Internal change.

PiperOrigin-RevId: 584674881

seqio/dataset_providers.py

@@ -61,6 +61,8 @@
 Feature = utils.Feature
 
 
+
+
 @dataclasses.dataclass(frozen=True)
 class ContinuousFeature(Feature):
   """A container for multi-modal output features of data providers."""
@@ -1458,6 +1460,8 @@ def cache_dir(self) -> Optional[str]:
           os.path.join(d, utils.get_task_dir_from_name(self.name))
           for d in utils.get_global_cache_dirs()
       ]
+
+
       for cache_dir in potential_cache_dirs:
         try:
           if tf.io.gfile.exists(os.path.join(cache_dir, "COMPLETED")):

seqio/dataset_providers_test.py

@@ -396,6 +396,18 @@ def test_requires_caching(self):
     ):
       task.get_dataset({"inputs": 512, "targets": 512}, use_cached=False)
 
+  def test_requires_caching_not_cached(self):
+    utils.set_global_cache_dirs([self.test_data_dir])
+
+    task = dataset_providers.Task(
+        "requires_cache",
+        output_features=self.DEFAULT_OUTPUT_FEATURES,
+        source=self.function_source,
+        preprocessors=[
+            dataset_providers.CacheDatasetPlaceholder(required=True),
+            preprocessors.tokenize,
+        ],
+    )
     # We haven't actually cached the task, so it still fails but with a
     # different error.
     with self.assertRaisesWithLiteralMatch(
@@ -488,6 +500,7 @@ def test_set_global_cache_dirs(self):
     utils.set_global_cache_dirs([self.test_data_dir])
     self.assertTrue(self.cached_task.cache_dir)
 
+
   def test_get_dataset_cached(self):
     self.verify_task_matches_fake_datasets(
         "cached_task", use_cached=True, token_preprocessed=False

seqio/utils.py

@@ -81,6 +81,8 @@ def add_global_cache_dirs(global_cache_dirs):
   _GLOBAL_CACHE_DIRECTORIES += global_cache_dirs
 
 
+
+
 def _validate_tfds_name(name: str) -> None:
   """Validates TFDS dataset name."""
   if (