X-Git-Url: https://wimlib.net/git/?a=blobdiff_plain;f=src%2Fdentry.c;h=7762e702caac6df5d8b975c64c4d5677255297d5;hb=af1a9f0d89f9d4428776238a561a6a5b6900f2d4;hp=bc1a524f8b1c3ee0acb07a3dbfb9ee322d152dc2;hpb=4d9742676317b6c0d62739bc7167ea70adf95440;p=wimlib

diff --git a/src/dentry.c b/src/dentry.c
index bc1a524f..7762e702 100644
--- a/src/dentry.c
+++ b/src/dentry.c
@@ -791,8 +791,12 @@ get_dentry_utf16le(WIMStruct *wim, const utf16lechar *path)
 	return cur_dentry;
 }
 
-/* Returns the dentry corresponding to the @path, or NULL if there is no such
- * dentry. */
+/*
+ * Returns the dentry in the currently selected WIM image named by @path
+ * starting from the root of the WIM image, or NULL if there is no such dentry.
+ *
+ * On Windows, the search is done case-insensitively.
+ */
 struct wim_dentry *
 get_dentry(WIMStruct *wim, const tchar *path)
 {
@@ -1082,6 +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)
@@ -1188,11 +1205,13 @@ do_free_dentry(struct wim_dentry *dentry, void *_lookup_table)
 /*
  * Unlinks and frees a dentry tree.
  *
- * @root: 		The root of the tree.
- * @lookup_table:  	The lookup table for dentries.  If non-NULL, the
- * 			reference counts in the lookup table for the lookup
- * 			table entries corresponding to the dentries will be
- * 			decremented.
+ * @root:
+ *	The root of the tree.
+ *
+ * @lookup_table:
+ *	The lookup table for dentries.  If non-NULL, the reference counts in the
+ *	lookup table for the lookup table entries corresponding to the dentries
+ *	will be decremented.
  */
 void
 free_dentry_tree(struct wim_dentry *root, struct wim_lookup_table *lookup_table)
@@ -1440,7 +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;
@@ -1487,6 +1506,15 @@ inode_add_ads_with_data(struct wim_inode *inode, const tchar *name,
 	return 0;
 }
 
+bool
+inode_has_named_stream(const struct wim_inode *inode)
+{
+	for (u16 i = 0; i < inode->i_num_ads; i++)
+		if (ads_entry_is_named_stream(&inode->i_ads_entries[i]))
+			return true;
+	return false;
+}
+
 /* Set the unnamed stream of a WIM inode, given a data buffer containing the
  * stream contents. */
 int
@@ -1527,6 +1555,15 @@ inode_remove_ads(struct wim_inode *inode, u16 idx,
 	inode->i_num_ads--;
 }
 
