From: Michael Tremer <michael.tremer@ipfire.org>
Date: Fri, 10 Mar 2023 16:34:48 +0000 (+0000)
Subject: archive: Be more efficient when reading single files
X-Git-Tag: 0.9.29~332
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=2a914a843da0f53cd6bb43074d55d8509d7a0c89;p=pakfire.git

archive: Be more efficient when reading single files

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
---

diff --git a/src/libpakfire/archive.c b/src/libpakfire/archive.c
index f840e1afd..69849ec76 100644
--- a/src/libpakfire/archive.c
+++ b/src/libpakfire/archive.c
@@ -525,13 +525,6 @@ static int64_t pakfire_archive_metadata_get_int64(
 	return json_object_get_int64(object);
 }
 
-struct pakfire_archive_read {
-	struct pakfire_archive* archive;
-	const char* path;
-	char** data;
-	size_t* length;
-};
-
 static int __pakfire_archive_filter_payload(struct pakfire* pakfire,
 		struct archive* a, struct archive_entry* entry, void* p) {
 	const char* path = archive_entry_pathname(entry);
@@ -556,44 +549,62 @@ static int __pakfire_archive_filter_payload(struct pakfire* pakfire,
 
 PAKFIRE_EXPORT int pakfire_archive_read(struct pakfire_archive* archive,
 		const char* path, char** data, size_t* length) {
+	int found = 0;
 	int r;
 
-	// Strip leading / from filenames, because tarballs don't use any leading slashes
-	path = pakfire_path_relpath("/", path);
+	// Check if path is absolute
+	if (!path || *path != '/') {
+		errno = EINVAL;
+		return 1;
 
-	// Check inputs
-	if (!path || !data || !length) {
+	// Check other inputs
+	} else if (!data || !length) {
 		errno = EINVAL;
 		return 1;
 	}
 
-	// Reset data
-	*data = NULL;
+	// Strip leading / from filenames, because tarballs don't use any leading slashes
+	path = pakfire_path_relpath("/", path);
+	if (!path)
+		return 1;
 
-	int __pakfire_archive_read(struct pakfire* pakfire, struct archive* a,
+	// Tries to find a matching file in the archive
+	int __pakfire_archive_read_filter(struct pakfire* pakfire, struct archive* a,
 			struct archive_entry* e, void* __data) {
-		// Fetch the path of this entry
+		// Stop reading the archive after we have found our file
+		if (found)
+			return PAKFIRE_WALK_END;
+
+		// Fetch path
 		const char* p = archive_entry_pathname(e);
 		if (!p)
-			return 1;
+			return PAKFIRE_WALK_ERROR;
 
-		if (strcmp(path, p) == 0) {
-			r = pakfire_archive_copy_data_to_buffer(pakfire, a, e, data, length);
-			if (r)
-				return r;
-		}
+		// We found a match
+		if (strcmp(path, p) == 0)
+			return PAKFIRE_WALK_OK;
 
-		return 0;
+		// Otherwise we skip the file
+		return PAKFIRE_WALK_SKIP;
+	}
+
+	// Reads a matching file into memory
+	int __pakfire_archive_read(struct pakfire* pakfire, struct archive* a,
+			struct archive_entry* e, void* __data) {
+		// We have found our file
+		found = 1;
+
+		return pakfire_archive_copy_data_to_buffer(pakfire, a, e, data, length);
 	}
 
 	// Walk through the archive
 	r = pakfire_archive_walk(archive,
-		__pakfire_archive_read, __pakfire_archive_filter_payload, NULL);
+		__pakfire_archive_read, __pakfire_archive_read_filter, NULL);
 	if (r)
 		return r;
 
 	// Nothing found
-	if (!*data) {
+	if (!found) {
 		// No such file or directory
 		errno = ENOENT;