Rework add-new-model-like with modular and make test filenames coherent (#39612)

* remove tf/flax

* fix

* style

* Update add_new_model_like.py

* work in progress

* continue

* more cleanup

* simplify and first final version

* fixes -> it works

* add linter checks

* Update add_new_model_like.py

* fix

* add modular conversion at the end

* Update add_new_model_like.py

* add video processor

* Update add_new_model_like.py

* Update add_new_model_like.py

* Update add_new_model_like.py

* fix

* Update image_processing_auto.py

* Update image_processing_auto.py

* fix post rebase

* start test filenames replacement

* rename all test_processor -> test_processing

* fix copied from

* add docstrings

* Update add_new_model_like.py

* fix regex

* improve wording

* Update add_new_model_like.py

* Update add_new_model_like.py

* Update add_new_model_like.py

* start adding test

* fix

* fix

* proper first test

* tests

* fix

* fix

* fix

* fix

* modular can be used from anywhere

* protect import

* fix

* Update add_new_model_like.py

* fix

conftest.py

@@ -29,7 +29,6 @@ from transformers.testing_utils import HfDoctestModule, HfDocTestParser
 NOT_DEVICE_TESTS = {
     "test_tokenization",
     "test_tokenization_mistral_common",
-    "test_processor",
     "test_processing",
     "test_beam_constraints",
     "test_configuration_utils",

