preprocessing.py 文件源码

def get_coocc_features(cacheddatapath, rawdatapath):
    try:
        with bz2.open(os.path.join(cacheddatapath, "coocc_features.pklz2"), "r") as infile:
            coocc_features = pickle.load(infile)
    except:
        df = get_preprocessed_df(cacheddatapath, rawdatapath)
        coocc_features = count_cooccurrences(df)
        with bz2.BZ2File(os.path.join(cacheddatapath, "coocc_features.pklz2"), "w") as outfile:
            pickle.dump(coocc_features, outfile, protocol=4)
    return coocc_features