file_utils.py 文件源码

def read_gzip(filepath):
    # handles buffering # TODO: profile whether this is fast.
    with gzip.open(filepath, 'rb') as f: # leave in binary mode (default), let TextIOWrapper decode
        with io.BufferedReader(f, buffer_size=2**18) as g: # 256KB buffer
            with io.TextIOWrapper(g) as h: # bytes -> unicode
                yield h