Fix sensitive_data_routing event_hook typing and tighten coverage

Normalize event_hook to a GuardrailEventHooks before passing it to the
CustomGuardrail base so mypy is satisfied with the declared type. Drop the
unused raw-config fallback in the initializer and read params directly from
litellm_params, and add tests covering the config model exposure, non-dict
messages, and the missing guardrail_name validation

https://claude.ai/code/session_01R2hGPr2jc5kxnSYLRgqfMs

litellm/proxy/guardrails/guardrail_hooks/sensitive_data_routing/__init__.py

@@ -1,6 +1,6 @@
 """Sensitive Data Routing guardrail: reroutes requests with sensitive data to an on-premise model."""
 
-from typing import TYPE_CHECKING, Any, List
+from typing import TYPE_CHECKING, List
 
 from litellm.types.guardrails import SupportedGuardrailIntegrations
 
@@ -10,21 +10,6 @@ if TYPE_CHECKING:
     from litellm.types.guardrails import Guardrail, LitellmParams
 
 
-def _get_param(
-    litellm_params: "LitellmParams",
-    guardrail: "Guardrail",
-    key: str,
-    default: Any = None,
-) -> Any:
-    value = getattr(litellm_params, key, None)
-    if value is not None:
-        return value
-    raw = guardrail.get("litellm_params")
-    if isinstance(raw, dict) and key in raw:
-        return raw[key]
-    return default
-
-
 def initialize_guardrail(
     litellm_params: "LitellmParams",
     guardrail: "Guardrail",
@@ -35,7 +20,7 @@ def initialize_guardrail(
     if not guardrail_name:
         raise ValueError("sensitive_data_routing guardrail requires a guardrail_name")
 
-    on_premise_model = _get_param(litellm_params, guardrail, "on_premise_model")
+    on_premise_model = getattr(litellm_params, "on_premise_model", None)
     if not on_premise_model:
         raise ValueError(
             "sensitive_data_routing guardrail requires 'on_premise_model' (the model_list "
@@ -45,17 +30,13 @@ def initialize_guardrail(
     instance = SensitiveDataRoutingGuardrail(
         guardrail_name=guardrail_name,
         on_premise_model=on_premise_model,
-        prebuilt_patterns=_get_param(litellm_params, guardrail, "prebuilt_patterns"),
-        regex_patterns=_get_param(litellm_params, guardrail, "regex_patterns"),
-        keywords=_get_param(litellm_params, guardrail, "keywords"),
-        sticky_session=bool(
-            _get_param(litellm_params, guardrail, "sticky_session", True)
-        ),
-        session_ttl_seconds=int(
-            _get_param(litellm_params, guardrail, "session_ttl_seconds", 14400)
-        ),
-        event_hook=_get_param(litellm_params, guardrail, "mode"),
-        default_on=bool(_get_param(litellm_params, guardrail, "default_on", False)),
+        prebuilt_patterns=getattr(litellm_params, "prebuilt_patterns", None),
+        regex_patterns=getattr(litellm_params, "regex_patterns", None),
+        keywords=getattr(litellm_params, "keywords", None),
+        sticky_session=bool(getattr(litellm_params, "sticky_session", True)),
+        session_ttl_seconds=int(getattr(litellm_params, "session_ttl_seconds", 14400)),
+        event_hook=getattr(litellm_params, "mode", None),
+        default_on=bool(getattr(litellm_params, "default_on", False)),
     )
     litellm.logging_callback_manager.add_litellm_callback(instance)
     return instance

litellm/proxy/guardrails/guardrail_hooks/sensitive_data_routing/sensitive_data_routing.py

@@ -49,7 +49,11 @@ class SensitiveDataRoutingGuardrail(CustomGuardrail):
         super().__init__(
             guardrail_name=guardrail_name or "sensitive_data_routing",
             supported_event_hooks=[GuardrailEventHooks.pre_call],
-            event_hook=event_hook or GuardrailEventHooks.pre_call,
+            event_hook=(
+                GuardrailEventHooks(event_hook)
+                if event_hook is not None
+                else GuardrailEventHooks.pre_call
+            ),
             default_on=default_on,
             **kwargs,
         )

tests/test_litellm/proxy/guardrails/guardrail_hooks/test_sensitive_data_routing.py

@@ -11,6 +11,9 @@ from litellm.proxy.guardrails.guardrail_hooks.sensitive_data_routing import (
     initialize_guardrail,
 )
 from litellm.types.guardrails import GuardrailEventHooks, LitellmParams
+from litellm.types.proxy.guardrails.guardrail_hooks.sensitive_data_routing import (
+    SensitiveDataRoutingConfigModel,
+)
 
 ON_PREM = "on-prem-model"
 USER_KEY = UserAPIKeyAuth()
@@ -87,6 +90,17 @@ async def test_keyword_match_is_case_insensitive():
     assert result is not None and result["model"] == ON_PREM
 
 
+@pytest.mark.asyncio
+async def test_non_dict_messages_are_skipped():
+    g = _make_guardrail()
+    data = {
+        "model": "gpt-4o",
+        "messages": ["not-a-dict", {"role": "user", "content": "ssn 123-45-6789"}],
+    }
+    result = await _hook(g, data)
+    assert result is not None and result["model"] == ON_PREM
+
+
 @pytest.mark.asyncio
 async def test_detects_text_in_content_parts_list():
     g = _make_guardrail()
@@ -180,6 +194,17 @@ def test_only_supports_pre_call_event_hook():
         _make_guardrail(event_hook=GuardrailEventHooks.post_call)
 
 
+def test_initializer_requires_guardrail_name():
+    params = LitellmParams(
+        guardrail="sensitive_data_routing", mode="pre_call", on_premise_model=ON_PREM
+    )
+    with pytest.raises(ValueError):
+        initialize_guardrail(
+            litellm_params=params,
+            guardrail={"guardrail_name": "", "litellm_params": {}},
+        )
+
+
 def test_initializer_requires_on_premise_model():
     params = LitellmParams(guardrail="sensitive_data_routing", mode="pre_call")
     with pytest.raises(ValueError):
@@ -189,6 +214,12 @@ def test_initializer_requires_on_premise_model():
         )
 
 
+def test_config_model_is_exposed_for_ui():
+    config_model = SensitiveDataRoutingGuardrail.get_config_model()
+    assert config_model is SensitiveDataRoutingConfigModel
+    assert config_model.ui_friendly_name() == "Sensitive Data Routing"
+
+
 def test_initializer_builds_guardrail_from_config():
     params = LitellmParams(
         guardrail="sensitive_data_routing",