Merge pull request #986 from afoucret/fix-thesaurus-max-rewrites

Fix synonym expansion to avoid huge queries.
Smile-SA · Jun 27, 2018 · 49bcbe5 · 49bcbe5
2 parents ed42a14 + ac14dcb
commit 49bcbe5
Show file tree

Hide file tree

Showing 4 changed files with 49 additions and 17 deletions.
diff --git a/src/module-elasticsuite-thesaurus/Config/ThesaurusConfig.php b/src/module-elasticsuite-thesaurus/Config/ThesaurusConfig.php
@@ -33,18 +33,35 @@ class ThesaurusConfig
      */
     private $expansionsConfig;
 
+    /**
+     * @var array
+     */
+    private $general;
+
     /**
      * Constructor.
      *
+     * @param array $general    General configuration.
      * @param array $synonyms   Synonyms configuration.
      * @param array $expansions Expansions configuration.
      */
-    public function __construct($synonyms = [], $expansions = [])
+    public function __construct($general = [], $synonyms = [], $expansions = [])
     {
+        $this->general          = $general;
         $this->synonymsConfig   = $synonyms;
         $this->expansionsConfig = $expansions;
     }
 
+    /**
+     * Max allowed rewrites for the synonym engine.
+     *
+     * @return int
+     */
+    public function getMaxRewrites()
+    {
+        return (int) $this->general['max_rewrites'];
+    }
+
     /**
      * Is the synonyms search enabled ?
      *

diff --git a/src/module-elasticsuite-thesaurus/Model/Index.php b/src/module-elasticsuite-thesaurus/Model/Index.php
@@ -122,19 +122,22 @@ private function computeQueryRewrites(ContainerConfigurationInterface $container
         $config   = $this->getConfig($containerConfig);
         $storeId  = $containerConfig->getStoreId();
         $rewrites = [];
+        $maxRewrites = $config->getMaxRewrites();
 
         if ($config->isSynonymSearchEnabled()) {
-            $synonymRewrites = $this->getSynonymRewrites($storeId, $queryText, ThesaurusInterface::TYPE_SYNONYM);
+            $thesaurusType   = ThesaurusInterface::TYPE_SYNONYM;
+            $synonymRewrites = $this->getSynonymRewrites($storeId, $queryText, $thesaurusType, $maxRewrites);
             $rewrites        = $this->getWeightedRewrites($synonymRewrites, $config->getSynonymWeightDivider());
         }
 
         if ($config->isExpansionSearchEnabled()) {
             $synonymRewrites = array_merge([$queryText => 1], $rewrites);
 
             foreach ($synonymRewrites as $currentQueryText => $currentWeight) {
-                $expansions        = $this->getSynonymRewrites($storeId, $currentQueryText, ThesaurusInterface::TYPE_EXPANSION);
+                $thesaurusType     = ThesaurusInterface::TYPE_EXPANSION;
+                $expansions        = $this->getSynonymRewrites($storeId, $currentQueryText, $thesaurusType, $maxRewrites);
                 $expansionRewrites = $this->getWeightedRewrites($expansions, $config->getExpansionWeightDivider(), $currentWeight);
-                $rewrites = array_merge($rewrites, $expansionRewrites);
+                $rewrites          = array_merge($rewrites, $expansionRewrites);
             }
         }
 
@@ -200,13 +203,14 @@ private function getConfig(ContainerConfigurationInterface $containerConfig)
     /**
      * Generates all possible synonym rewrites for a store and text query.
      *
-     * @param integer $storeId   Store id.
-     * @param string  $queryText Text query.
-     * @param string  $type      Substitution type (synonym or expansion).
+     * @param integer $storeId     Store id.
+     * @param string  $queryText   Text query.
+     * @param string  $type        Substitution type (synonym or expansion).
+     * @param integer $maxRewrites Max number of allowed rewrites.
      *
      * @return array
      */
-    private function getSynonymRewrites($storeId, $queryText, $type)
+    private function getSynonymRewrites($storeId, $queryText, $type, $maxRewrites)
     {
         $indexName = $this->getIndexAlias($storeId);
 