+bool
+inode_has_unix_data(const struct wim_inode *inode)
+{
+	for (u16 i = 0; i < inode->i_num_ads; i++)
+		if (ads_entry_is_unix_data(&inode->i_ads_entries[i]))
+			return true;
+	return false;
+}
+
 #ifndef __WIN32__
 int
 inode_get_unix_data(const struct wim_inode *inode,
@@ -1602,19 +1639,24 @@ inode_set_unix_data(struct wim_inode *inode, uid_t uid, gid_t gid, mode_t mode,
 /*
  * Reads the alternate data stream entries of a WIM dentry.
  *
- * @p:	Pointer to buffer that starts with the first alternate stream entry.
+ * @p:
+ *	Pointer to buffer that starts with the first alternate stream entry.
  *
- * @inode:	Inode to load the alternate data streams into.
- * 		@inode->i_num_ads must have been set to the number of
- * 		alternate data streams that are expected.
+ * @inode:
+ *	Inode to load the alternate data streams into.  @inode->i_num_ads must
+ *	have been set to the number of alternate data streams that are expected.
  *
- * @remaining_size:	Number of bytes of data remaining in the buffer pointed
- * 			to by @p.
+ * @remaining_size:
+ *	Number of bytes of data remaining in the buffer pointed to by @p.
  *
+ * On success, inode->i_ads_entries is set to an array of `struct
+ * wim_ads_entry's of length inode->i_num_ads.  On failure, @inode is not
+ * modified.
  *
- * Return 0 on success or nonzero on failure.  On success, inode->i_ads_entries
- * is set to an array of `struct wim_ads_entry's of length inode->i_num_ads.  On
- * failure, @inode is not modified.
+ * Return values:
+ *	WIMLIB_ERR_SUCCESS (0)
+ *	WIMLIB_ERR_INVALID_METADATA_RESOURCE
+ *	WIMLIB_ERR_NOMEM
  */
 static int
 read_ads_entries(const u8 * restrict p, struct wim_inode * restrict inode,
@@ -1718,7 +1760,7 @@ out_of_memory:
 	goto out_free_ads_entries;
 out_invalid:
 	ERROR("An alternate data stream entry is invalid");
-	ret = WIMLIB_ERR_INVALID_DENTRY;
+	ret = WIMLIB_ERR_INVALID_METADATA_RESOURCE;
 out_free_ads_entries:
 	if (ads_entries) {
 		for (u16 i = 0; i < num_ads; i++)
@@ -1739,7 +1781,7 @@ out:
  * @metadata_resource_len:
  *		Length of the metadata resource buffer, in bytes.
  *
- * @offset: 	Offset of the dentry within the metadata resource.
+ * @offset:	Offset of the dentry within the metadata resource.
  *
  * @dentry:	A `struct wim_dentry' that will be filled in by this function.
  *
@@ -1749,9 +1791,10 @@ out:
  * this was a special "end of directory" dentry and not a real dentry.  If
  * nonzero, this was a real dentry.
  *
- * Possible errors include:
- * 	WIMLIB_ERR_NOMEM
- * 	WIMLIB_ERR_INVALID_DENTRY
+ * Return values:
+ *	WIMLIB_ERR_SUCCESS (0)
+ *	WIMLIB_ERR_INVALID_METADATA_RESOURCE
+ *	WIMLIB_ERR_NOMEM
  */
 int
 read_dentry(const u8 * restrict metadata_resource, u64 metadata_resource_len,
@@ -1785,7 +1828,7 @@ read_dentry(const u8 * restrict metadata_resource, u64 metadata_resource_len,
 		ERROR("Directory entry starting at %"PRIu64" ends past the "
 		      "end of the metadata resource (size %"PRIu64")",
 		      offset, metadata_resource_len);
-		return WIMLIB_ERR_INVALID_DENTRY;
+		return WIMLIB_ERR_INVALID_METADATA_RESOURCE;
 	}
 	dentry->length = le64_to_cpu(disk_dentry->length);
 
@@ -1807,7 +1850,7 @@ read_dentry(const u8 * restrict metadata_resource, u64 metadata_resource_len,
 		      "%"PRIu64" ends past the end of the metadata resource "
 		      "(size %"PRIu64")",
 		      offset, dentry->length, metadata_resource_len);
-		return WIMLIB_ERR_INVALID_DENTRY;
+		return WIMLIB_ERR_INVALID_METADATA_RESOURCE;
 	}
 
 	/* Make sure the dentry length is at least as large as the number of
@@ -1815,7 +1858,7 @@ read_dentry(const u8 * restrict metadata_resource, u64 metadata_resource_len,
 	if (dentry->length < sizeof(struct wim_dentry_on_disk)) {
 		ERROR("Directory entry has invalid length of %"PRIu64" bytes",
 		      dentry->length);
-		return WIMLIB_ERR_INVALID_DENTRY;
+		return WIMLIB_ERR_INVALID_METADATA_RESOURCE;
 	}
 