src/transformers/models/auto/image_processing_auto.py

@@ -62,7 +62,7 @@ else:
             ("aimv2", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("aimv2_vision_model", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("align", ("EfficientNetImageProcessor", "EfficientNetImageProcessorFast")),
-            ("aria", ("AriaImageProcessor")),
+            ("aria", ("AriaImageProcessor", None)),
             ("beit", ("BeitImageProcessor", "BeitImageProcessorFast")),
             ("bit", ("BitImageProcessor", "BitImageProcessorFast")),
             ("blip", ("BlipImageProcessor", "BlipImageProcessorFast")),
@@ -72,7 +72,7 @@ else:
             ("chinese_clip", ("ChineseCLIPImageProcessor", "ChineseCLIPImageProcessorFast")),
             ("clip", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("clipseg", ("ViTImageProcessor", "ViTImageProcessorFast")),
-            ("cohere2_vision", ("Cohere2VisionImageProcessorFast",)),
+            ("cohere2_vision", (None, "Cohere2VisionImageProcessorFast")),
             ("conditional_detr", ("ConditionalDetrImageProcessor", "ConditionalDetrImageProcessorFast")),
             ("convnext", ("ConvNextImageProcessor", "ConvNextImageProcessorFast")),
             ("convnextv2", ("ConvNextImageProcessor", "ConvNextImageProcessorFast")),
@@ -84,52 +84,52 @@ else:
             ("deit", ("DeiTImageProcessor", "DeiTImageProcessorFast")),
             ("depth_anything", ("DPTImageProcessor", "DPTImageProcessorFast")),
             ("depth_pro", ("DepthProImageProcessor", "DepthProImageProcessorFast")),
-            ("deta", ("DetaImageProcessor",)),
+            ("deta", ("DetaImageProcessor", None)),
             ("detr", ("DetrImageProcessor", "DetrImageProcessorFast")),
             ("dinat", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("dinov2", ("BitImageProcessor", "BitImageProcessorFast")),
             ("donut-swin", ("DonutImageProcessor", "DonutImageProcessorFast")),
             ("dpt", ("DPTImageProcessor", "DPTImageProcessorFast")),
-            ("efficientformer", ("EfficientFormerImageProcessor",)),
+            ("efficientformer", ("EfficientFormerImageProcessor", None)),
-            ("efficientloftr", ("EfficientLoFTRImageProcessor",)),
+            ("efficientloftr", ("EfficientLoFTRImageProcessor", None)),
             ("efficientnet", ("EfficientNetImageProcessor", "EfficientNetImageProcessorFast")),
             ("eomt", ("EomtImageProcessor", "EomtImageProcessorFast")),
             ("flava", ("FlavaImageProcessor", "FlavaImageProcessorFast")),
             ("focalnet", ("BitImageProcessor", "BitImageProcessorFast")),
-            ("fuyu", ("FuyuImageProcessor",)),
+            ("fuyu", ("FuyuImageProcessor", None)),
             ("gemma3", ("Gemma3ImageProcessor", "Gemma3ImageProcessorFast")),
             ("gemma3n", ("SiglipImageProcessor", "SiglipImageProcessorFast")),
             ("git", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("glm4v", ("Glm4vImageProcessor", "Glm4vImageProcessorFast")),
-            ("glpn", ("GLPNImageProcessor",)),
+            ("glpn", ("GLPNImageProcessor", None)),
             ("got_ocr2", ("GotOcr2ImageProcessor", "GotOcr2ImageProcessorFast")),
             ("grounding-dino", ("GroundingDinoImageProcessor", "GroundingDinoImageProcessorFast")),
             ("groupvit", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("hiera", ("BitImageProcessor", "BitImageProcessorFast")),
-            ("idefics", ("IdeficsImageProcessor",)),
+            ("idefics", ("IdeficsImageProcessor", None)),
             ("idefics2", ("Idefics2ImageProcessor", "Idefics2ImageProcessorFast")),
             ("idefics3", ("Idefics3ImageProcessor", "Idefics3ImageProcessorFast")),
             ("ijepa", ("ViTImageProcessor", "ViTImageProcessorFast")),
-            ("imagegpt", ("ImageGPTImageProcessor",)),
+            ("imagegpt", ("ImageGPTImageProcessor", None)),
             ("instructblip", ("BlipImageProcessor", "BlipImageProcessorFast")),
-            ("instructblipvideo", ("InstructBlipVideoImageProcessor",)),
+            ("instructblipvideo", ("InstructBlipVideoImageProcessor", None)),
             ("janus", ("JanusImageProcessor", "JanusImageProcessorFast")),
             ("kosmos-2", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("layoutlmv2", ("LayoutLMv2ImageProcessor", "LayoutLMv2ImageProcessorFast")),
             ("layoutlmv3", ("LayoutLMv3ImageProcessor", "LayoutLMv3ImageProcessorFast")),
             ("levit", ("LevitImageProcessor", "LevitImageProcessorFast")),
-            ("lightglue", ("LightGlueImageProcessor",)),
+            ("lightglue", ("LightGlueImageProcessor", None)),
             ("llama4", ("Llama4ImageProcessor", "Llama4ImageProcessorFast")),
             ("llava", ("LlavaImageProcessor", "LlavaImageProcessorFast")),
             ("llava_next", ("LlavaNextImageProcessor", "LlavaNextImageProcessorFast")),
-            ("llava_next_video", ("LlavaNextVideoImageProcessor",)),
+            ("llava_next_video", ("LlavaNextVideoImageProcessor", None)),
             ("llava_onevision", ("LlavaOnevisionImageProcessor", "LlavaOnevisionImageProcessorFast")),
             ("mask2former", ("Mask2FormerImageProcessor", "Mask2FormerImageProcessorFast")),
             ("maskformer", ("MaskFormerImageProcessor", "MaskFormerImageProcessorFast")),
             ("mgp-str", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("mistral3", ("PixtralImageProcessor", "PixtralImageProcessorFast")),
             ("mlcd", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
-            ("mllama", ("MllamaImageProcessor",)),
+            ("mllama", ("MllamaImageProcessor", None)),
             ("mm-grounding-dino", ("GroundingDinoImageProcessor", "GroundingDinoImageProcessorFast")),
             ("mobilenet_v1", ("MobileNetV1ImageProcessor", "MobileNetV1ImageProcessorFast")),
             ("mobilenet_v2", ("MobileNetV2ImageProcessor", "MobileNetV2ImageProcessorFast")),
@@ -142,12 +142,12 @@ else:
             ("owlvit", ("OwlViTImageProcessor", "OwlViTImageProcessorFast")),
             ("paligemma", ("SiglipImageProcessor", "SiglipImageProcessorFast")),
             ("perceiver", ("PerceiverImageProcessor", "PerceiverImageProcessorFast")),
-            ("perception_lm", ("PerceptionLMImageProcessorFast",)),
+            ("perception_lm", (None, "PerceptionLMImageProcessorFast")),
-            ("phi4_multimodal", ("Phi4MultimodalImageProcessorFast",)),
+            ("phi4_multimodal", (None, "Phi4MultimodalImageProcessorFast")),
-            ("pix2struct", ("Pix2StructImageProcessor",)),
+            ("pix2struct", ("Pix2StructImageProcessor", None)),
             ("pixtral", ("PixtralImageProcessor", "PixtralImageProcessorFast")),
             ("poolformer", ("PoolFormerImageProcessor", "PoolFormerImageProcessorFast")),
-            ("prompt_depth_anything", ("PromptDepthAnythingImageProcessor",)),
+            ("prompt_depth_anything", ("PromptDepthAnythingImageProcessor", None)),
             ("pvt", ("PvtImageProcessor", "PvtImageProcessorFast")),
             ("pvt_v2", ("PvtImageProcessor", "PvtImageProcessorFast")),
             ("qwen2_5_vl", ("Qwen2VLImageProcessor", "Qwen2VLImageProcessorFast")),
@@ -157,39 +157,31 @@ else:
             ("rt_detr", ("RTDetrImageProcessor", "RTDetrImageProcessorFast")),
             ("sam", ("SamImageProcessor", "SamImageProcessorFast")),
             ("sam_hq", ("SamImageProcessor", "SamImageProcessorFast")),
-            ("segformer", ("SegformerImageProcessor",)),
             ("segformer", ("SegformerImageProcessor", "SegformerImageProcessorFast")),
-            ("seggpt", ("SegGptImageProcessor",)),
+            ("seggpt", ("SegGptImageProcessor", None)),
             ("shieldgemma2", ("Gemma3ImageProcessor", "Gemma3ImageProcessorFast")),
             ("siglip", ("SiglipImageProcessor", "SiglipImageProcessorFast")),
             ("siglip2", ("Siglip2ImageProcessor", "Siglip2ImageProcessorFast")),
             ("smolvlm", ("SmolVLMImageProcessor", "SmolVLMImageProcessorFast")),
-            ("superglue", ("SuperGlueImageProcessor",)),
+            ("superglue", ("SuperGlueImageProcessor", None)),
-            (
+            ("superpoint", ("SuperPointImageProcessor", "SuperPointImageProcessorFast")),
-                "superpoint",
-                (
-                    "SuperPointImageProcessor",
-                    "SuperPointImageProcessorFast",
-                ),
-            ),
             ("swiftformer", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("swin", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("swin2sr", ("Swin2SRImageProcessor", "Swin2SRImageProcessorFast")),
             ("swinv2", ("ViTImageProcessor", "ViTImageProcessorFast")),
-            ("table-transformer", ("DetrImageProcessor",)),
+            ("table-transformer", ("DetrImageProcessor", None)),
-            ("timesformer", ("VideoMAEImageProcessor",)),
+            ("timesformer", ("VideoMAEImageProcessor", None)),
-            ("timm_wrapper", ("TimmWrapperImageProcessor",)),
+            ("timm_wrapper", ("TimmWrapperImageProcessor", None)),
-            ("tvlt", ("TvltImageProcessor",)),
+            ("tvlt", ("TvltImageProcessor", None)),
-            ("tvp", ("TvpImageProcessor",)),
+            ("tvp", ("TvpImageProcessor", None)),
             ("udop", ("LayoutLMv3ImageProcessor", "LayoutLMv3ImageProcessorFast")),
-            ("udop", ("LayoutLMv3ImageProcessor",)),
             ("upernet", ("SegformerImageProcessor", "SegformerImageProcessorFast")),
             ("van", ("ConvNextImageProcessor", "ConvNextImageProcessorFast")),
-            ("videomae", ("VideoMAEImageProcessor",)),
+            ("videomae", ("VideoMAEImageProcessor", None)),
             ("vilt", ("ViltImageProcessor", "ViltImageProcessorFast")),
             ("vipllava", ("CLIPImageProcessor", "CLIPImageProcessorFast")),
             ("vit", ("ViTImageProcessor", "ViTImageProcessorFast")),
-            ("vit_hybrid", ("ViTHybridImageProcessor",)),
+            ("vit_hybrid", ("ViTHybridImageProcessor", None)),
             ("vit_mae", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("vit_msn", ("ViTImageProcessor", "ViTImageProcessorFast")),
             ("vitmatte", ("VitMatteImageProcessor", "VitMatteImageProcessorFast")),
@@ -199,18 +191,14 @@ else:
         ]
     )
 
