def test_prepare_df_from_dirs_one(tmp_path):
def check_filepath(filepath):
return filepath.startswith("act1") or filepath.startswith("act2")
_setup_dataset_fs(tmp_path)
root_path = tmp_path / "datasets/root1"
df = _prepare_df_from_dirs(str(root_path), "label")
assert df.shape[0] == 3
assert df.filepath.apply(check_filepath).sum().all()
assert df.label.isin(["act1", "act2"]).all()
def test_prepare_df_from_dirs_multi(tmp_path):
def check_filepath(filepath):
return filepath.startswith("root1/act1") or \
filepath.startswith("root1/act2") or \
filepath.startswith("root2/act1") or \
filepath.startswith("root2/act2")
_setup_dataset_fs(tmp_path)
ds_path = tmp_path / "datasets"
root_paths = ",".join([str(ds_path / "root1"), str(ds_path / "root2")])
df = _prepare_df_from_dirs(root_paths, "label")
assert df.shape[0] == 6
assert df.filepath.apply(check_filepath).sum().all()
assert df.label.isin(["act1", "act2"]).all()