test(proxy): expect litellm_error_type in Messages integration tests and type middleware tests

tests/integration/authorization/_hidden_alias_budget.py

@@ -151,9 +151,6 @@ def error_type(response: httpx.Response) -> str:
 
 
 def budget_refusal_error_type(route: str) -> str:
-    """The ``error.type`` a budget refusal carries on a route label from ``ROUTES``: the
-    Anthropic envelope on Messages (its budget refusal is raised in
-    ``user_api_key_auth``), the OpenAI ``budget_exceeded`` type on the other routes."""
     return "invalid_request_error" if route == "messages" else "budget_exceeded"
 
 

tests/integration/authorization/test_unknown_key_generic_401.py

@@ -55,6 +55,7 @@ class _AnthropicErrorDetail(BaseModel):
 
     type: str
     message: str
+    litellm_error_type: str
 
 
 class _AnthropicErrorResponse(BaseModel):
@@ -67,7 +68,9 @@ class _AnthropicErrorResponse(BaseModel):
 GENERIC_ERROR: Final = _ProxyError(message=GENERIC_MESSAGE, type="token_not_found_in_db", param="key", code="401")
 MESSAGES_PATH: Final = "/v1/messages"
 ANTHROPIC_ENVELOPE_PATHS: Final = frozenset({MESSAGES_PATH, "/anthropic/v1/messages"})
-ANTHROPIC_GENERIC_ERROR: Final = _AnthropicErrorDetail(type="authentication_error", message=GENERIC_MESSAGE)
+ANTHROPIC_GENERIC_ERROR: Final = _AnthropicErrorDetail(
+    type="authentication_error", message=GENERIC_MESSAGE, litellm_error_type=GENERIC_ERROR.type
+)
 
 
 @dataclass(frozen=True, slots=True)
@@ -146,7 +149,6 @@ def _assert_messages_refusal(response: httpx.Response, *secrets: str) -> None:
 
 
 def _assert_refusal(path: str, response: httpx.Response, *secrets: str) -> None:
-    """The Messages and Anthropic pass-through routes answer the Anthropic envelope, every other route the OpenAI one."""
     if path in ANTHROPIC_ENVELOPE_PATHS:
         _assert_messages_refusal(response, *secrets)
     else:
@@ -398,8 +400,6 @@ async def _send_async(client: httpx.AsyncClient, route: _Route, key: str) -> htt
 
 
 async def _burst(owned: OwnedProxy, keys: tuple[str, ...]) -> tuple[tuple[_Route, httpx.Response], ...]:
-    """Each key's response paired with the route it was sent on, so the caller asserts the
-    envelope the route actually answered with."""
     routes: Final = tuple(ROUTES[BURST_ROUTES[index % len(BURST_ROUTES)]] for index in range(len(keys)))
     async with httpx.AsyncClient(base_url=str(owned.gateway.client.base_url), timeout=60, trust_env=False) as client:
         responses: Final = await asyncio.gather(

tests/unit/proxy/middleware/test_request_body_decompression_middleware.py

@@ -683,7 +683,7 @@ async def test_non_http_scopes_reach_the_app_untouched(scope: Scope):
 )
 def test_a_messages_body_the_proxy_cannot_take_gets_the_anthropic_envelope(
     body: bytes, max_mb: float, status_code: int, anthropic_type: str, message: str | None
-):
+) -> None:
     response: Final = _post(
         _client(max_decompressed_body_size_mb=max_mb),
         ZSTD_FRAME_MAGIC + body if message is None else body,

tests/unit/proxy/middleware/test_request_size_limit_middleware.py

@@ -1,7 +1,9 @@
+from typing import Final
+
 import pytest
 from starlette.responses import JSONResponse
 from starlette.testclient import TestClient
-from starlette.types import Message
+from starlette.types import Message, Receive, Scope, Send
 
 from litellm.proxy.middleware.request_size_limit_middleware import (
     RequestSizeLimitMiddleware,
@@ -135,15 +137,15 @@ async def test_request_size_limit_middleware_rejects_streamed_body_without_conte
     assert received_body_bytes == 1024 * 1024
 
 
-def test_an_oversized_messages_request_gets_the_anthropic_request_too_large_error():
-    async def app(scope, receive, send):
+def test_an_oversized_messages_request_gets_the_anthropic_request_too_large_error() -> None:
+    async def app(scope: Scope, receive: Receive, send: Send) -> None:
         await JSONResponse({"ok": True})(scope, receive, send)
 
-    client = TestClient(
+    client: Final = TestClient(
         RequestSizeLimitMiddleware(app, get_max_request_size_mb=lambda: 1, is_request_size_limit_enabled=lambda: True)
     )
 
-    response = client.post(
+    response: Final = client.post(
         "/v1/messages", content=b"x" * (1024 * 1024 + 1), headers={"content-type": "application/json"}
     )