抑制 numpy 异常消息

Question

I have some code wrapped with try except that read .npz arrays

It can produce exceptions like:

Exception ignored in: <bound method NpzFile.__del__ of <numpy.lib.npyio.NpzFile object at 0x12dd65cf8>>
Traceback (most recent call last):
  File "/usr/local/lib/python3.6/site-packages/numpy/lib/npyio.py", line 226, in __del__
    self.close()
  File "/usr/local/lib/python3.6/site-packages/numpy/lib/npyio.py", line 217, in close
    if self.zip is not None:
AttributeError: 'NpzFile' object has no attribute 'zip'

Is it possible to suppress this message?

Code looks like:

video_dir_list = get_video_dir_list(input_dir)
for video_dir in tqdm(video_dir_list):
    try:
        img_filepath_list = get_filepaths_by_extension(video_dir, ['*.jpg'])
        for img_filepath in img_filepath_list:
            data = load_npz_files(img_filepath)
    except Exception as e:
        BROKEN_VIDEO_LIST.append(video_dir)

Answer 1

If you just want to passively handle an exception, add another except clause-- except AttributeError: #handle this attribute error

Also, is there a relationship between this exception, and the ignored one?

Answer 2

I encountered your same problem when I wanted to test if a given file was npz. Since the error is thrown when the file is closed, you will not catch it when opening the file. You can try closing the npz file within the try-except block. Don't forget to read out the data before closing if you would like to use the data.

def careful_loadz(path):
    data = None

    try:
        npz = np.load(path)
        data = [npz[f] for f in npz.files]
        npz.close()

    except (ValueError, BadZipFile, AttributeError):
        print(f"Error reading the input file {path}.")

    return data