Multi-Rag / src /tests /integration /test_chat_route.py
VashuTheGreat2's picture
Upload folder using huggingface_hub
644f326 verified
Raw History Blame Contribute Delete
6.81 kB
import json
from unittest.mock import AsyncMock, MagicMock, patch
import pytest
from fastapi.testclient import TestClient
from langchain_core.messages import HumanMessage, AIMessage, AIMessageChunk
from langchain_core.language_models.chat_models import BaseChatModel
from langchain_core.outputs import ChatGenerationChunk, ChatResult
from src.agents.graph import create_agent_graph
class FakeStreamingChatModel(BaseChatModel):
def _generate(self, messages, stop=None, run_manager=None, **kwargs):
raise NotImplementedError()
async def _agenerate(self, messages, stop=None, run_manager=None, **kwargs):
chunk_reasoning = ChatGenerationChunk(
message=AIMessageChunk(
content="",
additional_kwargs={"reasoning_content": "Analyzing user inquiry..."},
)
)
if run_manager:
await run_manager.on_llm_new_token("", chunk=chunk_reasoning)
tokens = ["<think>", "Internal deliberations", "</think>", "This is the final response."]
for t in tokens:
chunk = ChatGenerationChunk(message=AIMessageChunk(content=t))
if run_manager:
await run_manager.on_llm_new_token(t, chunk=chunk)
return ChatResult(
generations=[
ChatGenerationChunk(
message=AIMessageChunk(content="This is the final response.")
)
]
)
@property
def _llm_type(self):
return "fake_streaming_chat_model"
def parse_sse_events(sse_text: str):
"""Parse SSE raw text into a list of (event_type, json_data) tuples."""
events = []
current_event = "message"
for line in sse_text.splitlines():
line = line.strip()
if not line:
continue
if line.startswith("event:"):
current_event = line.replace("event:", "").strip()
elif line.startswith("data:"):
data_str = line.replace("data:", "").strip()
data = json.loads(data_str)
events.append((current_event, data))
return events
def test_chat_unauthenticated(client):
"""Chat endpoint should reject unauthenticated requests without user_id / thread_id."""
response = client.post(
"/api/v1/chat",
json={"message": "Hello"},
)
assert response.status_code == 401
def test_chat_streaming_success(client):
"""Chat endpoint should stream reasoning, tool_start/tool_end, token, and final SSE events."""
async def mock_planner(state):
return {"route_to": "retriever", "queries": "kubernetes ingress"}
async def mock_retriever(state):
return {"retreived_content": ["Document snippet about ingress"]}
fake_llm = FakeStreamingChatModel()
with patch("src.agents.graph.planner_node", side_effect=mock_planner), \
patch("src.agents.nodes.responder.get_llm", return_value=fake_llm) as mock_get_llm, \
patch("src.agents.graph.retreived_node", side_effect=mock_retriever):
test_graph = create_agent_graph()
with patch("api.routes.chat_router.get_graph", return_value=test_graph):
response = client.post(
"/api/v1/chat",
headers={"x-user-id": "usr_123", "x-thread-id": "thr_456"},
json={
"message": "How does ingress work?",
"stream": True,
"metadata": {"model": "openai/gpt-oss-20b"},
},
)
assert response.status_code == 200
assert "text/event-stream" in response.headers.get("content-type", "")
events = parse_sse_events(response.text)
payloads = [ev[1] for ev in events if isinstance(ev[1], dict)]
payload_types = [p.get("type") for p in payloads]
assert "reasoning" in payload_types
reasoning_contents = [p.get("content") for p in payloads if p.get("type") == "reasoning"]
assert any("analyzing" in c.lower() for c in reasoning_contents)
assert "tool_start" in payload_types
assert any(p.get("tool") == "retriever" for p in payloads if p.get("type") == "tool_start")
assert "tool_end" in payload_types
assert any(p.get("tool") == "retriever" for p in payloads if p.get("type") == "tool_end")
assert "token" in payload_types
token_contents = "".join([p.get("content", "") for p in payloads if p.get("type") == "token"])
assert "This is the final response." in token_contents
assert "final" in payload_types
final_payload = next(p for p in payloads if p.get("type") == "final")
assert "This is the final response." in final_payload.get("ai_response", "")
assert final_payload.get("state") == "completed"
mock_get_llm.assert_called_once_with(model="openai/gpt-oss-20b")
def test_chat_non_streaming_success(client):
"""Chat endpoint should support non-streaming JSON responses when stream=False."""
async def mock_planner(state):
return {"route_to": "responder"}
async def mock_responder(state):
return {
"messages": [
HumanMessage(content=state.get("user_message")),
AIMessage(content="Direct JSON answer"),
],
"retreived_content": [
type("FakeDocument", (), {"page_content": "Retrieved DevOps context"})()
],
}
with patch("src.agents.graph.planner_node", side_effect=mock_planner), \
patch("src.agents.graph.responder_node", side_effect=mock_responder):
test_graph = create_agent_graph()
with patch("api.routes.chat_router.get_graph", return_value=test_graph):
response = client.post(
"/api/v1/chat",
headers={"x-user-id": "usr_123", "x-thread-id": "thr_456"},
json={"message": "What is Python?", "stream": False},
)
assert response.status_code == 200
data = response.json()
assert data.get("success") is True
assert data.get("response") == "Direct JSON answer"
assert data.get("data", {}).get("retreived_content") == ["Retrieved DevOps context"]
def test_chat_graph_unconfigured(client):
"""Chat endpoint should handle missing graph configuration gracefully."""
with patch("api.routes.chat_router.get_graph", side_effect=Exception("Config error")):
response = client.post(
"/api/v1/chat",
headers={"x-user-id": "usr_123", "x-thread-id": "thr_456"},
json={"message": "Hello", "stream": False},
)
assert response.status_code == 500
assert "error" in response.json()