Beispiel #1
0
def unzip(filepath):
    z = ZipFile(filepath)
    unzip_dir = os.path.join(BIGDATA_PATH, filepath[:-4])
    if not os.path.isdir(unzip_dir) or not len(os.listdir(unzip_dir)) == len(
            z.filelist()):
        z.extractall(path=unzip_dir)
    w2v_paths = [
        os.path.join(BIGDATA_PATH, filename[:-4] + '.w2v.txt')
        for filename in os.listdir(unzip_dir)
    ]
    for (filename, word2vec_output_file) in zip(os.listdir(unzip_dir),
                                                w2v_paths):
        if filename.lower().endswith('.txt'):
            glove_input_file = os.path.join(unzip_dir, filename)
            glove2word2vec(glove_input_file=glove_input_file,
                           word2vec_output_file=word2vec_output_file)
    return w2v_paths