[Frontend] Consolidate online serving utils. (#44479)

Signed-off-by: wang.yuqi <yuqi.wang@daocloud.io>
This commit is contained in:
wang.yuqi
2026-06-04 06:49:31 +00:00
committed by GitHub
parent b4b4aaa70e
commit d01d0b4646
81 changed files with 466 additions and 435 deletions
+3 -3
View File
@@ -16,7 +16,6 @@ from vllm.config import ModelConfig
from vllm.engine.protocol import EngineClient
from vllm.entrypoints.chat_utils import ChatTemplateContentFormatOption
from vllm.entrypoints.generate.beam_search.online import BeamSearchOnlineMixin
from vllm.entrypoints.logger import RequestLogger
from vllm.entrypoints.openai.chat_completion.protocol import (
BatchChatCompletionRequest,
ChatCompletionRequest,
@@ -39,12 +38,13 @@ from vllm.entrypoints.serve.tokenize.protocol import (
TokenizeCompletionRequest,
TokenizeResponse,
)
from vllm.entrypoints.serve.utils.error_response import create_error_response
from vllm.entrypoints.serve.utils.request_logger import RequestLogger
from vllm.entrypoints.speech_to_text.transcription.protocol import (
TranscriptionRequest,
TranscriptionResponse,
)
from vllm.entrypoints.speech_to_text.translation.protocol import TranslationRequest
from vllm.entrypoints.utils import create_error_response
from vllm.inputs import EngineInput, PromptType
from vllm.logger import init_logger
from vllm.logprobs import Logprob, PromptLogprobs
@@ -153,7 +153,7 @@ class OpenAIServing(BeamSearchOnlineMixin):
# Computed once at startup (cached by ``vllm_config`` identity) and
# stamped on non-streaming responses. Streaming chunks deliberately
# omit it to avoid per-chunk overhead.
from vllm.entrypoints.openai.fingerprint import get_system_fingerprint
from vllm.entrypoints.serve.utils.fingerprint import get_system_fingerprint
try:
self.system_fingerprint: str | None = get_system_fingerprint(