Skip to content

:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.

2.5 KB
Raw Blame
from unittest.mock import MagicMock, patch

import pytest

from llm_gateway import flags


@pytest.fixture(autouse=True)
def clear_flag_caches() -> None:
    flags._flag_cache.clear()
    flags._flag_unavailable_cache.clear()


def flag_snapshot(values: dict[str, bool]) -> MagicMock:
    snapshot = MagicMock()
    snapshot.keys = list(values)
    snapshot.is_enabled.side_effect = values.__getitem__
    return snapshot


@pytest.mark.parametrize("enabled", [True, False])
async def test_evaluate_flag_caches_definitive_answers(enabled: bool) -> None:
    # Both definitive answers must cache — False (flag disabled) is the normal rollout state, and
    # caching only True would re-hit /flags on every GLM request while the ramp is off.
    client = MagicMock()
    client.evaluate_flags.return_value = flag_snapshot({"some-flag": enabled})
    with patch("llm_gateway.flags._get_client", return_value=client):
        assert await flags.evaluate_flag("some-flag", "user-a") is enabled
        assert await flags.evaluate_flag("some-flag", "user-a") is enabled
    # One roundtrip per (flag, user) within the TTL — flag checks sit on the GLM hot path.
    assert client.evaluate_flags.call_count == 1


async def test_evaluate_flags_batches_uncached_keys() -> None:
    client = MagicMock()
    client.evaluate_flags.return_value = flag_snapshot({"flag-a": True, "flag-b": False})

    with patch("llm_gateway.flags._get_client", return_value=client):
        assert await flags.evaluate_flags(["flag-a", "flag-b"], "user-a") == {
            "flag-a": True,
            "flag-b": False,
        }

    client.evaluate_flags.assert_called_once_with("user-a", flag_keys=["flag-a", "flag-b"])


@pytest.mark.parametrize("failure_mode", ["sdk_error", "sdk_none"])
async def test_evaluate_flag_backs_off_globally_when_unavailable(failure_mode: str) -> None:
    # An outage (exception) and an SDK None result (flag missing / evaluation unavailable) hit the
    # backoff independently; neither must stack one blocking roundtrip per new user — after the
    # first miss everyone gets the fallback answer until the backoff expires.
    client = MagicMock()
    if failure_mode == "sdk_error":
        client.evaluate_flags.side_effect = RuntimeError("posthog down")
    else:
        client.evaluate_flags.return_value = flag_snapshot({})
    with patch("llm_gateway.flags._get_client", return_value=client):
        assert await flags.evaluate_flag("some-flag", "user-a") is None
        assert await flags.evaluate_flag("some-flag", "user-b") is None
    assert client.evaluate_flags.call_count == 1