 	/* Allocate a `struct wim_inode' for this `struct wim_dentry'. */
@@ -1861,7 +1904,7 @@ read_dentry(const u8 * restrict metadata_resource, u64 metadata_resource_len,
 	if ((short_name_nbytes & 1) | (file_name_nbytes & 1))
 	{
 		ERROR("Dentry name is not valid UTF-16LE (odd number of bytes)!");
-		ret = WIMLIB_ERR_INVALID_DENTRY;
+		ret = WIMLIB_ERR_INVALID_METADATA_RESOURCE;
 		goto out_free_inode;
 	}
 
@@ -1878,7 +1921,7 @@ read_dentry(const u8 * restrict metadata_resource, u64 metadata_resource_len,
 		ERROR("Unexpected end of directory entry! (Expected "
 		      "at least %"PRIu64" bytes, got %"PRIu64" bytes.)",
 		      calculated_size, dentry->length);
-		ret = WIMLIB_ERR_INVALID_DENTRY;
+		ret = WIMLIB_ERR_INVALID_METADATA_RESOURCE;
 		goto out_free_inode;
 	}
 
@@ -1932,7 +1975,7 @@ read_dentry(const u8 * restrict metadata_resource, u64 metadata_resource_len,
 	 * be included in the dentry->length field for some reason.
 	 */
 	if (inode->i_num_ads != 0) {
-		ret = WIMLIB_ERR_INVALID_DENTRY;
+		ret = WIMLIB_ERR_INVALID_METADATA_RESOURCE;
 		if (offset + dentry->length > metadata_resource_len ||
 		    (ret = read_ads_entries(&metadata_resource[offset + dentry->length],
 					    inode,
@@ -1977,18 +2020,22 @@ dentry_get_file_type_string(const struct wim_dentry *dentry)
 /* Reads the children of a dentry, and all their children, ..., etc. from the
  * metadata resource and into the dentry tree.
  *
- * @metadata_resource:	An array that contains the uncompressed metadata
- * 			resource for the WIM file.
+ * @metadata_resource:
+ *	An array that contains the uncompressed metadata resource for the WIM
+ *	file.
  *
- * @metadata_resource_len:  The length of the uncompressed metadata resource, in
- * 			    bytes.
+ * @metadata_resource_len:
+ *	The length of the uncompressed metadata resource, in bytes.
  *
- * @dentry:	A pointer to a `struct wim_dentry' that is the root of the directory
- *		tree and has already been read from the metadata resource.  It
- *		does not need to be the real root because this procedure is
- *		called recursively.
+ * @dentry:
+ *	A pointer to a `struct wim_dentry' that is the root of the directory
+ *	tree and has already been read from the metadata resource.  It does not
+ *	need to be the real root because this procedure is called recursively.
  *
- * Returns zero on success; nonzero on failure.
+ * Return values:
+ *	WIMLIB_ERR_SUCCESS (0)
+ *	WIMLIB_ERR_INVALID_METADATA_RESOURCE
+ *	WIMLIB_ERR_NOMEM
  */
 int
 read_dentry_tree(const u8 * restrict metadata_resource,
@@ -2019,7 +2066,7 @@ read_dentry_tree(const u8 * restrict metadata_resource,
 			      "of \"%"TS"\" coincide with children of \"%"TS"\"",
 			      dentry_full_path(dentry),
 			      dentry_full_path(parent));
-			return WIMLIB_ERR_INVALID_DENTRY;
+			return WIMLIB_ERR_INVALID_METADATA_RESOURCE;
 		}
 	}
 
@@ -2113,7 +2160,7 @@ write_dentry(const struct wim_dentry * restrict dentry, u8 * restrict p)
 	wimlib_assert(((uintptr_t)p & 7) == 0); /* 8 byte aligned */
 	orig_p = p;
 
