litellm/tests/llm_translation/conftest.py
Cursor Agent b637d9f64a
test(vcr): classify cache verdicts, detect live calls, surface cost leaks
Convert the per-test VCR verdict line from a single 'NOOP / HIT / MISS /
PARTIAL' tag into a classified outcome that distinguishes the cases that
silently bill the live API on every CI run from the ones that don't:

  HIT                         pure replay
  PARTIAL                     mixed replay + new recordings
  MISS:RECORDED               new cassette saved to Redis (cached next run)
  MISS:OVERFLOW               cassette > MAX_EPISODES_PER_CASSETTE; persister
                              refused to save; re-bills every run
  MISS:NOT_PERSISTED          test failed; save_cassette skipped; re-bills
  NOOP                        VCR-marked but no HTTP traffic (mocked elsewhere)
  UNMARKED:LIVE_CALL          test bypassed VCR AND opened a TCP connection
                              to a known LLM provider host -> wasted spend
  UNMARKED:NO_TRAFFIC         test bypassed VCR but didn't call out

The UNMARKED:LIVE_CALL signal is what converts 'this test probably hits
live' into 'this test connected to api.openai.com'. We install a
socket.connect / socket.create_connection wrapper for the duration of
each non-VCR-marked test and record any outbound TCP to a known LLM
provider hostname. The probe sits below the httpx layer so vcrpy and
respx (which both patch above the socket) are unaffected.

Replace the file-level _RESPX_CONFLICTING_FILES blacklists in the
llm_translation and local_testing conftests with per-item respx
detection in apply_vcr_auto_marker_to_items. A test now skips VCR when
it actually carries @pytest.mark.respx or has respx_mock in its fixture
chain - not just because some other test in the same file imports
MockRouter. Items skipped by skip_files are split into respx_conflict
(real conflict, the module wires up respx) vs file_opt_out (dead skip-
list entry whose module never touches respx) so the session summary
makes pruning obvious.

Stabilize the AWS SigV4 fingerprint: the Authorization header on
Bedrock requests rotates its Credential date and Signature on every
call, which previously pushed every Bedrock test past the 50-episode
overflow threshold. Extract the access-key id only
('aws-sigv4:AKIA...') so two requests with the same identity match.

Always emit verdict logging when VCR is active (set
LITELLM_VCR_VERBOSE=0 to opt back into the legacy quiet mode). Add a
session-end classification summary that lists overflow tests, unmarked
live-call tests, and the skip-reason breakdown.

Wire the live-call probe + summary hook into every test directory that
already uses the Redis-backed VCR cache (audio_tests, guardrails_tests,
image_gen_tests, litellm_utils_tests, llm_responses_api_testing,
llm_translation, local_testing, logging_callback_tests, ocr_tests,
pass_through_unit_tests, router_unit_tests, search_tests,
unified_google_tests).

Add tests/llm_translation/test_vcr_classification.py covering the
verdict classifier, skip-reason tagging, AWS SigV4 fingerprint stability,
live-host classification, and session summary rendering.

Co-authored-by: Mateo Wang <mateo-berri@users.noreply.github.com>
2026-05-13 00:31:47 +00:00

