litellm/tests/test_litellm/proxy/client/test_model_groups.py
mateo-berri 4d5205c355 fix(proxy): give the remaining CLI clients a request timeout
The keys, credentials, models, model groups, and chat clients still sent
requests with no timeout, so a proxy that accepts the connection and
never answers pinned the caller forever. They now default to the same
30 seconds as their teams and users siblings, with chat on the OpenAI
SDK's 600 second default, and Client wires its timeout through to all of
them. S113 cannot see Session methods, so each client gets a
hanging-server regression test instead.
2026-08-29 13:32:39 -07:00

189 lines
5.8 KiB
Python

import time
import pytest
import requests
import responses
from litellm.proxy.client import Client, ModelGroupsManagementClient
from litellm.proxy.client.exceptions import UnauthorizedError
@pytest.fixture
def base_url():
return "http://localhost:8000"
@pytest.fixture
def api_key():
return "test-api-key"
@pytest.fixture
def client(base_url, api_key):
return ModelGroupsManagementClient(base_url=base_url, api_key=api_key)
def test_info_request_creation(client, base_url, api_key):
"""Test that info creates a request with correct URL and headers when return_request=True"""
request = client.info(return_request=True)
# Check request method
assert request.method == "GET"
# Check URL construction
expected_url = f"{base_url}/model_group/info"
assert request.url == expected_url
# Check authorization header
assert "Authorization" in request.headers
assert request.headers["Authorization"] == f"Bearer {api_key}"
def test_info_request_no_auth(base_url):
"""Test that info creates a request without auth header when no api_key is provided"""
client = ModelGroupsManagementClient(base_url=base_url) # No API key
request = client.info(return_request=True)
# Check URL is still correct
assert request.url == f"{base_url}/model_group/info"
# Check that there's no authorization header
assert "Authorization" not in request.headers
@pytest.mark.parametrize(
"base_url,expected",
[
("http://localhost:8000", "http://localhost:8000/model_group/info"),
(
"http://localhost:8000/",
"http://localhost:8000/model_group/info",
), # With trailing slash
("https://api.example.com", "https://api.example.com/model_group/info"),
("http://127.0.0.1:3000", "http://127.0.0.1:3000/model_group/info"),
],
)
def test_info_url_variants(base_url, expected):
"""Test that info handles different base URL formats correctly"""
client = ModelGroupsManagementClient(base_url=base_url)
request = client.info(return_request=True)
assert request.url == expected
@responses.activate
def test_info_with_mock_response(client):
"""Test the full info execution with a mocked response"""
mock_data = {
"data": [
{
"model_group_name": "gpt4-group",
"models": ["gpt-4", "gpt-4-32k"],
"litellm_params": {"timeout": 30, "max_retries": 3},
},
{
"model_group_name": "azure-group",
"models": ["azure-gpt-4", "azure-gpt-35"],
"litellm_params": {
"api_base": "https://azure-endpoint.com",
"api_version": "2023-05-15",
},
},
]
}
responses.add(
responses.GET,
f"{client._base_url}/model_group/info",
json=mock_data,
status=200,
)
response = client.info()
assert response == mock_data["data"]
assert len(response) == 2
assert response[0]["model_group_name"] == "gpt4-group"
assert response[1]["model_group_name"] == "azure-group"
@responses.activate
def test_info_unauthorized_error(client):
"""Test that info raises UnauthorizedError for 401 responses"""
responses.add(
responses.GET,
f"{client._base_url}/model_group/info",
status=401,
json={"error": "Invalid API key"},
)
with pytest.raises(UnauthorizedError) as exc_info:
client.info()
assert exc_info.value.orig_exception.response.status_code == 401
@responses.activate
def test_info_other_errors(client):
"""Test that info raises normal HTTPError for non-401 errors"""
responses.add(
responses.GET,
f"{client._base_url}/model_group/info",
status=500,
json={"error": "Internal Server Error"},
)
with pytest.raises(requests.exceptions.HTTPError) as exc_info:
client.info()
assert exc_info.value.response.status_code == 500
@pytest.mark.parametrize(
"api_key",
[
"", # Empty string
None, # None value
],
)
def test_info_invalid_api_keys(base_url, api_key):
"""Test that the client handles invalid API keys appropriately"""
client = ModelGroupsManagementClient(base_url=base_url, api_key=api_key)
request = client.info(return_request=True)
assert "Authorization" not in request.headers
def test_client_initialization_strips_trailing_slash():
"""Test that the client properly strips trailing slashes from base_url during initialization"""
client = ModelGroupsManagementClient(base_url="http://localhost:8000/////")
assert client._base_url == "http://localhost:8000"
def test_client_initialization(base_url, api_key):
"""Test that the Client properly initializes the model_groups client"""
client = Client(base_url=base_url, api_key=api_key)
# Check that model_groups client is properly initialized
assert isinstance(client.model_groups, ModelGroupsManagementClient)
assert client.model_groups._base_url == base_url
assert client.model_groups._api_key == api_key
def test_client_initialization_without_api_key(base_url):
"""Test that the client works without an API key"""
client = Client(base_url=base_url)
assert client._api_key is None
assert client.model_groups._api_key is None
def test_info_gives_up_at_the_timeout_instead_of_hanging(hanging_server):
"""
A proxy that accepts the connection but never answers used to pin the caller's
process forever, since the request carried no timeout at all.
"""
client = ModelGroupsManagementClient(base_url=hanging_server, api_key="sk-test", timeout=1)
started = time.monotonic()
with pytest.raises(requests.exceptions.Timeout):
client.info()
assert time.monotonic() - started < 10