ACP2E-989: "Minimum should match" with enquoted search term causes incorrect search query generation

Author: dhorytskyi

app/code/Magento/Elasticsearch/SearchAdapter/Query/Builder/MatchQuery.php

@@ -70,51 +70,6 @@ public function __construct(
         $this->config = $config;
     }
 
-    /**
-     * @inheritdoc
-     */
-    public function build(array $selectQuery, RequestQueryInterface $requestQuery, $conditionType)
-    {
-        $queryValue = $this->prepareQuery($requestQuery->getValue(), $conditionType);
-        $queries = $this->buildQueries($requestQuery->getMatches(), $queryValue);
-        $requestQueryBoost = $requestQuery->getBoost() ?: 1;
-        $minimumShouldMatch = $this->config->getElasticsearchConfigData('minimum_should_match');
-
-        foreach ($queries as $query) {
-            $queryBody = $query['body'];
-            $matchKey = array_keys($queryBody)[0];
-            foreach ($queryBody[$matchKey] as $field => $matchQuery) {
-                $matchQuery['boost'] = $requestQueryBoost + $matchQuery['boost'];
-                if ($minimumShouldMatch && $matchKey != 'match_phrase_prefix') {
-                    $matchQuery['minimum_should_match'] = $minimumShouldMatch;
-                }
-                $queryBody[$matchKey][$field] = $matchQuery;
-            }
-            $selectQuery['bool'][$query['condition']][] = $queryBody;
-        }
-
-        return $selectQuery;
-    }
-
-    /**
-     * Prepare query
-     *
-     * @param string $queryValue
-     * @param string $conditionType
-     * @return array
-     */
-    private function prepareQuery(string $queryValue, string $conditionType): array
-    {
-        $condition = $conditionType === BoolExpression::QUERY_CONDITION_NOT
-            ? self::QUERY_CONDITION_MUST_NOT
-            : $conditionType;
-
-        return [
-            'condition' => $condition,
-            'value' => $queryValue,
-        ];
-    }
-
     /**
      * Creates valid ElasticSearch search conditions from Match queries
      *
@@ -125,21 +80,24 @@ private function prepareQuery(string $queryValue, string $conditionType): array
      * The search query boost is an optional in the search query and therefore it will be set to 1 by default
      * if none passed with a match query.
      *
-     * @param array $matches
-     * @param array $queryValue
+     * @param array $selectQuery
+     * @param RequestQueryInterface $requestQuery
+     * @param string $conditionType
      * @return array
      */
-    private function buildQueries(array $matches, array $queryValue): array
+    public function build(array $selectQuery, RequestQueryInterface $requestQuery, $conditionType)
     {
-        $conditions = [];
+        $queryValue = $this->prepareQuery($requestQuery->getValue(), $conditionType);
+        $requestQueryBoost = $requestQuery->getBoost() ?: 1;
+        $minimumShouldMatch = $this->config->getElasticsearchConfigData('minimum_should_match');
 
         // Checking for quoted phrase \"phrase test\", trim escaped surrounding quotes if found
         $count = 0;
         $value = preg_replace('#^"(.*)"$#m', '$1', $queryValue['value'], -1, $count);
         $condition = ($count) ? 'match_phrase' : 'match';
         $transformedTypes = [];
 
-        foreach ($matches as $match) {
+        foreach ($requestQuery->getMatches() as $match) {
             $resolvedField = $this->fieldMapper->getFieldName(
                 $match['field'],
                 ['type' => FieldMapperInterface::TYPE_QUERY]
@@ -153,29 +111,62 @@ private function buildQueries(array $matches, array $queryValue): array
                 $transformedTypes[$valueTransformerHash] = $valueTransformer->transform($value);
             }
             $transformedValue = $transformedTypes[$valueTransformerHash];
-
             if (null === $transformedValue) {
                 //Value is incompatible with this field type.
                 continue;
             }
+
             $matchCondition = $match['matchCondition'] ?? $condition;
             $fields = [];
             $fields[$resolvedField] = [
                 'query' => $transformedValue,
-                'boost' => $match['boost'] ?? 1,
+                'boost' => $requestQueryBoost + $match['boost'] ?? 1,
             ];
 
