df1074b4e5
Two open atoms of #83390 (DeepSeek "This response_format type is unavailable now"): * `_call_fallback_candidate_sync/_async` only special-cased auth errors, so when the primary aux provider failed (timeout, rate limit, payment) and the fallback landed on a provider that rejects `json_schema`, the 400 re-raised and the whole task died — the primary-path rung from #89589 never applied there. Both fallback paths now retry once without `response_format`. * Every structured aux call (titles, kanban decomposer, goal judge, plugin structured calls) paid a guaranteed-fail request on providers that lack `json_schema` before the retry. A provider profile can now declare `unsupported_response_formats` (DeepSeek: json_schema, per https://api-docs.deepseek.com/guides/json_mode) and the recovery ladder remembers any route that rejected a type once (host:port scoped), so `_build_call_kwargs` — shared by the primary and fallback paths — omits the field before the first request. Dropping rather than downgrading to json_object matches the end state the retry already produced; json_object needs a JSON-mentioning prompt and some relays return empty content under it. New logic lives in agent/auxiliary_structured_output.py; the facade only gains the fallback rung next to the predicate it uses. tests/agent/conftest.py resets the process-level memo per test. Fixes #83390, #105191. Closes duplicates #84976, #88830, #102849, #113064. Co-authored-by: Legion-is-life <Legion-is-life@users.noreply.github.com>
48 lines
2.1 KiB
Python
48 lines
2.1 KiB
Python
"""Fast-path fixtures shared across tests/agent/.
|
|
|
|
Many tests in this directory exercise the retry/backoff paths in the
|
|
agent loop. Production code uses ``jittered_backoff(base_delay=5.0)``
|
|
with a ``while time.time() < sleep_end`` loop — a single retry test
|
|
spends 5+ seconds of real wall-clock time on backoff waits.
|
|
|
|
Mocking ``jittered_backoff`` to return 0.0 collapses the while-loop
|
|
to a no-op (``time.time() < time.time() + 0`` is false immediately),
|
|
which handles the most common case without touching ``time.sleep``.
|
|
|
|
We deliberately DO NOT mock ``time.sleep`` here — some tests
|
|
(test_interrupt_propagation, test_primary_runtime_restore, etc.) use
|
|
the real ``time.sleep`` for threading coordination or assert that it
|
|
was called with specific values. Tests that want to additionally
|
|
fast-path direct ``time.sleep(N)`` calls in production code should
|
|
monkeypatch ``run_agent.time.sleep`` locally (see
|
|
``test_anthropic_error_handling.py`` for the pattern).
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import pytest
|
|
|
|
|
|
@pytest.fixture(autouse=True)
|
|
def _fresh_structured_output_memo(monkeypatch):
|
|
"""The aux client remembers routes that rejected ``response_format`` for the whole process;
|
|
a rejection recorded by one test must not strip the field from the next test's request."""
|
|
from agent import auxiliary_structured_output
|
|
monkeypatch.setattr(auxiliary_structured_output, "_REJECTED_ROUTES", set())
|
|
|
|
|
|
@pytest.fixture(autouse=True)
|
|
def _fast_retry_backoff(request, monkeypatch):
|
|
"""Short-circuit retry backoff for all tests in this directory.
|
|
|
|
Tests that assert on the real backoff value opt out with
|
|
``@pytest.mark.real_retry_backoff``.
|
|
"""
|
|
if request.node.get_closest_marker("real_retry_backoff"):
|
|
return
|
|
# The agent.turn_* retry paths import ``jittered_backoff`` lazily from
|
|
# ``agent.retry_utils``; patch it there so rate-limit / invalid-response /
|
|
# server-error retries don't burn real wall-clock seconds.
|
|
from agent import retry_utils as _retry_utils
|
|
monkeypatch.setattr(_retry_utils, "jittered_backoff", lambda *a, **k: 0.0)
|