X-Git-Url: https://wimlib.net/git/?p=wimlib;a=blobdiff_plain;f=src%2Fdentry.c;h=af27c68776459c2d1272656b42a2b4b9e84039df;hp=c9a84d718eee2952197886e634cf1fcb12594335;hb=2799ba2de8e4409158ed47828a3be9ac26859947;hpb=61db93f82eca3fe9f7676355c709c58cc425a6ad

diff --git a/src/dentry.c b/src/dentry.c
index c9a84d71..af27c687 100644
--- a/src/dentry.c
+++ b/src/dentry.c
@@ -1086,6 +1086,19 @@ new_filler_directory(const tchar *name, struct wim_dentry **dentry_ret)
 	return 0;
 }
 
+static int
+dentry_clear_inode_visited(struct wim_dentry *dentry, void *_ignore)
+{
+	dentry->d_inode->i_visited = 0;
+	return 0;
+}
+
+void
+dentry_tree_clear_inode_visited(struct wim_dentry *root)
+{
+	for_dentry_in_tree(root, dentry_clear_inode_visited, NULL);
+}
+
 static int
 init_ads_entry(struct wim_ads_entry *ads_entry, const void *name,
 	       size_t name_nbytes, bool is_utf16le)
@@ -1446,7 +1459,7 @@ add_stream_from_data_buffer(const void *buffer, size_t size,
 	struct wim_lookup_table_entry *lte, *existing_lte;
 
 	sha1_buffer(buffer, size, hash);
-	existing_lte = __lookup_resource(lookup_table, hash);
+	existing_lte = lookup_resource(lookup_table, hash);
 	if (existing_lte) {
 		wimlib_assert(wim_resource_size(existing_lte) == size);
 		lte = existing_lte;
@@ -2295,12 +2308,14 @@ write_dentry_tree(const struct wim_dentry * restrict root, u8 * restrict p)
 static int
 init_wimlib_dentry(struct wimlib_dir_entry *wdentry,
 		   struct wim_dentry *dentry,
-		   const WIMStruct *wim)
+		   const WIMStruct *wim,
+		   int flags)
 {
 	int ret;
 	size_t dummy;
 	const struct wim_inode *inode = dentry->d_inode;
 	struct wim_lookup_table_entry *lte;
+	const u8 *hash;
 
 #if TCHAR_IS_UTF16LE
 	wdentry->filename = dentry->file_name;
@@ -2345,17 +2360,30 @@ init_wimlib_dentry(struct wimlib_dir_entry *wdentry,
 	wdentry->last_access_time = wim_timestamp_to_timespec(inode->i_last_access_time);
 
 	lte = inode_unnamed_lte(inode, wim->lookup_table);
-	if (lte)
+	if (lte) {
 		lte_to_wimlib_resource_entry(lte, &wdentry->streams[0].resource);
+	} else if (!is_zero_hash(hash = inode_unnamed_stream_hash(inode))) {
+		if (flags & WIMLIB_ITERATE_DIR_TREE_FLAG_RESOURCES_NEEDED)
+			return resource_not_found_error(inode, hash);
+		copy_hash(wdentry->streams[0].resource.sha1_hash, hash);
+		wdentry->streams[0].resource.is_missing = 1;
+	}
 
 	for (unsigned i = 0; i < inode->i_num_ads; i++) {
-		if (inode->i_ads_entries[i].stream_name == NULL)
+		if (!ads_entry_is_named_stream(&inode->i_ads_entries[i]))
 			continue;
 		lte = inode_stream_lte(inode, i + 1, wim->lookup_table);
 		wdentry->num_named_streams++;
 		if (lte) {
 			lte_to_wimlib_resource_entry(lte, &wdentry->streams[
 								wdentry->num_named_streams].resource);
+		} else if (!is_zero_hash(hash = inode_stream_hash(inode, i + 1))) {
+			if (flags & WIMLIB_ITERATE_DIR_TREE_FLAG_RESOURCES_NEEDED)
+				return resource_not_found_error(inode, hash);
+			copy_hash(wdentry->streams[
+				  wdentry->num_named_streams].resource.sha1_hash, hash);
+			wdentry->streams[
+				wdentry->num_named_streams].resource.is_missing = 1;
 		}
 	#if TCHAR_IS_UTF16LE
 		wdentry->streams[wdentry->num_named_streams].stream_name =
@@ -2424,7 +2452,7 @@ do_iterate_dir_tree(WIMStruct *wim,
 	if (!wdentry)
 		goto out;
 