-for model_type, image_processors in IMAGE_PROCESSOR_MAPPING_NAMES.items():
+# Override to None if the packages are not available
-    slow_image_processor_class, *fast_image_processor_class = image_processors
+for model_type, (slow_class, fast_class) in IMAGE_PROCESSOR_MAPPING_NAMES.items():
     if not is_vision_available():
-        slow_image_processor_class = None
+        slow_class = None
+    if not is_torchvision_available():
+        fast_class = None
 
-    # If the fast image processor is not defined, or torchvision is not available, we set it to None
+    IMAGE_PROCESSOR_MAPPING_NAMES[model_type] = (slow_class, fast_class)
-    if not fast_image_processor_class or fast_image_processor_class[0] is None or not is_torchvision_available():
-        fast_image_processor_class = None
-    else:
-        fast_image_processor_class = fast_image_processor_class[0]
-
-    IMAGE_PROCESSOR_MAPPING_NAMES[model_type] = (slow_image_processor_class, fast_image_processor_class)
 
 IMAGE_PROCESSOR_MAPPING = _LazyAutoMapping(CONFIG_MAPPING_NAMES, IMAGE_PROCESSOR_MAPPING_NAMES)
 

src/transformers/utils/__init__.py

@@ -179,6 +179,7 @@ from .import_utils import (
     is_keras_nlp_available,
     is_kernels_available,
     is_levenshtein_available,
+    is_libcst_available,
     is_librosa_available,
     is_liger_kernel_available,
     is_lomo_available,

src/transformers/utils/import_utils.py

@@ -120,6 +120,7 @@ _vptq_available, _vptq_version = _is_package_available("vptq", return_version=Tr
 _av_available = importlib.util.find_spec("av") is not None
 _decord_available = importlib.util.find_spec("decord") is not None
 _torchcodec_available = importlib.util.find_spec("torchcodec") is not None
+_libcst_available = _is_package_available("libcst")
 _bitsandbytes_available = _is_package_available("bitsandbytes")
 _eetq_available = _is_package_available("eetq")
 _fbgemm_gpu_available = _is_package_available("fbgemm_gpu")
@@ -379,6 +380,10 @@ def is_torch_available():
     return _torch_available
 
 
+def is_libcst_available():
+    return _libcst_available
+
+
 def is_accelerate_available(min_version: str = ACCELERATE_MIN_VERSION):
     return _accelerate_available and version.parse(_accelerate_version) >= version.parse(min_version)
 

tests/models/aria/test_processing_aria.py

@@ -95,7 +95,7 @@ class AriaProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     def tearDownClass(cls):
         shutil.rmtree(cls.tmpdirname, ignore_errors=True)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/aya_vision/test_processing_aya_vision.py

@@ -80,7 +80,7 @@ class AyaVisionProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     def tearDownClass(cls):
         shutil.rmtree(cls.tmpdirname, ignore_errors=True)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/chameleon/test_processing_chameleon.py

@@ -75,7 +75,7 @@ class ChameleonProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     def prepare_processor_dict():
         return {"image_seq_length": 2}  # fmt: skip
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/clvp/test_processing_clvp.py

@@ -35,15 +35,15 @@ class ClvpProcessorTest(unittest.TestCase):
         shutil.rmtree(self.tmpdirname)
         gc.collect()
 
-    # Copied from transformers.tests.models.whisper.test_processor_whisper.WhisperProcessorTest.get_tokenizer with Whisper->Clvp
+    # Copied from transformers.tests.models.whisper.test_processing_whisper.WhisperProcessorTest.get_tokenizer with Whisper->Clvp
     def get_tokenizer(self, **kwargs):
         return ClvpTokenizer.from_pretrained(self.checkpoint, **kwargs)
 
-    # Copied from transformers.tests.models.whisper.test_processor_whisper.WhisperProcessorTest.get_feature_extractor with Whisper->Clvp
+    # Copied from transformers.tests.models.whisper.test_processing_whisper.WhisperProcessorTest.get_feature_extractor with Whisper->Clvp
     def get_feature_extractor(self, **kwargs):
         return ClvpFeatureExtractor.from_pretrained(self.checkpoint, **kwargs)
 
