Python: Detect a rare strange GIF when extracting its frames

Question

I have collected a large-scale datasets of gifs(more than 100k) from the Internet, and I meet some rare strange GIFs when I try to extract frames of GIFs with python. Three common used packages(moviepy, PIL, imageio) provide totally different results of such rare strange gifs.

1. moviepy>=1.0.3 will block in VideoFileClip.iter_frames() loop at the second frame forever, and the code won't throw an exception.

from moviepy.video.io.VideoFileClip import VideoFileClip

video = VideoFileClip(path)
frame_iterator = video.iter_frames()

2. PIL>=7.1.2 will output multiple frames as same as the first frame.

from PIL import Image, ImageSequence
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

video = Image.open(path)
frame_iterator = ImageSequence.Iterator(video)

3. imageio>=2.6.1 can extract the frames correctly while the output frames are strange.

import imageio

frame_iterator = imageio.get_reader(path)

Then you can dump the frames from frame_iterator provided by these packages above:

def dump_video_frames(video_frames):
    root = 'data/frames'
    if os.path.exists(root):
        shutil.rmtree(root)
    os.makedirs(root)

    for i, frame in enumerate(video_frames):
        frame.save(os.path.join(root, '%d.jpg' % i))

frames = []
for frame in frame_iterator:
    if isinstance(frame, np.ndarray):
        frame = Image.fromarray(np.uint8(frame))
    frames.append(frame.convert('RGB'))

dump_video_frames(frames)

Here is an example:

Original GIF:

The output of PIL:

The output of imageio:

You can see PIL only get the first frame without any black area which is quite different with the output of imageio.

So my question is how to detect such a strange gif in python? Since I use moviepy first for its good performance in other gifs, I need to detect such a kind of GIF before the code use moviepy to extract its frames in order to avoid the infinite loop in VideoFileClip.iter_frames() which won't throw any exception. I can't get any information about such a rare gif from Google.

I will provide 2 more example GIFs below: