Enable Strict Mypy Type Checking (#67)

* Refactor libs/oci to meet mypy stardard

Removed unnecessary type: ignore comments, improved type annotations, and updated function signatures for better type safety across chat models, embeddings, and tests. Also enhanced mypy configuration for stricter type checking and plugin support.

* Refactor libs/oracledb to meet mypy standard

* Fix more lint in libs/oci

* Remove unnecessary type ignore

* Add type: ignore[call-arg] in /Users/panxia/Documents/GitHub/langchain-oracle/libs/oci/tests/unit_tests/chat_models/test_oci_data_science.py

* Pass response to HTTPError in mock raise_for_status

* Remove test dependencies not used in _lint.yml

Removes 'test' and 'test_integration' groups from the Poetry install step in the lint workflow, ensuring only linting and typing dependencies are installed during lint jobs.

* Revert "Remove test dependencies not used in _lint.yml"

This reverts commit 3170d1d.

* Add disk space cleanup to CI workflows

Introduced steps to free disk space in both lint and test GitHub Actions workflows by removing unused Android and CodeQL directories and pruning Docker images. This helps prevent CI failures due to insufficient disk space.

Author: paxiaatucsdedu

.github/workflows/_lint.yml

@@ -34,6 +34,12 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
+      - name: Free Disk Space
+        run: |
+          sudo rm -rf /usr/local/lib/android
+          sudo rm -rf /opt/hostedtoolcache/CodeQL
+          sudo docker image prune --all --force
+
       - name: Set up Python ${{ matrix.python-version }} + Poetry ${{ env.POETRY_VERSION }}
         uses: "./.github/actions/poetry_setup"
         with:

.github/workflows/_test.yml

@@ -26,6 +26,12 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
+      - name: Free Disk Space
+        run: |
+          sudo rm -rf /usr/local/lib/android
+          sudo rm -rf /opt/hostedtoolcache/CodeQL
+          sudo docker image prune --all --force
+
       - name: Set up Python ${{ matrix.python-version }} + Poetry ${{ env.POETRY_VERSION }}
         uses: "./.github/actions/poetry_setup"
         with:

libs/oci/langchain_oci/chat_models/oci_data_science.py

@@ -598,7 +598,7 @@ def with_structured_output(
         if method == "json_mode":
             llm = self.bind(response_format={"type": "json_object"})
             output_parser = (
-                PydanticOutputParser(pydantic_object=schema)  # type: ignore[type-var, arg-type]
+                PydanticOutputParser(pydantic_object=schema)
                 if is_pydantic_schema
                 else JsonOutputParser()
             )

libs/oci/langchain_oci/chat_models/oci_generative_ai.py

@@ -31,7 +31,6 @@
     AIMessage,
     AIMessageChunk,
     BaseMessage,
-    ChatMessage,
     HumanMessage,
     SystemMessage,
     ToolCall,
@@ -350,7 +349,7 @@ def get_role(self, message: BaseMessage) -> str:
         raise ValueError(f"Unknown message type: {type(message)}")
 
     def messages_to_oci_params(
-        self, messages: Sequence[ChatMessage], **kwargs: Any
+        self, messages: Sequence[BaseMessage], **kwargs: Any
     ) -> Dict[str, Any]:
         """
         Convert LangChain messages to OCI parameters for Cohere.
@@ -417,7 +416,7 @@ def messages_to_oci_params(
         current_turn = list(reversed(current_turn))
 
         # Process tool results from the current turn
-        oci_tool_results: List[Any] = []
+        oci_tool_results: Optional[List[Any]] = []
         for message in current_turn:
             if isinstance(message, ToolMessage):
                 tool_msg = message
@@ -434,7 +433,7 @@ def messages_to_oci_params(
                                 parameters=lc_tool_call["args"],
                             )
                             tool_result.outputs = [{"output": tool_msg.content}]
-                            oci_tool_results.append(tool_result)
+                            oci_tool_results.append(tool_result)  # type: ignore[union-attr]
         if not oci_tool_results:
             oci_tool_results = None
 
@@ -552,7 +551,7 @@ def process_stream_tool_calls(
         Returns:
             List of ToolCallChunk objects
         """
-        tool_call_chunks = []
+        tool_call_chunks: List[ToolCallChunk] = []
         tool_call_response = self.chat_stream_tool_calls(event_data)
 
         if not tool_call_response:
@@ -813,7 +812,7 @@ def _should_allow_more_tool_calls(
                 return False
 
             # Detect infinite loop: same tool called with same arguments in succession
-            recent_calls = []
+            recent_calls: list = []
             for msg in reversed(messages):
                 if hasattr(msg, "tool_calls") and msg.tool_calls:
                     for tc in msg.tool_calls:
@@ -895,7 +894,7 @@ def _process_message_content(
 
     def convert_to_oci_tool(
         self,
-        tool: Union[Type[BaseModel], Callable, BaseTool],
+        tool: Union[Dict[str, Any], Type[BaseModel], Callable, BaseTool],
     ) -> Dict[str, Any]:
         """Convert a BaseTool instance, TypedDict or BaseModel type
         to a OCI tool in Meta's format.
@@ -925,8 +924,8 @@ def convert_to_oci_tool(
                     "required": parameters.get("required", []),
                 },
             )
-        elif isinstance(tool, BaseTool):
-            return self.oci_function_definition(
+        elif isinstance(tool, BaseTool):  # type: ignore[unreachable]
+            return self.oci_function_definition(  # type: ignore[unreachable]
                 name=tool.name,
                 description=OCIUtils.remove_signature_from_tool_description(
                     tool.name, tool.description
@@ -1016,7 +1015,7 @@ def process_stream_tool_calls(
         Returns:
             List of ToolCallChunk objects
         """
-        tool_call_chunks = []
+        tool_call_chunks: List[ToolCallChunk] = []
         tool_call_response = self.chat_stream_tool_calls(event_data)
 
         if not tool_call_response:
@@ -1142,7 +1141,7 @@ def _prepare_request(
         stop: Optional[List[str]],
         stream: bool,
         **kwargs: Any,
-    ) -> Dict[str, Any]:
+    ) -> Any:
         """
         Prepare the OCI chat request from LangChain messages.
 
