[Renderer] Move InputPreprocessor into Renderer (1/2) (#34510)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: Cyrus Leung <cyrus.tl.leung@gmail.com>

vllm/engine/protocol.py

@@ -3,6 +3,7 @@
 
 from abc import ABC, abstractmethod
 from collections.abc import AsyncGenerator, Iterable, Mapping
+from dataclasses import dataclass
 from typing import TYPE_CHECKING, Any
 
 from vllm.config import ModelConfig, VllmConfig
@@ -10,7 +11,7 @@ from vllm.distributed.weight_transfer.base import (
     WeightTransferInitRequest,
     WeightTransferUpdateRequest,
 )
-from vllm.inputs.data import PromptType, StreamingInput
+from vllm.inputs.data import PromptType
 from vllm.lora.request import LoRARequest
 from vllm.outputs import PoolingRequestOutput, RequestOutput
 from vllm.plugins.io_processors import IOProcessor
@@ -26,6 +27,18 @@ if TYPE_CHECKING:
     from vllm.v1.engine import PauseMode
 
 
+@dataclass
+class StreamingInput:
+    """Input data for a streaming generation request.
+
+    This is used with generate() to support multi-turn streaming sessions
+    where inputs are provided via an async generator.
+    """
+
+    prompt: PromptType
+    sampling_params: SamplingParams | None = None
+
+
 class EngineClient(ABC):
     """Protocol class for Clients to Engine"""