fix tool handling in OpenAIServingChat (#18996)

Signed-off-by: Xinyuan Tong <xinyuantong.cs@gmail.com>

python/sglang/srt/entrypoints/openai/serving_chat.py

@@ -329,12 +329,12 @@ class OpenAIServingChat(OpenAIServingBase):
             request.skip_special_tokens = False
             if not isinstance(request.tool_choice, str):
                 tools = [
-                    item.function.model_dump()
+                    item.model_dump()
                     for item in request.tools
                     if item.function.name == request.tool_choice.function.name
                 ]
             else:
-                tools = [item.function.model_dump() for item in request.tools]
+                tools = [item.model_dump() for item in request.tools]
             if self.tool_call_parser:
                 parser = FunctionCallParser(request.tools, self.tool_call_parser)
                 tool_call_constraint = parser.get_structure_constraint(
@@ -472,11 +472,10 @@ class OpenAIServingChat(OpenAIServingBase):
                     return_dict=False,
                 )
             except Exception as e:
-                # If the first attempt fails, try transforming the tools format
-                # This handles models like Mistral that have a different tools input format
-                # that is not compatible with OpenAI's apply_chat_template tool_call format
+                # If the first attempt fails, try with flat function-only format.
+                # Some templates (e.g. Mistral) expect tools without the OpenAI wrapper.
                 tools = (
-                    [t if "function" in t else {"function": t} for t in tools]
+                    [t["function"] if "function" in t else t for t in tools]
                     if tools
                     else None
                 )

test/registered/openai_server/basic/test_serving_chat.py

@@ -133,6 +133,84 @@ class ServingChatTestCase(unittest.TestCase):
             self.assertFalse(adapted.stream)
             self.assertEqual(processed, self.basic_req)
 
+    def test_jinja_uses_openai_tool_schema_first(self):
+        """Ensure Jinja chat templates receive OpenAI-shaped tools by default."""
+        self.template_manager.chat_template_name = None
+        self.template_manager.jinja_template_content_format = "string"
+
+        req = ChatCompletionRequest(
+            model="x",
+            messages=[{"role": "user", "content": "What is 2+2?"}],
+            tools=[
+                {
+                    "type": "function",
+                    "function": {
+                        "name": "add",
+                        "description": "Add two numbers.",
+                        "parameters": {
+                            "type": "object",
+                            "properties": {
+                                "a": {"type": "integer"},
+                                "b": {"type": "integer"},
+                            },
+                            "required": ["a", "b"],
+                        },
+                    },
+                }
+            ],
+        )
+
+        self.chat._process_messages(req, is_multimodal=False)
+
+        expected_tools = [tool.model_dump() for tool in req.tools]
+        kwargs = self.tm.tokenizer.apply_chat_template.call_args.kwargs
+        self.assertEqual(kwargs["tools"], expected_tools)
+
+    def test_jinja_tool_schema_fallback_to_flat_function(self):
+        """Fallback to function-only schema when template rejects OpenAI wrapper."""
+        self.template_manager.chat_template_name = None
+        self.template_manager.jinja_template_content_format = "string"
+
+        req = ChatCompletionRequest(
+            model="x",
+            messages=[{"role": "user", "content": "What is 2+2?"}],
+            tools=[
+                {
+                    "type": "function",
+                    "function": {
+                        "name": "add",
+                        "description": "Add two numbers.",
+                        "parameters": {
+                            "type": "object",
+                            "properties": {
+                                "a": {"type": "integer"},
+                                "b": {"type": "integer"},
+                            },
+                            "required": ["a", "b"],
+                        },
+                    },
+                }
+            ],
+        )
+
+        self.tm.tokenizer.apply_chat_template.side_effect = [
+            RuntimeError("template expects flat tools format"),
+            [1, 2, 3],
+        ]
+
+        self.chat._process_messages(req, is_multimodal=False)
+
+        first_tools = self.tm.tokenizer.apply_chat_template.call_args_list[0].kwargs[
+            "tools"
+        ]
+        second_tools = self.tm.tokenizer.apply_chat_template.call_args_list[1].kwargs[
+            "tools"
+        ]
+        self.assertEqual(first_tools, [tool.model_dump() for tool in req.tools])
+        self.assertEqual(
+            second_tools, [tool.function.model_dump() for tool in req.tools]
+        )
+
     def test_stop_str_isolation_between_requests(self):
         """Test that stop strings from one request don't affect subsequent requests.