Code formatting

Author: koenvo

ingestify/application/dataset_store.py

@@ -61,19 +61,19 @@ def set_event_bus(self, event_bus: EventBus):
     def dispatch(self, event):
         if self.event_bus:
             self.event_bus.dispatch(event)
-            
+
     @contextmanager
     def with_file_cache(self):
         """Context manager to enable file caching during its scope.
-        
+
         Files loaded within this context will be cached and reused,
         avoiding multiple downloads of the same file.
-        
+
         Example:
             # Without caching (loads files twice)
             analyzer1 = StatsAnalyzer(store, dataset)
             analyzer2 = VisualizationTool(store, dataset)
-            
+
             # With caching (files are loaded once and shared)
             with store.with_file_cache():
                 analyzer1 = StatsAnalyzer(store, dataset)
@@ -82,7 +82,7 @@ def with_file_cache(self):
         # Enable caching for this thread
         self._thread_local.use_file_cache = True
         self._thread_local.file_cache = {}
-        
+
         try:
             yield
         finally:
@@ -419,11 +419,11 @@ def get_stream(file_):
             def make_loaded_file():
                 return LoadedFile(
                     stream_=get_stream if lazy else get_stream(file),
-                    **file.model_dump()
+                    **file.model_dump(),
                 )
 
             # Using getattr with a default value of False - simple one-liner
-            if getattr(self._thread_local, 'use_file_cache', False):
+            if getattr(self._thread_local, "use_file_cache", False):
                 key = (dataset.dataset_id, current_revision.revision_id, file.file_id)
                 if key not in self._thread_local.file_cache:
                     self._thread_local.file_cache[key] = make_loaded_file()

ingestify/tests/test_file_cache.py

@@ -13,10 +13,10 @@ def test_file_cache(config_file):
     # Get engine from the fixture
     engine = get_engine(config_file, "main")
     store = engine.store
-    
+
     # Create a timestamp for test data
     now = datetime.now(timezone.utc)
-    
+
     # Create a test file
     test_file = File(
         file_id="test_file_id",
@@ -30,18 +30,18 @@ def test_file_cache(config_file):
         modified_at=now,
         size=100,
         content_type="text/plain",
-        data_spec_version="v1"
+        data_spec_version="v1",
     )
-    
+
     # Create a test revision with the file
     revision = Revision(
         revision_id=1,
         created_at=now,
         description="Test revision",
         modified_files=[test_file],
-        source={"source_type": SourceType.MANUAL, "source_id": "test"}
+        source={"source_type": SourceType.MANUAL, "source_id": "test"},
     )
-    
+
     # Create a test dataset with the revision
     dataset = Dataset(
         bucket="test-bucket",
@@ -55,41 +55,44 @@ def test_file_cache(config_file):
         created_at=now,
         updated_at=now,
         last_modified_at=now,
-        revisions=[revision]
+        revisions=[revision],
     )
-    
+
     # Create a simple pass-through reader function to replace the gzip reader
     def simple_reader(stream):
         return stream
-        
+
     # Mock both the file repository and the _prepare_read_stream method
-    with patch.object(store.file_repository, 'load_content') as mock_load_content, \
-         patch.object(store, '_prepare_read_stream') as mock_prepare_read_stream:
-        
+    with patch.object(
+        store.file_repository, "load_content"
+    ) as mock_load_content, patch.object(
+        store, "_prepare_read_stream"
+    ) as mock_prepare_read_stream:
+
         # Set up the mocks
         mock_load_content.return_value = BytesIO(b"test content")
         mock_prepare_read_stream.return_value = (simple_reader, "")
-        
+
         # Test without caching - should load files twice
         store.load_files(dataset)
         store.load_files(dataset)
-        
+
         # Should have called load_content twice (without caching)
         assert mock_load_content.call_count == 2
-        
+
         # Reset the mock
         mock_load_content.reset_mock()
-        
+
         # Test with caching - should load files only once
         with store.with_file_cache():
             store.load_files(dataset)
             store.load_files(dataset)
-            
+
             # Should have called load_content only once (with caching)
             assert mock_load_content.call_count == 1
-            
+
         # After exiting context, caching should be disabled
         store.load_files(dataset)
-        
+
         # Should have called load_content again
         assert mock_load_content.call_count == 2