rename all test_processing_*.py to test_processor_*.py (#33878)

* rename all test_processing_*.py to test_processor_*.py ans fix duplicate test processor paligemma

* fix copies

* fix broken tests

* fix-copies

* fix test processor bridgetower

src/transformers/models/omdet_turbo/modeling_omdet_turbo.py

@@ -21,7 +21,7 @@
 from dataclasses import dataclass
 from functools import lru_cache
 from pathlib import Path
-from typing import Optional, Tuple, Union
+from typing import List, Optional, Tuple, Union
 
 import torch
 import torch.nn.functional as F
@@ -208,7 +208,10 @@ def load_cuda_kernels():
 
 # Copied from transformers.models.deformable_detr.modeling_deformable_detr.multi_scale_deformable_attention
 def multi_scale_deformable_attention(
-    value: Tensor, value_spatial_shapes: Tensor, sampling_locations: Tensor, attention_weights: Tensor
+    value: Tensor,
+    value_spatial_shapes: Union[Tensor, List[Tuple]],
+    sampling_locations: Tensor,
+    attention_weights: Tensor,
 ) -> Tensor:
     batch_size, _, num_heads, hidden_dim = value.shape
     _, num_queries, num_heads, num_levels, num_points, _ = sampling_locations.shape

tests/models/blip/test_processor_blip.py

@@ -152,7 +152,7 @@ def test_unstructured_kwargs_batched(self):
         self.skip_processor_without_typed_kwargs(processor)
 
         input_str = ["lower newer", "upper older longer string"]
-        image_input = self.prepare_image_inputs() * 2
+        image_input = self.prepare_image_inputs(batch_size=2)
         inputs = processor(
             text=input_str,
             images=image_input,

tests/models/blip_2/test_processor_blip_2.py

@@ -17,7 +17,7 @@
 
 import pytest
 
-from transformers.testing_utils import require_torch, require_vision
+from transformers.testing_utils import require_vision
 from transformers.utils import is_vision_available
 
 from ...test_processing_common import ProcessorTesterMixin
@@ -139,30 +139,3 @@ def test_model_input_names(self):
 
         # For now the processor supports only ['pixel_values', 'input_ids', 'attention_mask']
         self.assertCountEqual(list(inputs.keys()), ["input_ids", "pixel_values", "attention_mask"])
-
-    @require_torch
-    @require_vision
-    def test_unstructured_kwargs_batched(self):
-        if "image_processor" not in self.processor_class.attributes:
-            self.skipTest(f"image_processor attribute not present in {self.processor_class}")
-        image_processor = self.get_component("image_processor")
-        tokenizer = self.get_component("tokenizer")
-        if not tokenizer.pad_token:
-            tokenizer.pad_token = "[TEST_PAD]"
-        processor = self.processor_class(tokenizer=tokenizer, image_processor=image_processor)
-        self.skip_processor_without_typed_kwargs(processor)
-
-        input_str = ["lower newer", "upper older longer string"]
-        image_input = self.prepare_image_inputs() * 2
-        inputs = processor(
-            text=input_str,
-            images=image_input,
-            return_tensors="pt",
-            crop_size={"height": 214, "width": 214},
-            size={"height": 214, "width": 214},
-            padding="longest",
-            max_length=76,
-        )
-        self.assertEqual(inputs["pixel_values"].shape[2], 214)
-
-        self.assertEqual(len(inputs["input_ids"][0]), 11)

tests/models/bridgetower/test_processing_bridgetower.py → tests/models/bridgetower/test_processor_bridgetower.py

@@ -15,17 +15,13 @@
 import tempfile
 import unittest
 
-import numpy as np
-
 from transformers.testing_utils import require_torch, require_vision
 from transformers.utils import is_vision_available
 
 from ...test_processing_common import ProcessorTesterMixin
 
 
 if is_vision_available():
-    from PIL import Image
-
     from transformers import (
         AutoProcessor,
         BridgeTowerImageProcessor,
@@ -35,7 +31,7 @@
 
 
 @require_vision
-class Blip2ProcessorTest(ProcessorTesterMixin, unittest.TestCase):
+class BridgeTowerProcessorTest(ProcessorTesterMixin, unittest.TestCase):
     processor_class = BridgeTowerProcessor
 
     def setUp(self):
@@ -57,17 +53,6 @@ def get_image_processor(self, **kwargs):
     def tearDown(self):
         shutil.rmtree(self.tmpdirname)
 
-    def prepare_image_inputs(self):
-        """This function prepares a list of PIL images, or a list of numpy arrays if one specifies numpify=True,
-        or a list of PyTorch tensors if one specifies torchify=True.
-        """
-
-        image_inputs = [np.random.randint(255, size=(3, 30, 400), dtype=np.uint8)]
-
-        image_inputs = [Image.fromarray(np.moveaxis(x, 0, -1)) for x in image_inputs]
-
-        return image_inputs
-
     # Some kwargs tests are overriden from common tests to handle shortest_edge
     # and size_divisor behaviour
 
