[Renderer] Move InputPreprocessor into Renderer (2/2) (#34560)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2026-02-17 05:29:01 -08:00
committed by GitHub
parent c61a98f529
commit 574fe75245
32 changed files with 984 additions and 1054 deletions
+4 -22
View File
@@ -34,10 +34,10 @@ from vllm.entrypoints.openai.engine.serving import (
from vllm.entrypoints.openai.models.serving import OpenAIServingModels
from vllm.entrypoints.utils import get_max_tokens, should_include_usage
from vllm.exceptions import VLLMValidationError
from vllm.inputs.data import ProcessorInputs
from vllm.logger import init_logger
from vllm.logprobs import Logprob
from vllm.outputs import RequestOutput
from vllm.renderers.inputs import TokPrompt
from vllm.sampling_params import BeamSearchParams, SamplingParams
from vllm.tokenizers import TokenizerLike
from vllm.utils.async_utils import merge_async_iterators
@@ -80,7 +80,7 @@ class OpenAIServingCompletion(OpenAIServing):
async def render_completion_request(
self,
request: CompletionRequest,
) -> list[TokPrompt] | ErrorResponse:
) -> list[ProcessorInputs] | ErrorResponse:
"""
render completion request by validating and preprocessing inputs.
@@ -163,8 +163,6 @@ class OpenAIServingCompletion(OpenAIServing):
generators: list[AsyncGenerator[RequestOutput, None]] = []
try:
for i, engine_prompt in enumerate(engine_prompts):
prompt_text = self._extract_prompt_text(engine_prompt)
max_tokens = get_max_tokens(
max_model_len,
request.max_tokens,
@@ -208,29 +206,13 @@ class OpenAIServingCompletion(OpenAIServing):
trace_headers=trace_headers,
)
else:
tok_params = request.build_tok_params(self.model_config)
tokenization_kwargs = tok_params.get_encode_kwargs()
engine_request = self.input_processor.process_inputs(
request_id_item,
generator = self.engine_client.generate(
engine_prompt,
sampling_params,
lora_request=lora_request,
tokenization_kwargs=tokenization_kwargs,
trace_headers=trace_headers,
priority=request.priority,
data_parallel_rank=data_parallel_rank,
)
generator = self.engine_client.generate(
engine_request,
sampling_params,
request_id_item,
lora_request=lora_request,
trace_headers=trace_headers,
priority=request.priority,
prompt_text=prompt_text,
tokenization_kwargs=tokenization_kwargs,
data_parallel_rank=data_parallel_rank,
)
@@ -312,7 +294,7 @@ class OpenAIServingCompletion(OpenAIServing):
async def completion_stream_generator(
self,
request: CompletionRequest,
engine_prompts: list[TokPrompt],
engine_prompts: list[ProcessorInputs],
result_generator: AsyncIterator[tuple[int, RequestOutput]],
request_id: str,
created_time: int,