Merge pull request #3610 from 10up/feature/issue-2877

Create a default_search analyzer
10up · Sep 21, 2023 · efe8fde · efe8fde
2 parents 49033b7 + a93eaa4
commit efe8fde
Show file tree

Hide file tree

Showing 8 changed files with 119 additions and 58 deletions.
diff --git a/includes/classes/Feature/Search/Synonyms.php b/includes/classes/Feature/Search/Synonyms.php
@@ -379,8 +379,8 @@ public function add_search_synonyms( $mapping, $index ) {
 		}
 
 		// Ensure we have analyzers and that it is an array.
-		if ( ! isset( $mapping['settings']['analysis']['analyzer']['default']['filter'] )
-			|| ! is_array( $mapping['settings']['analysis']['analyzer']['default']['filter'] )
+		if ( ! isset( $mapping['settings']['analysis']['analyzer']['default_search']['filter'] )
+			|| ! is_array( $mapping['settings']['analysis']['analyzer']['default_search']['filter'] )
 		) {
 			return $mapping;
 		}
@@ -389,10 +389,10 @@ public function add_search_synonyms( $mapping, $index ) {
 		$mapping['settings']['analysis']['filter'][ $filter_name ] = $this->get_synonym_filter();
 
 		// Tell the analyzer to use our newly created filter.
-		$mapping['settings']['analysis']['analyzer']['default']['filter'] = array_values(
+		$mapping['settings']['analysis']['analyzer']['default_search']['filter'] = array_values(
 			array_merge(
 				[ $filter_name ],
-				$mapping['settings']['analysis']['analyzer']['default']['filter']
+				$mapping['settings']['analysis']['analyzer']['default_search']['filter']
 			)
 		);
 
@@ -464,7 +464,7 @@ public function update_synonyms() {
 			function( $success, $index ) {
 				$filter  = $this->get_synonym_filter();
 				$mapping = Elasticsearch::factory()->get_mapping( $index );
-				$filters = $mapping[ $index ]['settings']['index']['analysis']['analyzer']['default']['filter'];
+				$filters = $mapping[ $index ]['settings']['index']['analysis']['analyzer']['default_search']['filter'];
 
 				/*
 				 * Due to limitations in Elasticsearch, we can't remove the filter and analyzer
@@ -479,7 +479,7 @@ function( $success, $index ) {
 				$setting['index']['analysis']['filter']['ep_synonyms_filter'] = $filter;
 
 				// Add the analyzer.
-				$setting['index']['analysis']['analyzer']['default']['filter'] = array_values(
+				$setting['index']['analysis']['analyzer']['default_search']['filter'] = array_values(
 					array_unique(
 						array_merge(
 							[ $this->get_synonym_filter_name() ],

diff --git a/includes/mappings/post/5-2.php b/includes/mappings/post/5-2.php
@@ -48,27 +48,22 @@
 			'analyzer'   => array(
 				'default'          => array(
 					'tokenizer'   => 'standard',
-					/**
-					 * Filter Elasticsearch default analyzer's filters
-					 *
-					 * @since 3.6.2
-					 * @hook ep_default_analyzer_filters
-					 * @param  {array<string>} $filters Default filters
-					 * @return {array<string>} New filters
-					 */
+					/* This filter is documented in includes/mappings/post/7-0.php */
 					'filter'      => apply_filters( 'ep_default_analyzer_filters', array( 'standard', 'ewp_word_delimiter', 'lowercase', 'ep_stop', 'ewp_snowball' ) ),
-					/**
-					 * Filter Elasticsearch default analyzer's char_filter
-					 *
-					 * @since 4.2.2
-					 * @hook ep_default_analyzer_char_filters
-					 * @param  {array<string>} $char_filters Default filter
-					 * @return {array<string>} New filters
-					 */
+					/* This filter is documented in includes/mappings/post/7-0.php */
 					'char_filter' => apply_filters( 'ep_default_analyzer_char_filters', array( 'html_strip' ) ),
 					/* This filter is documented in includes/mappings/post/7-0.php */
 					'language'    => apply_filters( 'ep_analyzer_language', 'english', 'analyzer_default' ),
 				),
+				'default_search'   => array(
+					'tokenizer'   => 'standard',
+					/* This filter is documented in includes/mappings/post/7-0.php */
+					'filter'      => apply_filters( 'ep_default_search_analyzer_filters', array( 'standard', 'lowercase', 'ep_stop', 'ewp_snowball' ) ),
+					/* This filter is documented in includes/mappings/post/7-0.php */
+					'char_filter' => apply_filters( 'ep_default_search_analyzer_char_filters', array( 'html_strip' ) ),
+					/* This filter is documented in includes/mappings/post/7-0.php */
+					'language'    => apply_filters( 'ep_analyzer_language', 'english', 'analyzer_default' ),
+				),
 				'shingle_analyzer' => array(
 					'type'      => 'custom',
 					'tokenizer' => 'standard',

