tfds-nightly 4.9.9.dev202508080045__py3-none-any.whl → 4.9.9.dev202508090044__py3-none-any.whl

tensorflow_datasets/conftest.py

@@ -20,16 +20,21 @@ plugins (hooks and fixtures) common to all tests.
 
 See: https://docs.pytest.org/en/latest/writing_plugins.html
 """
+
 from __future__ import annotations
 
 import builtins
 import importlib
+import os
+import pathlib
 import sys
 import typing
 from typing import Iterator, Type
 
+from etils import epath
 import pytest
 from tensorflow_datasets import setup_teardown
+from tensorflow_datasets.core import constants
 
 if typing.TYPE_CHECKING:
   from tensorflow_datasets import testing
@@ -52,11 +57,9 @@ def disable_community_datasets():
   # visibility isn't automatically set.
   from tensorflow_datasets.core import visibility  # pylint: disable=g-import-not-at-top
 
-  visibility.set_availables(
-      [
-          visibility.DatasetType.TFDS_PUBLIC,
-      ]
-  )
+  visibility.set_availables([
+      visibility.DatasetType.TFDS_PUBLIC,
+  ])
 
 
 @pytest.fixture(scope='session', autouse=True)
@@ -144,3 +147,13 @@ def dummy_dataset(
   return _make_dataset(tmp_path_factory, testing.DummyDataset)
 
 
+@pytest.fixture(name='default_data_dir')
+def mock_default_data_dir(
+    monkeypatch: pytest.MonkeyPatch, tmp_path: pathlib.Path
+) -> epath.Path:
+  """Sets the default data dir to a temp dir."""
+  default_data_dir = epath.Path(tmp_path) / 'default_data_dir'
+  monkeypatch.setattr(constants, 'DATA_DIR', os.fspath(default_data_dir))
+  return default_data_dir
+
+

tensorflow_datasets/core/constants.py

@@ -18,6 +18,7 @@
 # IMPORTANT: when changing values here, update docstrings.
 
 import os
+from typing import Final
 
 # Directory in which datasets are declared within TFDS sources.
 DATASETS_TFDS_SRC_DIR = 'datasets'
@@ -27,7 +28,7 @@ SRC_BASE_URL = 'https://github.com/tensorflow/datasets/tree/master/'
 
 # Directory where to store processed datasets.
 # If modifying this, should also update `scripts/cli/build.py` `--data_dir`
-DATA_DIR = os.environ.get(
+DATA_DIR: Final[str] = os.environ.get(
     'TFDS_DATA_DIR',
     os.path.join(os.path.expanduser('~'), 'tensorflow_datasets'),
 )

tensorflow_datasets/core/utils/file_utils_test.py

@@ -62,16 +62,8 @@ def _assert_data_dir(
   assert data_dir == expected_data_dir
 
 
-@pytest.fixture(name='default_data_dir')
-def mock_default_data_dir(monkeypatch, tmp_path):
-  """Sets the default data dir to a temp dir."""
-  default_data_dir = tmp_path / 'default_data_dir'
-  monkeypatch.setattr(constants, 'DATA_DIR', default_data_dir)
-  return default_data_dir
-
-
 @pytest.fixture(name='other_data_dir')
-def mock_other_data_dir(default_data_dir):
+def mock_other_data_dir(default_data_dir: epath.Path):
   """Adds another data dir to the registered data dirs."""
   other_data_dir = default_data_dir.parent / 'other_data_dir'
   file_utils.add_data_dir(other_data_dir)

tensorflow_datasets/scripts/cli/build_test.py

@@ -13,14 +13,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-"""Tests for tensorflow_datasets.scripts.cli.build."""
-
+from collections.abc import Callable, Iterator
 import contextlib
 import dataclasses
+import functools
 import multiprocessing
 import os
-import pathlib
-from typing import Dict, Iterator, List, Optional
 from unittest import mock
 
 from etils import epath
@@ -45,7 +43,7 @@ class DummyDatasetNoGenerate(tfds.testing.DummyDataset):
 
   @utils.classproperty
   @classmethod
-  def url_infos(cls) -> Optional[Dict[str, download.checksums.UrlInfo]]:
+  def url_infos(cls) -> dict[str, download.checksums.UrlInfo] | None:
     return {
         'http://data.org/file1.zip': download.checksums.UrlInfo(
             size=42,
@@ -55,33 +53,21 @@ class DummyDatasetNoGenerate(tfds.testing.DummyDataset):
     }
 
 
-@pytest.fixture(scope='function', autouse=True)
-def mock_default_data_dir(tmp_path: pathlib.Path):
-  """Changes the default `--data_dir` to tmp_path."""
-  tmp_path = tmp_path / 'datasets'
-  default_data_dir = os.environ.get('TFDS_DATA_DIR')
-  try:
-    os.environ['TFDS_DATA_DIR'] = os.fspath(tmp_path)
-    yield tmp_path
-  finally:
-    if default_data_dir:
-      os.environ['TFDS_DATA_DIR'] = default_data_dir
-    else:
-      del os.environ['TFDS_DATA_DIR']
-
-
 @contextlib.contextmanager
 def mock_cwd(path: epath.PathLike) -> Iterator[None]:
   """Mock the current directory."""
-  path = pathlib.Path(path)
+  path = epath.Path(path)
   assert path.exists() and path.is_dir()  # Check given path is valid cwd dir
-  with mock.patch('os.getcwd', return_value=os.fspath(path)):
+  with mock.patch.object(os, 'getcwd', return_value=os.fspath(path)):
     yield
 
 
-def _build(cmd_flags: str, mock_download_and_prepare: bool = True) -> List[str]:
+def _build(
+    cmd_flags: str, data_dir: epath.Path, mock_download_and_prepare: bool = True
+) -> list[str]:
   """Executes `tfds build {cmd_flags}` and returns the list of generated ds."""
   # Execute the command
+  cmd_flags = f'--data_dir={data_dir} {cmd_flags}'
   args = main._parse_flags(f'tfds build {cmd_flags}'.split())
 
   original_dl_and_prepare = tfds.core.DatasetBuilder.download_and_prepare
@@ -101,9 +87,8 @@ def _build(cmd_flags: str, mock_download_and_prepare: bool = True) -> List[str]:
     else:
       return original_dl_and_prepare(self, *args, **kwargs)
 
-  with mock.patch(
-      'tensorflow_datasets.core.DatasetBuilder.download_and_prepare',
-      _download_and_prepare,
+  with mock.patch.object(
+      tfds.core.DatasetBuilder, 'download_and_prepare', _download_and_prepare
   ):
     main.main(args)
     queue.put(None)
@@ -115,31 +100,39 @@ def _build(cmd_flags: str, mock_download_and_prepare: bool = True) -> List[str]:
   return generated_ds_names
 
 
-def test_build_single():
-  assert _build('mnist') == ['mnist']
-  assert _build('mnist:3.0.1') == ['mnist']
+@pytest.fixture(name='build')
+def mock_build(
+    default_data_dir: epath.Path,
+) -> Callable[[str, bool], list[str]]:
+  """Returns a function to execute `tfds build`."""
+  return functools.partial(_build, data_dir=default_data_dir)
+
+
+def test_build_single(build):
+  assert build('mnist') == ['mnist']
+  assert build('mnist:3.0.1') == ['mnist']
   # Keyword arguments also possible
-  assert _build('--datasets mnist') == ['mnist']
+  assert build('--datasets mnist') == ['mnist']
 
   with pytest.raises(tfds.core.registered.DatasetNotFoundError):
-    _build('unknown_dataset')
+    build('unknown_dataset')
 
   with pytest.raises(AssertionError, match='cannot be loaded at version 1.0.0'):
-    _build('mnist:1.0.0')  # Can only built the last version
+    build('mnist:1.0.0')  # Can only built the last version
 
   with pytest.raises(ValueError, match='not have config'):
-    _build('mnist --config_idx 0')
+    build('mnist --config_idx 0')
 
 
-def test_build_multiple():
+def test_build_multiple(build):
   # Multiple datasets can be built in a single call
-  assert _build('mnist imagenet2012 cifar10') == [
+  assert build('mnist imagenet2012 cifar10') == [
       'mnist',
       'imagenet2012',
       'cifar10',
   ]
   # Keyword arguments also possible
-  assert _build('mnist --datasets imagenet2012 cifar10') == [
+  assert build('mnist --datasets imagenet2012 cifar10') == [
       'mnist',
       'imagenet2012',
       'cifar10',
@@ -147,9 +140,9 @@ def test_build_multiple():
 
 
 @pytest.mark.parametrize('num_processes', range(1, 4))
-def test_build_parallel(num_processes):
+def test_build_parallel(build, num_processes):
   # Order is not guaranteed
-  assert set(_build(f'trivia_qa --num-processes={num_processes}')) == set([
+  assert set(build(f'trivia_qa --num-processes={num_processes}')) == set([
       'trivia_qa/rc',
       'trivia_qa/rc.nocontext',
       'trivia_qa/unfiltered',
@@ -157,9 +150,9 @@ def test_build_parallel(num_processes):
   ])
 
 
-def test_build_dataset_configs():
+def test_build_dataset_configs(build):
   # By default, all configs are build
-  assert _build('trivia_qa') == [
+  assert build('trivia_qa') == [
       'trivia_qa/rc',
       'trivia_qa/rc.nocontext',
       'trivia_qa/unfiltered',
@@ -169,47 +162,47 @@ def test_build_dataset_configs():
   # If config is set, only the defined config is generated
 
   # --config_idx
-  assert _build('trivia_qa --config_idx=0') == ['trivia_qa/rc']
+  assert build('trivia_qa --config_idx=0') == ['trivia_qa/rc']
 
   # --config
-  assert _build('trivia_qa --config unfiltered.nocontext') == [
+  assert build('trivia_qa --config unfiltered.nocontext') == [
       'trivia_qa/unfiltered.nocontext',
   ]
 
   # --config Json
   config_json = '{"name":"my_config","description":"abcd"}'
-  assert _build(f'imdb_reviews --config {config_json}') == [
+  assert build(f'imdb_reviews --config {config_json}') == [
       'imdb_reviews/my_config',
   ]
 
   # name/config
-  assert _build('trivia_qa/unfiltered.nocontext') == [
+  assert build('trivia_qa/unfiltered.nocontext') == [
       'trivia_qa/unfiltered.nocontext'
   ]
 
   with pytest.raises(ValueError, match='Config should only be defined once'):
-    _build('trivia_qa/unfiltered.nocontext --config_idx=0')
+    build('trivia_qa/unfiltered.nocontext --config_idx=0')
 
   with pytest.raises(ValueError, match='greater than number of configs'):
-    _build('trivia_qa --config_idx 100')
+    build('trivia_qa --config_idx 100')
 
 
-def test_exclude_datasets():
+def test_exclude_datasets(build):
   # Exclude all datasets except 2
   all_ds = [b for b in tfds.list_builders() if b not in ('mnist', 'cifar10')]
   all_ds_str = ','.join(all_ds)
 
-  assert _build(f'--exclude_datasets {all_ds_str}') == [
+  assert build(f'--exclude_datasets {all_ds_str}') == [
       'cifar10',
       'mnist',
   ]
 
   with pytest.raises(ValueError, match="--exclude_datasets can't be used"):
-    _build('mnist --exclude_datasets cifar10')
+    build('mnist --exclude_datasets cifar10')
 
 
-def test_build_overwrite(mock_default_data_dir: pathlib.Path):  # pylint: disable=redefined-outer-name
-  data_dir = mock_default_data_dir / 'mnist/3.0.1'
+def test_build_overwrite(build, default_data_dir: epath.Path):
+  data_dir = default_data_dir / 'mnist/3.0.1'
   data_dir.mkdir(parents=True)
   metadata_path = tfds.core.tfds_path(
       'testing/test_data/dataset_info/mnist/3.0.1'
@@ -219,80 +212,82 @@ def test_build_overwrite(mock_default_data_dir: pathlib.Path):  # pylint: disabl
     data_dir.joinpath(f.name).write_text(f.read_text())
 
   # By default, will skip generation if the data already exists
-  assert _build('mnist') == ['mnist']  # Called, but no-op
+  assert build('mnist') == ['mnist']  # Called, but no-op
   assert data_dir.exists()
 
-  assert _build('mnist --overwrite') == ['mnist']
+  assert build('mnist --overwrite') == ['mnist']
   assert not data_dir.exists()  # Previous data-dir has been removed
 
 
-def test_max_examples_per_split_0(mock_default_data_dir: pathlib.Path):  # pylint: disable=redefined-outer-name
-  assert _build(
+def test_max_examples_per_split_0(build, default_data_dir: epath.Path):
+  assert build(
       'dummy_dataset_no_generate --max_examples_per_split 0',
       mock_download_and_prepare=False,
   ) == ['dummy_dataset_no_generate']
 
-  builder_path = mock_default_data_dir / 'dummy_dataset_no_generate/1.0.0'
+  builder_path = default_data_dir / 'dummy_dataset_no_generate/1.0.0'
   # Dataset has been generated
   assert builder_path.exists()
   # tf-records files have not been generated
-  assert sorted(builder_path.iterdir()) == [
-      builder_path / 'dataset_info.json',
-      builder_path / 'features.json',
+  assert sorted(p.name for p in builder_path.iterdir()) == [
+      'dataset_info.json',
+      'features.json',
   ]
 
 
-def test_build_files():
+def test_build_files(build):
   # Make sure DummyDataset isn't registered by default
   with pytest.raises(tfds.core.registered.DatasetNotFoundError):
-    _build('dummy_dataset')
+    build('dummy_dataset')
 
   with pytest.raises(FileNotFoundError, match='Could not find .* script'):
-    _build('')
+    build('')
 
   # cd .../datasets/dummy_dataset && tfds build
   with mock_cwd(_DUMMY_DATASET_PATH):
-    assert _build('') == ['dummy_dataset']
+    assert build('') == ['dummy_dataset']
 
   # cd .../datasets/dummy_dataset && tfds build dummy_dataset.py
   with mock_cwd(_DUMMY_DATASET_PATH):
-    assert _build('dummy_dataset.py') == ['dummy_dataset']
+    assert build('dummy_dataset.py') == ['dummy_dataset']
 
   # cd .../datasets/ && tfds build dummy_dataset
   with mock_cwd(_DUMMY_DATASET_PATH.parent):
-    assert _build('dummy_dataset') == ['dummy_dataset']
+    assert build('dummy_dataset') == ['dummy_dataset']
 
   # cd .../datasets/ && tfds build dummy_dataset --imports=xxx
   # --imports is passed. so do not load dataset from file
   with mock_cwd(_DUMMY_DATASET_PATH.parent):
     with pytest.raises(tfds.core.registered.DatasetNotFoundError):
-      assert _build('dummy_dataset --imports=os')
+      assert build('dummy_dataset --imports=os')
 
   # cd .../datasets/ && tfds build dummy_dataset/dummy_dataset
   with mock_cwd(_DUMMY_DATASET_PATH.parent):
-    assert _build('dummy_dataset/dummy_dataset') == ['dummy_dataset']
+    assert build('dummy_dataset/dummy_dataset') == ['dummy_dataset']
 
 
 # Somehow only with tf-nightly, `dummy_dataset` is already imported by
 # community/load_test.py (with `skip_registration()`). Thus import here have
 # no-effects.
 @pytest.mark.skip(reason='Conflict with `load_test.py`')
-def test_build_import():
+def test_build_import(build):
   # DummyDataset isn't registered by default
   with pytest.raises(tfds.core.registered.DatasetNotFoundError):
-    _build('dummy_dataset')
+    build('dummy_dataset')
 
   # --imports register the dataset
   ds_module = 'tensorflow_datasets.testing.dummy_dataset.dummy_dataset'
-  assert _build(f'dummy_dataset --imports {ds_module}') == ['dummy_dataset']
+  assert build(f'dummy_dataset --imports {ds_module}') == ['dummy_dataset']
 
 
-def test_download_only():
-  with mock.patch(
-      'tensorflow_datasets.download.DownloadManager.download'
+def test_download_only(build):
+  with mock.patch.object(
+      download.DownloadManager, 'download', autospec=True
   ) as mock_download:
-    assert not _build('dummy_dataset_no_generate --download_only')
-    mock_download.assert_called_with({'file0': 'http://data.org/file1.zip'})
+    assert not build('dummy_dataset_no_generate --download_only')
+    mock_download.assert_called_with(
+        mock.ANY, {'file0': 'http://data.org/file1.zip'}
+    )
 
 
 @pytest.mark.parametrize(

tensorflow_datasets/scripts/cli/cli_utils.py

@@ -16,18 +16,65 @@
 """Utility functions for TFDS CLI."""
 
 import argparse
+from collections.abc import Sequence
 import dataclasses
 import itertools
-import os
 import pathlib
 
 from absl import logging
+from absl.flags import argparse_flags
 from etils import epath
+import simple_parsing
+from tensorflow_datasets.core import constants
 from tensorflow_datasets.core import dataset_builder
 from tensorflow_datasets.core import download
 from tensorflow_datasets.core import file_adapters
 from tensorflow_datasets.core import naming
 from tensorflow_datasets.core.utils import file_utils
+from tensorflow_datasets.scripts.utils import flag_utils
+
+
+class ArgumentParser(
+    argparse_flags.ArgumentParser, simple_parsing.ArgumentParser
+):
+  """An `ArgumentParser` that handles both `simple_parsing` and `absl` flags.
+
+  This class is a workaround for the fact that `simple_parsing.ArgumentParser`
+  does not natively handle `absl.flags`. Without this, `absl` flags are not
+  correctly parsed, especially when they are mixed with positional arguments,
+  leading to errors.
+
+  The `absl.flags.argparse_flags.ArgumentParser` is designed to integrate `absl`
+  flags into an `argparse` setup. It does this by dynamically adding all
+  defined `absl` flags to the parser instance upon initialization.
+
+  By inheriting from both, we get the features of both:
+  - `simple_parsing.ArgumentParser`: Allows defining arguments from typed
+    dataclasses.
+  - `argparse_flags.ArgumentParser`: Adds support for `absl` flags.
+
+  The Method Resolution Order (MRO) is:
+  `ArgumentParser` -> `argparse_flags.ArgumentParser` ->
+  `simple_parsing.ArgumentParser` -> `argparse.ArgumentParser` -> `object`.
+
+  This order is important. `argparse_flags.ArgumentParser` is first so that it
+  can intercept arguments and handle `absl` flags before they are passed to
+  `simple_parsing.ArgumentParser`.
+  """
+
+  def parse_known_args(
+      self,
+      args: Sequence[str] | None = None,
+      namespace: argparse.Namespace | None = None,
+      attempt_to_reorder: bool = False,
+  ):
+    # `argparse_flags.ArgumentParser` does not support `attempt_to_reorder` that
+    # is used by `simple_parsing.ArgumentParser`. Since we don't need it, we can
+    # just ignore it.
+    del attempt_to_reorder
+    if args:
+      args = flag_utils.normalize_flags(args)
+    return super().parse_known_args(args, namespace)
 
 
 @dataclasses.dataclass
@@ -119,13 +166,7 @@ def add_path_argument_group(parser: argparse.ArgumentParser):
   path_group.add_argument(
       '--data_dir',
       type=epath.Path,
-      # Should match tfds.core.constant.DATA_DIR !!
-      default=epath.Path(
-          os.environ.get(
-              'TFDS_DATA_DIR',
-              os.path.join(os.path.expanduser('~'), 'tensorflow_datasets'),
-          )
-      ),
+      default=epath.Path(constants.DATA_DIR),
       help=(
           'Where to place datasets. Default to '
           '`~/tensorflow_datasets/` or `TFDS_DATA_DIR` environement variable.'

tensorflow_datasets/scripts/cli/conftest.py

@@ -15,10 +15,11 @@
 
 """CLI Fixtures."""
 
-import argparse
 from unittest import mock
 
 import pytest
+import simple_parsing
+from tensorflow_datasets.scripts.cli import cli_utils
 
 
 @pytest.fixture(scope='session', autouse=True)
@@ -29,7 +30,7 @@ def _mock_argparse_flags():
   # another test):
   # `flags.DEFINE_string('data_dir')` with `parser.add_argument('--data_dir')`
   # We patch argparse_flags during test, so absl flags are ignored.
-  with mock.patch(
-      'absl.flags.argparse_flags.ArgumentParser', argparse.ArgumentParser
+  with mock.patch.object(
+      cli_utils, 'ArgumentParser', simple_parsing.ArgumentParser
   ):
     yield

tensorflow_datasets/scripts/cli/main.py

@@ -29,24 +29,21 @@ from absl import app
 from absl import flags
 from absl import logging
 
-import simple_parsing
 import tensorflow_datasets.public_api as tfds
 
 # Import commands
 from tensorflow_datasets.scripts.cli import build
+from tensorflow_datasets.scripts.cli import cli_utils
 from tensorflow_datasets.scripts.cli import convert_format
 from tensorflow_datasets.scripts.cli import croissant
 from tensorflow_datasets.scripts.cli import new
-from tensorflow_datasets.scripts.utils import flag_utils
 
 FLAGS = flags.FLAGS
 
 
 def _parse_flags(argv: List[str]) -> argparse.Namespace:
   """Command lines flag parsing."""
-  argv = flag_utils.normalize_flags(argv)  # See b/174043007 for context.
-
-  parser = simple_parsing.ArgumentParser(
+  parser = cli_utils.ArgumentParser(
       description='Tensorflow Datasets CLI tool',
       allow_abbrev=False,
   )
@@ -67,22 +64,7 @@ def _parse_flags(argv: List[str]) -> argparse.Namespace:
   new.register_subparser(subparser)
   convert_format.register_subparser(subparser)
   croissant.register_subparser(subparser)
-
-  namespace, remaining_argv = parser.parse_known_args(argv[1:])
-
-  # Manually parse absl flags from the remaining arguments.
-  try:
-    # FLAGS requires the program name as the first argument.
-    positionals = FLAGS(argv[:1] + remaining_argv)
-  except flags.Error as e:
-    parser.error(str(e))
-
-  # There should be no positional arguments left, as they should have been
-  # handled by the sub-commands.
-  if len(positionals) > 1:
-    parser.error(f"unrecognized arguments: {' '.join(positionals[1:])}")
-
-  return namespace
+  return parser.parse_args(argv[1:])
 
 
 def main(args: argparse.Namespace) -> None:

tensorflow_datasets/scripts/utils/flag_utils.py

@@ -15,11 +15,11 @@
 
 """Utility for handling flags."""
 
+from collections.abc import Sequence
 import re
-from typing import List
 
 
-def normalize_flags(argv: List[str]) -> List[str]:
+def normalize_flags(argv: Sequence[str]) -> list[str]:
   """Returns normalized explicit bolean flags for `absl.flags` compatibility.
 
   Note: Boolean flags in `absl.flags` can be specified with --bool, --nobool,

{tfds_nightly-4.9.9.dev202508080045.dist-info → tfds_nightly-4.9.9.dev202508090044.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: tfds-nightly
-Version: 4.9.9.dev202508080045
+Version: 4.9.9.dev202508090044
 Summary: tensorflow/datasets is a library of datasets ready to use with TensorFlow.
 Home-page: https://github.com/tensorflow/datasets
 Download-URL: https://github.com/tensorflow/datasets/tags

{tfds_nightly-4.9.9.dev202508080045.dist-info → tfds_nightly-4.9.9.dev202508090044.dist-info}/RECORD

@@ -1,6 +1,6 @@
 tensorflow_datasets/__init__.py,sha256=uP6rUgdo-WeHzrZEHAsCREyxz9lO2IPCIVerl6Ea_JQ,3622
 tensorflow_datasets/community-datasets.toml,sha256=xHUi5yCwLSAbn1nMwImRviZ03F1ZmdR2cyLJ2LDqhIU,629
-tensorflow_datasets/conftest.py,sha256=FMsFSgifx-tVWv-smAejkKckdQAvnYl9nG9wmmi5G40,4831
+tensorflow_datasets/conftest.py,sha256=95ocbFU_SvVgWHI7NVkH1wXDrb-OGZrd_ihiVVkpk9g,5269
 tensorflow_datasets/import_public_api_test.py,sha256=NPjteyVJL2ZDPEznMWBpP6hHbCaWiaZPxOcgnyr-HqY,852
 tensorflow_datasets/import_test.py,sha256=6rtuyyIwF9QNPyu6c-s5t-aA0fKPy2R9ondIHJVOkhY,801
 tensorflow_datasets/import_without_tf_test.py,sha256=znenZUnnltG7Jh2-PhhMEl_APadgiz1qPXi5P3Z85xo,3459
@@ -59,7 +59,7 @@ tensorflow_datasets/core/as_dataframe.py,sha256=3-2ScAo2G6wwYWbz_w3Crb4QyBwcuIYh
 tensorflow_datasets/core/as_dataframe_test.py,sha256=cGgk3f9j87dDRA2EXedlYb11NpOLdew0dA_O0ZG-PLQ,2048
 tensorflow_datasets/core/beam_utils.py,sha256=0X2lE9ILqLWZe5Idg58-G5XtgITXEAxqVodDtCDo9Ro,5109
 tensorflow_datasets/core/beam_utils_test.py,sha256=5ZhntgG658uT5pF4kw_U5Epm8lu0tdg4cI-0viMItzg,2852
-tensorflow_datasets/core/constants.py,sha256=eAzxhRpS1vx5LMedCmes2nfSU7BL831yi1J8JZkhtXY,2769
+tensorflow_datasets/core/constants.py,sha256=5mWPx8lo4VFd-_HS07JquFc_4ajWvLkpbpU1_2izA0k,2806
 tensorflow_datasets/core/dataset_builder.py,sha256=GMPEtJ3vcELXg5IvqgYHLjTxdomMzLNZV7rW8YBJg2E,80616
 tensorflow_datasets/core/dataset_builder_beam_test.py,sha256=d7UsYNsAIY4FcANAERLcVMDcajIpAi0uMfrnQoe4yv4,8386
 tensorflow_datasets/core/dataset_builder_notfdv_test.py,sha256=eIWlOZijQfopdze85EkbcPY1I8lFmEBnedcoUoOAnRQ,1346
@@ -253,7 +253,7 @@ tensorflow_datasets/core/utils/dtype_utils_test.py,sha256=-Qe2fQzDO5sjS36ZL-dY9w
 tensorflow_datasets/core/utils/error_utils.py,sha256=lnquUa_VGRjn7-G_5x-PvWGgnnO6GAWsi9I7xeVuGxQ,3204
 tensorflow_datasets/core/utils/error_utils_test.py,sha256=Brt8X12ZlWCR4x3WLnSvq4X76eyU1yH3t5STPfAkxUs,2060
 tensorflow_datasets/core/utils/file_utils.py,sha256=vL-ulAVClrvkA71DvEvdGR2EdNmOR9Y9hNKUDwJ3Ll4,18534
-tensorflow_datasets/core/utils/file_utils_test.py,sha256=SCw_XFRhyxGCFEVjt9pOdupsoULPdi8iT38JBrnUuDM,13708
+tensorflow_datasets/core/utils/file_utils_test.py,sha256=N-_jZlE73h2OqDxFmQJQ0cdb5a3FBilNjeP0iU3JMwQ,13438
 tensorflow_datasets/core/utils/gcs_utils.py,sha256=8mBOgEepkah1Rw36F6DNIVhLzfXbR8iS8KMLQUM5sPk,5154
 tensorflow_datasets/core/utils/gcs_utils_test.py,sha256=Ig8S37AvFG2g7kNjYxqgmqNKlLPeXt31XD7RY4UzsDg,2578
 tensorflow_datasets/core/utils/huggingface_utils.py,sha256=NeYaUoO3vIFH8M0hZ8k4w7AchFZJIGsuV1XwKJVttfw,5325
@@ -1980,16 +1980,16 @@ tensorflow_datasets/scripts/cleanup/url_filename_recorder.py,sha256=iLcsT8UgbyNU
 tensorflow_datasets/scripts/cleanup/url_status_checker.py,sha256=Tr3LtLnGhI8ElDAS-ejmuAU3rs1lmqmYlU4figoVQg0,1967
 tensorflow_datasets/scripts/cli/__init__.py,sha256=Z8UWkv0wbzS4AzaLgSpYVGApYv5j57RWY0vN5Z553BQ,613
 tensorflow_datasets/scripts/cli/build.py,sha256=jZp7CaP62D2Usi4l-o9oCUqTHhnigX15PNUr9pOd4Wo,14961
-tensorflow_datasets/scripts/cli/build_test.py,sha256=UoDsSPDMdMhhx7eoZZfCP6beXgxUibiyUUDbWNudV_A,10822
+tensorflow_datasets/scripts/cli/build_test.py,sha256=xlFYScPSMcsUR27GQ-W5wdGdLdkXu_n0hM1rl20WWW8,10542
 tensorflow_datasets/scripts/cli/builder_templates.py,sha256=99SvH3skigkc2Qg737BV2OzhXL_Rgu4az8eVHsxKCLk,7985
 tensorflow_datasets/scripts/cli/builder_templates_test.py,sha256=HBNB-v2zlImKULPI8Webs9hXCkeFmWT29urxav-tDe8,2062
-tensorflow_datasets/scripts/cli/cli_utils.py,sha256=rMYMcQj1w46OTOeMyp3qf4y9v7ArOGh6u5NaCjBXal8,12313
-tensorflow_datasets/scripts/cli/conftest.py,sha256=cmvCCV-efT5ZXYPkCSGS1OxoKNPAfSsLcFTfYfe61S0,1233
+tensorflow_datasets/scripts/cli/cli_utils.py,sha256=zE-jLQw0dn_98PHOTLX6pMoFqjSCBOD7lh5dytJcphE,14049
+tensorflow_datasets/scripts/cli/conftest.py,sha256=3PNh_BbR013G4HyLAZOleUXsQ9mICrD03NaKwdHFMXs,1291
 tensorflow_datasets/scripts/cli/convert_format.py,sha256=02RDZQQCuXf_XFFpx0gmRVkYyJg534kY0fZwGKxtUL4,4197
 tensorflow_datasets/scripts/cli/convert_format_utils.py,sha256=U_q5WVgMNrjBkOc166U4Y_eca5KOS3Xb3jSDjp4XdK4,29078
 tensorflow_datasets/scripts/cli/convert_format_utils_test.py,sha256=9JGNu9TvUWzbuhe6DWwnO3V9Lia5S1Is64re-pceAWE,8823
 tensorflow_datasets/scripts/cli/croissant.py,sha256=6jzmOXt_i7aeJHUVX7_zpRRMEXId_PzU24zUDdExRUs,6112
-tensorflow_datasets/scripts/cli/main.py,sha256=FJJwyUtM1N9gNDsxGm850m5ejzzJ9mgESNW9Xz8E9_I,4383
+tensorflow_datasets/scripts/cli/main.py,sha256=qUKxjULaL2ilkPqh6vcsbtnKRJJhZCBn--ZkveFO2mA,3789
 tensorflow_datasets/scripts/cli/main_test.py,sha256=3zNaS_2FmxxLoZOX05iJ2riuP4Qv8cx6bhAI56tV8YI,1067
 tensorflow_datasets/scripts/cli/new.py,sha256=x_GQSEVva1XuMvFwL3rANjDxviwZviXKHCICY7P30Jc,7803
 tensorflow_datasets/scripts/cli/new_test.py,sha256=USr9So-FPtg8UzaQPPacXn0E1ukDIoew9oYkOn45oik,2655
@@ -2022,7 +2022,7 @@ tensorflow_datasets/scripts/documentation/templates/community_namespace_toc.md,s
 tensorflow_datasets/scripts/tools/__init__.py,sha256=rG-M1W3iDDSWXfKasKS3UhqroWicWvULibHch3yIuxc,623
 tensorflow_datasets/scripts/tools/compute_split_info.py,sha256=BmLQdoBVaODYABwCmUK4i6LW6m72E6N8T7Vu8QEsgmg,2791
 tensorflow_datasets/scripts/utils/__init__.py,sha256=rG-M1W3iDDSWXfKasKS3UhqroWicWvULibHch3yIuxc,623
-tensorflow_datasets/scripts/utils/flag_utils.py,sha256=X8H8j33jgKMZsNZYvR_P1EVMMAhQuOVkcsph2vxex-Q,1544
+tensorflow_datasets/scripts/utils/flag_utils.py,sha256=HJBbcfYJGcChSEUxFcfgIqYoTAq32K12hBeXGKn9qJU,1561
 tensorflow_datasets/scripts/utils/flag_utils_test.py,sha256=wHCZ7n5ppzQPHlDMMu6MI6zblCmRSXx3HAAz6K63oqw,1259
 tensorflow_datasets/structured/__init__.py,sha256=FaJdqKPZ6Ca4R1WybpoWLP6BL-frtAVA7Le5nU5bY0g,2468
 tensorflow_datasets/structured/amazon_us_reviews.py,sha256=ZZS7Qn3huPzcu5DVhd1aDpUTYFAB3jvIKQoQt9I-7iM,915
@@ -2468,10 +2468,10 @@ tensorflow_datasets/vision_language/wit/wit_test.py,sha256=PXS8DMNW-MDrT2p5oy4Ic
 tensorflow_datasets/vision_language/wit_kaggle/__init__.py,sha256=vGwSGeM8WE4Q-l0-eEE1sBojmk6YT0l1OO60AWa4Q40,719
 tensorflow_datasets/vision_language/wit_kaggle/wit_kaggle.py,sha256=q-vX_FBzIwsFxL4sY9vuyQ3UQD2PLM4yhUR4U6l-qao,16903
 tensorflow_datasets/vision_language/wit_kaggle/wit_kaggle_test.py,sha256=ZymHT1NkmD-pUnh3BmM3_g30c5afsWYnmqDD9dVyDSA,1778
-tfds_nightly-4.9.9.dev202508080045.dist-info/licenses/AUTHORS,sha256=nvBG4WwfgjuOu1oZkuQKw9kg7X6rve679ObS-YDDmXg,309
-tfds_nightly-4.9.9.dev202508080045.dist-info/licenses/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
-tfds_nightly-4.9.9.dev202508080045.dist-info/METADATA,sha256=Za3dYGEQJMn0dTM0uUOEd_2jik8ANJg6Lmtp94-hKQ0,11694
-tfds_nightly-4.9.9.dev202508080045.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-tfds_nightly-4.9.9.dev202508080045.dist-info/entry_points.txt,sha256=eHEL7nF5y1uCY2FgkuYIdE062epJXlAQTSdq89px4p4,73
-tfds_nightly-4.9.9.dev202508080045.dist-info/top_level.txt,sha256=bAevmk9209s_oxVZVlN6hSDIVS423qrMQvmcWSvW4do,20
-tfds_nightly-4.9.9.dev202508080045.dist-info/RECORD,,
+tfds_nightly-4.9.9.dev202508090044.dist-info/licenses/AUTHORS,sha256=nvBG4WwfgjuOu1oZkuQKw9kg7X6rve679ObS-YDDmXg,309
+tfds_nightly-4.9.9.dev202508090044.dist-info/licenses/LICENSE,sha256=z8d0m5b2O9McPEK1xHG_dWgUBT6EfBDz6wA0F7xSPTA,11358
+tfds_nightly-4.9.9.dev202508090044.dist-info/METADATA,sha256=zSR51MIaMyUvB8ZEeJ5C4RzIQ2BAjy_VKE5aO6IsHJc,11694
+tfds_nightly-4.9.9.dev202508090044.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+tfds_nightly-4.9.9.dev202508090044.dist-info/entry_points.txt,sha256=eHEL7nF5y1uCY2FgkuYIdE062epJXlAQTSdq89px4p4,73
+tfds_nightly-4.9.9.dev202508090044.dist-info/top_level.txt,sha256=bAevmk9209s_oxVZVlN6hSDIVS423qrMQvmcWSvW4do,20
+tfds_nightly-4.9.9.dev202508090044.dist-info/RECORD,,