179 lines
5.4 KiB
Python

# conftest.py
#
# xdist-compatible test isolation for llm_translation tests.
# Mirrors the pattern in tests/local_testing/conftest.py:
# - Function-scoped fixture resets litellm globals to true defaults
# - Module-scoped reload only in single-process mode
import asyncio
import importlib
import os
import sys
import pytest
sys.path.insert(
0, os.path.abspath("../..")
) # Adds the parent directory to the system path
import litellm # noqa: E402
from tests._vcr_conftest_common import ( # noqa: E402
VerboseReporterState,
apply_vcr_auto_marker_to_items,
emit_cassette_cache_session_banner,
emit_vcr_classification_summary,
install_live_call_probe,
record_vcr_outcome,
register_persister_if_enabled,
vcr_config_dict,
)
# Per-item respx detection (``apply_vcr_auto_marker_to_items``) handles
# the vast majority of respx-vs-vcrpy conflicts automatically. The only
# entry below is the persister's own unit-test file, which exercises
# ``save_cassette`` / ``load_cassette`` against fakeredis and must not
# itself run under a live cassette context.
_VCR_AUTO_MARKER_SKIP_FILES = frozenset({"test_vcr_redis_persister.py"})
# Tests that observe live cross-call provider state (e.g. prompt-cache
# warm-up between two consecutive calls); replay can't reproduce that state.
_VCR_INCOMPATIBLE_NODEID_SUFFIXES = (
"::test_prompt_caching",
"TestBedrockInvokeNovaJson::test_json_response_pydantic_obj",
"::test_bedrock_converse__streaming_passthrough",
)
_verbose_state = VerboseReporterState()
@pytest.fixture(scope="module")
def vcr_config():
return vcr_config_dict()
def pytest_recording_configure(config, vcr):
register_persister_if_enabled(vcr)
@pytest.hookimpl(hookwrapper=True)
def pytest_runtest_makereport(item, call):
outcome = yield
rep = outcome.get_result()
setattr(item, f"rep_{rep.when}", rep)
@pytest.fixture(autouse=True)
def _vcr_outcome_gate(request, vcr):
install_live_call_probe(request, vcr)
yield
record_vcr_outcome(request, vcr)
def pytest_configure(config):
_verbose_state.remember_pluginmanager(config)
def pytest_runtest_logreport(report):
_verbose_state.maybe_emit_verdict(report)
def pytest_terminal_summary(terminalreporter, exitstatus, config):
emit_cassette_cache_session_banner(terminalreporter)
emit_vcr_classification_summary(terminalreporter)
# ---------------------------------------------------------------------------
# Capture TRUE defaults at conftest import time (before test modules pollute).
# ---------------------------------------------------------------------------
_SCALAR_DEFAULTS = {
"num_retries": getattr(litellm, "num_retries", None),
"set_verbose": getattr(litellm, "set_verbose", False),
"cache": getattr(litellm, "cache", None),
"allowed_fails": getattr(litellm, "allowed_fails", 3),
"disable_aiohttp_transport": getattr(litellm, "disable_aiohttp_transport", False),
"force_ipv4": getattr(litellm, "force_ipv4", False),
"drop_params": getattr(litellm, "drop_params", None),
"modify_params": getattr(litellm, "modify_params", False),
"api_base": getattr(litellm, "api_base", None),
"api_key": getattr(litellm, "api_key", None),
"cohere_key": getattr(litellm, "cohere_key", None),
}
@pytest.fixture(scope="session")
def event_loop():
try:
loop = asyncio.get_running_loop()
except RuntimeError:
loop = asyncio.new_event_loop()
yield loop
loop.close()
@pytest.fixture(scope="function", autouse=True)
def setup_and_teardown(event_loop): # Add event_loop as a dependency
sys.path.insert(0, os.path.abspath("../.."))
import litellm
# ---- Save current state (for teardown restore) ----
original_state = {}
for attr in (
"callbacks",
"success_callback",
"failure_callback",
"_async_success_callback",
"_async_failure_callback",
):
if hasattr(litellm, attr):
val = getattr(litellm, attr)
original_state[attr] = val.copy() if val else []
for attr in _SCALAR_DEFAULTS:
if hasattr(litellm, attr):
original_state[attr] = getattr(litellm, attr)
# ---- Reset to true defaults before the test ----
from litellm.litellm_core_utils.logging_worker import GLOBAL_LOGGING_WORKER
asyncio.run(GLOBAL_LOGGING_WORKER.clear_queue())
importlib.reload(litellm)
# Set the event loop from the fixture
asyncio.set_event_loop(event_loop)
yield
# ---- Teardown ----
for attr, original_value in original_state.items():
if hasattr(litellm, attr):
setattr(litellm, attr, original_value)
# Clean up any pending tasks
pending = asyncio.all_tasks(event_loop)
for task in pending:
task.cancel()
# Run the event loop until all tasks are cancelled
if pending:
event_loop.run_until_complete(asyncio.gather(*pending, return_exceptions=True))
def pytest_collection_modifyitems(config, items):
apply_vcr_auto_marker_to_items(
items,
skip_files=_VCR_AUTO_MARKER_SKIP_FILES,
skip_nodeid_suffixes=_VCR_INCOMPATIBLE_NODEID_SUFFIXES,
)
custom_logger_tests = [
item for item in items if "custom_logger" in item.parent.name
]
other_tests = [item for item in items if "custom_logger" not in item.parent.name]
custom_logger_tests.sort(key=lambda x: x.name)
other_tests.sort(key=lambda x: x.name)
items[:] = custom_logger_tests + other_tests