[tests] add the missing `require_sacremoses` decorator (#29504)

* add sacremoses check

* fix style

* for FlaubertTokenizer

* HerbertTokenizer fix

* add typeHint

* Update src/transformers/testing_utils.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* make less skipped

* make quality

* remove import

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

src/transformers/__init__.py

@@ -1078,6 +1078,7 @@ _import_structure = {
         "is_psutil_available",
         "is_py3nvml_available",
         "is_pyctcdecode_available",
+        "is_sacremoses_available",
         "is_safetensors_available",
         "is_scipy_available",
         "is_sentencepiece_available",
@@ -5882,6 +5883,7 @@ if TYPE_CHECKING:
         is_psutil_available,
         is_py3nvml_available,
         is_pyctcdecode_available,
+        is_sacremoses_available,
         is_safetensors_available,
         is_scipy_available,
         is_sentencepiece_available,

src/transformers/testing_utils.py

@@ -90,6 +90,7 @@ from .utils import (
     is_pytest_available,
     is_pytorch_quantization_available,
     is_rjieba_available,
+    is_sacremoses_available,
     is_safetensors_available,
     is_scipy_available,
     is_sentencepiece_available,
@@ -562,6 +563,13 @@ def require_sentencepiece(test_case):
     return unittest.skipUnless(is_sentencepiece_available(), "test requires SentencePiece")(test_case)
 
 
+def require_sacremoses(test_case):
+    """
+    Decorator marking a test that requires Sacremoses. These tests are skipped when Sacremoses isn't installed.
+    """
+    return unittest.skipUnless(is_sacremoses_available(), "test requires Sacremoses")(test_case)
+
+
 def require_seqio(test_case):
     """
     Decorator marking a test that requires SentencePiece. These tests are skipped when SentencePiece isn't installed.

tests/models/biogpt/test_modeling_biogpt.py

@@ -17,7 +17,7 @@
 import math
 import unittest
 
-from transformers import BioGptConfig, is_torch_available
+from transformers import BioGptConfig, is_sacremoses_available, is_torch_available
 from transformers.testing_utils import require_torch, slow, torch_device
 
 from ...generation.test_utils import GenerationTesterMixin
@@ -294,7 +294,7 @@ class BioGptModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMix
             "token-classification": BioGptForTokenClassification,
             "zero-shot": BioGptForSequenceClassification,
         }
-        if is_torch_available()
+        if is_torch_available() and is_sacremoses_available()
         else {}
     )
     test_pruning = False

tests/models/biogpt/test_tokenization_biogpt.py

@@ -19,11 +19,12 @@ import os
 import unittest
 
 from transformers.models.biogpt.tokenization_biogpt import VOCAB_FILES_NAMES, BioGptTokenizer
-from transformers.testing_utils import slow
+from transformers.testing_utils import require_sacremoses, slow
 
 from ...test_tokenization_common import TokenizerTesterMixin
 
 
+@require_sacremoses
 class BioGptTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
     tokenizer_class = BioGptTokenizer
     test_rust_tokenizer = False

tests/models/flaubert/test_modeling_flaubert.py

@@ -16,7 +16,7 @@ import os
 import tempfile
 import unittest
 
-from transformers import FlaubertConfig, is_torch_available
+from transformers import FlaubertConfig, is_sacremoses_available, is_torch_available
 from transformers.testing_utils import require_torch, require_torch_accelerator, slow, torch_device
 
 from ...test_configuration_common import ConfigTester
@@ -386,7 +386,7 @@ class FlaubertModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase
             "token-classification": FlaubertForTokenClassification,
             "zero-shot": FlaubertForSequenceClassification,
         }
-        if is_torch_available()
+        if is_torch_available() and is_sacremoses_available()
         else {}
     )
 

tests/models/herbert/test_tokenization_herbert.py

@@ -20,11 +20,12 @@ import unittest
 
 from transformers import HerbertTokenizer, HerbertTokenizerFast
 from transformers.models.herbert.tokenization_herbert import VOCAB_FILES_NAMES
-from transformers.testing_utils import get_tests_dir, require_tokenizers, slow
+from transformers.testing_utils import get_tests_dir, require_sacremoses, require_tokenizers, slow
 
 from ...test_tokenization_common import TokenizerTesterMixin
 
 
+@require_sacremoses
 @require_tokenizers
 class HerbertTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
     tokenizer_class = HerbertTokenizer