EZP-31248: Fixed handling non-printable characters for Search

eZ/Publish/API/Repository/Tests/SearchServiceLocationTest.php

@@ -9,6 +9,7 @@
 namespace eZ\Publish\API\Repository\Tests;
 
 use eZ\Publish\API\Repository\Tests\SetupFactory\LegacyElasticsearch;
+use eZ\Publish\API\Repository\Values\Content\Content;
 use eZ\Publish\Core\Repository\Values\Content\Location;
 use eZ\Publish\API\Repository\Values\Content\LocationQuery;
 use eZ\Publish\API\Repository\Values\Content\Query\Criterion;
@@ -54,7 +55,7 @@ public function testFindFacetedLocation(LocationQuery $query, $fixture)
     /**
      * Create test Content with ezcountry field having multiple countries selected.
      *
-     * @return \eZ\Publish\API\Repository\Values\Content\Content
+     * @return Content
      */
     protected function createMultipleCountriesContent()
     {
@@ -101,6 +102,30 @@ protected function createMultipleCountriesContent()
         return $content;
     }
 
+    protected function createFolderWithNonPrintableUtf8Characters(): Content
+    {
+        $repository = $this->getRepository();
+        $contentTypeService = $repository->getContentTypeService();
+        $contentService = $repository->getContentService();
+
+        $contentType = $contentTypeService->loadContentTypeByIdentifier('folder');
+        $createStruct = $contentService->newContentCreateStruct($contentType, 'eng-GB');
+        $createStruct->remoteId = 'non-printable-char-folder-123';
+        $createStruct->alwaysAvailable = false;
+        $createStruct->setField(
+            'name',
+            utf8_decode("Non\x09Printable\x0EFolder")
+        );
+
+        $locationCreateStruct = $repository->getLocationService()->newLocationCreateStruct(2);
+        $draft = $contentService->createContent($createStruct, [$locationCreateStruct]);
+        $content = $contentService->publishVersion($draft->getVersionInfo());
+
+        $this->refreshSearch($repository);
+
+        return $content;
+    }
+
     /**
      * Test for the findLocations() method.
      *
@@ -157,6 +182,33 @@ public function testFieldCollectionContainsNoMatch()
         $this->assertEquals(0, $result->totalCount);
     }
 
+    /**
+     * @covers \eZ\Publish\API\Repository\SearchService::findLocations
+     */
+    public function testNonPrintableUtf8Characters(): void
+    {
+        $folder = $this->createFolderWithNonPrintableUtf8Characters();
+        $query = new LocationQuery(
+            [
+                'query' => new Criterion\Field(
+                    'name',
+                    Criterion\Operator::EQ,
+                    utf8_decode("Non\x09Printable\x0EFolder")
+                ),
+            ]
+        );
+
+        $repository = $this->getRepository();
+        $searchService = $repository->getSearchService();
+        $result = $searchService->findLocations($query);
+
+        $this->assertEquals(1, $result->totalCount);
+        $this->assertEquals(
+            $folder->contentInfo->mainLocationId,
+            $result->searchHits[0]->valueObject->id
+        );
+    }
+
     /**
      * @expectedException \eZ\Publish\API\Repository\Exceptions\InvalidArgumentException
      */

eZ/Publish/Core/Search/Common/FieldValueMapper/StringMapper.php

@@ -17,6 +17,10 @@
  */
 class StringMapper extends FieldValueMapper
 {
+    public const REPLACE_WITH_SPACE_PATTERN = '([\x09\x0B\x0C]+)';
+    public const REMOVE_PATTERN = '([\x00-\x08\x0E-\x1F]+)';
+
+
     /**
      * Check if field can be mapped.
      *
@@ -51,9 +55,16 @@ public function map(Field $field)
      */
     protected function convert($value)
     {
-        // Remove non-printable characters
+        // Replace tab, vertical tab, form-feed chars to single space.
+        $value = preg_replace(
+            self::REPLACE_WITH_SPACE_PATTERN,
+            ' ',
+            (string)$value
+        );
+
+        // Remove non-printable characters.
         return preg_replace(
-            '([\x00-\x09\x0B\x0C\x1E\x1F]+)',
+            self::REMOVE_PATTERN,
             '',
             (string)$value
         );