diff --git a/tests/renderers/test_process_multi_modal_uuids.py b/tests/renderers/test_process_multi_modal_uuids.py index 797d390934..9f3440fc2e 100644 --- a/tests/renderers/test_process_multi_modal_uuids.py +++ b/tests/renderers/test_process_multi_modal_uuids.py @@ -1,11 +1,14 @@ # SPDX-License-Identifier: Apache-2.0 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project +from http import HTTPStatus + import pytest from vllm.assets.image import ImageAsset from vllm.assets.video import VideoAsset from vllm.config import CacheConfig, ModelConfig, VllmConfig +from vllm.entrypoints.serve.utils.error_response import create_error_response from vllm.multimodal.parse import parse_mm_uuids from vllm.renderers.hf import HfRenderer from vllm.tokenizers.registry import cached_tokenizer_from_config @@ -35,6 +38,33 @@ def _build_renderer( ) +def _build_text_only_renderer() -> HfRenderer: + model_config = ModelConfig(model="openai-community/gpt2", max_model_len=128) + + return HfRenderer( + VllmConfig(model_config=model_config), + cached_tokenizer_from_config(model_config), + ) + + +def test_text_only_model_mm_data_maps_to_bad_request(): + """Sending multimodal data to a text-only model is a client mistake, so it + must surface as a ValueError and reach the client as HTTP 400, not 500.""" + renderer = _build_text_only_renderer() + + with pytest.raises(ValueError, match="text-only") as exc_info: + renderer._process_multimodal( + prompt="What is in this image?", + mm_data={"image": [cherry_pil_image]}, + mm_uuids=None, + mm_processor_kwargs=None, + tokenization_kwargs=None, + ) + + error_response = create_error_response(exc_info.value) + assert error_response.error.code == HTTPStatus.BAD_REQUEST + + def test_multi_modal_uuids_length_mismatch_raises(): renderer = _build_renderer() diff --git a/vllm/renderers/base.py b/vllm/renderers/base.py index a0f2508ccc..bb02c115ca 100644 --- a/vllm/renderers/base.py +++ b/vllm/renderers/base.py @@ -735,13 +735,13 @@ class BaseRenderer(ABC, Generic[_T]): *, skip_mm_cache: bool = False, ) -> "MultiModalInput": - mm_req_id = f"renderer{self.api_process_rank}-mm-{self._mm_req_counter.inc(1)}" - if skip_mm_cache and self._readonly_mm_processor is not None: mm_processor = self._readonly_mm_processor else: mm_processor = self.get_mm_processor() + mm_req_id = f"renderer{self.api_process_rank}-mm-{self._mm_req_counter.inc(1)}" + mm_data_items = mm_processor.info.parse_mm_data(mm_data) mm_uuid_items = parse_mm_uuids(mm_uuids)