@@ -149,7 +134,7 @@ def test_unstructured_kwargs_batched(self):
         self.skip_processor_without_typed_kwargs(processor)
 
         input_str = ["lower newer", "upper older longer string"]
-        image_input = self.prepare_image_inputs() * 2
+        image_input = self.prepare_image_inputs(batch_size=2)
         inputs = processor(
             text=input_str,
             images=image_input,

tests/models/donut/test_processing_donut.py → tests/models/donut/test_processor_donut.py

@@ -18,10 +18,6 @@
 import unittest
 
 from transformers import DonutImageProcessor, DonutProcessor, XLMRobertaTokenizerFast
-from transformers.testing_utils import (
-    require_torch,
-    require_vision,
-)
 
 from ...test_processing_common import ProcessorTesterMixin
 
@@ -65,30 +61,3 @@ def test_token2json(self):
         actual_json = self.processor.token2json(sequence)
 
         self.assertDictEqual(actual_json, expected_json)
-
-    @require_torch
-    @require_vision
-    def test_unstructured_kwargs_batched(self):
-        if "image_processor" not in self.processor_class.attributes:
-            self.skipTest(f"image_processor attribute not present in {self.processor_class}")
-        image_processor = self.get_component("image_processor")
-        tokenizer = self.get_component("tokenizer")
-        if not tokenizer.pad_token:
-            tokenizer.pad_token = "[TEST_PAD]"
-        processor = self.processor_class(tokenizer=tokenizer, image_processor=image_processor)
-        self.skip_processor_without_typed_kwargs(processor)
-
-        input_str = ["lower newer", "upper older longer string"]
-        image_input = self.prepare_image_inputs() * 2
-        inputs = processor(
-            text=input_str,
-            images=image_input,
-            return_tensors="pt",
-            crop_size={"height": 214, "width": 214},
-            size={"height": 214, "width": 214},
-            padding="longest",
-            max_length=76,
-        )
-        self.assertEqual(inputs["pixel_values"].shape[2], 214)
-
-        self.assertEqual(len(inputs["input_ids"][0]), 7)

tests/models/musicgen_melody/test_processor_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_processing_musicgen.MusicgenProcessorTest with Musicgen->MusicgenMelody, Encodec->MusicgenMelody, padding_mask->attention_mask, input_values->input_features
+# Copied from tests.models.musicgen.test_processor_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_processor_paligemma.py

@@ -16,19 +16,15 @@
 import tempfile
 import unittest
 
-from transformers import GemmaTokenizer
+from transformers import GemmaTokenizer, PaliGemmaProcessor
 from transformers.testing_utils import get_tests_dir, require_torch, require_vision
 from transformers.utils import is_vision_available
 
 from ...test_processing_common import ProcessorTesterMixin
 
 
 if is_vision_available():
-    from transformers import (
-        PaliGemmaProcessor,
-        SiglipImageProcessor,
-        is_vision_available,
-    )
+    from transformers import SiglipImageProcessor
 
 SAMPLE_VOCAB = get_tests_dir("fixtures/test_sentencepiece.model")
 
@@ -61,3 +57,37 @@ def test_image_seq_length(self):
             text=input_str, images=image_input, return_tensors="pt", max_length=112, padding="max_length"
         )
         self.assertEqual(len(inputs["input_ids"][0]), 112 + 14)
+
+    def test_text_with_image_tokens(self):
+        image_processor = self.get_component("image_processor")
+        tokenizer = self.get_component("tokenizer")
+
+        processor = self.processor_class(tokenizer=tokenizer, image_processor=image_processor)
+        text_multi_images = "<image><image><bos>Dummy text!"
+        text_single_image = "<image><bos>Dummy text!"
+        text_no_image = "Dummy text!"
+
+        image = self.prepare_image_inputs()
+
+        out_noimage = processor(text=text_no_image, images=image, return_tensors="np")
+        out_singlimage = processor(text=text_single_image, images=image, return_tensors="np")
+        for k in out_noimage:
+            self.assertTrue(out_noimage[k].tolist() == out_singlimage[k].tolist())
+
+        out_multiimages = processor(text=text_multi_images, images=[image, image], return_tensors="np")
+        out_noimage = processor(text=text_no_image, images=[[image, image]], return_tensors="np")
+
+        # We can't be sure what is users intention, whether user want "one text + two images" or user forgot to add the second text
+        with self.assertRaises(ValueError):
+            out_noimage = processor(text=text_no_image, images=[image, image], return_tensors="np")
+
+        for k in out_noimage:
+            self.assertTrue(out_noimage[k].tolist() == out_multiimages[k].tolist())
+
+        text_batched = ["Dummy text!", "Dummy text!"]
+        text_batched_with_image = ["<image><bos>Dummy text!", "<image><bos>Dummy text!"]
+        out_images = processor(text=text_batched_with_image, images=[image, image], return_tensors="np")
+        out_noimage_nested = processor(text=text_batched, images=[[image], [image]], return_tensors="np")
+        out_noimage = processor(text=text_batched, images=[image, image], return_tensors="np")
+        for k in out_noimage:
+            self.assertTrue(out_noimage[k].tolist() == out_images[k].tolist() == out_noimage_nested[k].tolist())