preprocessing.py 文件源码

def get_preprocessed_df(cacheddatapath=None, rawdatapath=None):
    try:
        with gzip.open(os.path.join(cacheddatapath, "preprocessed_df.pklz"), "rb") as infile:
            df = pickle.load(infile)
    except:
        df = preprocess(rawdatapath)
        if rawdatapath is None:
            pass
            # needs an io error for missing rawdatapath
        with gzip.open(os.path.join(cacheddatapath, "preprocessed_df.pklz"), "wb") as outfile:
            pickle.dump(df, outfile, protocol=4)
    return df