mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-25 05:54:53 +00:00
44d99475a5
* test(integration): stop GET /__observations from draining the upstream log GET /__observations used to empty the queue, so a second read in a test saw nothing and a "never reached upstream" check could pass whether or not the request was forwarded. GET now returns every request since the last reset and DELETE /__observations is the reset. An autouse fixture resets the worker upstream before each test, which also covers tests that read without the gateway fixture. Discarded reads used as resets become DELETEs, helpers that accumulated drained reads now read one snapshot, and tests with several phases reset between them. The three vacuous negative checks (rag allowlist, guardrail opt-out chaos, xAI batch output) now assert the absence on the same read that shows a request which did reach the upstream. * style(integration): space the observations unit test like its neighbours
71 lines
2.5 KiB
Python
71 lines
2.5 KiB
Python
import json
|
|
import uuid
|
|
from hashlib import sha256
|
|
from pathlib import Path
|
|
from typing import Final
|
|
|
|
import httpx
|
|
import yaml
|
|
from pydantic import JsonValue
|
|
|
|
from integration._support.client import Gateway, Scenario, object_value
|
|
from integration._support.database import read_rows
|
|
from integration._support.wire import Reply, Request
|
|
|
|
MANAGEMENT_ROUTES: Final = ["/key/*", "/team/new", "/team/update", "/v1/chat/completions"]
|
|
|
|
|
|
def denying_guardrail(request: Request) -> Reply:
|
|
assert request.target == "/beta/litellm_basic_guardrail_api"
|
|
return Reply(body=json.dumps({"action": "BLOCKED", "blocked_reason": "synthetic policy denial"}).encode())
|
|
|
|
|
|
def guardrail_config(policy_url: str, path: Path) -> Path:
|
|
config: Final = yaml.safe_load(Path("tests/integration/proxy_config.yaml").read_text())
|
|
config["guardrails"] = [
|
|
{
|
|
"guardrail_name": "guardrail" + uuid.uuid4().hex,
|
|
"litellm_params": {
|
|
"guardrail": "generic_guardrail_api",
|
|
"mode": "pre_call",
|
|
"default_on": True,
|
|
"api_base": policy_url,
|
|
"api_key": "synthetic-guardrail-key",
|
|
},
|
|
}
|
|
]
|
|
path.write_text(yaml.safe_dump(config))
|
|
return path
|
|
|
|
|
|
def stored_metadata(token: str) -> dict[str, object]:
|
|
rows: Final = read_rows(
|
|
'SELECT metadata FROM "LiteLLM_VerificationToken" WHERE token = %s', (sha256(token.encode()).hexdigest(),)
|
|
)
|
|
assert len(rows) == 1, rows
|
|
return rows[0]["metadata"]
|
|
|
|
|
|
def non_admin_caller(scenario: Scenario, member: str, team: str, model: str) -> str:
|
|
return scenario.key(user_id=member, team_id=team, models=[model], allowed_routes=MANAGEMENT_ROUTES)
|
|
|
|
|
|
def chat(candidate: Gateway, model: str, key: str, marker: str, *, stream: bool = False) -> httpx.Response:
|
|
return candidate.request(
|
|
"POST",
|
|
"/v1/chat/completions",
|
|
{"model": model, "messages": [{"role": "user", "content": marker}], "stream": stream},
|
|
key=key,
|
|
)
|
|
|
|
|
|
def upstream_observations(gateway: Gateway) -> tuple[dict[str, JsonValue], ...]:
|
|
with httpx.Client(timeout=5, trust_env=False) as client:
|
|
observed: Final = object_value(client.get(f"{gateway.upstream_url}/__observations").json())
|
|
requests: Final = observed["requests"]
|
|
assert isinstance(requests, list)
|
|
return tuple(object_value(entry) for entry in requests)
|
|
|
|
|
|
def upstream_hits(gateway: Gateway, marker: str) -> int:
|
|
return sum(1 for entry in upstream_observations(gateway) if marker in json.dumps(entry.get("body")))
|