Fix ES' query string queries, refs #12186

Splits multi-field string queries into boolean queries with a string query clause per field. This allows ES to use the field's analyzer on each clause instead of the default analyzer and therefore it generates better results, specially working with stop-words in a multi-culture index. Additionally: - Simplify retrieval of all fields by not processing the enabled languages until the query generation. - Modify and simplify boosting process, now required as part of the boolean query clauses and not appended to the fields. - Remove unused parse function from arElasticSearchPlugin class.
artefactual · Mar 18, 2021 · 34adf96 · 34adf96
1 parent e55cc29
commit 34adf96
Show file tree

Hide file tree

Showing 11 changed files with 214 additions and 366 deletions.
diff --git a/apps/qubit/modules/default/actions/moveAction.class.php b/apps/qubit/modules/default/actions/moveAction.class.php
@@ -124,13 +124,13 @@ public function execute($request)
 
     if (isset($request->query))
     {
-      $query = new \Elastica\Query\QueryString(arElasticSearchPluginUtil::escapeTerm($request->query));
-      $query->setDefaultOperator('AND');
-      $query->setFields(array(
-        'identifier',
-        'referenceCode',
-        sprintf('i18n.%s.title', sfContext::getInstance()->user->getCulture())));
-      $this->queryBool->addMust($query);
+      $fields = array(
+        'identifier' => 1,
+        'referenceCode' => 1,
+        sprintf('i18n.%s.title', sfContext::getInstance()->user->getCulture()) => 1);
+      $this->queryBool->addMust(
+        arElasticSearchPluginUtil::generateBoolQueryString($request->query, $fields)
+      );
     }
     else
     {

diff --git a/apps/qubit/modules/informationobject/actions/autocompleteAction.class.php b/apps/qubit/modules/informationobject/actions/autocompleteAction.class.php
@@ -51,13 +51,12 @@ public function execute($request)
     }
     else
     {
-      $queryString = new \Elastica\Query\QueryString(arElasticSearchPluginUtil::escapeTerm($request->query));
-      $queryString->setDefaultOperator('AND');
+      $fields = array('i18n.'.$culture.'.title.autocomplete' => 1);
 
       // Search for referenceCode or identifier, and title
       if (1 == sfConfig::get('app_inherit_code_informationobject', 1))
       {
-        $queryString->setFields(array('i18n.'.$culture.'.title.autocomplete', 'referenceCode.autocomplete'));
+        $fields['referenceCode.autocomplete'] = 1;
 
         // Change sort order
         $this->query->setSort(array(
@@ -67,10 +66,12 @@ public function execute($request)
       }
       else
       {
-        $queryString->setFields(array('i18n.'.$culture.'.title.autocomplete', 'identifier'));
+        $fields['identifier'] = 1;
       }
 
-      $this->queryBool->addMust($queryString);
+      $this->queryBool->addMust(
+        arElasticSearchPluginUtil::generateBoolQueryString($request->query, $fields)
+      );
     }
 
     // Filter results by parent

diff --git a/apps/qubit/modules/repository/actions/browseAction.class.php b/apps/qubit/modules/repository/actions/browseAction.class.php
@@ -122,11 +122,11 @@ public function execute($request)
     }
     else
     {
-      $queryText = new \Elastica\Query\QueryString(arElasticSearchPluginUtil::escapeTerm($request->subquery));
-      $queryText->setDefaultOperator('OR');
-      arElasticSearchPluginUtil::setFields($queryText, 'repository');
-
-      $this->search->queryBool->addMust($queryText);
+      $this->search->queryBool->addMust(
+        arElasticSearchPluginUtil::generateBoolQueryString(
+          $request->subquery, arElasticSearchPluginUtil::getAllFields('repository')
+        )
+      );
     }
 
     $i18n = sprintf('i18n.%s.', $this->selectedCulture);

diff --git a/apps/qubit/modules/search/actions/indexAction.class.php b/apps/qubit/modules/search/actions/indexAction.class.php
@@ -28,11 +28,11 @@ public function execute($request)
   {
     parent::execute($request);
 
-    $queryText = new \Elastica\Query\QueryString(arElasticSearchPluginUtil::escapeTerm($request->query));
-    $queryText->setDefaultOperator('AND');
-    arElasticSearchPluginUtil::setFields($queryText, 'informationObject');
-
-    $this->search->queryBool->addMust($queryText);
+    $this->search->queryBool->addMust(
+      arElasticSearchPluginUtil::generateBoolQueryString(
+        $request->query, arElasticSearchPluginUtil::getAllFields('informationObject')
+      )
+    );
 
     // Realm filter
     if (isset($request->repos) && ctype_digit($request->repos))

diff --git a/apps/qubit/modules/taxonomy/actions/indexAction.class.php b/apps/qubit/modules/taxonomy/actions/indexAction.class.php
@@ -170,33 +170,30 @@ public function execute($request)
 
     if (1 !== preg_match('/^[\s\t\r\n]*$/', $request->subquery))
     {
-      $queryString = new \Elastica\Query\QueryString(arElasticSearchPluginUtil::escapeTerm($request->subquery));
-
       switch ($request->subqueryField)
       {
         case 'preferredLabel':
-          $queryString->setFields(arElasticSearchPluginUtil::getI18nFieldNames('i18n.%s.name'));
+          $fields = array('i18n.%s.name' => 1);
 
           break;
 
         case 'useForLabels':
-          $queryString->setFields(arElasticSearchPluginUtil::getI18nFieldNames('useFor.i18n.%s.name'));
+          $fields = array('useFor.i18n.%s.name' => 1);
 
           break;
 
         case 'allLabels':
         default:
           // Search over preferred label (boosted by five) and "Use for" labels
-          $fields = array('i18n.%s.name', 'useFor.i18n.%s.name');
-          $boost = array('i18n.%s.name' => 5);
-          $queryString->setFields(arElasticSearchPluginUtil::getI18nFieldNames($fields, null, $boost));
-          $queryString->setDefaultOperator('AND');
+          $fields = array('i18n.%s.name' => 5, 'useFor.i18n.%s.name' => 1);
 
           break;
       }
 
       // Filter results by subquery
-      $this->queryBool->addMust($queryString);
+      $this->queryBool->addMust(
+        arElasticSearchPluginUtil::generateBoolQueryString($request->subquery, $fields)
+      );
     }
 
     // Set query

diff --git a/plugins/arElasticSearchPlugin/lib/arElasticSearchPlugin.class.php b/plugins/arElasticSearchPlugin/lib/arElasticSearchPlugin.class.php
@@ -520,22 +520,6 @@ public function partialUpdate($object, $data)
     }
   }
 
-  /**
-   * Function helper to parse query strings
-   */
-  public function parse(string $query)
-  {
-    if (empty($query))
-    {
-      throw new Exception('No search terms specified.');
-    }
-
-    $query = new \Elastica\Query\QueryString(arElasticSearchPluginUtil::escapeTerm($query));
-    $query->setDefaultOperator('AND');
-
-    return $query;
-  }
-
   // ---------------------------------------------------------------------------
 
   public function delete($object)