookup_table_entry(struct wim_lookup_table_entry *lte, void *fp)
 {
-	print_lookup_table_entry(lte);
+	print_lookup_table_entry(lte, (FILE*)fp);
 	return 0;
 }
 
 /*
- * Prints the lookup table of a WIM file. 
+ * Prints the lookup table of a WIM file.
  */
-WIMLIBAPI void wimlib_print_lookup_table(WIMStruct *w)
+WIMLIBAPI void
+wimlib_print_lookup_table(WIMStruct *w)
 {
-	for_lookup_table_entry(w->lookup_table, 
+	for_lookup_table_entry(w->lookup_table,
 			       do_print_lookup_table_entry,
-			       NULL);
+			       stdout);
 }
 
-/* 
- * Looks up an entry in the lookup table.
- */
-struct lookup_table_entry *
-__lookup_resource(const struct lookup_table *table, const u8 hash[])
+/* 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[])
 {
 	size_t i;
-	struct lookup_table_entry *lte;
+	struct wim_lookup_table_entry *lte;
 	struct hlist_node *pos;
 
+	wimlib_assert(table != NULL);
+	wimlib_assert(hash != NULL);
+
 	i = *(size_t*)hash % table->capacity;
 	hlist_for_each_entry(lte, pos, &table->array[i], hash_list)
 		if (hashes_equal(hash, lte->hash))
@@ -458,49 +720,52 @@ __lookup_resource(const struct lookup_table *table, const u8 hash[])
 	return NULL;
 }
 
-/* 
+#ifdef WITH_FUSE
+/*
  * Finds the dentry, lookup table entry, and stream index for a WIM file stream,
  * given a path name.
  *
  * This is only for pre-resolved inodes.
  */
-int lookup_resource(WIMStruct *w, const char *path,
-		    int lookup_flags,
-		    struct dentry **dentry_ret,
-		    struct lookup_table_entry **lte_ret,
-		    u16 *stream_idx_ret)
-{
-	struct dentry *dentry;
-	struct lookup_table_entry *lte;
+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)
+{
+	struct wim_dentry *dentry;
+	struct wim_lookup_table_entry *lte;
 	u16 stream_idx;
-	const char *stream_name = NULL;
-	struct inode *inode;
-	char *p = NULL;
+	const tchar *stream_name = NULL;
+	struct wim_inode *inode;
+	tchar *p = NULL;
 
 	if (lookup_flags & LOOKUP_FLAG_ADS_OK) {
 		stream_name = path_stream_name(path);
 		if (stream_name) {
-			p = (char*)stream_name - 1;
-			*p = '\0';
+			p = (tchar*)stream_name - 1;
+			*p = T('\0');
 		}
 	}
 
 	dentry = get_dentry(w, path);
 	if (p)
-		*p = ':';
+		*p = T(':');
 	if (!dentry)
-		return -ENOENT;
+		return -errno;
 
-	inode = dentry->inode;
+	inode = dentry->d_inode;
 
-	wimlib_assert(inode->resolved);
+	wimlib_assert(inode->i_resolved);
 
 	if (!(lookup_flags & LOOKUP_FLAG_DIRECTORY_OK)
 	      && inode_is_directory(inode))
 		return -EISDIR;
 
 	if (stream_name) {
-		struct ads_entry *ads_entry;
+		struct wim_ads_entry *ads_entry;
 		u16 ads_idx;
 		ads_entry = inode_get_ads_entry(inode, stream_name,
 						&ads_idx);
@@ -512,7 +777,7 @@ int lookup_resource(WIMStruct *w, const char *path,
 			return -ENOENT;
 		}
 	} else {
-		lte = inode->lte;
+		lte = inode->i_lte;
 		stream_idx = 0;
 	}
 out:
@@ -524,26 +789,9 @@ out:
 		*stream_idx_ret = stream_idx;
 	return 0;
 }
+#endif
 
