IBX-11146: Updated filtering logic to use LanguageCode criterion with {terms} clause (#107)

* IBX-1146: Updated filtering logic to use LanguageCode criterion with {terms} clause

* rewrote `matchAlwaysAvailable === true` part to also use {terms} query parser

* added phpdoc

Author: ViniTou

phpstan-baseline.neon

@@ -996,12 +996,6 @@ parameters:
 			count: 1
 			path: src/lib/Query/Common/CriterionVisitor/Field/FieldRelation.php
 
-		-
-			message: '#^Parameter \#2 \$array of function array_map expects array, array\<bool\|float\|int\|string\>\|bool\|float\|int\|string given\.$#'
-			identifier: argument.type
-			count: 1
-			path: src/lib/Query/Common/CriterionVisitor/LanguageCodeIn.php
-
 		-
 			message: '#^Method Ibexa\\Solr\\Query\\Common\\CriterionVisitor\\MapLocation\:\:getSearchFields\(\) return type has no value type specified in iterable type array\.$#'
 			identifier: missingType.iterableValue

src/lib/CoreFilter/NativeCoreFilter.php

@@ -9,6 +9,7 @@
 use Ibexa\Contracts\Core\Repository\Values\Content\Query;
 use Ibexa\Contracts\Core\Repository\Values\Content\Query\Criterion;
 use Ibexa\Contracts\Core\Repository\Values\Content\Query\Criterion\CustomField;
+use Ibexa\Contracts\Core\Repository\Values\Content\Query\Criterion\LanguageCode;
 use Ibexa\Contracts\Core\Repository\Values\Content\Query\Criterion\LogicalAnd;
 use Ibexa\Contracts\Core\Repository\Values\Content\Query\Criterion\LogicalNot;
 use Ibexa\Contracts\Core\Repository\Values\Content\Query\Criterion\LogicalOr;
@@ -182,7 +183,7 @@ private function getLanguageFilter(array $languageCodes)
                     [
                         $condition,
                         new LogicalNot(
-                            new CustomField(self::FIELD_LANGUAGES, Operator::IN, $excluded)
+                            new LanguageCode($excluded, false)
                         ),
                     ]
                 );

src/lib/Query/Common/CriterionVisitor/LanguageCodeIn.php

@@ -30,19 +30,25 @@ public function canVisit(Criterion $criterion)
 
     public function visit(Criterion $criterion, ?CriterionVisitor $subVisitor = null): string
     {
-        $languageCodeExpressions = array_map(
-            static function ($value) {
-                return 'content_language_codes_ms:"' . $value . '"';
-            },
-            $criterion->value
+        /** @var Criterion\LanguageCode $criterion */
+        $values = is_array($criterion->value) ? $criterion->value : [$criterion->value];
+        // content_language_codes_ms is a string field which uses LowerCaseFilter.
+        // Since {!terms} bypasses analysis, we must manually lowercase the values to match the index.
+        /** @param bool|float|int|string $value */
+        $values = array_map(static function ($value): string {
+            return strtolower((string)$value);
+        }, $values);
+
+        $termsQuery = sprintf(
+            '_query_:"{!terms f=content_language_codes_ms}%s"',
+            implode(',', $values)
         );
 
-        /** @var Criterion\LanguageCode $criterion */
         if ($criterion->matchAlwaysAvailable) {
-            $languageCodeExpressions[] = 'content_always_available_b:true';
+            return '(' . $termsQuery . ' OR content_always_available_b:true)';
         }
 
-        return '(' . implode(' OR ', $languageCodeExpressions) . ')';
+        return $termsQuery;
     }
 }
 

tests/lib/Search/Query/Common/CriterionVisitor/LanguageCodeInTest.php

@@ -0,0 +1,53 @@
+<?php
+
+/**
+ * @copyright Copyright (C) Ibexa AS. All rights reserved.
+ * @license For full copyright and license information view LICENSE file distributed with this source code.
+ */
+declare(strict_types=1);
+
+namespace Ibexa\Tests\Solr\Search\Query\Common\CriterionVisitor;
+
+use Ibexa\Contracts\Core\Repository\Values\Content\Query\Criterion\LanguageCode;
+use Ibexa\Contracts\Solr\Query\CriterionVisitor;
+use Ibexa\Solr\Query\Common\CriterionVisitor\LanguageCodeIn;
+use Ibexa\Tests\Solr\Search\Query\BaseCriterionVisitorTestCase;
+
+final class LanguageCodeInTest extends BaseCriterionVisitorTestCase
+{
+    protected function getVisitor(): CriterionVisitor
+    {
+        return new LanguageCodeIn();
+    }
+
+    protected function getSupportedCriterion(): LanguageCode
+    {
+        return new LanguageCode('eng-GB');
+    }
+
+    /**
+     * @return iterable<string, array{0: string, 1: \Ibexa\Contracts\Core\Repository\Values\Content\Query\Criterion\LanguageCode}>
+     */
+    public function provideDataForTestVisit(): iterable
+    {
+        yield 'Single language, match always available' => [
+            '(_query_:"{!terms f=content_language_codes_ms}eng-gb" OR content_always_available_b:true)',
+            new LanguageCode('eng-GB', true),
+        ];
+
+        yield 'Multiple languages, match always available' => [
+            '(_query_:"{!terms f=content_language_codes_ms}eng-gb,pol-pl" OR content_always_available_b:true)',
+            new LanguageCode(['eng-GB', 'pol-PL'], true),
+        ];
+
+        yield 'Multiple languages, do NOT match always available (optimization)' => [
+            '_query_:"{!terms f=content_language_codes_ms}eng-gb,pol-pl"',
+            new LanguageCode(['eng-GB', 'pol-PL'], false),
+        ];
+
+        yield 'Single language, do NOT match always available (optimization)' => [
+            '_query_:"{!terms f=content_language_codes_ms}eng-gb"',
+            new LanguageCode('eng-GB', false),
+        ];
+    }
+}