Fix WXRReader reentrancy tests by correctly matching '*' breadcrumbs to namespaced tags

Author: adamziel

components/DataLiberation/EntityReader/WXREntityReader.php

@@ -171,15 +171,7 @@ class WXREntityReader implements EntityReader {
 	 * @since WP_VERSION
 	 * @var int
 	 */
-	private $last_xml_byte_offset_outside_of_entity;
-
-	/**
-	 * The XML processor cursor of the last entity opener.
-	 *
-	 * @since WP_VERSION
-	 * @var string|null
-	 */
-	private $last_xml_cursor_outside_of_entity;
+	private $entity_opener_byte_offset;
 
 	/**
 	 * Whether the current entity has been emitted.
@@ -431,13 +423,13 @@ public function get_reentrancy_cursor() {
 		 */
 		$xml_cursor                             = $this->xml->get_reentrancy_cursor();
 		$xml_cursor                             = json_decode( base64_decode( $xml_cursor ), true );
-		$xml_cursor['upstream_bytes_forgotten'] = $this->last_xml_byte_offset_outside_of_entity;
+		$xml_cursor['upstream_bytes_forgotten'] = $this->entity_opener_byte_offset;
 		$xml_cursor                             = base64_encode( json_encode( $xml_cursor ) );
 
 		return json_encode(
 			array(
 				'xml'             => $xml_cursor,
-				'upstream'        => $this->last_xml_byte_offset_outside_of_entity,
+				'upstream'        => $this->entity_opener_byte_offset,
 				'last_post_id'    => $this->last_post_id,
 				'last_comment_id' => $this->last_comment_id,
 			)
@@ -663,8 +655,7 @@ private function read_next_entity() {
 			}
 
 			if ( count( $breadcrumbs ) <= 2 && $this->xml->is_tag_opener() ) {
-				$this->last_xml_byte_offset_outside_of_entity = $this->xml->get_token_byte_offset_in_the_input_stream();
-				$this->last_xml_cursor_outside_of_entity      = $this->xml->get_reentrancy_cursor();
+				$this->entity_opener_byte_offset = $this->xml->get_token_byte_offset_in_the_input_stream();
 			}
 
 			$tag_with_namespace = $this->xml->get_tag_name_with_namespace();
@@ -697,8 +688,7 @@ private function read_next_entity() {
 				// the previous entity is finished.
 				if ( $this->xml->is_tag_opener() ) {
 					$this->set_entity_tag( $tag_with_namespace );
-					$this->last_xml_byte_offset_outside_of_entity = $this->xml->get_token_byte_offset_in_the_input_stream();
-					$this->last_xml_cursor_outside_of_entity      = $this->xml->get_reentrancy_cursor();
+					$this->entity_opener_byte_offset = $this->xml->get_token_byte_offset_in_the_input_stream();
 				}
 				continue;
 			}
@@ -753,8 +743,9 @@ private function read_next_entity() {
 					$this->xml->matches_breadcrumbs( array( 'rss', 'channel', '*' ) ) &&
 					array_key_exists( $this->xml->get_tag_name_with_namespace(), static::KNOWN_SITE_OPTIONS )
 				);
+				
 				if ( $is_site_option_opener ) {		
-					$this->last_xml_byte_offset_outside_of_entity = $this->xml->get_token_byte_offset_in_the_input_stream();
+					$this->entity_opener_byte_offset = $this->xml->get_token_byte_offset_in_the_input_stream();
 				}
 				continue;
 			}

components/XML/Tests/XMLProcessorTest.php

@@ -1461,6 +1461,31 @@ public function test_matches_breadcrumbs() {
 		$this->assertTrue( $processor->matches_breadcrumbs( array( array( 'w.org', 'post' ), '*', 'image' ) ) );
 	}
 
+	/**
+	 *
+	 * @return void
+	 */
+	public function test_matches_breadcrumbs_wildcard_namespace() {
+		// Initialize the XMLProcessor with the given XML string
+		$processor = XMLProcessor::create_from_string(
+<<<XML
+<?xml version="1.0" encoding="UTF-8" ?>
+<rss xmlns:wp="http://wordpress.org/export/1.2/">
+    <channel>
+        <wp:base_site_url>http://wordpress.com/</wp:base_site_url>
+	</channel>	
+</rss>	
+XML
+);
+
+		$this->assertTrue( $processor->next_tag() ); // rss
+		$this->assertTrue( $processor->next_tag() ); // channel
+		$this->assertTrue( $processor->next_tag() ); // wp:base_site_url
+
+		// '*' should match wp:base_site_url
+		$this->assertTrue( $processor->matches_breadcrumbs( array( 'rss', 'channel', '*' ) ), 'A wildcard breadcrumb segment did not match the namespaced wp:base_site_url tag.' );
+	}
+
 	/**
 	 *
 	 * @return void

components/XML/XMLProcessor.php

@@ -3914,7 +3914,11 @@ public function matches_breadcrumbs( $breadcrumbs ) {
 
 			// Normalize crumb to [namespace, local_name]
 			if ( ! is_array( $crumb ) ) {
-				$crumb = array( '', $crumb );
+				if ( '*' === $crumb ) {
+					$crumb = ['*','*'];
+				} else {
+					$crumb = array( '', $crumb );
+				}
 			}
 			list( $namespace, $local_name ) = $crumb;