-static void inode_resolve_ltes(struct inode *inode, struct lookup_table *table)
-{
-	struct lookup_table_entry *lte;
-
-	/* Resolve the default file stream */
-	lte = __lookup_resource(table, inode->hash);
-	inode->lte = lte;
-	inode->resolved = true;
-
-	/* Resolve the alternate data streams */
-	for (u16 i = 0; i < inode->num_ads; i++) {
-		struct ads_entry *cur_entry = inode->ads_entries[i];
-
-		lte = __lookup_resource(table, cur_entry->hash);
-		cur_entry->lte = lte;
-	}
-}
-
-/* Resolve a dentry's lookup table entries 
+/* 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
@@ -552,13 +800,64 @@ static void inode_resolve_ltes(struct inode *inode, struct lookup_table *table)
  * This function always succeeds; unresolved lookup table entries are given a
  * NULL pointer.
  */
-int dentry_resolve_ltes(struct dentry *dentry, void *table)
+void
+inode_resolve_ltes(struct wim_inode *inode, struct wim_lookup_table *table)
 {
-	if (!dentry->inode->resolved)
-		inode_resolve_ltes(dentry->inode, table);
-	return 0;
+
+	if (!inode->i_resolved) {
+		struct wim_lookup_table_entry *lte;
+		/* Resolve the default file stream */
+		lte = __lookup_resource(table, inode->i_hash);
+		inode->i_lte = lte;
+		inode->i_resolved = 1;
+
+		/* Resolve the alternate data streams */
+		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;
+		}
+	}
+}
+
+void
+inode_unresolve_ltes(struct wim_inode *inode)
+{
+	if (inode->i_resolved) {
+		if (inode->i_lte)
+			copy_hash(inode->i_hash, inode->i_lte->hash);
+		else
+			zero_out_hash(inode->i_hash);
+
+		for (u16 i = 0; i < inode->i_num_ads; i++) {
+			if (inode->i_ads_entries[i].lte)
+				copy_hash(inode->i_ads_entries[i].hash,
+					  inode->i_ads_entries[i].lte->hash);
+			else
+				zero_out_hash(inode->i_ads_entries[i].hash);
+		}
+		inode->i_resolved = 0;
+	}
 }
 
+/*
+ * Returns the lookup table entry for stream @stream_idx of the inode, where
+ * stream_idx = 0 means the default un-named file stream, and stream_idx >= 1
+ * corresponds to an alternate data stream.
+ *
+ * This works for both resolved and un-resolved inodes.
+ */
+struct wim_lookup_table_entry *
+inode_stream_lte(const struct wim_inode *inode, unsigned stream_idx,
+		 const struct wim_lookup_table *table)
+{
+	if (inode->i_resolved)
+		return inode_stream_lte_resolved(inode, stream_idx);
+	else
+		return inode_stream_lte_unresolved(inode, stream_idx, table);
+}
+
+
 /* Return the lookup table entry for the unnamed data stream of an inode, or
  * NULL if there is none.
  *
@@ -568,20 +867,110 @@ int dentry_resolve_ltes(struct dentry *dentry, void *table)
  * data stream in one of the alternate data streams instead of inside the WIM
  * dentry itself.  So we need to check the alternate data streams too.
  *
- * Also, note that a dentry may appear to have than one unnamed stream, but if
- * the SHA1 message digest is all 0's then the corresponding stream does not
+ * Also, note that a dentry may appear to have more than one unnamed stream, but
+ * if the SHA1 message digest is all 0's then the corresponding stream does not
  * really "count" (this is the case for the inode's own file stream when the
  * file stream that should be there is actually in one of the alternate stream
  * entries.).  This is despite the fact that we may need to extract such a
  * missing entry as an empty file or empty named data stream.
  */
-struct lookup_table_entry *
-inode_unnamed_lte(const struct inode *inode,
-		   const struct lookup_table *table)
+struct wim_lookup_table_entry *
+inode_unnamed_lte(const struct wim_inode *inode,
+		  const struct wim_lookup_table *table)
 {
-	if (inode->resolved)
+	if (inode->i_resolved)
 		return inode_unnamed_lte_resolved(inode);
 	else
 		return inode_unnamed_lte_unresolved(inode, table);
 }
 
+static int
+lte_add_stream_size(struct wim_lookup_table_entry *lte, void *total_bytes_p)
+{
+	*(u64*)total_bytes_p += lte->resource_entry.size;
+	return 0;
+}
+
+u64
+lookup_table_total_stream_size(struct wim_lookup_table *table)
+{
+	u64 total_size = 0;
+	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;
+}
+