Fix read from code when config does not exists

PiperOrigin-RevId: 347997824

tensorflow_datasets/core/load.py

@@ -171,7 +171,10 @@ def builder(
   raise not_found_error
 
 
-def _try_load_from_files_first(cls, **builder_kwargs) -> bool:
+def _try_load_from_files_first(
+    cls: Optional[Type[dataset_builder.DatasetBuilder]],
+    **builder_kwargs: Any,
+) -> bool:
   """Returns True if files should be used rather than code."""
   if set(builder_kwargs) - {'version', 'config', 'data_dir'}:
     return False  # Has extra kwargs, require original code.
@@ -181,6 +184,12 @@ def _try_load_from_files_first(cls, **builder_kwargs) -> bool:
     return True  # Code does not exists
   elif 'version' in builder_kwargs:
     return True  # Version explicitly given (unlock backward compatibility)
+  elif (
+      'config' in builder_kwargs
+      and isinstance(builder_kwargs['config'], str)
+      and builder_kwargs['config'] not in cls.builder_configs
+  ):
+    return True  # Requested config isn't found in the code
   else:
     return False  # Code exists and no version given, use code.
 

tensorflow_datasets/core/read_only_builder_test.py

@@ -86,6 +86,26 @@ def test_builder_files_exists(code_builder: dataset_builder.DatasetBuilder):
   assert not isinstance(builder, read_only_builder.ReadOnlyBuilder)
 
 
+def test_builder_config(code_builder: dataset_builder.DatasetBuilder):
+  """Tests that code found but config not loads from files."""
+  if not code_builder.BUILDER_CONFIGS:
+    return
+
+  # Remove the registered configs
+  with mock.patch.object(type(code_builder), 'BUILDER_CONFIGS', []), \
+       mock.patch.object(type(code_builder), 'builder_configs', {}):
+    # Config isn't present in the code anymore
+    with pytest.raises(ValueError, match='BuilderConfig .* not found'):
+      load.builder(
+          f'{code_builder.name}/dummy_config', data_dir='/tmp/path/not-exists'
+      )
+
+    # But previously generated configs still be loaded from disk
+    builder = load.builder(f'{code_builder.name}/dummy_config')
+    assert not isinstance(builder, type(code_builder))
+    assert isinstance(builder, read_only_builder.ReadOnlyBuilder)
+
+
 def test_builder_code_not_found(code_builder: dataset_builder.DatasetBuilder):
   """If the code isn't found, use files instead."""
 

tensorflow_datasets/core/registered_test.py

@@ -42,6 +42,8 @@ def as_dataset(self, **kwargs):
     self.as_dataset_kwargs = kwargs
     return self
 
+  builder_configs = {}
+
 
 class UnregisteredBuilder(EmptyDatasetBuilder):