wehub-resource-sync
129 lines
3.8 KiB
Python
129 lines
3.8 KiB
Python
"""Retry-After header threading: adapter 429/5xx errors carry parsed seconds.
|
|
|
|
Fully offline via ``httpx.MockTransport`` — same pattern as the golden
|
|
harness. Synthetic credentials and models only.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
from typing import Any
|
|
|
|
import httpx
|
|
import pytest
|
|
|
|
from opensquilla.provider.selector import build_provider
|
|
from opensquilla.provider.types import ChatConfig, ErrorEvent, Message
|
|
|
|
FAKE_API_KEY = "sk-test-000"
|
|
|
|
|
|
def _patch_transport(
|
|
monkeypatch: pytest.MonkeyPatch, response_factory: Any
|
|
) -> None:
|
|
transport = httpx.MockTransport(lambda request: response_factory())
|
|
real_async_client = httpx.AsyncClient
|
|
|
|
def patched_async_client(*args: Any, **kwargs: Any) -> httpx.AsyncClient:
|
|
kwargs["transport"] = transport
|
|
return real_async_client(*args, **kwargs)
|
|
|
|
monkeypatch.setattr(httpx, "AsyncClient", patched_async_client)
|
|
|
|
|
|
async def _chat_errors(provider: Any) -> list[ErrorEvent]:
|
|
messages = [Message(role="user", content="ping")]
|
|
events = [
|
|
event
|
|
async for event in provider.chat(messages, tools=None, config=ChatConfig())
|
|
]
|
|
return [event for event in events if isinstance(event, ErrorEvent)]
|
|
|
|
|
|
async def test_openai_compat_429_carries_retry_after_seconds(
|
|
monkeypatch: pytest.MonkeyPatch,
|
|
) -> None:
|
|
_patch_transport(
|
|
monkeypatch,
|
|
lambda: httpx.Response(
|
|
429,
|
|
headers={"Retry-After": "7"},
|
|
json={"error": {"message": "synthetic rate limit"}},
|
|
),
|
|
)
|
|
provider = build_provider("openai", "test-chat-model", api_key=FAKE_API_KEY)
|
|
errors = await _chat_errors(provider)
|
|
assert errors
|
|
assert errors[0].code == "429"
|
|
assert errors[0].retry_after_s == 7.0
|
|
|
|
|
|
async def test_openai_compat_503_carries_retry_after_seconds(
|
|
monkeypatch: pytest.MonkeyPatch,
|
|
) -> None:
|
|
_patch_transport(
|
|
monkeypatch,
|
|
lambda: httpx.Response(
|
|
503,
|
|
headers={"Retry-After": "12"},
|
|
json={"error": {"message": "synthetic overload"}},
|
|
),
|
|
)
|
|
provider = build_provider("openai", "test-chat-model", api_key=FAKE_API_KEY)
|
|
errors = await _chat_errors(provider)
|
|
assert errors
|
|
assert errors[0].code == "503"
|
|
assert errors[0].retry_after_s == 12.0
|
|
|
|
|
|
async def test_openai_compat_400_never_carries_retry_after(
|
|
monkeypatch: pytest.MonkeyPatch,
|
|
) -> None:
|
|
_patch_transport(
|
|
monkeypatch,
|
|
lambda: httpx.Response(
|
|
400,
|
|
headers={"Retry-After": "7"},
|
|
json={"error": {"message": "synthetic bad request"}},
|
|
),
|
|
)
|
|
provider = build_provider("openai", "test-chat-model", api_key=FAKE_API_KEY)
|
|
errors = await _chat_errors(provider)
|
|
assert errors
|
|
assert errors[0].code == "400"
|
|
assert errors[0].retry_after_s is None
|
|
|
|
|
|
async def test_anthropic_429_carries_retry_after_seconds(
|
|
monkeypatch: pytest.MonkeyPatch,
|
|
) -> None:
|
|
_patch_transport(
|
|
monkeypatch,
|
|
lambda: httpx.Response(
|
|
429,
|
|
headers={"Retry-After": "9"},
|
|
json={"error": {"type": "rate_limit_error", "message": "synthetic"}},
|
|
),
|
|
)
|
|
provider = build_provider("anthropic", "claude-sonnet-4-6", api_key=FAKE_API_KEY)
|
|
errors = await _chat_errors(provider)
|
|
assert errors
|
|
assert errors[0].code == "429"
|
|
assert errors[0].retry_after_s == 9.0
|
|
|
|
|
|
async def test_missing_header_leaves_retry_after_none(
|
|
monkeypatch: pytest.MonkeyPatch,
|
|
) -> None:
|
|
_patch_transport(
|
|
monkeypatch,
|
|
lambda: httpx.Response(
|
|
429,
|
|
json={"error": {"message": "synthetic rate limit"}},
|
|
),
|
|
)
|
|
provider = build_provider("openai", "test-chat-model", api_key=FAKE_API_KEY)
|
|
errors = await _chat_errors(provider)
|
|
assert errors
|
|
assert errors[0].code == "429"
|
|
assert errors[0].retry_after_s is None
|