             if (isset($match['analyzer'])) {
                 $fields[$resolvedField]['analyzer'] = $match['analyzer'];
             }
-            $conditions[] = [
-                'condition' => $queryValue['condition'],
-                'body' => [
-                    $matchCondition => $fields,
-                ],
-            ];
+
+            if ($minimumShouldMatch && $this->isConditionSupportMinimumShouldMatch($matchCondition)) {
+                $fields[$resolvedField]['minimum_should_match'] = $minimumShouldMatch;
+            }
+
+            $selectQuery['bool'][$queryValue['condition']][] = [$matchCondition => $fields];
         }
 
-        return $conditions;
+        return $selectQuery;
+    }
+
+    /**
+     * Prepare query
+     *
+     * @param string $queryValue
+     * @param string $conditionType
+     * @return array
+     */
+    private function prepareQuery(string $queryValue, string $conditionType): array
+    {
+        $condition = $conditionType === BoolExpression::QUERY_CONDITION_NOT
+            ? self::QUERY_CONDITION_MUST_NOT
+            : $conditionType;
+
+        return [
+            'condition' => $condition,
+            'value' => $queryValue,
+        ];
+    }
+
+    /**
+     * Check does condition support the minimum_should_match field
+     *
+     * @param string $condition
+     * @return bool
+     */
+    private function isConditionSupportMinimumShouldMatch(string $condition): bool
+    {
+        return !in_array($condition, [
+            'match_phrase_prefix',
+            'match_phrase',
+        ]);
     }
 }

app/code/Magento/Elasticsearch/Test/Unit/SearchAdapter/Query/Builder/MatchQueryTest.php

@@ -16,7 +16,6 @@
 use Magento\Elasticsearch\SearchAdapter\Query\ValueTransformerInterface;
 use Magento\Elasticsearch\SearchAdapter\Query\ValueTransformerPool;
 use Magento\Framework\Search\Request\Query\MatchQuery;
-use Magento\Framework\TestFramework\Unit\Helper\ObjectManager;
 use PHPUnit\Framework\MockObject\MockObject;
 use PHPUnit\Framework\TestCase;
 
@@ -39,12 +38,14 @@ class MatchQueryTest extends TestCase
      * @var MatchQueryBuilder
      */
     private $matchQueryBuilder;
+
     /**
-     * @var MockObject
+     * @var Config|MockObject
      */
     private $config;
+
     /**
-     * @var MockObject
+     * @var FieldMapperInterface|MockObject
      */
     private $fieldMapper;
 
@@ -65,16 +66,13 @@ protected function setUp(): void
             ->willReturn($valueTransformerMock);
         $valueTransformerMock->method('transform')
             ->willReturnArgument(0);
-        $this->matchQueryBuilder = (new ObjectManager($this))->getObject(
-            MatchQueryBuilder::class,
-            [
-                'fieldMapper' => $this->fieldMapper,
-                'preprocessorContainer' => [],
-                'attributeProvider' => $this->attributeProvider,
-                'fieldTypeResolver' => $this->fieldTypeResolver,
-                'valueTransformerPool' => $valueTransformerPoolMock,
-                'config' => $this->config,
-            ]
+
+        $this->matchQueryBuilder = new MatchQueryBuilder(
+            $this->fieldMapper,
+            $this->attributeProvider,
+            $this->fieldTypeResolver,
+            $valueTransformerPoolMock,
+            $this->config
         );
     }
 
@@ -182,6 +180,7 @@ public function buildDataProvider(): array
                 ],
                 '2<75%'
             ],
+            //[match_phrase] query does not support [minimum_should_match]
             'match_phrase query with minimum_should_match' => [
                 '"fitness bottle"',
                 [
@@ -196,14 +195,12 @@ public function buildDataProvider(): array
                             'name' => [
                                 'query' => 'fitness bottle',
                                 'boost' => 6,
-                                'minimum_should_match' => '2<75%',
                             ],
                         ],
                     ],
                 ],
                 '2<75%'
             ],
-
         ];
     }