Remove using kwargs to swallow unknown factory config parameters. Unknown config parameters for a given implementation will throw an exception.

Author: dworthen

packages/graphrag-storage/README.md

@@ -29,7 +29,7 @@ from typing import Any
 from graphrag_storage import Storage, StorageConfig, create_storage, register_storage
 
 class MyStorage(Storage):
-    def __init__(self, some_setting: str, **kwargs: Any):
+    def __init__(self, some_setting: str):
         # Validate settings and initialize
         ...
 

packages/graphrag-storage/graphrag_storage/azure_blob_storage.py

@@ -33,29 +33,18 @@ class AzureBlobStorage(Storage):
 
     def __init__(
         self,
-        base_dir: str | None = None,
-        connection_string: str | None = None,
+        container_name: str,
         storage_account_blob_url: str | None = None,
-        container_name: str | None = None,
+        connection_string: str | None = None,
+        base_dir: str | None = None,
         encoding: str = "utf-8",
-        **kwargs: Any,
     ) -> None:
         """Create a new BlobStorage instance."""
-        if connection_string is None and storage_account_blob_url is None:
-            msg = "AzureBlobStorage requires either a connection_string or storage_account_blob_url to be specified."
-            logger.error(msg)
-            raise ValueError(msg)
-
         if connection_string is not None and storage_account_blob_url is not None:
             msg = "AzureBlobStorage requires only one of connection_string or storage_account_blob_url to be specified, not both."
             logger.error(msg)
             raise ValueError(msg)
 
-        if container_name is None:
-            msg = "AzureBlobStorage requires a container_name to be specified."
-            logger.error(msg)
-            raise ValueError(msg)
-
         _validate_blob_container_name(container_name)
 
         logger.info(
@@ -70,6 +59,11 @@ def __init__(
                 account_url=storage_account_blob_url,
                 credential=DefaultAzureCredential(),
             )
+        else:
+            msg = "AzureBlobStorage requires either a connection_string or storage_account_blob_url to be specified."
+            logger.error(msg)
+            raise ValueError(msg)
+
         self._encoding = encoding
         self._container_name = container_name
         self._connection_string = connection_string

packages/graphrag-storage/graphrag_storage/azure_cosmos_storage.py

@@ -41,12 +41,11 @@ class AzureCosmosStorage(Storage):
 
     def __init__(
         self,
-        base_dir: str | None = None,
-        container_name: str | None = None,
+        base_dir: str,
+        container_name: str,
         connection_string: str | None = None,
         cosmosdb_account_url: str | None = None,
         encoding: str = "utf-8",
-        **kwargs: Any,
     ) -> None:
         """Create a CosmosDB storage instance."""
         logger.info("Creating cosmosdb storage")
@@ -56,28 +55,23 @@ def __init__(
             logger.error(msg)
             raise ValueError(msg)
 
-        if connection_string is None and cosmosdb_account_url is None:
-            msg = "CosmosDB Storage requires either a connection_string or cosmosdb_account_url to be specified."
-            logger.error(msg)
-            raise ValueError(msg)
-
         if connection_string is not None and cosmosdb_account_url is not None:
             msg = "CosmosDB Storage requires either a connection_string or cosmosdb_account_url to be specified, not both."
             logger.error(msg)
             raise ValueError(msg)
 
-        if container_name is None:
-            msg = "CosmosDB Storage requires a container_name to be specified."
-            logger.error(msg)
-            raise ValueError(msg)
-
         if connection_string:
             self._cosmos_client = CosmosClient.from_connection_string(connection_string)
         elif cosmosdb_account_url:
             self._cosmos_client = CosmosClient(
                 url=cosmosdb_account_url,
                 credential=DefaultAzureCredential(),
             )
+        else:
+            msg = "CosmosDB Storage requires either a connection_string or cosmosdb_account_url to be specified."
+            logger.error(msg)
+            raise ValueError(msg)
+
         self._encoding = encoding
         self._database_name = database_name
         self._connection_string = connection_string

packages/graphrag-storage/graphrag_storage/file_storage.py

@@ -30,15 +30,8 @@ class FileStorage(Storage):
     _base_dir: Path
     _encoding: str
 
-    def __init__(
-        self, base_dir: str | None = None, encoding: str = "utf-8", **kwargs: Any
-    ) -> None:
+    def __init__(self, base_dir: str, encoding: str = "utf-8") -> None:
         """Create a file based storage."""
-        if base_dir is None:
-            msg = "FileStorage requires a base_dir to be specified."
-            logger.error(msg)
-            raise ValueError(msg)
-
         self._base_dir = Path(base_dir).resolve()
         self._encoding = encoding
         logger.info("Creating file storage at [%s]", self._base_dir)

packages/graphrag-storage/graphrag_storage/memory_storage.py

@@ -22,6 +22,7 @@ def __init__(self, **kwargs: Any) -> None:
             "base_dir": "",
             **kwargs,
         }
+        kwargs.pop("type", None)
         super().__init__(**kwargs)
         self._storage = {}
 

packages/graphrag-storage/graphrag_storage/storage.py

@@ -13,10 +13,6 @@
 class Storage(ABC):
     """Provide a storage interface."""
 
-    @abstractmethod
-    def __init__(self, **kwargs: Any) -> None:
-        """Create a storage instance."""
-
     @abstractmethod
     def find(
         self,

packages/graphrag/graphrag/cache/factory.py

@@ -30,18 +30,21 @@ class CacheFactory(Factory[PipelineCache]):
 # --- register built-in cache implementations ---
 def create_file_cache(**kwargs) -> PipelineCache:
     """Create a file-based cache implementation."""
+    kwargs.pop("type", None)
     storage = FileStorage(**kwargs)
     return JsonPipelineCache(storage)
 
 
 def create_blob_cache(**kwargs) -> PipelineCache:
     """Create a blob storage-based cache implementation."""
+    kwargs.pop("type", None)
     storage = AzureBlobStorage(**kwargs)
     return JsonPipelineCache(storage)
 
 
 def create_cosmosdb_cache(**kwargs) -> PipelineCache:
     """Create a CosmosDB-based cache implementation."""
+    kwargs.pop("type", None)
     storage = AzureCosmosStorage(**kwargs)
     return JsonPipelineCache(storage)
 

tests/integration/storage/test_file_storage.py

@@ -43,7 +43,7 @@ async def test_get_creation_date():
 
 
 async def test_child():
-    storage = FileStorage()
+    storage = FileStorage(base_dir="")
     storage = storage.child("tests/fixtures/text/input")
     items = list(storage.find(re.compile(r".*\.txt$")))
     assert items == [str(Path("dulce.txt"))]