Merge branch '2.10.x' into 2.11.x

Author: rbayet

src/module-elasticsuite-core/Search/Request/Query/Fulltext/QueryBuilder.php

@@ -71,7 +71,7 @@ public function __construct(
      * Create the fulltext search query.
      *
      * @param ContainerConfigurationInterface $containerConfig Search request container configuration.
-     * @param string                          $queryText       The text query.
+     * @param string|array                    $queryText       The text query.
      * @param string                          $spellingType    The type of spellchecked applied.
      * @param float                           $boost           Boost of the created query.
      * @param int                             $depth           Call depth of the create method. Can be used to avoid/prevent cycles.

src/module-elasticsuite-thesaurus/Plugin/QueryRewrite.php

@@ -22,6 +22,7 @@
 use Smile\ElasticsuiteThesaurus\Model\Index;
 use Smile\ElasticsuiteCore\Api\Search\SpellcheckerInterface;
 use Smile\ElasticsuiteCore\Search\Request\QueryInterface;
+use Smile\ElasticsuiteCore\Helper\Text as TextHelper;
 
 /**
  * Plugin that handle query rewriting (synonym substitution) during fulltext query building phase.
@@ -47,6 +48,11 @@ class QueryRewrite
      */
     private $index;
 
+    /**
+     * @var Text
+     */
+    private $textHelper;
+
     /**
      * @var array
      */
@@ -58,15 +64,18 @@ class QueryRewrite
      * @param QueryFactory           $queryFactory           Search request query factory.
      * @param ThesaurusConfigFactory $thesaurusConfigFactory Thesaurus configuration factory.
      * @param Index                  $index                  Synonym index.
+     * @param TextHelper             $textHelper             Text helper.
      */
     public function __construct(
         QueryFactory $queryFactory,
         ThesaurusConfigFactory $thesaurusConfigFactory,
-        Index $index
+        Index $index,
+        TextHelper $textHelper
     ) {
         $this->queryFactory           = $queryFactory;
         $this->thesaurusConfigFactory = $thesaurusConfigFactory;
         $this->index                  = $index;
+        $this->textHelper             = $textHelper;
     }
 
     /**
@@ -77,7 +86,7 @@ public function __construct(
      * @param QueryBuilder                    $subject         Original query builder.
      * @param \Closure                        $proceed         Original create func.
      * @param ContainerConfigurationInterface $containerConfig Search request container config.
-     * @param string                          $queryText       Current query text.
+     * @param string|array                    $queryText       Current query text.
      * @param string                          $spellingType    Spelling type of the query.
      * @param float                           $boost           Original query boost.
      * @param int                             $depth           Call depth of the create method. Can be used to avoid/prevent cycles.
@@ -109,13 +118,16 @@ public function aroundCreate(
             if ($depth === 0) {
                 $rewrites = $this->getWeightedRewrites($queryText, $containerConfig, $boost);
             }
+            $originalSpellingType = $spellingType;
             // Set base query as SPELLING_TYPE_EXACT if synonyms/expansions are found.
+            // This is to prevent possible fuzzy matches on that original query's terms and doing so, prioritize the rewritten queries.
             $spellingType = empty($rewrites) ? $spellingType : SpellcheckerInterface::SPELLING_TYPE_EXACT;
             $query        = $proceed($containerConfig, $queryText, $spellingType, $boost, $depth);
 
             if (!empty($rewrites)) {
                 $synonymQueries           = [$query];
-                $synonymQueriesSpellcheck = SpellcheckerInterface::SPELLING_TYPE_EXACT;
+                // Do not enforce SPELLING_TYPE_EXACT systematically for alternative queries.
+                $synonymQueriesSpellcheck = $this->getRewritesSpellingType($queryText, $originalSpellingType);
 
                 foreach ($rewrites as $rewrittenQuery => $weight) {
                     $synonymQueries[] = $proceed($containerConfig, $rewrittenQuery, $synonymQueriesSpellcheck, $weight, $depth + 1);
@@ -161,6 +173,40 @@ private function getWeightedRewrites($queryText, $containerConfig, $originalBoos
         return $rewrites;
     }
 
