refactor(sap): inject http client into deployment_url to drop internal patch

Replace the test's patch of litellm.module_level_client with dependency
injection via a new optional _http_client attribute, clearing the TQ008
test-quality violation. Production still falls back to the module-level
client when none is injected.

litellm/llms/sap/chat/transformation.py

@@ -18,6 +18,7 @@ if TYPE_CHECKING:
     import tiktoken
 
     from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj
+    from litellm.llms.custom_httpx.http_handler import HTTPHandler
 
     LiteLLMLoggingObj = _LiteLLMLoggingObj
 else:
@@ -136,6 +137,7 @@ class GenAIHubOrchestrationConfig(OpenAIGPTConfig):
         self.token_creator = None
         self._base_url = None
         self._resource_group = None
+        self._http_client: HTTPHandler | None = None
 
     def run_env_setup(self, service_key: str | None = None) -> None:
         try:
@@ -169,9 +171,7 @@ class GenAIHubOrchestrationConfig(OpenAIGPTConfig):
 
     @cached_property
     def deployment_url(self) -> str:
-        # Keep a short, tight client lifecycle here to avoid fd leaks
-        client: Final = litellm.module_level_client
-        # with httpx.Client(timeout=30) as client:
+        client: Final = self._http_client if self._http_client is not None else litellm.module_level_client
         deployments: Final = client.get(f"{self.base_url}/lm/deployments", headers=self.headers).json()
         valid: Final[list[tuple[str, str]]] = []
         for dep in deployments.get("resources", []):

tests/test_litellm/llms/sap/chat/test_sap_stream_chunk_validation.py

@@ -10,7 +10,7 @@ Regression tests for:
   configured resource group contains no orchestration deployment.
 """
 
-from unittest.mock import MagicMock, patch
+from unittest.mock import MagicMock
 
 import pytest
 
@@ -128,10 +128,10 @@ def test_deployment_url_raises_404_when_no_orchestration_deployment():
 
     mock_client = MagicMock()
     mock_client.get.return_value.json.return_value = {"resources": []}
+    config._http_client = mock_client
 
-    with patch("litellm.module_level_client", mock_client):
-        with pytest.raises(GenAIHubOrchestrationError) as exc_info:
-            _ = config.deployment_url
+    with pytest.raises(GenAIHubOrchestrationError) as exc_info:
+        _ = config.deployment_url
 
     assert exc_info.value.status_code == 404
     assert "fake-group" in exc_info.value.message