api_router.py 1.01 KB
Newer Older
1
2
3
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project

4
from fastapi import APIRouter, Depends, Request
5
from fastapi.responses import Response
6
7

from vllm.entrypoints.openai.utils import validate_json_request
8
from vllm.entrypoints.pooling.classify.protocol import ClassificationRequest
9
from vllm.entrypoints.pooling.classify.serving import ServingClassification
10
11
12
13
from vllm.entrypoints.utils import (
    load_aware_call,
    with_cancellation,
)
14
15
16
17
18

router = APIRouter()


def classify(request: Request) -> ServingClassification | None:
19
    return request.app.state.serving_classification
20
21
22
23
24


@router.post("/classify", dependencies=[Depends(validate_json_request)])
@with_cancellation
@load_aware_call
25
26
async def create_classify(
    request: ClassificationRequest, raw_request: Request
27
) -> Response:
28
29
    handler = classify(raw_request)
    if handler is None:
30
        raise NotImplementedError("The model does not support Classification API")
31

32
    return await handler(request, raw_request)