[Chore] Remove redundant input parsing methods (#33542)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

vllm/entrypoints/openai/chat_completion/serving.py

@@ -68,7 +68,6 @@ from vllm.entrypoints.openai.parser.harmony_utils import (
 from vllm.entrypoints.openai.utils import maybe_filter_parallel_tool_calls
 from vllm.entrypoints.utils import get_max_tokens, should_include_usage
 from vllm.inputs.data import EmbedsPrompt, TokensPrompt
-from vllm.inputs.parse import get_prompt_components
 from vllm.logger import init_logger
 from vllm.logprobs import Logprob
 from vllm.outputs import CompletionOutput, RequestOutput
@@ -359,7 +358,7 @@ class OpenAIServingChat(OpenAIServing):
         generators: list[AsyncGenerator[RequestOutput, None]] = []
         try:
             for i, engine_prompt in enumerate(engine_prompts):
-                prompt_text, _, _ = get_prompt_components(engine_prompt)
+                prompt_text = engine_prompt.get("prompt")
 
                 # If we are creating sub requests for multiple prompts, ensure that they
                 # have unique request ids.