- 	inode = dentry->d_inode;
+	inode = dentry->d_inode;
 	disk_dentry = (struct wim_dentry_on_disk*)p;
 
 	disk_dentry->attributes = cpu_to_le32(inode->i_attributes);
@@ -2259,7 +2306,7 @@ write_dentry_tree(const struct wim_dentry * restrict root, u8 * restrict p)
 
 
 static int
-init_wimlib_dentry(struct wimlib_wim_dentry *wdentry,
+init_wimlib_dentry(struct wimlib_dir_entry *wdentry,
 		   struct wim_dentry *dentry,
 		   const WIMStruct *wim)
 {
@@ -2312,7 +2359,7 @@ init_wimlib_dentry(struct wimlib_wim_dentry *wdentry,
 
 	lte = inode_unnamed_lte(inode, wim->lookup_table);
 	if (lte)
-		wdentry->streams[0].stream_size = wim_resource_size(lte);
+		lte_to_wimlib_resource_entry(lte, &wdentry->streams[0].resource);
 
 	for (unsigned i = 0; i < inode->i_num_ads; i++) {
 		if (inode->i_ads_entries[i].stream_name == NULL)
@@ -2320,8 +2367,8 @@ init_wimlib_dentry(struct wimlib_wim_dentry *wdentry,
 		lte = inode_stream_lte(inode, i + 1, wim->lookup_table);
 		wdentry->num_named_streams++;
 		if (lte) {
-			wdentry->streams[wdentry->num_named_streams].stream_size =
-					wim_resource_size(lte);
+			lte_to_wimlib_resource_entry(lte, &wdentry->streams[
+								wdentry->num_named_streams].resource);
 		}
 	#if TCHAR_IS_UTF16LE
 		wdentry->streams[wdentry->num_named_streams].stream_name =
@@ -2332,7 +2379,7 @@ init_wimlib_dentry(struct wimlib_wim_dentry *wdentry,
 		ret = utf16le_to_tstr(inode->i_ads_entries[i].stream_name,
 				      inode->i_ads_entries[i].stream_name_nbytes,
 				      (tchar**)&wdentry->streams[
-				      		wdentry->num_named_streams].stream_name,
+						wdentry->num_named_streams].stream_name,
 				      &dummy);
 		if (ret)
 			return ret;
@@ -2342,7 +2389,7 @@ init_wimlib_dentry(struct wimlib_wim_dentry *wdentry,
 }
 
 static void
-free_wimlib_dentry(struct wimlib_wim_dentry *wdentry)
+free_wimlib_dentry(struct wimlib_dir_entry *wdentry)
 {
 #if !TCHAR_IS_UTF16LE
 	FREE((tchar*)wdentry->filename);
@@ -2380,14 +2427,13 @@ do_iterate_dir_tree(WIMStruct *wim,
 		    wimlib_iterate_dir_tree_callback_t cb,
 		    void *user_ctx)
 {
-	u32 level;
-	struct wimlib_wim_dentry *wdentry;
+	struct wimlib_dir_entry *wdentry;
 	int ret = WIMLIB_ERR_NOMEM;
 
 
-	wdentry = CALLOC(1, sizeof(struct wimlib_wim_dentry) +
-			       	  (1 + dentry->d_inode->i_num_ads) *
-				  	sizeof(struct wimlib_stream_entry));
+	wdentry = CALLOC(1, sizeof(struct wimlib_dir_entry) +
+				  (1 + dentry->d_inode->i_num_ads) *
+					sizeof(struct wimlib_stream_entry));
 	if (!wdentry)
 		goto out;
 
@@ -2438,13 +2484,12 @@ image_do_iterate_dir_tree(WIMStruct *wim)
 	return do_iterate_dir_tree(wim, dentry, ctx->flags, ctx->cb, ctx->user_ctx);
 }
 
+/* API function documented in wimlib.h  */
 WIMLIBAPI int
 wimlib_iterate_dir_tree(WIMStruct *wim, int image, const tchar *path,
 			int flags,
 			wimlib_iterate_dir_tree_callback_t cb, void *user_ctx)
 {
-	int ret;
-	struct wim_dentry *dentry;
 	struct image_iterate_dir_tree_ctx ctx = {
 		.path = path,
 		.flags = flags,
@@ -2454,3 +2499,211 @@ 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;
+}