[Misc] Reorganize inputs (#35182)

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

tests/entrypoints/openai/chat_completion/test_chat_error.py

@@ -105,7 +105,7 @@ def _build_serving_chat(engine: AsyncLLM) -> OpenAIServingChat:
     )
 
     async def _fake_preprocess_chat(*args, **kwargs):
-        # return conversation, engine_prompts
+        # return conversation, engine_inputs
         return (
             [{"role": "user", "content": "Test"}],
             [{"prompt_token_ids": [1, 2, 3]}],

tests/entrypoints/openai/chat_completion/test_serving_chat.py

@@ -958,14 +958,14 @@ async def test_serving_chat_did_set_correct_cache_salt(model_type):
     serving_chat = _build_serving_chat(mock_engine)
 
     orig_render_chat_request = serving_chat.render_chat_request
-    captured_prompts = []
+    captured_inputs = []
 
     async def render_chat_request(request):
         result = await orig_render_chat_request(request)
 
         assert isinstance(result, tuple)
-        conversation, engine_prompts = result
-        captured_prompts.extend(engine_prompts)
+        conversation, engine_inputs = result
+        captured_inputs.extend(engine_inputs)
 
         return result
 
@@ -981,18 +981,18 @@ async def test_serving_chat_did_set_correct_cache_salt(model_type):
     with suppress(Exception):
         await serving_chat.create_chat_completion(req)
 
-    assert len(captured_prompts) == 1
-    assert "cache_salt" not in captured_prompts[0]
+    assert len(captured_inputs) == 1
+    assert "cache_salt" not in captured_inputs[0]
 
-    captured_prompts.clear()
+    captured_inputs.clear()
 
     # Test with certain cache_salt
     req.cache_salt = "test_salt"
     with suppress(Exception):
         await serving_chat.create_chat_completion(req)
 
-    assert len(captured_prompts) == 1
-    assert captured_prompts[0]["cache_salt"] == "test_salt"
+    assert len(captured_inputs) == 1
+    assert captured_inputs[0]["cache_salt"] == "test_salt"
 
 
 @pytest.mark.asyncio

tests/entrypoints/openai/responses/test_serving_responses.py

@@ -37,7 +37,7 @@ from vllm.entrypoints.openai.responses.serving import (
 from vllm.entrypoints.openai.responses.streaming_events import (
     StreamingState,
 )
-from vllm.inputs.data import TokensPrompt
+from vllm.inputs import tokens_input
 from vllm.outputs import CompletionOutput, RequestOutput
 from vllm.sampling_params import SamplingParams
 
@@ -258,20 +258,20 @@ class TestValidateGeneratorInput:
         """Test _validate_generator_input with valid prompt length"""
         # Create an engine prompt with valid length (less than max_model_len)
         valid_prompt_token_ids = list(range(5))  # 5 tokens < 100 max_model_len
-        engine_prompt = TokensPrompt(prompt_token_ids=valid_prompt_token_ids)
+        engine_input = tokens_input(valid_prompt_token_ids)
 
         # Call the method
-        result = serving_responses_instance._validate_generator_input(engine_prompt)
+        result = serving_responses_instance._validate_generator_input(engine_input)
 
         # Should return None for valid input
         assert result is None
 
         # create an invalid engine prompt
         invalid_prompt_token_ids = list(range(200))  # 100 tokens >= 100 max_model_len
-        engine_prompt = TokensPrompt(prompt_token_ids=invalid_prompt_token_ids)
+        engine_input = tokens_input(invalid_prompt_token_ids)
 
         # Call the method
-        result = serving_responses_instance._validate_generator_input(engine_prompt)
+        result = serving_responses_instance._validate_generator_input(engine_input)
 
         # Should return an ErrorResponse
         assert result is not None

tests/entrypoints/serve/render/test_launch_render.py

@@ -73,20 +73,6 @@ async def test_chat_render_multi_turn(client):
     assert len(data["token_ids"]) > 0
 
 
-@pytest.mark.asyncio
-async def test_chat_render_invalid_model(client):
-    response = await client.post(
-        "/v1/chat/completions/render",
-        json={
-            "model": "nonexistent-model",
-            "messages": [{"role": "user", "content": "Hello"}],
-        },
-    )
-
-    assert response.status_code == 404
-    assert "error" in response.json()
-
-
 # -- Completion Render --
 
 

tests/entrypoints/test_chat_utils.py

@@ -16,7 +16,7 @@ from vllm.entrypoints.chat_utils import (
     parse_chat_messages,
     parse_chat_messages_async,
 )
-from vllm.multimodal import MultiModalDataDict, MultiModalUUIDDict
+from vllm.inputs import MultiModalDataDict, MultiModalUUIDDict
 from vllm.multimodal.utils import (
     encode_audio_url,
     encode_image_url,

tests/models/multimodal/generation/test_pixtral.py

@@ -13,8 +13,8 @@ from mistral_common.tokens.tokenizers.multimodal import image_from_chunk
 from transformers import AutoProcessor
 
 from vllm import SamplingParams, TextPrompt, TokensPrompt
+from vllm.inputs import MultiModalDataBuiltins
 from vllm.logprobs import Logprob, SampleLogprobs
-from vllm.multimodal import MultiModalDataBuiltins
 from vllm.platforms import current_platform
 
