Fix excerpt output on search

jaymcp · jaymcp · commit e3a886fc71fd · 2025-05-20T15:04:59.000+01:00
diff --git a/wp-content/themes/humanity-theme/includes/features/search/post-excerpt.php b/wp-content/themes/humanity-theme/includes/features/search/post-excerpt.php
@@ -15,17 +15,37 @@ function get_first_paragraph( string $html ): string {
 			return '';
 		}
 
-		$doc = new DOMDocument();
+		$cache_key = md5( $html );
+		$cached    = wp_cache_get( $cache_key );
+
+		if ( $cached ) {
+			return $cached;
+		}
+
+		$doc = new DOMDocument( '1.0', 'utf-8' );
 
 		// phpcs:disable WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase
 		$doc->formatOutput       = false;
 		$doc->substituteEntities = false;
-		// phpcs:enable WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase
+		$doc->preserveWhiteSpace = true;
+		$doc->validateOnParse    = false;
+
+		// ensure string is utf8
+		$encoded_content = mb_convert_encoding( $html, 'UTF-8' );
+		// encode everything
+		$encoded_content = htmlentities( $encoded_content, ENT_NOQUOTES, 'UTF-8' );
+		// decode "standard" characters
+		$encoded_content = htmlspecialchars_decode( $encoded_content, ENT_NOQUOTES );
+		// convert left side of ISO-8859-1 to HTML numeric character reference
+		// this was taken from PHP docs for mb_encode_numericentity   vvvvvvvvvvvvvvvvvvvvvvvvv
+		$encoded_content = mb_encode_numericentity( $encoded_content, [ 0x80, 0x10FFFF, 0, ~0 ], 'UTF-8' );
 
 		libxml_use_internal_errors( true );
 		$doc->loadHTML(
-			$html,
-			LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD | LIBXML_NOXMLDECL | LIBXML_NOERROR | LIBXML_NOWARNING
+			'<head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"/></head><body>' .
+			$encoded_content .
+			'</body>',
+			LIBXML_HTML_NODEFDTD | LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NOENT
 		);
 		libxml_use_internal_errors( false );
 
@@ -39,7 +59,10 @@ function get_first_paragraph( string $html ): string {
 				continue;
 			}
 
-			return wp_kses_post( sprintf( '<p>%s</p>', $innards ) );
+			$output = sprintf( '<p>%s</p>', $innards );
+
+			wp_cache_add( $cache_key, $output );
+			return wp_kses_post( $output );
 		}
 
 		return '';
diff --git a/wp-content/themes/humanity-theme/patterns/search.php b/wp-content/themes/humanity-theme/patterns/search.php
@@ -111,7 +111,7 @@
 					<!-- wp:group {"tagName":"div","className":"post-excerpt wp-block-post-excerpt"} -->
 					<div class="wp-block-group post-excerpt wp-block-post-excerpt">
 						<!-- wp:paragraph {"className":"wp-block-post-excerpt__excerpt"} -->
-						<p class="wp-block-paragraph wp-block-post-excerpt__excerpt"><?php echo wp_kses_post( get_the_excerpt() ); ?></p>
+						<p class="wp-block-paragraph wp-block-post-excerpt__excerpt"><?php echo wp_kses_post( get_first_paragraph( get_the_content() ) ); ?></p>
 						<!-- /wp:paragraph -->
 					</div>
 					<!-- /wp:group -->
