[Refactor] Consolidate sequence normalization and enc-dec parsing (vllm-project#33928)

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

Author: DarkLight1337

tests/entrypoints/openai/test_chat_error.py

@@ -54,6 +54,7 @@ class MockModelConfig:
     generation_config: str = "auto"
     media_io_kwargs: dict[str, dict[str, Any]] = field(default_factory=dict)
     skip_tokenizer_init = False
+    is_encoder_decoder: bool = False
 
     def get_diff_sampling_param(self):
         return self.diff_sampling_param or {}

tests/entrypoints/openai/test_completion_error.py

@@ -53,6 +53,7 @@ class MockModelConfig:
     generation_config: str = "auto"
     media_io_kwargs: dict[str, dict[str, Any]] = field(default_factory=dict)
     skip_tokenizer_init = False
+    is_encoder_decoder: bool = False
 
     def get_diff_sampling_param(self):
         return self.diff_sampling_param or {}

tests/entrypoints/openai/test_lora_resolvers.py

@@ -52,6 +52,7 @@ class MockModelConfig:
     encoder_config = None
     generation_config: str = "auto"
     skip_tokenizer_init: bool = False
+    is_encoder_decoder: bool = False
 
     def get_diff_sampling_param(self):
         return self.diff_sampling_param or {}

tests/entrypoints/openai/test_serving_chat.py

@@ -529,6 +529,7 @@ class MockModelConfig:
     generation_config: str = "auto"
     media_io_kwargs: dict[str, dict[str, Any]] = field(default_factory=dict)
     skip_tokenizer_init: bool = False
+    is_encoder_decoder: bool = False
 
     def get_diff_sampling_param(self):
         return self.diff_sampling_param or {}

tests/renderers/inputs/__init__.py

@@ -0,0 +1,41 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+from vllm.renderers.inputs.preprocess import prompt_to_seq
+
+
+def test_empty_input():
+    assert prompt_to_seq([]) == []
+    assert prompt_to_seq([[]]) == [[]]
+    assert prompt_to_seq([[], []]) == [[], []]
+
+
+def test_text_input():
+    assert prompt_to_seq("foo") == ["foo"]
+    assert prompt_to_seq(["foo"]) == ["foo"]
+    assert prompt_to_seq(["foo", "bar"]) == ["foo", "bar"]
+
+
+def test_token_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():
+    assert prompt_to_seq([[1, 2], "foo"]) == [[1, 2], "foo"]
+    assert prompt_to_seq(["foo", [1, 2]]) == ["foo", [1, 2]]
+
+
+def test_bytes_input():
+    assert prompt_to_seq(b"foo") == [b"foo"]
+    assert prompt_to_seq([b"foo"]) == [b"foo"]
+    assert prompt_to_seq([b"foo", b"bar"]) == [b"foo", b"bar"]
+
+
+def test_dict_input():
+    assert prompt_to_seq({"prompt": "foo"}) == [{"prompt": "foo"}]
+    assert prompt_to_seq([{"prompt": "foo"}]) == [{"prompt": "foo"}]
+    assert prompt_to_seq([{"prompt": "foo"}, {"prompt_token_ids": [1, 2]}]) == [
+        {"prompt": "foo"},
+        {"prompt_token_ids": [1, 2]},
+    ]