tfds.testing.dummy_croissant_file

Yields temporary path to a dummy Croissant file.

@contextlib.contextmanager
tfds.testing.dummy_croissant_file(
    dataset_name: str = 'DummyDataset',
    entries: (Sequence[dict[str, Any]] | None) = None,
    raw_data_filename: epath.PathLike = 'raw_data.jsonl',
    croissant_filename: epath.PathLike = 'croissant.json',
    split_names: (Sequence[str] | None) = None,
    version: str = '1.2.0'
) -> Iterator[epath.Path]

The function creates a temporary directory that stores raw data files and the Croissant JSON-LD.

dataset_name The name of the dataset.
entries A list of dictionaries representing the dataset's entries. Each dictionary should contain an 'index', a 'text', and a split key. If None, the function will create two entries with indices 0 and 1 and dummy text, and with the first entry belonging to the split train and the second to test.
raw_data_filename Filename of the raw data file. If split_names is True, the function will create a raw data file for each split, including the split name before the file extension.
croissant_filename Filename of the Croissant JSON-LD file.
split_names A list of split names to populate the split record set with. If split_names are defined, they must match the split key in the entries. If None, the function will create a split record set with the default split names train and test. If split_names is defined, the split key in the entries must match one of the split names.
version The version of the dataset. Defaults to 1.2.0.