预处理数据集

import tarfile
from pathlib import Path

def init_dataset(datapath):
    if Path(datapath).is_file() and datapath.endswith('.tar.gz'):
        parent=Path(datapath).parent
        tar = tarfile.open(datapath)
        dirname=tar.getnames()[0]
        tar.extractall(path=parent.as_posix())
        datapath=(parent/dirname).as_posix()
        os.environ['DATAPATH']=datapath
    return datapath