fix(models): handle function_response in lite_llm payload check

src/google/adk/models/lite_llm.py

@@ -468,6 +468,8 @@ def _part_has_payload(part: types.Part) -> bool:
     return True
   if part.inline_data and part.inline_data.data:
     return True
+  if part.function_response:
+    return True
   if part.file_data and (part.file_data.file_uri or part.file_data.data):
     return True
   return False
@@ -1293,12 +1295,15 @@ def _model_response_to_chunk(
           if not func_name and not func_args:
             continue
 
-          yield FunctionChunk(
-              id=tool_call.id,
-              name=func_name,
-              args=func_args,
-              index=func_index,
-          ), finish_reason
+          yield (
+              FunctionChunk(
+                  id=tool_call.id,
+                  name=func_name,
+                  args=func_args,
+                  index=func_index,
+              ),
+              finish_reason,
+          )
 
     if finish_reason and not (message_content or tool_calls):
       yield None, finish_reason
@@ -1310,12 +1315,17 @@ def _model_response_to_chunk(
   # finish_reason set. But this is not the case we are observing from litellm.
   # So we are sending it as a separate chunk to be set on the llm_response.
   if response.get("usage", None):
-    yield UsageMetadataChunk(
-        prompt_tokens=response["usage"].get("prompt_tokens", 0),
-        completion_tokens=response["usage"].get("completion_tokens", 0),
-        total_tokens=response["usage"].get("total_tokens", 0),
-        cached_prompt_tokens=_extract_cached_prompt_tokens(response["usage"]),
-    ), None
+    yield (
+        UsageMetadataChunk(
+            prompt_tokens=response["usage"].get("prompt_tokens", 0),
+            completion_tokens=response["usage"].get("completion_tokens", 0),
+            total_tokens=response["usage"].get("total_tokens", 0),
+            cached_prompt_tokens=_extract_cached_prompt_tokens(
+                response["usage"]
+            ),
+        ),
+        None,
+    )
 
 
 def _model_response_to_generate_content_response(
@@ -1708,7 +1718,12 @@ def _warn_gemini_via_litellm(model_string: str) -> None:
   # Check if warning should be suppressed via environment variable
   if os.environ.get(
       "ADK_SUPPRESS_GEMINI_LITELLM_WARNINGS", ""
-  ).strip().lower() in ("1", "true", "yes", "on"):
+  ).strip().lower() in (
+      "1",
+      "true",
+      "yes",
+      "on",
+  ):
     return
 
   warnings.warn(
@@ -1812,9 +1827,12 @@ async def generate_content_async(
     logger.debug(_build_request_log(llm_request))
 
     effective_model = llm_request.model or self.model
-    messages, tools, response_format, generation_params = (
-        await _get_completion_inputs(llm_request, effective_model)
-    )
+    (
+        messages,
+        tools,
+        response_format,
+        generation_params,
+    ) = await _get_completion_inputs(llm_request, effective_model)
     normalized_messages = _normalize_ollama_chat_messages(
         messages,
         model=effective_model,

tests/unittests/models/test_litellm_function_response.py

@@ -0,0 +1,15 @@
+from google.adk.models.lite_llm import _part_has_payload
+from google.genai import types
+import pytest
+
+
+def test_part_has_payload_with_function_response():
+  part = types.Part.from_function_response(
+      name="test_fn", response={"result": "success"}
+  )
+  assert _part_has_payload(part) is True
+
+
+def test_part_has_payload_without_payload():
+  part = types.Part()
+  assert _part_has_payload(part) is False