populate service_type and purpose on pre-completion availability records

Author: randy-concepcion

src/mlpa/core/auth/authorize.py

@@ -129,9 +129,7 @@ async def authorize_chat_request(
             AvailabilityReason.INVALID_SERVICE_TYPE_FOR_MODEL,
             model=chat_request.model,
             service_type=service_type.value,
-            # `purpose` is not yet resolved here
-            # This serves as a placeholder until a validated purpose is available
-            purpose="",
+            purpose=(purpose or "").strip(),
         )
         raise HTTPException(
             status_code=400,

src/mlpa/core/middleware/request_size.py

@@ -21,14 +21,15 @@ async def check_request_size_middleware(request: Request, call_next):
                     logger.warning(
                         f"Request size {size} bytes exceeds maximum {env.MAX_REQUEST_SIZE_BYTES} bytes"
                     )
-                    # Placeholders for `model`, `service_type`, and `purpose`
-                    # are used since values are not set until the body is parsed
-                    # and auth has run.
+                    # `model` is in the request body, which we don't read here.
+                    # We reject on the Content-Length header without parsing it.
                     record_chat_availability_for(
                         AvailabilityReason.PAYLOAD_TOO_LARGE,
                         model="",
-                        service_type="",
-                        purpose="",
+                        service_type=(
+                            request.headers.get("service-type") or ""
+                        ).strip(),
+                        purpose=(request.headers.get("purpose") or "").strip(),
                     )
                     return JSONResponse(
                         status_code=413,

src/tests/integration/test_request_size_limit.py

@@ -67,8 +67,8 @@ def test_request_size_over_limit(mocked_client_integration, metrics_spy):
             outcome="excluded",
             reason="payload_too_large",
             model="",
-            service_type="",
-            purpose="",
+            service_type="ai",
+            purpose="chat",
         )
         == 1
     )

src/tests/unit/test_precompletion_availability.py

@@ -119,6 +119,7 @@ async def test_wrapper_invalid_service_type_records_excluded(metrics_spy):
             AvailabilityReason.INVALID_SERVICE_TYPE_FOR_MODEL,
             model="exa",
             service_type="ai",
+            purpose="chat",
         )
         == 1
     )