Use Drupal’s mime type guesser service to determine local mime type (#…

…4122)

modules/metastore/metastore.services.yml

@@ -42,6 +42,7 @@ services:
       - '@dkan.metastore.storage'
       - '@dkan.metastore.url_generator'
       - '@http_client'
+      - '@file.mime_type.guesser.extension'
     calls:
       - [setLoggerFactory, ['@logger.factory']]
 

modules/metastore/src/Reference/Referencer.php

@@ -14,6 +14,7 @@
 
 use GuzzleHttp\Client;
 use GuzzleHttp\Exception\GuzzleException;
+use Symfony\Component\Mime\MimeTypeGuesserInterface;
 
 /**
  * Metastore referencer service.
@@ -50,20 +51,40 @@ class Referencer {
    */
   private Client $httpClient;
 
+  /**
+   * The MIME type guesser.
+   *
+   * @var \Symfony\Component\Mime\MimeTypeGuesserInterface
+   */
+  protected $mimeTypeGuesser;
+
   /**
    * Constructor.
+   *
+   * @param \Drupal\Core\Config\ConfigFactoryInterface $configService
+   *   Drupal config factory service.
+   * @param \Contracts\FactoryInterface $storageFactory
+   *   DKAN contracts factory.
+   * @param \Drupal\metastore\Reference\MetastoreUrlGenerator $metastoreUrlGenerator
+   *   DKAN metastore url generator.
+   * @param \GuzzleHttp\Client $httpClient
+   *   Guzzle http client.
+   * @param \Symfony\Component\Mime\MimeTypeGuesserInterface $mimeTypeGuesser
+   *   The MIME type guesser.
    */
   public function __construct(
     ConfigFactoryInterface $configService,
     FactoryInterface $storageFactory,
     MetastoreUrlGenerator $metastoreUrlGenerator,
-    Client $httpClient
+    Client $httpClient,
+    MimeTypeGuesserInterface $mimeTypeGuesser
   ) {
     $this->setConfigService($configService);
     $this->storageFactory = $storageFactory;
     $this->setLoggerFactory(\Drupal::service('logger.factory'));
     $this->metastoreUrlGenerator = $metastoreUrlGenerator;
     $this->httpClient = $httpClient;
+    $this->mimeTypeGuesser = $mimeTypeGuesser;
   }
 
   /**
@@ -312,27 +333,13 @@ private function getFileMapper(): ResourceMapper {
    *   The detected mime type or NULL on failure.
    */
   private function getLocalMimeType(string $downloadUrl): ?string {
-    $mime_type = NULL;
+    // Use Drupal's mime type guesser service to get the mime type.
+    $mime_type = $this->mimeTypeGuesser->guessMimeType($downloadUrl);
 
-    // Retrieve and decode the file name from the supplied download URL's path.
-    $filename = \Drupal::service('file_system')->basename($downloadUrl);
-    $filename = urldecode($filename);
-
-    // Attempt to load the file by file name.
-    /** @var \Drupal\file\FileInterface[] $files */
-    $files = \Drupal::entityTypeManager()
-      ->getStorage('file')
-      ->loadByProperties(['filename' => $filename]);
-    $file = reset($files);
-
-    // If a valid file was found for the given file name, extract the file's
-    // mime type...
-    if ($file !== FALSE) {
-      $mime_type = $file->getMimeType();
-    }
-    // Otherwise, log an error notifying the user that a file was not found.
-    else {
-      $this->log('value_referencer', 'Unable to determine mime type of file with name "@name", because no file was found with that name.', [
+    // If we couldn't find a mime type, log an error notifying the user.
+    if (is_null($mime_type)) {
+      $filename = basename($downloadUrl);
+      $this->log('value_referencer', 'Unable to determine mime type of file with name "@name".', [
         '@name' => $filename,
       ]);
     }

modules/metastore/tests/src/Unit/Reference/ReferencerTest.php

@@ -8,6 +8,7 @@
 use Drupal\Core\Entity\EntityTypeManager;
 use Drupal\Core\Field\FieldItemListInterface;
 use Drupal\Core\File\FileSystem;
+use Drupal\Core\File\MimeType\ExtensionMimeTypeGuesser;
 use Drupal\Core\Logger\LoggerChannelFactory;
 use Drupal\Core\StreamWrapper\PublicStream;
 use Drupal\Core\StreamWrapper\StreamWrapperManager;
@@ -35,6 +36,7 @@
 use Symfony\Component\DependencyInjection\Container;
 use Symfony\Component\HttpFoundation\Request;
 use Symfony\Component\HttpFoundation\RequestStack;
+use Symfony\Component\Mime\MimeTypeGuesserInterface;
 
 /**
  * @covers \Drupal\metastore\Reference\Referencer
@@ -119,11 +121,16 @@ private function mockReferencer($existing = TRUE) {
       ->add(MetastoreUrlGenerator::class, 'uriFromUrl', 'dkan://metastore/schemas/data-dictionary/items/111')
       ->getMock();
 
+    $mimeTypeGuesser = (new Chain($this))
+      ->add(MimeTypeGuesserInterface::class, 'guessMimeType', self::MIME_TYPE)
+      ->getMock();
+
     return new Referencer(
       $configService,
       $storageFactory,
       $urlGenerator,
-      new Client()
+      new Client(),
+      $mimeTypeGuesser
     );
   }
 
@@ -394,11 +401,16 @@ public function getMimeType() { return ReferencerTest::MIME_TYPE; }
       ->add(MetastoreUrlGenerator::class, 'uriFromUrl', '')
       ->getMock();
 
+    $mimeTypeGuesser = (new Chain($this))
+      ->add(MimeTypeGuesserInterface::class, 'guessMimeType', self::MIME_TYPE)
+      ->getMock();
+
     $referencer = new Referencer(
       $configService,
       $storageFactory,
       $urlGenerator,
-      new Client()
+      new Client(),
+      $mimeTypeGuesser
     );
 
     // Test Mime Type detection using the resource `mediaType` property.
@@ -454,11 +466,16 @@ public function testDistributionHandlingDataDict($distribution, $describedBy) {
       ->disableOriginalConstructor()
       ->getMock();
 
+    $mimeTypeGuesser = (new Chain($this))
+      ->add(MimeTypeGuesserInterface::class, 'guessMimeType', self::MIME_TYPE)
+      ->getMock();
+
     $referencer = new Referencer(
       $configService,
       $storageFactory,
       $urlGenerator,
-      $http_client
+      $http_client,
+      $mimeTypeGuesser
     );
 
     if ($describedBy instanceof \Exception) {

tests/src/Functional/DatasetBTBTest.php

@@ -514,8 +514,7 @@ private function getData(string $identifier, string $title, array $downloadUrls)
       $distribution = new \stdClass();
       $distribution->title = 'Distribution #' . $key . ' for ' . $identifier;
       $distribution->downloadURL = $this->getDownloadUrl($downloadUrl);
-      $distribution->format = 'csv';
-      $distribution->mediaType = 'text/csv';
+      // Don't provide mime type or format fields since they're not required.
 
       $data->distribution[] = $distribution;
     }