-    # Copied from transformers.tests.models.whisper.test_processor_whisper.WhisperProcessorTest.test_save_load_pretrained_default with Whisper->Clvp
+    # Copied from transformers.tests.models.whisper.test_processing_whisper.WhisperProcessorTest.test_save_load_pretrained_default with Whisper->Clvp
     def test_save_load_pretrained_default(self):
         tokenizer = self.get_tokenizer()
         feature_extractor = self.get_feature_extractor()
@@ -59,7 +59,7 @@ class ClvpProcessorTest(unittest.TestCase):
         self.assertEqual(processor.feature_extractor.to_json_string(), feature_extractor.to_json_string())
         self.assertIsInstance(processor.feature_extractor, ClvpFeatureExtractor)
 
-    # Copied from transformers.tests.models.whisper.test_processor_whisper.WhisperProcessorTest.test_feature_extractor with Whisper->Clvp,processor(raw_speech->processor(raw_speech=raw_speech
+    # Copied from transformers.tests.models.whisper.test_processing_whisper.WhisperProcessorTest.test_feature_extractor with Whisper->Clvp,processor(raw_speech->processor(raw_speech=raw_speech
     def test_feature_extractor(self):
         feature_extractor = self.get_feature_extractor()
         tokenizer = self.get_tokenizer()
@@ -74,7 +74,7 @@ class ClvpProcessorTest(unittest.TestCase):
         for key in input_feat_extract:
             self.assertAlmostEqual(input_feat_extract[key].sum(), input_processor[key].sum(), delta=1e-2)
 
-    # Copied from transformers.tests.models.whisper.test_processor_whisper.WhisperProcessorTest.test_tokenizer with Whisper->Clvp
+    # Copied from transformers.tests.models.whisper.test_processing_whisper.WhisperProcessorTest.test_tokenizer with Whisper->Clvp
     def test_tokenizer(self):
         feature_extractor = self.get_feature_extractor()
         tokenizer = self.get_tokenizer()
@@ -90,7 +90,7 @@ class ClvpProcessorTest(unittest.TestCase):
         for key in encoded_tok:
             self.assertListEqual(encoded_tok[key], encoded_processor[key])
 
-    # Copied from transformers.tests.models.whisper.test_processor_whisper.WhisperProcessorTest.test_tokenizer_decode with Whisper->Clvp
+    # Copied from transformers.tests.models.whisper.test_processing_whisper.WhisperProcessorTest.test_tokenizer_decode with Whisper->Clvp
     def test_tokenizer_decode(self):
         feature_extractor = self.get_feature_extractor()
         tokenizer = self.get_tokenizer()

tests/models/colpali/test_processing_colpali.py

@@ -54,7 +54,7 @@ class ColPaliProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     def tearDownClass(cls):
         shutil.rmtree(cls.tmpdirname, ignore_errors=True)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/colqwen2/test_processing_colqwen2.py

@@ -57,7 +57,7 @@ class ColQwen2ProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     def tearDownClass(cls):
         shutil.rmtree(cls.tmpdirname)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/emu3/test_processing_emu3.py

@@ -91,7 +91,7 @@ class Emu3ProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         # For an image where pixels go from 0 to 255 the diff can be 1 due to some numerical precision errors when scaling and unscaling
         self.assertTrue(np.abs(orig_image - unnormalized_images).max() >= 1)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/fuyu/test_processing_fuyu.py

@@ -64,7 +64,7 @@ class FuyuProcessingTest(ProcessorTesterMixin, unittest.TestCase):
     def get_image_processor(self, **kwargs):
         return AutoProcessor.from_pretrained(self.tmpdirname, **kwargs).image_processor
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/gemma3/test_processing_gemma3.py

@@ -58,7 +58,7 @@ class Gemma3ProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         processor.save_pretrained(cls.tmpdirname)
         cls.image_token = processor.boi_token
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/grounding_dino/test_processing_grounding_dino.py

@@ -94,17 +94,17 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         return [labels, labels_longer] + [labels] * (batch_size - 2)
 
     @classmethod
-    # Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.get_tokenizer with CLIP->Bert
+    # Copied from tests.models.clip.test_processing_clip.CLIPProcessorTest.get_tokenizer with CLIP->Bert
     def get_tokenizer(cls, **kwargs):
         return BertTokenizer.from_pretrained(cls.tmpdirname, **kwargs)
 
     @classmethod
-    # Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.get_rust_tokenizer with CLIP->Bert
+    # Copied from tests.models.clip.test_processing_clip.CLIPProcessorTest.get_rust_tokenizer with CLIP->Bert
     def get_rust_tokenizer(cls, **kwargs):
         return BertTokenizerFast.from_pretrained(cls.tmpdirname, **kwargs)
 
