def test_temporary_directory__no_permission():
# create a local temporary directory using Python tempfile
with tempfile.TemporaryDirectory() as dir:
os.chmod(dir, 0o444) # make it read-only
with pytest.raises(PermissionError):
with temporary_directory(dir=dir):
pass # pragma: no cover
def test_temporary_directory(dir):
prefix = "prefix-"
suffix = "-suffix"
with temporary_directory(suffix=suffix, prefix=prefix, dir=dir) as tmpdir:
if tmpdir.startswith("file:///"):
tmpdir = tmpdir[7:]
dir = Path(tmpdir)
assert dir.exists()
assert dir.name.startswith(prefix)
assert dir.name.endswith(suffix)
with open(dir / "file.txt", "w") as file:
file.write("Hello")
assert not dir.exists()
def test_non_local_filesystem(mocker):
# mock out fsspec calls
mock = mocker.patch("fsspec.filesystem")
myfs = mocker.MagicMock()
mock.return_value = myfs
# call function
with temporary_directory(
prefix="mytmp", dir="myfs://path/file", storage_options=dict(a="b")
):
pass
# check expected called were made
fsspec.filesystem.assert_called_once_with("myfs", a="b")
myfs.mkdir.assert_called_once_with(StartsWith("myfs://path/file/mytmp"))
myfs.rm.assert_called_once_with(
StartsWith("myfs://path/file/mytmp"), recursive=True
)
input: Union[PathType, Sequence[PathType]],
output: Union[PathType, MutableMapping[str, bytes]],
regions: Union[None, Sequence[str], Sequence[Optional[Sequence[str]]]],
chunk_length: int = 10_000,
chunk_width: int = 1_000,
temp_chunk_length: Optional[int] = None,
tempdir: Optional[PathType] = None,
tempdir_storage_options: Optional[Dict[str, str]] = None,
) -> None:
"""Convert specified regions of one or more VCF files to zarr files, then concat, rechunk, write to zarr"""
if temp_chunk_length is None:
temp_chunk_length = chunk_length
with temporary_directory(
prefix="vcf_to_zarr_", dir=tempdir, storage_options=tempdir_storage_options
) as tmpdir:
paths = vcf_to_zarrs(
input,
tmpdir,
regions,
temp_chunk_length,
chunk_width,
tempdir_storage_options,
)
ds = zarrs_to_dataset(paths, chunk_length, chunk_width, tempdir_storage_options)
# Ensure Dask task graph is efficient, see https://github.com/dask/dask/issues/5105
with dask.config.set({"optimization.fuse.ave-width": 50}):