fix: DEV-2687: Display source filename for S3 objects (HumanSignal#2555)

* fix: DEV-2687: Display source filename for S3 objects * Move storage filename to separate field (DEV-2687)
SimoneNardi · Jul 5, 2022 · 8b693b9 · 8b693b9
1 parent fb62892
commit 8b693b9
Show file tree

Hide file tree

Showing 5 changed files with 37 additions and 7 deletions.
diff --git a/label_studio/core/settings/base.py b/label_studio/core/settings/base.py
@@ -438,6 +438,14 @@
 USE_ENFORCE_CSRF_CHECKS = get_bool_env('USE_ENFORCE_CSRF_CHECKS', True)  # False is for tests
 CLOUD_FILE_STORAGE_ENABLED = False
 
+IO_STORAGES_IMPORT_LINK_NAMES = [
+    'io_storages_s3importstoragelink',
+    'io_storages_gcsimportstoragelink',
+    'io_storages_azureblobimportstoragelink',
+    'io_storages_localfilesimportstoragelink',
+    'io_storages_redisimportstoragelink',
+]
+
 CREATE_ORGANIZATION = 'organizations.functions.create_organization'
 GET_OBJECT_WITH_CHECK_AND_LOG = 'core.utils.get_object.get_object_with_check_and_log'
 SAVE_USER = 'users.functions.save_user'

diff --git a/label_studio/data_manager/functions.py b/label_studio/data_manager/functions.py
@@ -219,10 +219,21 @@ def get_all_columns(project, *_):
         },
         {
             'id': 'file_upload',
-            'title': "Source filename",
+            'title': "Upload filename",
             'type': "String",
             'target': 'tasks',
-            'help': 'Source filename from import step',
+            'help': 'Filename of uploaded file',
+            'visibility_defaults': {
+                'explore': False,
+                'labeling': False
+            }
+        },
+        {
+            'id': 'storage_filename',
+            'title': "Storage filename",
+            'type': "String",
+            'target': 'tasks',
+            'help': 'Filename from import storage',
             'visibility_defaults': {
                 'explore': False,
                 'labeling': False

diff --git a/label_studio/data_manager/serializers.py b/label_studio/data_manager/serializers.py
@@ -174,6 +174,7 @@ class DataManagerTaskSerializer(TaskSerializer):
     predictions_results = serializers.SerializerMethodField(required=False)
     predictions_score = serializers.FloatField(required=False)
     file_upload = serializers.SerializerMethodField(required=False)
+    storage_filename = serializers.SerializerMethodField(required=False)
     annotations_ids = serializers.SerializerMethodField(required=False)
     predictions_model_versions = serializers.SerializerMethodField(required=False)
     avg_lead_time = serializers.FloatField(required=False)
@@ -232,10 +233,14 @@ def get_predictions(self, task):
 
     @staticmethod
     def get_file_upload(task):
-        if not hasattr(task, 'file_upload_field'):
-            return None
-        file_upload = task.file_upload_field
-        return os.path.basename(task.file_upload_field) if file_upload else None
+        if hasattr(task, 'file_upload_field'):
+            file_upload = task.file_upload_field
+            return os.path.basename(task.file_upload_field) if file_upload else None
+        return None
+
+    @staticmethod
+    def get_storage_filename(task):
+        return task.storage_filename
 
     @staticmethod
     def get_updated_by(obj):

diff --git a/label_studio/tasks/models.py b/label_studio/tasks/models.py
@@ -137,6 +137,12 @@ def has_lock(self, user=None):
     def num_locks(self):
         return self.locks.filter(expire_at__gt=now()).count()
 
+    @property
+    def storage_filename(self):
+        for link_name in settings.IO_STORAGES_IMPORT_LINK_NAMES:
+            if hasattr(self, link_name):
+                return getattr(self, link_name).key
+
     def get_lock_ttl(self):
         if settings.TASK_LOCK_TTL is not None:
             return settings.TASK_LOCK_TTL

diff --git a/label_studio/tests/test_api.py b/label_studio/tests/test_api.py
@@ -178,7 +178,7 @@ def test_delete_annotations(business_client, configured_project):
          'overlap': 1, 'file_upload': None, 'annotations_ids': '', 'annotations_results': '',
          'annotators': [], 'completed_at': None, 'predictions_model_versions': '',
          'predictions_results': '', 'predictions_score': None, 'total_annotations': 0, 'total_predictions': 0,
-         'avg_lead_time': None, 'cancelled_annotations': 0, 'inner_id': 0},
+         'avg_lead_time': None, 'cancelled_annotations': 0, 'inner_id': 0,'storage_filename': None},
         200
     )
 ])