Change the stop filter depending on the language

10up · Jul 7, 2023 · 239b4ea · 239b4ea
1 parent 8ed79a2
commit 239b4ea
Show file tree

Hide file tree

Showing 3 changed files with 40 additions and 22 deletions.
diff --git a/includes/classes/Indexable/Post/SyncManager.php b/includes/classes/Indexable/Post/SyncManager.php
@@ -76,9 +76,9 @@ public function setup() {
 		add_action( 'deleted_term_relationships', array( $this, 'action_deleted_term_relationships' ), 10, 3 );
 
 		// Clear field limit cache
-		add_action( 'ep_update_index_settings', [ $this, 'clear_total_fields_limit_cache' ] );
-		add_action( 'ep_sync_put_mapping', [ $this, 'clear_total_fields_limit_cache' ] );
-		add_action( 'ep_saved_weighting_configuration', [ $this, 'clear_total_fields_limit_cache' ] );
+		add_action( 'ep_update_index_settings', [ $this, 'clear_index_settings_cache' ] );
+		add_action( 'ep_sync_put_mapping', [ $this, 'clear_index_settings_cache' ] );
+		add_action( 'ep_saved_weighting_configuration', [ $this, 'clear_index_settings_cache' ] );
 
 		// Clear distinct meta field per post type cache
 		add_action( 'wp_insert_post', [ $this, 'clear_meta_keys_db_per_post_type_cache_by_post_id' ] );
@@ -743,13 +743,28 @@ public function action_create_blog_index( $blog ) {
 	 * @since 4.4.0
 	 */
 	public function clear_total_fields_limit_cache() {
+		_deprecated_function( __METHOD__, '4.5.0', '\ElasticPress\Indexable\Post\SyncManager::clear_index_settings_cache()' );
+
+		$this->clear_index_settings_cache();
+	}
+
+	/**
+	 * Clear the cache of specific index settings
+	 *
+	 * @since 4.7.0
+	 */
+	public function clear_index_settings_cache() {
 		$indexable = Indexables::factory()->get( $this->indexable_slug );
-		$cache_key = 'ep_total_fields_limit_' . $indexable->get_index_name();
 
-		if ( defined( 'EP_IS_NETWORK' ) && EP_IS_NETWORK ) {
-			delete_site_transient( $cache_key );
-		} else {
-			delete_transient( $cache_key );
+		$settings = [ 'total_fields_limit', 'default_analyzer_language', 'snowball_filter_language' ];
+		foreach ( $settings as $setting ) {
+			$cache_key = 'ep_' . $setting . '_' . $indexable->get_index_name();
+
+			if ( defined( 'EP_IS_NETWORK' ) && EP_IS_NETWORK ) {
+				delete_site_transient( $cache_key );
+			} else {
+				delete_transient( $cache_key );
+			}
 		}
 	}
 

diff --git a/includes/dashboard.php b/includes/dashboard.php
@@ -746,13 +746,13 @@ function use_language_in_setting( $language = 'english', $context = '' ) {
 	require_once ABSPATH . 'wp-admin/includes/translation-install.php';
 	$translations = wp_get_available_translations();
 
-	// Bail early if not in the array of available translations.
-	if ( empty( $translations[ $ep_language ]['english_name'] ) ) {
-		return $language;
+	// Default to en_US if not in the array of available translations.
+	if ( ! empty( $translations[ $ep_language ]['english_name'] ) ) {
+		$wp_language = $translations[ $ep_language ]['language'];
+	} else {
+		$wp_language = 'en_US';
 	}
 
-	$wp_language = $translations[ $ep_language ]['language'];
-
 	/**
 	 * Languages supported in Elasticsearch mappings.
 	 * Array format: Elasticsearch analyzer name => WordPress language package name
@@ -848,6 +848,10 @@ function use_language_in_setting( $language = 'english', $context = '' ) {
 		return $es_snowball_similar[ $uc_first_language ] ?? 'English';
 	}
 
+	if ( 'filter_ep_stop' === $context ) {
+		return "_{$language}_";
+	}
+
 	return $language;
 }
 

diff --git a/includes/mappings/post/7-0.php b/includes/mappings/post/7-0.php
@@ -72,7 +72,7 @@
 					 * @param  {array<string>} $filters Default filters
 					 * @return {array<string>} New filters
 					 */
-					'filter'      => apply_filters( 'ep_default_analyzer_filters', array( 'ewp_word_delimiter', 'lowercase', 'stop', 'ewp_snowball' ) ),
+					'filter'      => apply_filters( 'ep_default_analyzer_filters', array( 'ewp_word_delimiter', 'lowercase', 'ep_stop', 'ewp_snowball' ) ),
 					/**
 					 * Filter Elasticsearch default analyzer's char_filter
 					 *
@@ -115,14 +115,7 @@
 				),
 				'ewp_snowball'       => array(
 					'type'     => 'snowball',
-					/**
-					 * Filter Elasticsearch default language in mapping
-					 *
-					 * @hook ep_analyzer_language
-					 * @param  {string} $lang Default language
-					 * @param {string} $lang_context Language context
-					 * @return {string} New language
-					 */
+					/* This filter is documented above */
 					'language' => apply_filters( 'ep_analyzer_language', 'english', 'filter_ewp_snowball' ),
 				),
 				'edge_ngram'         => array(
@@ -131,6 +124,12 @@
 					'min_gram' => 3,
 					'type'     => 'edge_ngram',
 				),
+				'ep_stop'            => array(
+					'type'        => 'stop',
+					'ignore_case' => true,
+					/* This filter is documented above */
+					'stopwords'   => apply_filters( 'ep_analyzer_language', 'english', 'filter_ep_stop' ),
+				),
 			),
 			'normalizer' => array(
 				'lowerasciinormalizer' => array(