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', 'ep_stop', 'ewp_snowball' ) ),
+					'filter'      => apply_filters( 'ep_default_analyzer_filters', array( 'lowercase', 'ep_stop', 'ewp_snowball' ) ),
 					/**
 					 * Filter Elasticsearch default analyzer's char_filter
 					 *
@@ -92,6 +92,29 @@
 					 */
 					'language'    => apply_filters( 'ep_analyzer_language', 'english', 'analyzer_default' ),
 				),
+				'default_search'   => array(
+					'tokenizer'   => 'standard',
+					/**
+					 * Filter Elasticsearch default analyzer's filters
+					 *
+					 * @since 5.0.0
+					 * @hook ep_default_search_analyzer_filters
+					 * @param  {array<string>} $filters Default filters
+					 * @return {array<string>} New filters
+					 */
+					'filter'      => apply_filters( 'ep_default_search_analyzer_filters', array( 'lowercase', 'ep_stop', 'ewp_snowball' ) ),
+					/**
+					 * Filter Elasticsearch default analyzer's char_filter
+					 *
+					 * @since 5.0.0
+					 * @hook ep_default_search_analyzer_char_filters
+					 * @param  {array<string>} $char_filters Default filter
+					 * @return {array<string>} New filters
+					 */
+					'char_filter' => apply_filters( 'ep_default_search_analyzer_char_filters', array( 'html_strip' ) ),
+					/* This filter is documented above */
+					'language'    => apply_filters( 'ep_analyzer_language', 'english', 'analyzer_default' ),
+				),
 				'shingle_analyzer' => array(
 					'type'      => 'custom',
 					'tokenizer' => 'standard',
@@ -104,27 +127,23 @@
 				),
 			),
 			'filter'     => array(
-				'shingle_filter'     => array(
+				'shingle_filter' => array(
 					'type'             => 'shingle',
 					'min_shingle_size' => 2,
 					'max_shingle_size' => 5,
 				),
-				'ewp_word_delimiter' => array(
-					'type'              => 'word_delimiter_graph',
-					'preserve_original' => true,
-				),
-				'ewp_snowball'       => array(
+				'ewp_snowball'   => array(
 					'type'     => 'snowball',
 					/* This filter is documented in includes/mappings/post/7-0.php */
 					'language' => apply_filters( 'ep_analyzer_language', 'english', 'filter_ewp_snowball' ),
 				),
-				'edge_ngram'         => array(
+				'edge_ngram'     => array(
 					'side'     => 'front',
 					'max_gram' => 10,
 					'min_gram' => 3,
 					'type'     => 'edge_ngram',
 				),
-				'ep_stop'            => [
+				'ep_stop'        => [
 					'type'        => 'stop',
 					'ignore_case' => true,
 					/* This filter is documented in includes/mappings/post/7-0.php */

diff --git a/tests/cypress/integration/features/instant-results.cy.js b/tests/cypress/integration/features/instant-results.cy.js
@@ -354,10 +354,10 @@ describe('Instant Results Feature', { tags: '@slow' }, () => {
 				/**
 				 * Perform a search.
 				 */
-				cy.intercept('*search=wordpre*').as('apiRequest');
+				cy.intercept('*search=wordpless*').as('apiRequest');
 				cy.visit('/');
 				cy.get('.wp-block-search').last().as('searchBlock');
-				cy.get('@searchBlock').find('input[type="search"]').type('wordpre');
+				cy.get('@searchBlock').find('input[type="search"]').type('wordpless');
 				cy.get('@searchBlock').find('button').click();
 				cy.get('.ep-search-modal').should('be.visible');
 				cy.wait('@apiRequest');

diff --git a/tests/cypress/integration/features/woocommerce.cy.js b/tests/cypress/integration/features/woocommerce.cy.js
@@ -75,26 +75,35 @@ describe('WooCommerce Feature', { tags: '@slow' }, () => {
 		cy.activatePlugin('woocommerce', 'wpCli');
 		cy.maybeEnableFeature('woocommerce');
 
