test: minor adjustments and encoding tests

10up · Dec 20, 2024 · f857bd5 · f857bd5
1 parent 2b55b2a
commit f857bd5
Show file tree

Hide file tree

Showing 2 changed files with 44 additions and 8 deletions.
diff --git a/wp/headless-wp/includes/classes/Integrations/Gutenberg.php b/wp/headless-wp/includes/classes/Integrations/Gutenberg.php
@@ -59,8 +59,6 @@ protected function bypass_block_attributes( string $block_name, WP_Block $block_
 	 */
 	public function process_block_with_dom_document_api( $html, $block_name, $block_attrs_serialized, $block, $block_instance ) {
 		try {
-			libxml_use_internal_errors( true );
-
 			return $this->bypass_block_attributes( $block_name, $block_instance )
 				? $this->process_dom_document_bypassed_block( $html )
 				: $this->process_dom_document_block( $html, $block_name, $block_attrs_serialized, $block, $block_instance );
@@ -125,7 +123,7 @@ public function process_dom_document_block(
 		array $block,
 		WP_Block $block_instance
 	): string {
-		$document = $this->read_decoded_dom_document( $html );
+		$document = $this->read_converted_dom_document( $html );
 
 		// phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase
 		$root_node = $document->documentElement;
@@ -161,13 +159,13 @@ public function process_dom_document_block(
 	 * @return string
 	 */
 	public function process_dom_document_bypassed_block( string $html ): string {
-		$document  = $this->read_decoded_dom_document( "<body>{$html}</body>" );
+		$document  = $this->read_converted_dom_document( "<body>{$html}</body>" );
 		$body      = $document->getElementsByTagName( 'body' )->item( 0 );
 		$node_html = [];
 
 		// phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase
 		foreach ( $body->childNodes as $child ) {
-			$block = new DOMDocument();
+			$block = new DOMDocument( '1.0', 'UTF-8' );
 			$block->appendChild( $block->importNode( $child, true ) );
 
 			$child_html   = $block->saveHTML();
@@ -190,9 +188,13 @@ public function process_dom_document_bypassed_block( string $html ): string {
 	 *
 	 * @return DOMDocument
 	 */
-	protected function read_decoded_dom_document( string $html ) {
-		$document = new DomDocument( '1.0', 'UTF-8' );
-		$document->loadHTML( htmlspecialchars_decode( htmlentities( mb_convert_encoding( $html, 'HTML-ENTITIES', 'UTF-8' ) ) ), LIBXML_HTML_NODEFDTD | LIBXML_HTML_NOIMPLIED );
+	protected function read_converted_dom_document( string $html ) {
+		$converted_html = htmlspecialchars_decode( htmlentities( mb_convert_encoding( $html, 'HTML-ENTITIES', 'UTF-8' ) ) );
+		$document       = new DomDocument( '1.0', 'UTF-8' );
+
+		libxml_use_internal_errors( true );
+		$document->loadHTML( $converted_html, LIBXML_HTML_NODEFDTD | LIBXML_HTML_NOIMPLIED );
+		libxml_clear_errors();
 
 		// phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase
 		if ( null === $document->documentElement ) {

diff --git a/wp/headless-wp/tests/php/tests/TestGutenbergIntegration.php b/wp/headless-wp/tests/php/tests/TestGutenbergIntegration.php
@@ -152,6 +152,40 @@ public function test_render_classic_block() {
 		);
 	}
 
+	/**
+	 * Test to ensure the parser handles both HTML and Multi-byte encodings properly
+	 *
+	 * @return void
+	 */
+	public function test_handle_multi_byte_html_encoding() {
+		[ 'html' => $html, 'parsed_block' => $block, 'instance' => $instance ] =
+			$this->core_render_block_from_markup(
+				<<<MARKUP
+				<!-- wp:paragraph -->
+				<p>The temperature is 23°C ☀️ (sun emoji) and © (copyright symbol). HTML entity for Degrees: &#176;.</p>
+				<!-- /wp:paragraph -->
+				MARKUP
+			);
+		$dom_expected          = <<<RESULT
+			<p data-wp-block='{"dropCap":false}' data-wp-block-name="core/paragraph">The temperature is 23&deg;C &#9728;&#65039; (sun emoji) and &copy; (copyright symbol). HTML entity for Degrees: &deg;.</p>
+			RESULT;
+		$html_tag_api_expected = <<<RESULT
+			<p data-wp-block="{&quot;dropCap&quot;:false}" data-wp-block-name="core/paragraph">The temperature is 23&deg;C &#9728;&#65039; (sun emoji) and &copy; (copyright symbol). HTML entity for Degrees: &deg;.</p>
+			RESULT;
+
+		$dom_output = $this->parser->render_block( $html, $block, $instance );
+
+		$this->assertSame( trim( $dom_expected ), trim( $dom_output ), 'Gutenberg | DOM Document | Test HTML Encoding' );
+
+		add_filter( 'tenup_headless_wp_render_block_use_tag_processor', '__return_true' );
+
+		$html_api_output = $this->parser->render_block( $html, $block, $instance );
+
+		remove_filter( 'tenup_headless_wp_render_block_use_tag_processor', '__return_true' );
+
+		$this->assertSame( trim( $html_tag_api_expected ), trim( $html_api_output ), 'Gutenberg | HTML Tag API | Test HTML Encoding' );
+	}
+
 	/**
 	 * Tests rendering classic block with the HTML tag api
 	 *