[Bugfix]: Fix the incompatibility issue with stream when Thinking is disabled (#19135)

Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>

tests/entrypoints/openai/test_completion_with_function_calling.py

 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 
+from typing import NamedTuple
+
 import openai  # use the official client for correctness check
 import pytest
 import pytest_asyncio
@@ -22,7 +24,9 @@ def server():  # noqa: F811
         "--guided-decoding-backend",
         "xgrammar",
         "--tool-call-parser",
-        "hermes"
+        "hermes",
+        "--reasoning-parser",
+        "qwen3",
     ]
 
     with RemoteOpenAIServer(MODEL_NAME, args) as remote_server:
@@ -35,9 +39,53 @@ async def client(server):
         yield async_client
 
 
+class TestCase(NamedTuple):
+    model_name: str
+    stream: bool
+    tool_choice: str
+    enable_thinking: bool
+
+
 @pytest.mark.asyncio
-@pytest.mark.parametrize("model_name", [MODEL_NAME])
-async def test_required_tool_use(client: openai.AsyncOpenAI, model_name: str):
+@pytest.mark.parametrize(
+    "test_case",
+    [
+        TestCase(model_name=MODEL_NAME,
+                 stream=True,
+                 tool_choice="auto",
+                 enable_thinking=False),
+        TestCase(model_name=MODEL_NAME,
+                 stream=False,
+                 tool_choice="auto",
+                 enable_thinking=False),
+        TestCase(model_name=MODEL_NAME,
+                 stream=True,
+                 tool_choice="required",
+                 enable_thinking=False),
+        TestCase(model_name=MODEL_NAME,
+                 stream=False,
+                 tool_choice="required",
+                 enable_thinking=False),
+        TestCase(model_name=MODEL_NAME,
+                 stream=True,
+                 tool_choice="auto",
+                 enable_thinking=True),
+        TestCase(model_name=MODEL_NAME,
+                 stream=False,
+                 tool_choice="auto",
+                 enable_thinking=True),
+        TestCase(model_name=MODEL_NAME,
+                 stream=True,
+                 tool_choice="required",
+                 enable_thinking=True),
+        TestCase(model_name=MODEL_NAME,
+                 stream=False,
+                 tool_choice="required",
+                 enable_thinking=True),
+    ],
+)
+async def test_function_tool_use(client: openai.AsyncOpenAI,
+                                 test_case: TestCase):
     tools = [
         {
             "type": "function",
@@ -126,26 +174,34 @@ async def test_required_tool_use(client: openai.AsyncOpenAI, model_name: str):
             "forecast for the next 5 days, in fahrenheit?",
         },
     ]
-
+    if not test_case.stream:
         # Non-streaming test
         chat_completion = await client.chat.completions.create(
             messages=messages,
-        model=model_name,
+            model=test_case.model_name,
             tools=tools,
-        tool_choice="required",
-    )
+            tool_choice=test_case.tool_choice,
+            extra_body={
+                "chat_template_kwargs": {
+                    "enable_thinking": test_case.enable_thinking
+                }
+            })
 
         assert chat_completion.choices[0].message.tool_calls is not None
         assert len(chat_completion.choices[0].message.tool_calls) > 0
-
+    else:
         # Streaming test
         stream = await client.chat.completions.create(
             messages=messages,
-        model=model_name,
+            model=test_case.model_name,
             tools=tools,
-        tool_choice="required",
+            tool_choice=test_case.tool_choice,
             stream=True,
-    )
+            extra_body={
+                "chat_template_kwargs": {
+                    "enable_thinking": test_case.enable_thinking
+                }
+            })
 
         output = []
         async for chunk in stream:

vllm/entrypoints/openai/serving_chat.py

@@ -689,7 +689,21 @@ class OpenAIServingChat(OpenAIServing):
                                     current_token_ids,
                                     output.token_ids,
                                 ))
-
+                            # When encountering think end id in prompt_token_ids
+                            # i.e {"enable_thinking": False},
+                            # set reasoning status to end.
+                            # Remove the text and token ids related
+                            # to 'reasoning_content'.
+                            if res.prompt_token_ids and \
+                                reasoning_parser.is_reasoning_end(
+                                    list(res.prompt_token_ids)):
+                                reasoning_end_arr[i] = True
+                                current_token_ids = list(output.token_ids)
+                                if delta_message and delta_message.content:
+                                    current_text = delta_message.content
+                                    delta_message.content = None
+                                else:
+                                    current_text = ""
                             # When encountering think end id in delta_token_ids,
                             # set reasoning status to end.
                             # Remove the text and token ids related