     @classmethod
-    # Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.get_image_processor with CLIP->GroundingDino
+    # Copied from tests.models.clip.test_processing_clip.CLIPProcessorTest.get_image_processor with CLIP->GroundingDino
     def get_image_processor(cls, **kwargs):
         return GroundingDinoImageProcessor.from_pretrained(cls.tmpdirname, **kwargs)
 
@@ -145,7 +145,7 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         expected_box_slice = torch.tensor([0.6908, 0.4354, 1.0737, 1.3947])
         torch.testing.assert_close(post_processed[0]["boxes"][0], expected_box_slice, rtol=1e-4, atol=1e-4)
 
-    # Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.test_save_load_pretrained_default with CLIP->GroundingDino,GroundingDinoTokenizer->BertTokenizer
+    # Copied from tests.models.clip.test_processing_clip.CLIPProcessorTest.test_save_load_pretrained_default with CLIP->GroundingDino,GroundingDinoTokenizer->BertTokenizer
     def test_save_load_pretrained_default(self):
         tokenizer_slow = self.get_tokenizer()
         tokenizer_fast = self.get_rust_tokenizer()
@@ -171,7 +171,7 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         self.assertIsInstance(processor_slow.image_processor, GroundingDinoImageProcessor)
         self.assertIsInstance(processor_fast.image_processor, GroundingDinoImageProcessor)
 
-    # Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.test_save_load_pretrained_additional_features with CLIP->GroundingDino,GroundingDinoTokenizer->BertTokenizer
+    # Copied from tests.models.clip.test_processing_clip.CLIPProcessorTest.test_save_load_pretrained_additional_features with CLIP->GroundingDino,GroundingDinoTokenizer->BertTokenizer
     def test_save_load_pretrained_additional_features(self):
         with tempfile.TemporaryDirectory() as tmpdir:
             processor = GroundingDinoProcessor(
@@ -194,7 +194,7 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         self.assertEqual(processor.image_processor.to_json_string(), image_processor_add_kwargs.to_json_string())
         self.assertIsInstance(processor.image_processor, GroundingDinoImageProcessor)
 
-    # Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.test_image_processor with CLIP->GroundingDino
+    # Copied from tests.models.clip.test_processing_clip.CLIPProcessorTest.test_image_processor with CLIP->GroundingDino
     def test_image_processor(self):
         image_processor = self.get_image_processor()
         tokenizer = self.get_tokenizer()
@@ -209,7 +209,7 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         for key in input_image_proc:
             self.assertAlmostEqual(input_image_proc[key].sum(), input_processor[key].sum(), delta=1e-2)
 
-    # Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.test_tokenizer with CLIP->GroundingDino
+    # Copied from tests.models.clip.test_processing_clip.CLIPProcessorTest.test_tokenizer with CLIP->GroundingDino
     def test_tokenizer(self):
         image_processor = self.get_image_processor()
         tokenizer = self.get_tokenizer()
@@ -244,7 +244,7 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         with pytest.raises(ValueError):
             processor()
 
-    # Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.test_tokenizer_decode with CLIP->GroundingDino
+    # Copied from tests.models.clip.test_processing_clip.CLIPProcessorTest.test_tokenizer_decode with CLIP->GroundingDino
     def test_tokenizer_decode(self):
         image_processor = self.get_image_processor()
         tokenizer = self.get_tokenizer()
@@ -258,7 +258,7 @@ class GroundingDinoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
 
         self.assertListEqual(decoded_tok, decoded_processor)
 
-    # Copied from tests.models.clip.test_processor_clip.CLIPProcessorTest.test_model_input_names with CLIP->GroundingDino
+    # Copied from tests.models.clip.test_processing_clip.CLIPProcessorTest.test_model_input_names with CLIP->GroundingDino
     def test_model_input_names(self):
         image_processor = self.get_image_processor()
         tokenizer = self.get_tokenizer()

tests/models/idefics3/test_processing_idefics3.py

@@ -84,7 +84,7 @@ class Idefics3ProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     def prepare_processor_dict():
         return {"image_seq_len": 2}
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 
@@ -284,7 +284,7 @@ class Idefics3ProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         self.assertEqual(np.array(inputs["pixel_values"]).shape, (2, 2, 3, 364, 364))
         self.assertEqual(np.array(inputs["pixel_attention_mask"]).shape, (2, 2, 364, 364))
 
