Fix _process_params method when body is present (#429)

fastagency/api/openapi/client.py

@@ -98,15 +98,16 @@ def _process_params(
 
         url = self.servers[0]["url"] + expanded_path
 
-        body_dict = (
-            {
-                "json": kwargs[body].model_dump()
-                if hasattr(kwargs[body], "model_dump")
-                else kwargs[body].dict()
-            }
-            if body and body in kwargs
-            else {}
-        )
+        body_dict = {}
+        if body and body in kwargs:
+            body_value = kwargs[body]
+            if isinstance(body_value, dict):
+                body_dict = {"json": body_value}
+            elif hasattr(body_value, "model_dump"):
+                body_dict = {"json": body_value.model_dump()}
+            else:
+                body_dict = {"json": body_value.dict()}
+
         body_dict["headers"] = {"Content-Type": "application/json"}
         if security:
             q_params, body_dict = kwargs["security"].add_security(q_params, body_dict)

tests/api/openapi/test_endpoint_with_body.py

@@ -0,0 +1,70 @@
+from typing import Any
+
+import pytest
+from autogen import ConversableAgent, UserProxyAgent
+from fastapi import FastAPI
+from pydantic import BaseModel
+
+from fastagency.api.openapi.client import OpenAPI
+
+
+def create_fastapi_app_with_body(host: str, port: int) -> FastAPI:
+    class Item(BaseModel):
+        name: str
+        price: float
+
+    app = FastAPI(
+        servers=[
+            {"url": f"http://{host}:{port}", "description": "Local development server"}
+        ]
+    )
+
+    @app.post("/items")
+    async def create_item(item: Item) -> str:
+        return "Item created"
+
+    return app
+
+
+@pytest.mark.azure_oai
+@pytest.mark.parametrize(
+    "fastapi_openapi_url",
+    [(create_fastapi_app_with_body)],
+    indirect=["fastapi_openapi_url"],
+)
+def test_end2end(
+    fastapi_openapi_url: str,
+    azure_gpt35_turbo_16k_llm_config: dict[str, Any],
+) -> None:
+    api = OpenAPI.create(openapi_url=fastapi_openapi_url)
+
+    agent = ConversableAgent(name="agent", llm_config=azure_gpt35_turbo_16k_llm_config)
+    user_proxy = UserProxyAgent(
+        name="user_proxy",
+        llm_config=azure_gpt35_turbo_16k_llm_config,
+        human_input_mode="NEVER",
+    )
+
+    api._register_for_llm(agent)
+    api._register_for_execution(user_proxy)
+
+    message = "Add item with name 'apple', price 1.0"
+    user_proxy.initiate_chat(
+        agent,
+        message=message,
+        summary_method="reflection_with_llm",
+        max_turns=3,
+    )
+
+    message_existed = False
+    expected_message = "Item created"
+    for message in agent.chat_messages[user_proxy]:
+        if (
+            isinstance(message, dict)
+            and "content" in message
+            and isinstance(message["content"], str)
+            and message["content"] == expected_message
+        ):
+            message_existed = True
+            break
+    assert message_existed, f"Expected message '{expected_message}' not found"