dataload.py 文件源码

def anyfile(infile, mode='r', encoding="utf8"):
    '''
    return a file handler with the support for gzip/zip comppressed files
    if infile is a two value tuple, then first one is the compressed file;
      the second one is the actual filename in the compressed file.
      e.g., ('a.zip', 'aa.txt')

    '''
    if isinstance(infile, tuple):
        infile, rawfile = infile[:2]
    else:
        rawfile = os.path.splitext(infile)[0]
    filetype = os.path.splitext(infile)[1].lower()
    if filetype == '.gz':
        import gzip
        in_f = io.TextIOWrapper(gzip.GzipFile(infile, 'r'),encoding=encoding)
    elif filetype == '.zip':
        import zipfile
        in_f = io.TextIOWrapper(zipfile.ZipFile(infile, 'r').open(rawfile, 'r'),encoding=encoding)
    else:
        in_f = open(infile, mode, encoding=encoding)
    return in_f