-    # Copied from tests.models.idefics2.test_processor_idefics2.Idefics2ProcessorTest.test_process_interleaved_images_prompts_image_error
+    # Copied from tests.models.idefics2.test_processing_idefics2.Idefics2ProcessorTest.test_process_interleaved_images_prompts_image_error
     def test_process_interleaved_images_prompts_image_error(self):
         processor = self.get_processor()
 

tests/models/internvl/test_processing_internvl.py

@@ -97,7 +97,7 @@ class InternVLProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     def tearDownClass(cls):
         shutil.rmtree(cls.tmpdirname, ignore_errors=True)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/llava_next/test_processing_llava_next.py

@@ -66,7 +66,7 @@ class LlavaNextProcessorTest(ProcessorTesterMixin, unittest.TestCase):
             "vision_feature_select_strategy": "default"
         }  # fmt: skip
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 
@@ -79,7 +79,7 @@ class LlavaNextProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         self.assertTrue("num_image_patches" in output)
         self.assertEqual(len(output["num_image_patches"]), 3)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_chat_template_is_saved
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_chat_template_is_saved
     def test_chat_template_is_saved(self):
         processor_loaded = self.processor_class.from_pretrained(self.tmpdirname)
         processor_dict_loaded = json.loads(processor_loaded.to_json_string())

tests/models/llava_next_video/test_processing_llava_next_video.py

@@ -75,7 +75,7 @@ class LlavaNextVideoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
             "vision_feature_select_strategy": "default",
         }
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 
@@ -88,7 +88,7 @@ class LlavaNextVideoProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         self.assertTrue("num_image_patches" in output)
         self.assertEqual(len(output["num_image_patches"]), 3)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_chat_template_is_saved
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_chat_template_is_saved
     def test_chat_template_is_saved(self):
         processor_loaded = self.processor_class.from_pretrained(self.tmpdirname)
         processor_dict_loaded = json.loads(processor_loaded.to_json_string())

tests/models/llava_onevision/test_processing_llava_onevision.py

@@ -79,7 +79,7 @@ class LlavaOnevisionProcessorTest(ProcessorTesterMixin, unittest.TestCase):
             "vision_feature_select_strategy": "default"
         }  # fmt: skip
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 
@@ -92,7 +92,7 @@ class LlavaOnevisionProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         self.assertTrue("num_image_patches" in output)
         self.assertEqual(len(output["num_image_patches"]), 3)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_chat_template_is_saved
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_chat_template_is_saved
     def test_chat_template_is_saved(self):
         processor_loaded = self.processor_class.from_pretrained(self.tmpdirname)
         processor_dict_loaded = json.loads(processor_loaded.to_json_string())

tests/models/musicgen_melody/test_processing_musicgen_melody.py

@@ -50,7 +50,7 @@ def floats_list(shape, scale=1.0, rng=None, name=None):
 @require_torch
 @require_sentencepiece
 @require_torchaudio
-# Copied from tests.models.musicgen.test_processor_musicgen.MusicgenProcessorTest with Musicgen->MusicgenMelody, Encodec->MusicgenMelody, padding_mask->attention_mask, input_values->input_features
+# Copied from tests.models.musicgen.test_processing_musicgen.MusicgenProcessorTest with Musicgen->MusicgenMelody, Encodec->MusicgenMelody, padding_mask->attention_mask, input_values->input_features
 class MusicgenMelodyProcessorTest(unittest.TestCase):
     def setUp(self):
         # Ignore copy

tests/models/paligemma/test_processing_paligemma.py

@@ -48,7 +48,7 @@ class PaliGemmaProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     def tearDownClass(cls):
         shutil.rmtree(cls.tmpdirname, ignore_errors=True)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/qwen2_5_vl/test_processing_qwen2_5_vl.py

@@ -65,7 +65,7 @@ class Qwen2_5_VLProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     def tearDownClass(cls):
         shutil.rmtree(cls.tmpdirname, ignore_errors=True)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/qwen2_vl/test_processing_qwen2_vl.py

@@ -68,7 +68,7 @@ class Qwen2VLProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     def tearDownClass(cls):
         shutil.rmtree(cls.tmpdirname, ignore_errors=True)
 
