1 files changed, 124 insertions, 0 deletions
diff --git a/sci-libs/datasets/files/datasets-2.17.1-tests.patch b/sci-libs/datasets/files/datasets-2.17.1-tests.patch
new file mode 100644
index 000000000000..14ae50602d10
--- /dev/null
+++ b/sci-libs/datasets/files/datasets-2.17.1-tests.patch
@@ -0,0 +1,124 @@
+--- a/tests/test_arrow_dataset.py	2024-02-20 21:53:24.248470991 +0100
++++ b/tests/test_arrow_dataset.py	2024-02-20 21:53:29.441804737 +0100
+@@ -4016,7 +4016,6 @@
+     [
+         "relative/path",
+         "/absolute/path",
+-        "s3://bucket/relative/path",
+         "hdfs://relative/path",
+         "hdfs:///absolute/path",
+     ],
+--- a/tests/test_load.py	2024-02-20 22:12:13.699209107 +0100
++++ b/tests/test_load.py	2024-02-20 22:13:10.862626708 +0100
+@@ -388,6 +388,7 @@
+             hf_modules_cache=self.hf_modules_cache,
+         )
+ 
++    @pytest.mark.skip(reason="")
+     def test_HubDatasetModuleFactoryWithScript_dont_trust_remote_code(self):
+         # "lhoestq/test" has a dataset script
+         factory = HubDatasetModuleFactoryWithScript(
+@@ -403,6 +404,7 @@
+         )
+         self.assertRaises(ValueError, factory.get_module)
+ 
++    @pytest.mark.skip()
+     def test_HubDatasetModuleFactoryWithScript_with_github_dataset(self):
+         # "wmt_t2t" has additional imports (internal)
+         factory = HubDatasetModuleFactoryWithScript(
+@@ -412,6 +414,7 @@
+         assert importlib.import_module(module_factory_result.module_path) is not None
+         assert module_factory_result.builder_kwargs["base_path"].startswith(config.HF_ENDPOINT)
+ 
++    @pytest.mark.skip()
+     def test_GithubMetricModuleFactory_with_internal_import(self):
+         # "squad_v2" requires additional imports (internal)
+         factory = GithubMetricModuleFactory(
+@@ -420,6 +423,7 @@
+         module_factory_result = factory.get_module()
+         assert importlib.import_module(module_factory_result.module_path) is not None
+ 
++    @pytest.mark.skip()
+     @pytest.mark.filterwarnings("ignore:GithubMetricModuleFactory is deprecated:FutureWarning")
+     def test_GithubMetricModuleFactory_with_external_import(self):
+         # "bleu" requires additional imports (external from github)
+@@ -1033,6 +1037,7 @@
+         datasets.load_dataset_builder(SAMPLE_DATASET_TWO_CONFIG_IN_METADATA, "non-existing-config")
+ 
+ 
++@pytest.mark.skip()
+ @pytest.mark.parametrize("serializer", [pickle, dill])
+ def test_load_dataset_builder_with_metadata_configs_pickable(serializer):
+     builder = datasets.load_dataset_builder(SAMPLE_DATASET_SINGLE_CONFIG_IN_METADATA)
+@@ -1154,6 +1159,7 @@
+     assert len(builder.config.data_files["test"]) > 0
+ 
+ 
++@pytest.mark.skip()
+ def test_load_dataset_builder_fail():
+     with pytest.raises(DatasetNotFoundError):
+         datasets.load_dataset_builder("blabla")
+@@ -1169,6 +1175,7 @@
+     assert isinstance(next(iter(dataset["train"])), dict)
+ 
+ 
++@pytest.mark.skip()
+ def test_load_dataset_cached_local_script(dataset_loading_script_dir, data_dir, caplog):
+     dataset = load_dataset(dataset_loading_script_dir, data_dir=data_dir)
+     assert isinstance(dataset, DatasetDict)
+--- a/tests/test_hf_gcp.py	2024-02-21 09:59:26.918397895 +0100
++++ b/tests/test_hf_gcp.py	2024-02-21 09:59:46.335100597 +0100
+@@ -45,6 +45,7 @@
+         ]
+ 
+ 
++@pytest.mark.skip("network")
+ @parameterized.named_parameters(list_datasets_on_hf_gcp_parameters(with_config=True))
+ class TestDatasetOnHfGcp(TestCase):
+     dataset = None
+--- a/tests/test_inspect.py	2024-02-21 10:03:32.315520016 +0100
++++ b/tests/test_inspect.py	2024-02-21 10:03:50.345553490 +0100
+@@ -49,6 +49,7 @@
+     assert list(info.splits.keys()) == expected_splits
+ 
+ 
++@pytest.mark.skip(reason="require network")
+ def test_get_dataset_config_info_private(hf_token, hf_private_dataset_repo_txt_data):
+     info = get_dataset_config_info(hf_private_dataset_repo_txt_data, config_name="default", token=hf_token)
+     assert list(info.splits.keys()) == ["train"]
+--- a/tests/test_data_files.py	2024-02-21 20:22:57.536160356 +0100
++++ b/tests/test_data_files.py	2024-02-21 20:25:00.153052174 +0100
+@@ -378,6 +378,7 @@
+         assert len(hub_dataset_repo_patterns_results[pattern]) == 0
+ 
+ 
++@pytest.mark.skip(reason="network")
+ def test_DataFilesList_from_patterns_locally_with_extra_files(complex_data_dir, text_file):
+     data_files_list = DataFilesList.from_patterns([_TEST_URL, text_file.as_posix()], complex_data_dir)
+     assert list(data_files_list) == [_TEST_URL, text_file.as_posix()]
+@@ -467,6 +468,7 @@
+         assert Hasher.hash(data_files1) != Hasher.hash(data_files2)
+ 
+ 
++@pytest.mark.skip(reason="network")
+ def test_DataFilesDict_from_patterns_locally_or_remote_hashing(text_file):
+     patterns = {"train": [_TEST_URL], "test": [str(text_file)]}
+     data_files1 = DataFilesDict.from_patterns(patterns)
+--- a/tests/packaged_modules/test_folder_based_builder.py	2024-02-21 21:30:20.718922523 +0100
++++ b/tests/packaged_modules/test_folder_based_builder.py	2024-02-21 21:31:46.309061287 +0100
+@@ -382,6 +382,7 @@
+         assert example[column] is not None
+ 
+ 
++@pytest.mark.skip(reason="network")
+ @pytest.mark.parametrize("remote", [True, False])
+ @pytest.mark.parametrize("drop_labels", [None, True, False])
+ def test_data_files_with_different_levels_no_metadata(
+@@ -405,6 +406,7 @@
+         assert all(example.keys() == {"base", "label"} for _, example in generator)
+ 
+ 
++@pytest.mark.skip(reason="network")
+ @pytest.mark.parametrize("remote", [False, True])
+ @pytest.mark.parametrize("drop_labels", [None, True, False])
+ def test_data_files_with_one_label_no_metadata(data_files_with_one_label_no_metadata, drop_labels, remote, cache_dir):