[Refactor] Relocate endpoint tests to mirror serving code directory structure (#37504)

Signed-off-by: sfeng33 <4florafeng@gmail.com>

tests/entrypoints/anthropic/test_messages.py

@@ -5,7 +5,7 @@ import anthropic
 import pytest
 import pytest_asyncio
 
-from ...utils import RemoteOpenAIServer
+from tests.utils import RemoteOpenAIServer
 
 MODEL_NAME = "Qwen/Qwen3-0.6B"
 

tests/entrypoints/openai/models/test_models.py

@@ -5,7 +5,7 @@ import openai  # use the official client for correctness check
 import pytest
 import pytest_asyncio
 
-from ...utils import RemoteOpenAIServer
+from tests.utils import RemoteOpenAIServer
 
 # any model with a chat template should work here
 MODEL_NAME = "Qwen/Qwen3-0.6B"

tests/entrypoints/openai/realtime/test_realtime_validation.py

@@ -11,11 +11,10 @@ import pybase64 as base64
 import pytest
 import websockets
 
+from tests.entrypoints.openai.conftest import add_attention_backend
+from tests.utils import ROCM_ENV_OVERRIDES, ROCM_EXTRA_ARGS, RemoteOpenAIServer
 from vllm.assets.audio import AudioAsset
 
-from ...utils import ROCM_ENV_OVERRIDES, ROCM_EXTRA_ARGS, RemoteOpenAIServer
-from .conftest import add_attention_backend
-
 MISTRAL_FORMAT_ARGS = [
     "--tokenizer_mode",
     "mistral",

tests/entrypoints/openai/responses/test_mcp_tools.py

@@ -42,7 +42,7 @@ class TestMCPToolServerUnit:
     Note: The wildcard "*" is normalized to None by
     _extract_allowed_tools_from_mcp_requests before reaching this layer,
     so we only test None and specific tool filtering here.
-    See test_serving_responses.py for "*" normalization tests.
+    See responses/test_serving_responses.py for "*" normalization tests.
     """
 
     def test_get_tool_description(self):

tests/entrypoints/openai/speech_to_text/test_transcription_validation.py

@@ -6,8 +6,8 @@ import json
 
 import pytest
 
-from ...utils import ROCM_ENV_OVERRIDES, ROCM_EXTRA_ARGS, RemoteOpenAIServer
-from .conftest import add_attention_backend
+from tests.entrypoints.openai.conftest import add_attention_backend
+from tests.utils import ROCM_ENV_OVERRIDES, ROCM_EXTRA_ARGS, RemoteOpenAIServer
 
 MISTRAL_FORMAT_ARGS = [
     "--tokenizer_mode",

tests/entrypoints/openai/speech_to_text/test_transcription_validation_whisper.py

@@ -13,7 +13,7 @@ import pytest
 import pytest_asyncio
 import soundfile as sf
 
-from ...utils import RemoteOpenAIServer
+from tests.utils import RemoteOpenAIServer
 
 MODEL_NAME = "openai/whisper-large-v3-turbo"
 

tests/entrypoints/openai/speech_to_text/test_translation_validation.py

@@ -14,8 +14,8 @@ import pytest
 import pytest_asyncio
 import soundfile as sf
 
-from ...utils import RemoteOpenAIServer
-from .conftest import add_attention_backend
+from tests.entrypoints.openai.conftest import add_attention_backend
+from tests.utils import RemoteOpenAIServer
 
 SERVER_ARGS = ["--enforce-eager"]
 

tests/entrypoints/serve/tokenize/test_tokenization.py

@@ -5,10 +5,9 @@ import pytest
 import pytest_asyncio
 import requests
 
+from tests.utils import RemoteOpenAIServer
 from vllm.tokenizers import get_tokenizer
 
-from ...utils import RemoteOpenAIServer
-
 # any model with a chat template should work here
 MODEL_NAME = "HuggingFaceH4/zephyr-7b-beta"
 

tests/entrypoints/serve/tokenize/test_tokenization_vlm.py

@@ -13,7 +13,7 @@ import json
 import pytest
 import requests
 
-from ...utils import RemoteOpenAIServer
+from tests.utils import RemoteOpenAIServer
 
 MODEL_NAME = "Qwen/Qwen2.5-VL-3B-Instruct"