@@ -228,24 +232,25 @@ private function getSynonymRewrites($storeId, $queryText, $type)
             }
         }
 
-        return $this->combineSynonyms($queryText, $synonymByPositions);
+        return $this->combineSynonyms($queryText, $synonymByPositions, $maxRewrites);
     }
 
     /**
      * Combine analysis result to provides all possible synonyms substitution comination.
      *
-     * @param string $queryText          Original query text
-     * @param array  $synonymByPositions Synonyms array by positions.
-     * @param int    $substitutions      Number of substitutions in the current query.
-     * @param int    $offset             Offset of previous substitutions.
+     * @param string  $queryText          Original query text
+     * @param array   $synonymByPositions Synonyms array by positions.
+     * @param integer $maxRewrites        Max number of allowed rewrites.
+     * @param int     $substitutions      Number of substitutions in the current query.
+     * @param int     $offset             Offset of previous substitutions.
      *
      * @return array
      */
-    private function combineSynonyms($queryText, $synonymByPositions, $substitutions = 0, $offset = 0)
+    private function combineSynonyms($queryText, $synonymByPositions, $maxRewrites, $substitutions = 0, $offset = 0)
     {
         $combinations = [];
 
-        if (!empty($synonymByPositions)) {
+        if (!empty($synonymByPositions) && $substitutions < $maxRewrites) {
             $currentPositionSynonyms = current($synonymByPositions);
             $remainingSynonyms = array_slice($synonymByPositions, 1);
 
@@ -259,15 +264,15 @@ private function combineSynonyms($queryText, $synonymByPositions, $substitutions
                 if (!empty($remainingSynonyms)) {
                     $combinations = array_merge(
                         $combinations,
-                        $this->combineSynonyms($rewrittenQueryText, $remainingSynonyms, $substitutions + 1, $newOffset)
+                        $this->combineSynonyms($rewrittenQueryText, $remainingSynonyms, $maxRewrites, $substitutions + 1, $newOffset)
                     );
                 }
             }
 
             if (!empty($remainingSynonyms)) {
                 $combinations = array_merge(
                     $combinations,
-                    $this->combineSynonyms($queryText, $remainingSynonyms, $substitutions, $offset)
+                    $this->combineSynonyms($queryText, $remainingSynonyms, $maxRewrites, $substitutions, $offset)
                 );
             }
         }

diff --git a/src/module-elasticsuite-thesaurus/etc/adminhtml/elasticsuite_relevance.xml b/src/module-elasticsuite-thesaurus/etc/adminhtml/elasticsuite_relevance.xml
@@ -21,6 +21,13 @@
             <label>Thesaurus Configuration</label>
             <resource>Smile_ElasticsuiteCore::manage_relevance</resource>
 
+            <group id="general" translate="label" type="text" sortOrder="20" showInDefault="1" showInContainer="1" showInStore="1">
+                <label>General Configuration</label>
+                <field id="max_rewrites" translate="label" type="text" sortOrder="10" showInDefault="1" showInContainer="1" showInStore="1">
+                    <label>Max Allowed Rewrites</label>
+                </field>
+            </group>
+
             <group id="synonyms" translate="label" type="text" sortOrder="40" showInDefault="1" showInContainer="1" showInStore="1">
                 <label>Synonyms Configuration</label>
                 <field id="enable" translate="label" type="select" sortOrder="10" showInDefault="1" showInContainer="1" showInStore="1">

diff --git a/src/module-elasticsuite-thesaurus/etc/elasticsuite_relevance.xml b/src/module-elasticsuite-thesaurus/etc/elasticsuite_relevance.xml
@@ -16,6 +16,9 @@
           xsi:noNamespaceSchemaLocation="urn:magento:module:Smile_ElasticsuiteCore:etc/elasticsuite_relevance_initial_config.xsd">
     <default>
         <thesaurus>
+            <general>
+                <max_rewrites>2</max_rewrites>
+            </general>
             <synonyms>
                 <enable>1</enable>
                 <weight_divider>10</weight_divider>