 from ....utils import VLLM_PATH, large_gpu_test

tests/models/multimodal/processing/test_common.py

@@ -15,13 +15,11 @@ from vllm.config.multimodal import (
     ImageDummyOptions,
     VideoDummyOptions,
 )
-from vllm.multimodal import MULTIMODAL_REGISTRY, MultiModalDataDict
+from vllm.inputs import MultiModalDataDict, MultiModalInput
+from vllm.multimodal import MULTIMODAL_REGISTRY
 from vllm.multimodal.cache import MultiModalProcessorOnlyCache
-from vllm.multimodal.inputs import MultiModalInputs, batched_tensors_equal
-from vllm.multimodal.processing import (
-    BaseMultiModalProcessor,
-    InputProcessingContext,
-)
+from vllm.multimodal.inputs import batched_tensors_equal
+from vllm.multimodal.processing import BaseMultiModalProcessor, InputProcessingContext
 from vllm.tokenizers import TokenizerLike, cached_tokenizer_from_config
 from vllm.utils.mistral import is_mistral_tokenizer
 
@@ -420,8 +418,8 @@ def test_processing_correctness(
 
 
 def _assert_inputs_equal(
-    a: MultiModalInputs,
-    b: MultiModalInputs,
+    a: MultiModalInput,
+    b: MultiModalInput,
     *,
     ignore_mm_keys: set[str] | None = None,
     msg: str = "",

tests/plugins/bge_m3_sparse_plugin/bge_m3_sparse_processor/sparse_embeddings_processor.py

@@ -6,11 +6,9 @@ from collections.abc import Sequence
 from vllm.config import ModelConfig, PoolerConfig, VllmConfig
 from vllm.entrypoints.openai.engine.protocol import UsageInfo
 from vllm.entrypoints.pooling.base.protocol import EmbedRequestMixin
-from vllm.inputs.data import PromptType
+from vllm.inputs import PromptType
 from vllm.outputs import PoolingRequestOutput
-from vllm.plugins.io_processors.interface import (
-    IOProcessor,
-)
+from vllm.plugins.io_processors.interface import IOProcessor
 from vllm.pooling_params import PoolingParams
 from vllm.renderers import BaseRenderer
 from vllm.tokenizers.detokenizer_utils import convert_ids_list_to_tokens

tests/plugins/prithvi_io_processor_plugin/prithvi_io_processor/prithvi_processor.py

@@ -18,7 +18,7 @@ from einops import rearrange
 from terratorch.datamodules import Sen1Floods11NonGeoDataModule
 
 from vllm.config import VllmConfig
-from vllm.inputs.data import PromptType
+from vllm.inputs import PromptType
 from vllm.logger import init_logger
 from vllm.outputs import PoolingRequestOutput
 from vllm.plugins.io_processors.interface import IOProcessor

tests/plugins_tests/test_io_processor_plugins.py

@@ -6,7 +6,7 @@ from unittest.mock import MagicMock, patch
 import pytest
 
 from vllm.config import VllmConfig
-from vllm.inputs.data import PromptType
+from vllm.inputs import PromptType
 from vllm.outputs import PoolingRequestOutput
 from vllm.plugins.io_processors import get_io_processor
 from vllm.plugins.io_processors.interface import IOProcessor

tests/renderers/inputs/test_preprocess.py

@@ -15,13 +15,13 @@ def test_text_input():
     assert prompt_to_seq(["foo", "bar"]) == ["foo", "bar"]
 
 
-def test_token_input():
+def test_tokens_input():
     assert prompt_to_seq([1, 2]) == [[1, 2]]
     assert prompt_to_seq([[1, 2]]) == [[1, 2]]
     assert prompt_to_seq([[1, 2], [3, 4]]) == [[1, 2], [3, 4]]
 
 
-def test_text_token_input():
+def test_text_tokens_input():
     assert prompt_to_seq([[1, 2], "foo"]) == [[1, 2], "foo"]
     assert prompt_to_seq(["foo", [1, 2]]) == ["foo", [1, 2]]
 

tests/renderers/test_completions.py

@@ -129,7 +129,7 @@ class TestValidatePrompt:
 
 
 class TestRenderPrompt:
-    def test_token_input(self):
+    def test_tokens_input(self):
         renderer = _build_renderer(MockModelConfig())
 
         tokens = [101, 7592, 2088]
@@ -339,7 +339,7 @@ class TestRenderPrompt:
                 TokenizeParams(max_total_tokens=100),
             )
 
-    def test_token_input_with_needs_detokenization(self):
+    def test_tokens_input_with_needs_detokenization(self):
         renderer = _build_renderer(MockModelConfig())
 
         tokens = [1, 2, 3, 4]

tests/v1/shutdown/test_processor_error.py

@@ -9,7 +9,7 @@ import pytest
 from tests.v1.shutdown.utils import SHUTDOWN_TEST_TIMEOUT_SEC
 from vllm import SamplingParams
 from vllm.engine.arg_utils import AsyncEngineArgs
-from vllm.inputs.data import TokensPrompt
+from vllm.inputs import TokensPrompt
 from vllm.sampling_params import RequestOutputKind
 from vllm.v1.engine.async_llm import AsyncLLM
 from vllm.v1.engine.exceptions import EngineGenerateError