-    # Copied from tests.models.llava.test_processor_llava.LlavaProcessorTest.test_get_num_vision_tokens
+    # Copied from tests.models.llava.test_processing_llava.LlavaProcessorTest.test_get_num_vision_tokens
     def test_get_num_vision_tokens(self):
         "Tests general functionality of the helper used internally in vLLM"
 

tests/models/seamless_m4t/test_processing_seamless_m4t.py

@@ -76,7 +76,7 @@ class SeamlessM4TProcessorTest(unittest.TestCase):
         tokenizer_instance = isinstance(processor.tokenizer, (SeamlessM4TTokenizerFast, SeamlessM4TTokenizer))
         self.assertTrue(tokenizer_instance)
 
-    # Copied from test.models.whisper.test_processor_whisper.WhisperProcessorTest.test_feature_extractor with Whisper->SeamlessM4T
+    # Copied from test.models.whisper.test_processing_whisper.WhisperProcessorTest.test_feature_extractor with Whisper->SeamlessM4T
     def test_feature_extractor(self):
         feature_extractor = self.get_feature_extractor()
         tokenizer = self.get_tokenizer()
@@ -91,7 +91,7 @@ class SeamlessM4TProcessorTest(unittest.TestCase):
         for key in input_feat_extract:
             self.assertAlmostEqual(input_feat_extract[key].sum(), input_processor[key].sum(), delta=1e-2)
 
-    # Copied from test.models.whisper.test_processor_whisper.WhisperProcessorTest.test_tokenizer with Whisper->SeamlessM4T
+    # Copied from test.models.whisper.test_processing_whisper.WhisperProcessorTest.test_tokenizer with Whisper->SeamlessM4T
     def test_tokenizer(self):
         feature_extractor = self.get_feature_extractor()
         tokenizer = self.get_tokenizer()
@@ -107,7 +107,7 @@ class SeamlessM4TProcessorTest(unittest.TestCase):
         for key in encoded_tok:
             self.assertListEqual(encoded_tok[key], encoded_processor[key])
 
-    # Copied from test.models.whisper.test_processor_whisper.WhisperProcessorTest.test_tokenizer_decode with Whisper->SeamlessM4T
+    # Copied from test.models.whisper.test_processing_whisper.WhisperProcessorTest.test_tokenizer_decode with Whisper->SeamlessM4T
     def test_tokenizer_decode(self):
         feature_extractor = self.get_feature_extractor()
         tokenizer = self.get_tokenizer()

tests/models/smolvlm/test_processing_smolvlm.py

@@ -297,7 +297,7 @@ class SmolVLMProcessorTest(ProcessorTesterMixin, unittest.TestCase):
         self.assertEqual(np.array(inputs["pixel_values"]).shape, (2, 2, 3, 512, 512))
         self.assertEqual(np.array(inputs["pixel_attention_mask"]).shape, (2, 2, 512, 512))
 
-    # Copied from tests.models.idefics2.test_processor_idefics2.Idefics2ProcessorTest.test_process_interleaved_images_prompts_image_error
+    # Copied from tests.models.idefics2.test_processing_idefics2.Idefics2ProcessorTest.test_process_interleaved_images_prompts_image_error
     def test_process_interleaved_images_prompts_image_error(self):
         processor = self.get_processor()
 

utils/modular_model_converter.py

@@ -17,12 +17,12 @@ import glob
 import importlib
 import os
 import re
+import subprocess
 from abc import ABC, abstractmethod
 from collections import Counter, defaultdict, deque
 from typing import Optional, Union
 
 import libcst as cst
-from check_copies import run_ruff
 from create_dependency_mapping import find_priority_list
 from libcst import ClassDef, CSTVisitor
 from libcst import matchers as m
@@ -1676,6 +1676,16 @@ def create_modules(modular_mapper: ModularFileMapper) -> dict[str, cst.Module]:
     return files
 
 
+def run_ruff(code, check=False):
+    if check:
+        command = ["ruff", "check", "-", "--fix", "--exit-zero"]
+    else:
+        command = ["ruff", "format", "-", "--config", "pyproject.toml", "--silent"]
+    process = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE)
+    stdout, _ = process.communicate(input=code.encode())
+    return stdout.decode()
+
+
 def convert_modular_file(modular_file):
     pattern = re.search(r"modular_(.*)(?=\.py$)", modular_file)
     output = {}