[openai api] log exception in exception handler (1/N) (#31164)
Signed-off-by: Andy Xie <andy.xning@gmail.com>
This commit is contained in:
@@ -49,10 +49,7 @@ router = APIRouter()
|
||||
async def tokenize(request: TokenizeRequest, raw_request: Request):
|
||||
handler = tokenization(raw_request)
|
||||
|
||||
try:
|
||||
generator = await handler.create_tokenize(request, raw_request)
|
||||
except Exception as e:
|
||||
generator = handler.create_error_response(e)
|
||||
generator = await handler.create_tokenize(request, raw_request)
|
||||
|
||||
if isinstance(generator, ErrorResponse):
|
||||
return JSONResponse(
|
||||
|
||||
@@ -3,7 +3,6 @@
|
||||
from dataclasses import dataclass
|
||||
from typing import Any, Final
|
||||
|
||||
import jinja2
|
||||
from fastapi import Request
|
||||
|
||||
from vllm.engine.protocol import EngineClient
|
||||
@@ -37,13 +36,11 @@ class OpenAIServingTokenization(OpenAIServing):
|
||||
chat_template: str | None,
|
||||
chat_template_content_format: ChatTemplateContentFormatOption,
|
||||
trust_request_chat_template: bool = False,
|
||||
log_error_stack: bool = False,
|
||||
) -> None:
|
||||
super().__init__(
|
||||
engine_client=engine_client,
|
||||
models=models,
|
||||
request_logger=request_logger,
|
||||
log_error_stack=log_error_stack,
|
||||
)
|
||||
|
||||
self.chat_template = chat_template
|
||||
@@ -61,40 +58,36 @@ class OpenAIServingTokenization(OpenAIServing):
|
||||
|
||||
request_id = f"tokenize-{self._base_request_id(raw_request)}"
|
||||
|
||||
try:
|
||||
lora_request = self._maybe_get_adapters(request)
|
||||
lora_request = self._maybe_get_adapters(request)
|
||||
|
||||
if isinstance(request, TokenizeChatRequest):
|
||||
tool_dicts = (
|
||||
None
|
||||
if request.tools is None
|
||||
else [tool.model_dump() for tool in request.tools]
|
||||
)
|
||||
error_check_ret = self._validate_chat_template(
|
||||
request_chat_template=request.chat_template,
|
||||
chat_template_kwargs=request.chat_template_kwargs,
|
||||
trust_request_chat_template=self.trust_request_chat_template,
|
||||
)
|
||||
if error_check_ret is not None:
|
||||
return error_check_ret
|
||||
if isinstance(request, TokenizeChatRequest):
|
||||
tool_dicts = (
|
||||
None
|
||||
if request.tools is None
|
||||
else [tool.model_dump() for tool in request.tools]
|
||||
)
|
||||
error_check_ret = self._validate_chat_template(
|
||||
request_chat_template=request.chat_template,
|
||||
chat_template_kwargs=request.chat_template_kwargs,
|
||||
trust_request_chat_template=self.trust_request_chat_template,
|
||||
)
|
||||
if error_check_ret is not None:
|
||||
return error_check_ret
|
||||
|
||||
_, engine_prompts = await self._preprocess_chat(
|
||||
request,
|
||||
request.messages,
|
||||
default_template=self.chat_template,
|
||||
default_template_content_format=self.chat_template_content_format,
|
||||
default_template_kwargs=None,
|
||||
tool_dicts=tool_dicts,
|
||||
)
|
||||
else:
|
||||
engine_prompts = await self._preprocess_completion(
|
||||
request,
|
||||
prompt_input=request.prompt,
|
||||
prompt_embeds=None,
|
||||
)
|
||||
except (ValueError, TypeError, jinja2.TemplateError) as e:
|
||||
logger.exception("Error in preprocessing prompt inputs")
|
||||
return self.create_error_response(f"{e} {e.__cause__}")
|
||||
_, engine_prompts = await self._preprocess_chat(
|
||||
request,
|
||||
request.messages,
|
||||
default_template=self.chat_template,
|
||||
default_template_content_format=self.chat_template_content_format,
|
||||
default_template_kwargs=None,
|
||||
tool_dicts=tool_dicts,
|
||||
)
|
||||
else:
|
||||
engine_prompts = await self._preprocess_completion(
|
||||
request,
|
||||
prompt_input=request.prompt,
|
||||
prompt_embeds=None,
|
||||
)
|
||||
|
||||
input_ids: list[int] = []
|
||||
for engine_prompt in engine_prompts:
|
||||
@@ -152,12 +145,9 @@ class OpenAIServingTokenization(OpenAIServing):
|
||||
self,
|
||||
) -> TokenizerInfoResponse | ErrorResponse:
|
||||
"""Get comprehensive tokenizer information."""
|
||||
try:
|
||||
tokenizer = self.renderer.get_tokenizer()
|
||||
info = TokenizerInfo(tokenizer, self.chat_template).to_dict()
|
||||
return TokenizerInfoResponse(**info)
|
||||
except Exception as e:
|
||||
return self.create_error_response(f"Failed to get tokenizer info: {str(e)}")
|
||||
tokenizer = self.renderer.get_tokenizer()
|
||||
info = TokenizerInfo(tokenizer, self.chat_template).to_dict()
|
||||
return TokenizerInfoResponse(**info)
|
||||
|
||||
|
||||
@dataclass
|
||||
|
||||
Reference in New Issue
Block a user