[Renderer] Move MM Hash parsing into Renderer (#34711)

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

vllm/model_executor/models/deepseek_vl2.py

@@ -24,13 +24,13 @@ from vllm.multimodal.inputs import (
     MultiModalDataDict,
     MultiModalFieldConfig,
     MultiModalKwargsItems,
-    MultiModalUUIDDict,
 )
 from vllm.multimodal.parse import (
     ImageEmbeddingItems,
     ImageProcessorItems,
     ImageSize,
     MultiModalDataItems,
+    MultiModalUUIDItems,
 )
 from vllm.multimodal.processing import BaseDummyInputsBuilder
 from vllm.multimodal.processing.processor import (
@@ -313,9 +313,9 @@ class DeepseekVL2MultiModalProcessor(
         self,
         prompt: str | list[int],
         mm_data_items: MultiModalDataItems,
+        mm_uuid_items: MultiModalUUIDItems | None,
         hf_processor_mm_kwargs: Mapping[str, object],
         tokenization_kwargs: Mapping[str, object],
-        mm_uuids: MultiModalUUIDDict | None = None,
     ) -> tuple[list[int], MultiModalProcessingInfo, bool]:
         # The processor logic is different for len(images) <= 2 vs > 2
         # Since the processing cache assumes that the processor output is
@@ -325,17 +325,17 @@ class DeepseekVL2MultiModalProcessor(
             return self._apply_hf_processor(
                 prompt=prompt,
                 mm_data_items=mm_data_items,
+                mm_uuid_items=mm_uuid_items,
                 hf_processor_mm_kwargs=hf_processor_mm_kwargs,
                 tokenization_kwargs=tokenization_kwargs,
-                mm_uuids=mm_uuids,
             )
 
         return super()._cached_apply_hf_processor(
             prompt=prompt,
             mm_data_items=mm_data_items,
+            mm_uuid_items=mm_uuid_items,
             hf_processor_mm_kwargs=hf_processor_mm_kwargs,
             tokenization_kwargs=tokenization_kwargs,
-            mm_uuids=mm_uuids,
         )