test(router): type the retry-cap tests this PR adds or touches

tests/router_unit_tests/test_router_helper_utils.py

@@ -11,7 +11,7 @@ import litellm
 from unittest.mock import patch, MagicMock, AsyncMock
 from create_mock_standard_logging_payload import create_standard_logging_payload
 from litellm.types.utils import StandardLoggingPayload
-from litellm.types.router import Deployment, LiteLLM_Params, ModelInfo
+from litellm.types.router import Deployment, DeploymentTypedDict, LiteLLM_Params, ModelInfo
 from litellm.constants import DEFAULT_AUTO_ROUTER_MAX_INPUT_CHARS
 
 
@@ -630,7 +630,7 @@ def test_deployment_callback_respects_cooldown_time(model_list):
 
 
 @pytest.mark.parametrize("metadata_key", ["metadata", "litellm_metadata"])
-def test_log_retry(model_list, metadata_key):
+def test_log_retry(model_list: list[DeploymentTypedDict], metadata_key: str) -> None:
     """log_retry appends one flat record per failed attempt, copies neither the request kwargs nor the
     request metadata into it, counts every failed attempt of the request independently of the
     per-hop attempted_retries, and never trusts a negative count planted before the first failure"""

tests/test_litellm/proxy/test_litellm_pre_call_utils.py

@@ -7353,7 +7353,7 @@ _PLANTED_STAMPS = {
 
 
 @pytest.mark.asyncio
-async def test_add_litellm_data_to_request_strips_router_reserved_stamps_from_both_buckets():
+async def test_add_litellm_data_to_request_strips_router_reserved_stamps_from_both_buckets() -> None:
     """attempted_fallbacks and original_model_group are router-written facts the spend row
     reads back; a client planting them in either bucket is dropped at the boundary so the
     router never sees a reserved key it did not write."""
@@ -7384,7 +7384,7 @@ async def test_add_litellm_data_to_request_strips_router_reserved_stamps_from_bo
 
 
 @pytest.mark.asyncio
-async def test_add_litellm_data_to_request_strips_router_reserved_stamps_from_json_string_litellm_metadata():
+async def test_add_litellm_data_to_request_strips_router_reserved_stamps_from_json_string_litellm_metadata() -> None:
     from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request
 
     data = {
@@ -7410,7 +7410,7 @@ async def test_add_litellm_data_to_request_strips_router_reserved_stamps_from_js
 
 
 @pytest.mark.asyncio
-async def test_add_litellm_data_to_request_strips_router_reserved_stamps_despite_pricing_override_opt_in():
+async def test_add_litellm_data_to_request_strips_router_reserved_stamps_despite_pricing_override_opt_in() -> None:
     """The pricing strip is gated on allow_client_pricing_override; the reserved-stamp strip
     is not, because no key or team setting makes a client-written fallback count valid."""
     from litellm.proxy.litellm_pre_call_utils import add_litellm_data_to_request

tests/test_litellm/test_router.py

@@ -11066,7 +11066,7 @@ async def test_num_retries_per_request_stops_retries_at_caps_above_four(monkeypa
     ]
 
 
-def _failing_group_with_healthy_fallback_router(num_retries):
+def _failing_group_with_healthy_fallback_router(num_retries: int) -> litellm.Router:
     return litellm.Router(
         model_list=[
             {
@@ -11094,8 +11094,8 @@ def _failing_group_with_healthy_fallback_router(num_retries):
     ids=["cap-spent-before-the-hop", "cap-not-reached-by-the-hop", "planted-negative-count-does-not-lift-the-cap"],
 )
 async def test_num_retries_per_request_counts_retries_across_fallback_hops(
-    monkeypatch, cap, planted_count, hop_refused
-):
+    monkeypatch: pytest.MonkeyPatch, cap: int, planted_count: int | None, hop_refused: bool
+) -> None:
     """num_retries_per_request caps the retries of one request, fallback hops included. Each hop starts a
     fresh per-hop attempted_retries at zero, so a cap read from that counter let every hop retry from zero
     and a request could spend far more retries than the cap allows. A caller who plants a negative count

tests/test_litellm/test_utils.py

@@ -4099,7 +4099,9 @@ def _capped_completion_kwargs(metadata_key: str, metadata: object) -> dict[str,
 
 @pytest.mark.parametrize("metadata_key", ["metadata", "litellm_metadata"])
 @pytest.mark.parametrize("cap, metadata, refused", _RETRY_CAP_CASES)
-def test_num_retries_per_request_reads_request_retry_count_sync(monkeypatch, metadata_key, cap, metadata, refused):
+def test_num_retries_per_request_reads_request_retry_count_sync(
+    monkeypatch: pytest.MonkeyPatch, metadata_key: str, cap: int, metadata: object, refused: bool
+) -> None:
     monkeypatch.setattr(litellm, "num_retries_per_request", cap)
     kwargs: Final = _capped_completion_kwargs(metadata_key, metadata)
     if refused:
@@ -4112,7 +4114,9 @@ def test_num_retries_per_request_reads_request_retry_count_sync(monkeypatch, met
 @pytest.mark.asyncio
 @pytest.mark.parametrize("metadata_key", ["metadata", "litellm_metadata"])
 @pytest.mark.parametrize("cap, metadata, refused", _RETRY_CAP_CASES)
-async def test_num_retries_per_request_reads_request_retry_count_async(monkeypatch, metadata_key, cap, metadata, refused):
+async def test_num_retries_per_request_reads_request_retry_count_async(
+    monkeypatch: pytest.MonkeyPatch, metadata_key: str, cap: int, metadata: object, refused: bool
+) -> None:
     monkeypatch.setattr(litellm, "num_retries_per_request", cap)
     kwargs: Final = _capped_completion_kwargs(metadata_key, metadata)
     if refused: