def test_file_format_save_restore(
tmp_path: pathlib.Path,
file_format: file_adapters.FileFormat,
):
builder = testing.DummyDataset(data_dir=tmp_path, file_format=file_format)
assert isinstance(builder.info.file_format, file_adapters.FileFormat)
assert builder.info.file_format is file_format
builder.download_and_prepare()
# When restoring the builder, we do not provide the `file_format=`
# yet it is correctly restored
builder2 = testing.DummyDataset(data_dir=tmp_path)
assert builder2.info.file_format is file_format
# Explicitly passing the correct format is accepted.
builder3 = testing.DummyDataset(data_dir=tmp_path, file_format=file_format)
assert builder3.info.file_format is file_format
# Providing an inconsistent format is rejected.
with pytest.raises(ValueError, match="File format is already set to"):
different_file_format = {
file_adapters.FileFormat.TFRECORD:
file_adapters.FileFormat.RIEGELI,
file_adapters.FileFormat.RIEGELI:
file_adapters.FileFormat.TFRECORD,
}[file_format]
testing.DummyDataset(data_dir=tmp_path,
file_format=different_file_format)
def test_file_format_values(tmp_path: pathlib.Path):
# Default file format
builder = testing.DummyDataset(data_dir=tmp_path, file_format=None)
assert builder.info.file_format == file_adapters.FileFormat.TFRECORD
# str accepted
builder = testing.DummyDataset(data_dir=tmp_path, file_format="riegeli")
assert builder.info.file_format == file_adapters.FileFormat.RIEGELI
# file_adapters.FileFormat accepted
builder = testing.DummyDataset(
data_dir=tmp_path, file_format=file_adapters.FileFormat.RIEGELI)
assert builder.info.file_format == file_adapters.FileFormat.RIEGELI
# Unknown value
with pytest.raises(ValueError, match="is not a valid FileFormat"):
testing.DummyDataset(data_dir=tmp_path, file_format="arrow")
def test_compute_split_info(tmp_path):
builder = testing.DummyDataset(data_dir=tmp_path)
builder.download_and_prepare()
split_infos = compute_split_utils.compute_split_info(
data_dir=tmp_path / builder.info.full_name,
out_dir=tmp_path,
)
assert [s.to_proto() for s in split_infos
] == [s.to_proto() for s in builder.info.splits.values()]
# Split info are correctly saved
split_path = tmp_path / compute_split_utils._out_filename('train')
split_info = compute_split_utils._split_info_from_path(split_path)
assert builder.info.splits['train'].to_proto() == split_info.to_proto()
def test_write_metadata(
tmp_path: pathlib.Path,
file_format,
):
tmp_path = utils.as_path(tmp_path)
src_builder = testing.DummyDataset(
data_dir=tmp_path / 'origin',
file_format=file_format,
)
src_builder.download_and_prepare()
dst_dir = tmp_path / 'copy'
dst_dir.mkdir()
# Copy all the tfrecord files, but not the dataset info
for f in src_builder.data_path.iterdir():
if naming.FilenameInfo.is_valid(f.name):
f.copy(dst_dir / f.name)
metadata_path = dst_dir / 'dataset_info.json'
if file_format is None:
split_infos = list(src_builder.info.splits.values())
else:
split_infos = None # Auto-compute split infos
assert not metadata_path.exists()
write_metadata_utils.write_metadata(
data_dir=dst_dir,
features=src_builder.info.features,
split_infos=split_infos,
description='my test description.')
assert metadata_path.exists()
# After metadata are written, builder can be restored from the directory
builder = read_only_builder.builder_from_directory(dst_dir)
assert builder.name == 'dummy_dataset'
assert builder.version == '1.0.0'
assert set(builder.info.splits) == {'train'}
assert builder.info.splits['train'].num_examples == 3
assert builder.info.description == 'my test description.'
# Values are the same
src_ds = src_builder.as_dataset(split='train')
ds = builder.as_dataset(split='train')
assert list(src_ds.as_numpy_iterator()) == list(ds.as_numpy_iterator())
def test_compute_split_info(tmp_path):
builder = testing.DummyDataset(data_dir=tmp_path)
builder.download_and_prepare()
filename_template = naming.ShardedFileTemplate(
dataset_name=builder.name,
data_dir=builder.data_dir,
filetype_suffix=builder.info.file_format.file_suffix)
split_infos = compute_split_utils.compute_split_info(
out_dir=tmp_path,
filename_template=filename_template,
)
assert [s.to_proto() for s in split_infos
] == [s.to_proto() for s in builder.info.splits.values()]
# Split info are correctly saved
filename_template = filename_template.replace(
data_dir=tmp_path, split='train')
split_info = compute_split_utils._split_info_from_path(filename_template)
assert builder.info.splits['train'].to_proto() == split_info.to_proto()
def dummy_builder(tmp_path_factory):
"""Dummy dataset shared accross tests."""
data_dir = tmp_path_factory.mktemp('datasets')
builder = testing.DummyDataset(data_dir=data_dir)
builder.download_and_prepare()
yield builder