-	ret = init_wimlib_dentry(wdentry, dentry, wim);
+	ret = init_wimlib_dentry(wdentry, dentry, wim, flags);
 	if (ret)
 		goto out_free_wimlib_dentry;
 
@@ -2486,3 +2514,211 @@ wimlib_iterate_dir_tree(WIMStruct *wim, int image, const tchar *path,
 	wim->private = &ctx;
 	return for_image(wim, image, image_do_iterate_dir_tree);
 }
+
+/* Returns %true iff the metadata of @inode and @template_inode are reasonably
+ * consistent with them being the same, unmodified file.  */
+static bool
+inode_metadata_consistent(const struct wim_inode *inode,
+			  const struct wim_inode *template_inode,
+			  const struct wim_lookup_table *template_lookup_table)
+{
+	/* Must have exact same creation time and last write time.  */
+	if (inode->i_creation_time != template_inode->i_creation_time ||
+	    inode->i_last_write_time != template_inode->i_last_write_time)
+		return false;
+
+	/* Last access time may have stayed the same or increased, but certainly
+	 * shouldn't have decreased.  */
+	if (inode->i_last_access_time < template_inode->i_last_access_time)
+		return false;
+
+	/* Must have same number of alternate data stream entries.  */
+	if (inode->i_num_ads != template_inode->i_num_ads)
+		return false;
+
+	/* If the stream entries for the inode are for some reason not resolved,
+	 * then the hashes are already available and the point of this function
+	 * is defeated.  */
+	if (!inode->i_resolved)
+		return false;
+
+	/* Iterate through each stream and do some more checks.  */
+	for (unsigned i = 0; i <= inode->i_num_ads; i++) {
+		const struct wim_lookup_table_entry *lte, *template_lte;
+
+		lte = inode_stream_lte_resolved(inode, i);
+		template_lte = inode_stream_lte(template_inode, i,
+						template_lookup_table);
+
+		/* Compare stream sizes.  */
+		if (lte && template_lte) {
+			if (wim_resource_size(lte) != wim_resource_size(template_lte))
+				return false;
+
+			/* If hash happens to be available, compare with template.  */
+			if (!lte->unhashed && !template_lte->unhashed &&
+			    !hashes_equal(lte->hash, template_lte->hash))
+				return false;
+
+		} else if (lte && wim_resource_size(lte)) {
+			return false;
+		} else if (template_lte && wim_resource_size(template_lte)) {
+			return false;
+		}
+	}
+
+	/* All right, barring a full checksum and given that the inodes share a
+	 * path and the user isn't trying to trick us, these inodes most likely
+	 * refer to the same file.  */
+	return true;
+}
+
+/**
+ * Given an inode @inode that has been determined to be "the same" as another
+ * inode @template_inode in either the same WIM or another WIM, retrieve some
+ * useful stream information (e.g. checksums) from @template_inode.
+ *
+ * This assumes that the streams for @inode have been resolved (to point
+ * directly to the appropriate `struct wim_lookup_table_entry's)  but do not
+ * necessarily have checksum information filled in.
+ */
+static int
+inode_copy_checksums(struct wim_inode *inode,
+		     struct wim_inode *template_inode,
+		     WIMStruct *wim,
+		     WIMStruct *template_wim)
+{
+	for (unsigned i = 0; i <= inode->i_num_ads; i++) {
+		struct wim_lookup_table_entry *lte, *template_lte;
+		struct wim_lookup_table_entry *replace_lte;
+
+		lte = inode_stream_lte_resolved(inode, i);
+		template_lte = inode_stream_lte(template_inode, i,
+						template_wim->lookup_table);
+
+		/* Only take action if both entries exist, the entry for @inode