-		cy.updateWeighting({
-			product: {
-				'meta._variations_skus.value': {
-					weight: 1,
-					enabled: true,
-				},
-			},
+		cy.updateFeatures('search', {
+			active: 1,
+			highlight_enabled: true,
+			highlight_excerpt: true,
+			highlight_tag: 'mark',
+			highlight_color: '#157d84',
+			decaying_enabled: 'disabled_includes_products',
 		}).then(() => {
-			cy.wpCli('elasticpress sync --setup --yes').then(() => {
-				/**
-				 * Give Elasticsearch some time. Apparently, if the visit happens right after the index, it won't find anything.
-				 *
-				 */
-				// eslint-disable-next-line cypress/no-unnecessary-waiting
-				cy.wait(2000);
-				cy.visit('/?s=awesome-aluminum-shoes-variation-sku');
-				cy.contains(
-					'.site-content article:nth-of-type(1) h2',
-					'Awesome Aluminum Shoes',
-				).should('exist');
+			cy.updateWeighting({
+				product: {
+					'meta._variations_skus.value': {
+						weight: 1,
+						enabled: true,
+					},
+				},
+			}).then(() => {
+				cy.wpCli('elasticpress sync --setup --yes').then(() => {
+					/**
+					 * Give Elasticsearch some time. Apparently, if the visit happens right after the index, it won't find anything.
+					 *
+					 */
+					// eslint-disable-next-line cypress/no-unnecessary-waiting
+					cy.wait(2000);
+					cy.visit('/?s=awesome-aluminum-shoes-variation-sku');
+					cy.contains(
+						'.site-content article:nth-of-type(1) h2',
+						'Awesome Aluminum Shoes',
+					).should('exist');
+				});
 			});
 		});
 	});

diff --git a/tests/cypress/support/commands.js b/tests/cypress/support/commands.js
@@ -169,13 +169,16 @@ Cypress.Commands.add('publishPost', (postData, viewPost) => {
 	cy.wait(2000);
 });
 
-Cypress.Commands.add('updateFeatures', (newFeaturesValues = {}) => {
-	const features = Object.assign({}, cy.elasticPress.defaultFeatures, ...newFeaturesValues);
-
-	const escapedFeatures = JSON.stringify(features);
+Cypress.Commands.add('updateFeatures', (featureName, newValues) => {
+	const escapedNewValues = JSON.stringify(newValues);
 
 	cy.wpCliEval(
-		`$features = json_decode( '${escapedFeatures}', true ); update_option( 'ep_feature_settings', $features );`,
+		`
+		$feature_settings = get_option( 'ep_feature_settings', [] );
+
+		$feature_settings['${featureName}'] = json_decode( '${escapedNewValues}', true );
+		update_option( 'ep_feature_settings', $feature_settings );
+		`,
 	);
 });
 

diff --git a/tests/cypress/wordpress-files/test-docs/content-example.xml b/tests/cypress/wordpress-files/test-docs/content-example.xml
@@ -4834,7 +4834,7 @@
 sneakers, tennis shoes, trainers, runners
 
 # Defined alternatives (explicit mappings).
-shoes => sneaker, sandal, boots, high heels]]></content:encoded>
+shoes => shoes, sneaker, sandal, boots, high heels]]></content:encoded>
 		<excerpt:encoded><![CDATA[]]></excerpt:encoded>
 		<wp:post_id>2002</wp:post_id>
 		<wp:post_date><![CDATA[2021-02-17 15:48:58]]></wp:post_date>

diff --git a/tests/php/features/TestSynonyms.php b/tests/php/features/TestSynonyms.php
@@ -125,4 +125,39 @@ public function testValidateSynonyms() {
 		$this->assertEquals( 'foo, bar', $instance->validate_synonym( ' foo, bar ' ) );
 		$this->assertEquals( 'foo => bar', $instance->validate_synonym( ' foo => bar ' ) );
 	}
+
+	/**
+	 * Test synonyms with spaces
+	 *
+	 * @since 5.0.0
+	 * @group synonyms
+	 * @group skip-on-single-site
+	 */
+	public function test_synonyms_with_spaces() {
+		$instance = $this->getFeature();
+
+		wp_insert_post(
+			[
+				'ID'           => $instance->get_synonym_post_id(),
+				'post_content' => 'internet of things, IoT',
+				'post_type'    => $instance::POST_TYPE_NAME,
+			],
+			true
+		);
+		$instance->update_synonyms();
+
+		$post_id = $this->ep_factory->post->create( [ 'post_content' => 'IoT' ] );
+
+		ElasticPress\Elasticsearch::factory()->refresh_indices();
+
+		$query = new \WP_Query(
+			[
+				's'      => 'internet of things',
+				'fields' => 'ids',
+			]
+		);
+
+		$this->assertTrue( $query->elasticsearch_success );
+		$this->assertSame( $post_id, $query->posts['0'] );
+	}
 }