+    /**
+     * Returns the spelling type to use for rewritten queries.
+     * For multi terms queries, considering that at least one term has been replaced,
+     * but there could be at least one mistyped term that might not have been replaced.
+     * So "elevate" the spelling type a bit.
+     * Ideally a new spellcheck query should be run.
+     *
+     * @param string|array $originalQueryText    The original query text.
+     * @param int          $originalSpellingType The original spelling type.
+     *
+     * @return int
+     */
+    private function getRewritesSpellingType($originalQueryText, $originalSpellingType)
+    {
+        if (is_array($originalQueryText)) {
+            // Expected to be SPELLING_TYPE_EXACT as enforced by the request builder.
+            return $originalSpellingType;
+        }
+
+        if ($this->textHelper->mbWordCount($originalQueryText) === 1) {
+            return SpellcheckerInterface::SPELLING_TYPE_EXACT;
+        }
+
+        $spellingType = $originalSpellingType;
+
+        if (SpellcheckerInterface::SPELLING_TYPE_FUZZY === $originalSpellingType) {
+            $spellingType = SpellcheckerInterface::SPELLING_TYPE_MOST_FUZZY;
+        } elseif (SpellcheckerInterface::SPELLING_TYPE_PURE_STOPWORDS === $originalSpellingType) {
+            $spellingType = SpellcheckerInterface::SPELLING_TYPE_MOST_EXACT;
+        }
+
+        return $spellingType;
+    }
+
     /**
      * Return thesaurus/relevance configuration.
      *

src/module-elasticsuite-thesaurus/Test/Unit/Plugin/QueryRewriteTest.php

@@ -95,12 +95,12 @@ public function testMultipleSearchQueryDepthBuilder()
         $maxRewrittenQueries = 0;
 
         $thesaurusConfigFactory = $this->getThesaurusConfigFactoryMock($maxRewrittenQueries);
-
         $thesaurusIndex = $this->getMockBuilder(ThesaurusIndex::class)
             ->disableOriginalConstructor()
             ->getMock();
+        $textHelper = $this->getRealTextHelper();
 
-        $queryRewritePlugin = new QueryRewrite($queryFactory, $thesaurusConfigFactory, $thesaurusIndex);
+        $queryRewritePlugin = new QueryRewrite($queryFactory, $thesaurusConfigFactory, $thesaurusIndex, $textHelper);
         $queryBuilderInterceptor = $this->getQueryBuilderWithPlugin($queryFactory, $queryRewritePlugin);
 
         /*
@@ -162,12 +162,12 @@ public function testMultipleSearchQueryDepthBuilderWithRewrites()
         $maxRewrittenQueries = 0;
 
         $thesaurusConfigFactory = $this->getThesaurusConfigFactoryMock($maxRewrittenQueries);
-
         $thesaurusIndex = $this->getMockBuilder(ThesaurusIndex::class)
             ->disableOriginalConstructor()
             ->getMock();
+        $textHelper = $this->getRealTextHelper();
 
-        $queryRewritePlugin = new QueryRewrite($queryFactory, $thesaurusConfigFactory, $thesaurusIndex);
+        $queryRewritePlugin = new QueryRewrite($queryFactory, $thesaurusConfigFactory, $thesaurusIndex, $textHelper);
         $queryBuilderInterceptor = $this->getQueryBuilderWithPlugin($queryFactory, $queryRewritePlugin);
 
         /*
@@ -224,13 +224,13 @@ public function testSingleSearchQueryLimitedRewrites()
         $maxRewrittenQueries = 1;
 
         $thesaurusConfigFactory = $this->getThesaurusConfigFactoryMock($maxRewrittenQueries);
-
         $thesaurusIndex = $this->getMockBuilder(ThesaurusIndex::class)
             ->disableOriginalConstructor()
             ->getMock();
+        $textHelper = $this->getRealTextHelper();
 
-        // Passing the mock Query Factory to the plugin to count the occurrence of calls to 'create'.
-        $queryRewritePlugin = new QueryRewrite($queryFactoryFullMock, $thesaurusConfigFactory, $thesaurusIndex);
+        // Passing the mock Query Factory to the plugin to count the occurence of calls to 'create'.
+        $queryRewritePlugin = new QueryRewrite($queryFactoryFullMock, $thesaurusConfigFactory, $thesaurusIndex, $textHelper);
         // But passing the real Query Factory (with mocked factories) to the query builder itself.
         $queryBuilderInterceptor = $this->getQueryBuilderWithPlugin($queryFactory, $queryRewritePlugin);