Issue #3437068 by phenaproxima: When importing file entities, check for duplicate files

(cherry picked from commit 67ac9bc2f75f3f284427a00208aa0c019c8e6459)

core/lib/Drupal/Core/DefaultContent/Importer.php

@@ -87,29 +87,45 @@ public function importContent(Finder $content): void {
       // If a file exists in the same folder, copy it to the designated
       // target URI.
       if ($entity instanceof FileInterface) {
-        /** @var string $uri */
-        $uri = $entity->getFileUri();
-        $file_source = $path . '/' . $entity->getFilename();
-        if (file_exists($file_source)) {
-          $target_directory = dirname($uri);
-          $this->fileSystem->prepareDirectory($target_directory, FileSystemInterface::CREATE_DIRECTORY);
-          // @todo https://drupal.org/i/3437068 Don't copy the file if it
-          //   already exists.
-          $new_uri = $this->fileSystem->copy($file_source, $uri);
-          $entity->setFileUri($new_uri);
-        }
-        else {
-          $this->logger?->warning("File entity %name was imported, but the associated file (@path) was not found.", [
-            '%name' => $entity->getFilename(),
-            '@path' => $file_source,
-          ]);
-        }
+        $this->copyFileAssociatedWithEntity($path, $entity);
       }
       $entity->save();
     }
     $this->accountSwitcher->switchBack();
   }
 
+  private function copyFileAssociatedWithEntity(string $path, FileInterface $entity): void {
+    $destination = $entity->getFileUri();
+    assert(is_string($destination));
+
+    // If the source file doesn't exist, there's nothing we can do.
+    $source = $path . '/' . basename($destination);
+    if (!file_exists($source)) {
+      $this->logger?->warning("File entity %name was imported, but the associated file (@path) was not found.", [
+        '%name' => $entity->label(),
+        '@path' => $source,
+      ]);
+      return;
+    }
+
+    if (file_exists($destination)) {
+      $source_hash = hash_file('sha256', $source);
+      assert(is_string($source_hash));
+      $destination_hash = hash_file('sha256', $destination);
+      assert(is_string($destination_hash));
+
+      // If we already have the file, we don't need to do anything else.
+      if (hash_equals($source_hash, $destination_hash)) {
+        return;
+      }
+    }
+
+    $target_directory = dirname($destination);
+    $this->fileSystem->prepareDirectory($target_directory, FileSystemInterface::CREATE_DIRECTORY);
+    $uri = $this->fileSystem->copy($source, $destination);
+    $entity->setFileUri($uri);
+  }
+
   /**
    * Converts an array of content entity data to a content entity object.
    *

core/tests/Drupal/FunctionalTests/DefaultContent/ContentImportTest.php

@@ -163,6 +163,21 @@ private function assertContentWasImported(): void {
     $this->assertSame('druplicon.png', $file->getFilename());
     $this->assertSame('d8404562-efcc-40e3-869e-40132d53fe0b', $file->uuid());
 
+    // Another file entity referencing an existing file should have the same
+    // file URI -- in other words, it should have reused the existing file.
+    $duplicate_file = $entity_repository->loadEntityByUuid('file', '23a7f61f-1db3-407d-a6dd-eb4731995c9f');
+    $this->assertInstanceOf(FileInterface::class, $duplicate_file);
+    $this->assertSame('druplicon-duplicate.png', $duplicate_file->getFilename());
+    $this->assertSame($file->getFileUri(), $duplicate_file->getFileUri());
+
+    // Another file entity that references a file with the same name as, but
+    // different contents than, an existing file, should be imported and the
+    // file should be renamed.
+    $different_file = $entity_repository->loadEntityByUuid('file', 'a6b79928-838f-44bd-a8f0-44c2fff9e4cc');
+    $this->assertInstanceOf(FileInterface::class, $different_file);
+    $this->assertSame('druplicon-different.png', $different_file->getFilename());
+    $this->assertStringEndsWith('/druplicon_0.png', (string) $different_file->getFileUri());
+
     // Our node should have a menu link, and it should use the path alias we
     // included with the recipe.
     $menu_link = $entity_repository->loadEntityByUuid('menu_link_content', '3434bd5a-d2cd-4f26-bf79-a7f6b951a21b');

core/tests/fixtures/default_content/a6b79928-838f-44bd-a8f0-44c2fff9e4cc.yml

+_meta:
+  version: '1.0'
+  entity_type: file
+  uuid: a6b79928-838f-44bd-a8f0-44c2fff9e4cc
+  default_langcode: en
+default:
+  uid:
+    -
+      target_id: 1
+  filename:
+    -
+      value: druplicon-different.png
+  uri:
+    -
+      value: 'public://2024-03/druplicon.png'
+  filemime:
+    -
+      value: text/plain
+  filesize:
+    -
+      value: 11
+  status:
+    -
+      value: true
+  created:
+    -
+      value: 1711121742

core/tests/fixtures/default_content/druplicon.png

+Not a PNG.

core/tests/fixtures/default_content/file/23a7f61f-1db3-407d-a6dd-eb4731995c9f.yml

+_meta:
+  version: '1.0'
+  entity_type: file
+  uuid: 23a7f61f-1db3-407d-a6dd-eb4731995c9f
+  default_langcode: en
+default:
+  uid:
+    -
+      target_id: 1
+  filename:
+    -
+      value: druplicon-duplicate.png
+  uri:
+    -
+      value: 'public://2024-03/druplicon.png'
+  filemime:
+    -
+      value: image/png
+  filesize:
+    -
+      value: 3905
+  status:
+    -
+      value: true
+  created:
+    -
+      value: 1711121742