From fe4d4109c24063fb5e60ede9d48a1c0b3e9e4cea Mon Sep 17 00:00:00 2001 From: "wang.yuqi" Date: Fri, 11 Sep 2026 13:33:26 +0800 Subject: [PATCH] [Frontend][last/N] Move all non-OpenAI content out of the OpenAI folder. (#56369) Signed-off-by: wang.yuqi --- .../openai/test_async_tokenization.py | 82 ------------------- .../utils}/test_sse_keep_alive.py | 2 +- .../openai/chat_completion/api_router.py | 2 +- .../openai/completion/api_router.py | 2 +- .../{openai => serve/utils}/sse_keep_alive.py | 0 5 files changed, 3 insertions(+), 85 deletions(-) delete mode 100644 tests/entrypoints/openai/test_async_tokenization.py rename tests/entrypoints/{openai => serve/utils}/test_sse_keep_alive.py (99%) rename vllm/entrypoints/{openai => serve/utils}/sse_keep_alive.py (100%) diff --git a/tests/entrypoints/openai/test_async_tokenization.py b/tests/entrypoints/openai/test_async_tokenization.py deleted file mode 100644 index 1d3d110d30..0000000000 --- a/tests/entrypoints/openai/test_async_tokenization.py +++ /dev/null @@ -1,82 +0,0 @@ -# SPDX-License-Identifier: Apache-2.0 -# SPDX-FileCopyrightText: Copyright contributors to the vLLM project - -import asyncio -import random -from collections.abc import Callable - -import openai -import pytest -import pytest_asyncio - -from tests.utils import RemoteOpenAIServer - -MODEL_NAME = "Qwen/Qwen2.5-1.5B-Instruct" - - -@pytest.fixture(scope="module") -def server(): - args = [ - # use half precision for speed and memory savings in CI environment - "--dtype", - "bfloat16", - "--max-model-len", - "8192", - "--enforce-eager", - "--max-num-seqs", - "128", - "--load-format", - "dummy", - ] - - with RemoteOpenAIServer(MODEL_NAME, args) as remote_server: - yield remote_server - - -@pytest_asyncio.fixture -async def client(server): - async with server.get_async_client() as async_client: - yield async_client - - -@pytest.mark.asyncio -@pytest.mark.parametrize( - ids=["completion", "chat"], - argnames=["create_func_gen", "content_body"], - argvalues=[ - (lambda x: x.completions.create, {"prompt": " ".join(["A"] * 10_000)}), - ( - lambda x: x.chat.completions.create, - {"messages": [{"role": "user", "content": " ".join(["A"] * 10_000)}]}, - ), - ], -) -async def test_with_and_without_truncate( - server: RemoteOpenAIServer, - client: openai.AsyncOpenAI, - create_func_gen: Callable, - content_body: dict, -): - create_func = create_func_gen(client) - body = {"model": MODEL_NAME, **content_body, "max_tokens": 10} - - num_requests = 10 - truncate_prompt_tokens = [1000] * (num_requests // 2) + [None] * ( - num_requests - num_requests // 2 - ) - random.shuffle(truncate_prompt_tokens) - - bodies = [ - {**body, "extra_body": {"truncate_prompt_tokens": t}} - for t in truncate_prompt_tokens - ] - - async def get_status_code(**kwargs): - try: - await create_func(**kwargs) - return 200 - except openai.APIStatusError as e: - return e.status_code - - responses = await asyncio.gather(*[get_status_code(**b) for b in bodies]) - assert 500 not in responses diff --git a/tests/entrypoints/openai/test_sse_keep_alive.py b/tests/entrypoints/serve/utils/test_sse_keep_alive.py similarity index 99% rename from tests/entrypoints/openai/test_sse_keep_alive.py rename to tests/entrypoints/serve/utils/test_sse_keep_alive.py index 74f5ee3ba2..519b9bb4a1 100644 --- a/tests/entrypoints/openai/test_sse_keep_alive.py +++ b/tests/entrypoints/serve/utils/test_sse_keep_alive.py @@ -6,7 +6,7 @@ import asyncio import pytest from starlette.responses import StreamingResponse -from vllm.entrypoints.openai.sse_keep_alive import ( +from vllm.entrypoints.serve.utils.sse_keep_alive import ( SSE_KEEP_ALIVE_COMMENT, with_sse_keep_alive, ) diff --git a/vllm/entrypoints/openai/chat_completion/api_router.py b/vllm/entrypoints/openai/chat_completion/api_router.py index 291907a77a..aecf5bba44 100644 --- a/vllm/entrypoints/openai/chat_completion/api_router.py +++ b/vllm/entrypoints/openai/chat_completion/api_router.py @@ -14,7 +14,6 @@ from vllm.entrypoints.openai.chat_completion.protocol import ( ChatCompletionResponse, ) from vllm.entrypoints.openai.chat_completion.serving import OpenAIServingChat -from vllm.entrypoints.openai.sse_keep_alive import with_sse_keep_alive from vllm.entrypoints.serve.engine.protocol import ErrorResponse from vllm.entrypoints.serve.utils.api_utils import ( load_aware_call, @@ -22,6 +21,7 @@ from vllm.entrypoints.serve.utils.api_utils import ( with_cancellation, ) from vllm.entrypoints.serve.utils.orca_metrics import metrics_header +from vllm.entrypoints.serve.utils.sse_keep_alive import with_sse_keep_alive from vllm.logger import init_logger logger = init_logger(__name__) diff --git a/vllm/entrypoints/openai/completion/api_router.py b/vllm/entrypoints/openai/completion/api_router.py index 7b33fcdb6c..fe909c7bae 100644 --- a/vllm/entrypoints/openai/completion/api_router.py +++ b/vllm/entrypoints/openai/completion/api_router.py @@ -12,7 +12,6 @@ from vllm.entrypoints.openai.completion.protocol import ( CompletionResponse, ) from vllm.entrypoints.openai.completion.serving import OpenAIServingCompletion -from vllm.entrypoints.openai.sse_keep_alive import with_sse_keep_alive from vllm.entrypoints.serve.engine.protocol import ErrorResponse from vllm.entrypoints.serve.utils.api_utils import ( load_aware_call, @@ -20,6 +19,7 @@ from vllm.entrypoints.serve.utils.api_utils import ( with_cancellation, ) from vllm.entrypoints.serve.utils.orca_metrics import metrics_header +from vllm.entrypoints.serve.utils.sse_keep_alive import with_sse_keep_alive from vllm.logger import init_logger logger = init_logger(__name__) diff --git a/vllm/entrypoints/openai/sse_keep_alive.py b/vllm/entrypoints/serve/utils/sse_keep_alive.py similarity index 100% rename from vllm/entrypoints/openai/sse_keep_alive.py rename to vllm/entrypoints/serve/utils/sse_keep_alive.py