__init__.py 文件源码

def _mnist_raw_data():
    fname = resource_filename('sudokuextract.data', "train-images-idx3-ubyte.gz")
    if resource_exists('sudokuextract.data', "train-images-idx3-ubyte.gz"):
        f = gzip.open(fname, mode='rb')
        data = f.read()
        f.close()
    else:
        sio = StringIO(urlopen(_url_to_mnist_train_data).read())
        sio.seek(0)
        f = gzip.GzipFile(fileobj=sio, mode='rb')
        data = f.read()
        f.close()
        try:
            sio.seek(0)
            with open(fname, 'wb') as f:
                f.write(sio.read())
        except:
            pass

    correct_magic_number = 2051
    magic_number = _toS32(data[:4])
    if magic_number != correct_magic_number:
        raise ValueError("Error parsing images file. Read magic number {0} != {1}!".format(
            magic_number, correct_magic_number))
    n_images = _toS32(data[4:8])
    n_rows = _toS32(data[8:12])
    n_cols = _toS32(data[12:16])
    images = np.fromstring(data[16:], 'uint8').reshape(n_images, n_rows*n_cols)

    return [imrow.reshape(28, 28) for imrow in images]