embed-multi with CLIP - error message not very helpful in pinpointing the file that causes the error

I ran llm embed-multi gallery --files pictures '**/*.png' --binary -m clip to embed hundreds of PNG files with the CLIP model, and the model encountered error on an image and threw a Traceback like this:

...
- image_processor: CLIPImageProcessor {
  "crop_size": {
    "height": 224,
    "width": 224
  },
  "do_center_cr...
  File "/Users/tscp/.local/pipx/venvs/llm/lib/python3.11/site-packages/transformers/image_processing_utils.py", line 552, in __call__
    return self.preprocess(images, **kwargs)
           │               │         └ {'return_tensors': 'pt', 'padding': True}
           │               └ [<PIL.PngImagePlugin.PngImageFile image mode=RGB size=4032x3024 at 0x130EFD750>, <PIL.PngImagePlugin.PngImageFile image mode=RGB...
           └ CLIPImageProcessor {
  "crop_size": {
    "height": 224,
    "width": 224
  },
  "do_center_crop": true,
  "do_convert_rgb": tru...
  File "/Users/tscp/.local/pipx/venvs/llm/lib/python3.11/site-packages/transformers/models/clip/image_processing_clip.py", line 283, in preprocess
    images = [convert_to_rgb(image) for image in images]
    │         │                                  └ [<PIL.PngImagePlugin.PngImageFile image mode=RGB size=4032x3024 at 0x130EFD750>, <PIL.PngImagePlugin.PngImageFile image mode=RGB...
    │         └ <function convert_to_rgb at 0x13050fc40>
    └ [<PIL.PngImagePlugin.PngImageFile image mode=RGB size=4032x3024 at 0x130EFD750>, <PIL.PngImagePlugin.PngImageFile image mode=RGB...
  File "/Users/tscp/.local/pipx/venvs/llm/lib/python3.11/site-packages/transformers/models/clip/image_processing_clip.py", line 283, in <listcomp>
    images = [convert_to_rgb(image) for image in images]
              │              │          └ <PIL.PngImagePlugin.PngImageFile image mode=RGB size=3024x4032 at 0x130EFADD0>
              │              └ <PIL.PngImagePlugin.PngImageFile image mode=RGB size=3024x4032 at 0x130EFADD0>
              └ <function convert_to_rgb at 0x13050fc40>
  File "/Users/tscp/.local/pipx/venvs/llm/lib/python3.11/site-packages/transformers/image_transforms.py", line 762, in convert_to_rgb
    image = image.convert("RGB")
    │       └ <PIL.PngImagePlugin.PngImageFile image mode=RGB size=3024x4032 at 0x130EFADD0>
    └ <PIL.PngImagePlugin.PngImageFile image mode=RGB size=3024x4032 at 0x130EFADD0>
  File "/Users/tscp/.local/pipx/venvs/llm/lib/python3.11/site-packages/PIL/Image.py", line 922, in convert
    self.load()
    └ <PIL.PngImagePlugin.PngImageFile image mode=RGB size=3024x4032 at 0x130EFADD0>
  File "/Users/tscp/.local/pipx/venvs/llm/lib/python3.11/site-packages/PIL/ImageFile.py", line 278, in load
    raise OSError(msg) from e
                  └ 'image file is truncated'
OSError: image file is truncated

I can see the image that caused the error was <PIL.PngImagePlugin.PngImageFile image mode=RGB size=3024x4032 at 0x130EFADD0>, but I didn't know the path of that particular image.

Do you think it'd be a good idea to handle such exceptions when running embedding with more than 1 file by also outputing the path that causes the error?

Start at the llm embed-multi command and trace how multiple files are loaded and passed to the CLIP embedding model. Reproduce the failure with the command shown, then ensure an exception identifies the source path for the failing image; done means the error output names that path.