def test_check_suffix_disabled_repack():
path_in_o = retrieve_data("fmt-hdf5_polygon_gate_2021.zip")
path_in = path_in_o.with_suffix("")
path_in_o.rename(path_in)
assert path_in.suffix == ""
with pytest.raises(ValueError, match="Unsupported file type"):
cli.repack(path_in=path_in,
path_out=path_in.with_name("repacked.rtdc"))
# but this should work:
cli.repack(path_in=path_in,
path_out=path_in.with_name("repacked2.rtdc"),
check_suffix=False)
def test_repack_user_metadata():
path_in = retrieve_data("fmt-hdf5_mask-contour_2018.zip")
with h5py.File(path_in, "a") as h5:
h5.attrs["user:peter"] = "hans"
# same directory (will be cleaned up with path_in)
path_out = path_in.with_name("repacked.rtdc")
cli.repack(path_out=path_out, path_in=path_in)
with new_dataset(path_out) as ds:
assert ds.config["user"]["peter"] == "hans"
def test_repack_strip_logs():
path_in = retrieve_data("fmt-hdf5_mask-contour_2018.zip")
# same directory (will be cleaned up with path_in)
path_out = path_in.with_name("repacked.rtdc")
# write some logs
with h5py.File(path_in, "a") as h5:
hw = rtdc_dataset.RTDCWriter(h5)
hw.store_log("test_log", ["peter", "hans"])
cli.repack(path_out=path_out, path_in=path_in, strip_logs=True)
with new_dataset(path_out) as dsj, new_dataset(path_in) as ds0:
assert ds0.logs
assert not dsj.logs
def test_repack_basic():
path_in = retrieve_data("fmt-hdf5_mask-contour_2018.zip")
# same directory (will be cleaned up with path_in)
path_out = path_in.with_name("repacked.rtdc")
cli.repack(path_out=path_out, path_in=path_in)
with new_dataset(path_out) as dsj, new_dataset(path_in) as ds0:
assert len(dsj)
assert len(dsj) == len(ds0)
for feat in ds0.features_innate:
if feat in ds0.features_scalar:
assert np.all(dsj[feat] == ds0[feat]), feat
for ii in range(len(ds0)):
assert np.all(dsj["contour"][ii] == ds0["contour"][ii])
assert np.all(dsj["image"][ii] == ds0["image"][ii])
assert np.all(dsj["mask"][ii] == ds0["mask"][ii])
def test_repack_remove_secrets():
path_in = retrieve_data("fmt-hdf5_mask-contour_2018.zip")
# same directory (will be cleaned up with path_in)
path_out = path_in.with_name("repacked.rtdc")
with h5py.File(path_in, "a") as h5:
h5.attrs["experiment:sample"] = "my dirty secret"
with h5py.File(path_in, "a") as h5:
h5.attrs["experiment:sample"] = "sunshine"
# test whether the dirty secret is still there
with open(str(path_in), "rb") as fd:
data = fd.read()
assert str(data).count("my dirty secret")
# now repack
cli.repack(path_out=path_out, path_in=path_in)
# clean?
with open(str(path_out), "rb") as fd:
data = fd.read()
assert not str(data).count("my dirty secret")