Add new sorter

packages/playground/data-liberation/phpunit.xml

@@ -15,7 +15,7 @@
       <file>tests/WPXMLProcessorTests.php</file>
       <file>tests/UrldecodeNTests.php</file>
       <file>tests/WPStreamImporterTests.php</file>
-      <file>tests/WPTopologicalSorterTests.php</file>
+      <file>tests/WPWXRSortedReaderTests.php</file>
     </testsuite>
   </testsuites>
 </phpunit>

packages/playground/data-liberation/plugin.php

@@ -64,17 +64,20 @@ function data_liberation_init() {
 add_action( 'init', 'data_liberation_init' );
 
 function data_liberation_activate() {
-	// Activate the topological sorter. Create tables and options.
-	WP_Topological_Sorter::activate();
-	update_option( WP_Topological_Sorter::OPTION_NAME, WP_Topological_Sorter::DB_VERSION );
+	// Create tables and option.
+	WP_WXR_Sorted_Reader::create_or_update_db();
+	update_option( 'data_liberation_db_version', WP_WXR_Sorted_Reader::DB_VERSION );
 }
 
 // Run when the plugin is activated.
 register_activation_hook( __FILE__, 'data_liberation_activate' );
 
 function data_liberation_deactivate() {
-	// Deactivate the topological sorter. Flush away all data.
-	WP_Topological_Sorter::deactivate();
+	// Flush away all data.
+	WP_WXR_Sorted_Reader::delete_db();
+
+	// Delete the option.
+	delete_option( 'data_liberation_db_version' );
 
 	// @TODO: Cancel any active import sessions and cleanup other data.
 }
@@ -83,10 +86,10 @@ function data_liberation_deactivate() {
 register_deactivation_hook( __FILE__, 'data_liberation_deactivate' );
 
 function data_liberation_load() {
-	if ( WP_Topological_Sorter::DB_VERSION !== (int) get_site_option( WP_Topological_Sorter::OPTION_NAME ) ) {
+	if ( WP_WXR_Sorted_Reader::DB_VERSION !== (int) get_site_option( 'data_liberation_db_version' ) ) {
 		// Update the database with dbDelta, if needed in the future.
-		WP_Topological_Sorter::activate();
-		update_option( WP_Topological_Sorter::OPTION_NAME, WP_Topological_Sorter::DB_VERSION );
+		WP_WXR_Sorted_Reader::create_or_update_db();
+		update_option( 'data_liberation_db_version', WP_WXR_Sorted_Reader::DB_VERSION );
 	}
 }
 
@@ -458,15 +461,15 @@ function data_liberation_create_importer( $import ) {
 			}
 			$importer = WP_Stream_Importer::create_for_wxr_file(
 				$wxr_path,
-				array(),
+				$import,
 				$import['cursor'] ?? null
 			);
 			break;
 
 		case 'wxr_url':
 			$importer = WP_Stream_Importer::create_for_wxr_url(
 				$import['wxr_url'],
-				array(),
+				$import,
 				$import['cursor'] ?? null
 			);
 			break;

packages/playground/data-liberation/src/entity-readers/WP_WXR_Entity_Reader.php

@@ -133,7 +133,7 @@ class WP_WXR_Entity_Reader extends WP_Entity_Reader {
 	 * @since WP_VERSION
 	 * @var WP_XML_Processor
 	 */
-	private $xml;
+	protected $xml;
 
 	/**
 	 * The name of the XML tag containing information about the WordPress entity
@@ -206,23 +206,23 @@ class WP_WXR_Entity_Reader extends WP_Entity_Reader {
 	 * @since WP_VERSION
 	 * @var int|null
 	 */
-	private $last_post_id = null;
+	protected $last_post_id = null;
 
 	/**
 	 * The ID of the last processed comment.
 	 *
 	 * @since WP_VERSION
 	 * @var int|null
 	 */
-	private $last_comment_id = null;
+	protected $last_comment_id = null;
 
 	/**
 	 * The ID of the last processed term.
 	 *
 	 * @since WP_VERSION
 	 * @var int|null
 	 */
-	private $last_term_id = null;
+	protected $last_term_id = null;
 
 	/**
 	 * Buffer for accumulating text content between tags.
@@ -367,7 +367,7 @@ class WP_WXR_Entity_Reader extends WP_Entity_Reader {
 		),
 	);
 
-	public static function create( WP_Byte_Reader $upstream = null, $cursor = null ) {
+	public static function create( WP_Byte_Reader $upstream = null, $cursor = null, $options = array() ) {
 		$xml_cursor = null;
 		if ( null !== $cursor ) {
 			$cursor = json_decode( $cursor, true );
@@ -383,7 +383,7 @@ public static function create( WP_Byte_Reader $upstream = null, $cursor = null )
 		}
 
 		$xml    = WP_XML_Processor::create_for_streaming( '', $xml_cursor );
-		$reader = new WP_WXR_Entity_Reader( $xml );
+		$reader = new static( $xml );
 		if ( null !== $cursor ) {
 			$reader->last_post_id    = $cursor['last_post_id'];
 			$reader->last_comment_id = $cursor['last_comment_id'];
@@ -416,10 +416,6 @@ protected function __construct( WP_XML_Processor $xml ) {
 		$this->xml = $xml;
 	}
 
-	public function get_last_xml_byte_offset_outside_of_entity() {
-		return $this->last_xml_byte_offset_outside_of_entity;
-	}
-
 	public function get_reentrancy_cursor() {
 		/**
 		 * @TODO: Instead of adjusting the XML cursor internals, adjust the get_reentrancy_cursor()
@@ -593,7 +589,7 @@ public function next_entity() {
 	 *
 	 * @return bool Whether another entity was found.
 	 */
-	private function read_next_entity() {
+	protected function read_next_entity() {
 		if ( $this->xml->is_finished() ) {
 			$this->after_entity();
 			return false;