+		 * has no checksum calculated, but the entry for @template_inode
+		 * does.  */
+		if (!lte || !template_lte ||
+		    !lte->unhashed || template_lte->unhashed)
+			continue;
+
+		wimlib_assert(lte->refcnt == inode->i_nlink);
+
+		/* If the WIM of the template image is the same as the WIM of
+		 * the new image, then @template_lte can be used directly.
+		 *
+		 * Otherwise, look for a stream with the same hash in the WIM of
+		 * the new image.  If found, use it; otherwise re-use the entry
+		 * being discarded, filling in the hash.  */
+
+		if (wim == template_wim)
+			replace_lte = template_lte;
+		else
+			replace_lte = lookup_resource(wim->lookup_table,
+						      template_lte->hash);
+
+		list_del(&lte->unhashed_list);
+		if (replace_lte) {
+			free_lookup_table_entry(lte);
+		} else {
+			copy_hash(lte->hash, template_lte->hash);
+			lte->unhashed = 0;
+			lookup_table_insert(wim->lookup_table, lte);
+			lte->refcnt = 0;
+			replace_lte = lte;
+		}
+
+		if (i == 0)
+			inode->i_lte = replace_lte;
+		else
+			inode->i_ads_entries[i - 1].lte = replace_lte;
+
+		replace_lte->refcnt += inode->i_nlink;
+	}
+	return 0;
+}
+
+struct reference_template_args {
+	WIMStruct *wim;
+	WIMStruct *template_wim;
+};
+
+static int
+dentry_reference_template(struct wim_dentry *dentry, void *_args)
+{
+	int ret;
+	struct wim_dentry *template_dentry;
+	struct wim_inode *inode, *template_inode;
+	struct reference_template_args *args = _args;
+	WIMStruct *wim = args->wim;
+	WIMStruct *template_wim = args->template_wim;
+
+	if (dentry->d_inode->i_visited)
+		return 0;
+
+	ret = calculate_dentry_full_path(dentry);
+	if (ret)
+		return ret;
+
+	template_dentry = get_dentry(template_wim, dentry->_full_path);
+	if (!template_dentry) {
+		DEBUG("\"%"TS"\": newly added file", dentry->_full_path);
+		return 0;
+	}
+
+	inode = dentry->d_inode;
+	template_inode = template_dentry->d_inode;
+
+	if (inode_metadata_consistent(inode, template_inode,
+				      template_wim->lookup_table)) {
+		/*DEBUG("\"%"TS"\": No change detected", dentry->_full_path);*/
+		ret = inode_copy_checksums(inode, template_inode,
+					   wim, template_wim);
+		inode->i_visited = 1;
+	} else {
+		DEBUG("\"%"TS"\": change detected!", dentry->_full_path);
+		ret = 0;
+	}
+	return ret;
+}
+
+/* API function documented in wimlib.h  */
+WIMLIBAPI int
+wimlib_reference_template_image(WIMStruct *wim, int new_image,
+				WIMStruct *template_wim, int template_image,
+				int flags, wimlib_progress_func_t progress_func)
+{
+	int ret;
+	struct wim_image_metadata *new_imd;
+
+	if (wim == NULL || template_wim == NULL)
+		return WIMLIB_ERR_INVALID_PARAM;
+
+	if (wim == template_wim && new_image == template_image)
+		return WIMLIB_ERR_INVALID_PARAM;
+
+	if (new_image < 1 || new_image > wim->hdr.image_count)
+		return WIMLIB_ERR_INVALID_IMAGE;
+
+	if (!wim_has_metadata(wim))
+		return WIMLIB_ERR_METADATA_NOT_FOUND;
+
+	new_imd = wim->image_metadata[new_image - 1];
+	if (!new_imd->modified)
+		return WIMLIB_ERR_INVALID_PARAM;
+
+	ret = select_wim_image(template_wim, template_image);
+	if (ret)
+		return ret;
+
+	struct reference_template_args args = {
+		.wim = wim,
+		.template_wim = template_wim,
+	};
+
+	ret = for_dentry_in_tree(new_imd->root_dentry,
+				 dentry_reference_template, &args);
+	dentry_tree_clear_inode_visited(new_imd->root_dentry);
+	return ret;
+}