@@ -1299,8 +1298,8 @@ def with_structured_output(
             tool_name = getattr(self._provider.convert_to_oci_tool(schema), "name")
             if is_pydantic_schema:
                 output_parser: OutputParserLike = PydanticToolsParser(
-                    tools=[schema],  # type: ignore[list-item]
-                    first_tool_only=True,  # type: ignore[list-item]
+                    tools=[schema],
+                    first_tool_only=True,
                 )
             else:
                 output_parser = JsonOutputKeyToolsParser(
@@ -1309,15 +1308,15 @@ def with_structured_output(
         elif method == "json_mode":
             llm = self.bind(response_format={"type": "JSON_OBJECT"})
             output_parser = (
-                PydanticOutputParser(pydantic_object=schema)  # type: ignore[type-var, arg-type]
+                PydanticOutputParser(pydantic_object=schema)
                 if is_pydantic_schema
                 else JsonOutputParser()
             )
         elif method == "json_schema":
-            json_schema_dict = (
+            json_schema_dict: Dict[str, Any] = (
                 schema.model_json_schema()  # type: ignore[union-attr]
                 if is_pydantic_schema
-                else schema
+                else schema  # type: ignore[assignment]
             )
 
             response_json_schema = self._provider.oci_response_json_schema(

libs/oci/langchain_oci/embeddings/oci_data_science_model_deployment_endpoint.py

@@ -104,12 +104,9 @@ def _completion_with_retry(**kwargs: Any) -> Any:
                 response.raise_for_status()
                 return response
             except requests.exceptions.HTTPError as http_err:
-                if response.status_code == 401 and self._refresh_signer():
-                    raise TokenExpiredError() from http_err
-                else:
-                    raise ValueError(
-                        f"Server error: {str(http_err)}. Message: {response.text}"
-                    ) from http_err
+                raise ValueError(
+                    f"Server error: {str(http_err)}. Message: {response.text}"
+                ) from http_err
             except Exception as e:
                 raise ValueError(f"Error occurs by inference endpoint: {str(e)}") from e
 

libs/oci/langchain_oci/embeddings/oci_generative_ai.py

@@ -125,7 +125,7 @@ def validate_environment(cls, values: Dict) -> Dict:  # pylint: disable=no-self-
                 client_kwargs.pop("signer", None)
             elif values["auth_type"] == OCIAuthType(2).name:
 
-                def make_security_token_signer(oci_config):  # type: ignore[no-untyped-def]
+                def make_security_token_signer(oci_config):
                     pk = oci.signer.load_private_key_from_file(
                         oci_config.get("key_file"), None
                     )

libs/oci/langchain_oci/llms/oci_generative_ai.py

@@ -151,7 +151,7 @@ def validate_environment(cls, values: Dict) -> Dict:
                 client_kwargs.pop("signer", None)
             elif values["auth_type"] == OCIAuthType(2).name:
 
-                def make_security_token_signer(oci_config):  # type: ignore[no-untyped-def]
+                def make_security_token_signer(oci_config):
                     pk = oci.signer.load_private_key_from_file(
                         oci_config.get("key_file"), None
                     )

libs/oci/pyproject.toml

@@ -77,20 +77,26 @@ ignore = [
 ]
 
 [tool.mypy]
-ignore_missing_imports = "True"
-
-# Disable specific error codes that are causing issues
-disallow_untyped_defs = "False"
-disable_error_code = ["attr-defined", "assignment", "var-annotated", "override", "union-attr", "arg-type"]
-
-# TODO: LangChain Google settings
-# plugins = ["pydantic.mypy"]
-# strict = true
-# disallow_untyped_defs = true
-
-# # TODO: activate for 'strict' checking
-# disallow_any_generics = false
-# warn_return_any = false
+plugins = ["pydantic.mypy"]
+check_untyped_defs = true
+error_summary = false
+pretty = true
+show_column_numbers = true
+show_error_codes = true
+show_error_context = true
+warn_redundant_casts = true
+warn_unreachable = true
+warn_unused_configs = true
+warn_unused_ignores = true
+
+# Ignore missing imports only for specific untyped packages
+[[tool.mypy.overrides]]
+module = [
+    "oci.*",
+    "ads.*",
+    "langchain_openai.*",
+]
+ignore_missing_imports = true
 
 [tool.coverage.run]
 omit = ["tests/*"]

libs/oci/tests/unit_tests/chat_models/test_oci_data_science.py

@@ -80,7 +80,7 @@ def __init__(self, json_data: Dict, status_code: int = 200):
     def raise_for_status(self) -> None:
         """Mocked raise for status."""
         if 400 <= self.status_code < 600:
-            raise HTTPError()  # type: ignore[call-arg]
+            raise HTTPError(response=self)  # type: ignore[arg-type]
 
     def json(self) -> Dict:
         """Returns mocked json data."""
@@ -155,7 +155,7 @@ def test_stream_vllm(*args: Any) -> None:
         if output is None:
             output = chunk
         else:
-            output += chunk
+            output += chunk  # type: ignore[assignment]
         count += 1
     assert count == 5
     assert output is not None