Fixed linting problems

libs/vertexai/langchain_google_vertexai/functions_utils.py

@@ -55,7 +55,7 @@ def _format_tool_to_vertex_function(tool: BaseTool) -> FunctionDescription:
 
 
 def _format_tools_to_vertex_tool(
-    tools: List[Union[BaseTool, Type[BaseModel]]],
+    tools: List[Union[BaseTool, Type[BaseModel], dict]],
 ) -> List[VertexTool]:
     "Format tool into the Vertex Tool instance."
     function_declarations = []
@@ -64,12 +64,14 @@ def _format_tools_to_vertex_tool(
             func = _format_tool_to_vertex_function(tool)
         elif isinstance(tool, type) and issubclass(tool, BaseModel):
             func = _format_pydantic_to_vertex_function(tool)
-        else:
+        elif isinstance(tool, dict):
             func = {
                 "name": tool["name"],
-                "description": tool.get("description"),
+                "description": tool.pop("description"),
                 "parameters": _get_parameters_from_schema(tool["parameters"]),
             }
+        else:
+            raise ValueError(f"Unsupported tool call type {tool}")
         function_declarations.append(FunctionDeclaration(**func))
 
     return [VertexTool(function_declarations=function_declarations)]

libs/vertexai/tests/integration_tests/test_tools.py

@@ -95,9 +95,12 @@ def test_tools() -> None:
 
 @pytest.mark.extended
 def test_custom_tool() -> None:
-    from langchain.agents import AgentExecutor, create_openai_functions_agent, tool
+    from langchain.agents import AgentExecutor, tool
+    from langchain.agents.format_scratchpad import (
+        format_to_openai_function_messages,
+    )
 
-    @tool
+    @tool("search", return_direct=True)
     def search(query: str) -> str:
         """Look up things online."""
         return "LangChain"
@@ -115,8 +118,19 @@ def search(query: str) -> str:
             MessagesPlaceholder("agent_scratchpad"),
         ]
     )
+    llm_with_tools = llm.bind(functions=tools)
 
-    agent = create_openai_functions_agent(llm, tools, prompt)
+    agent: Any = (
+        {
+            "input": lambda x: x["input"],
+            "agent_scratchpad": lambda x: format_to_openai_function_messages(
+                x["intermediate_steps"]
+            ),
+        }
+        | prompt
+        | llm_with_tools
+        | _TestOutputParser()
+    )
     agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
 
     response = agent_executor.invoke({"input": "What is LangChain?"})