Is there a way to add image height and width in theme automagically

[aledeg]

When checking for accessibility, one of the things that bothers me its the warning about image sizes. I search how to do that without including it manually on all included images. I was wondering if there is something on theme level I can do to address that.

[Kwpolska]

Nikola can’t do that automagically, and that probably wouldn’t be done on the theme level. It could be done by post compilers, I don’t think reST has an option for this. You could try extending your input format of choice with that feature, or perhaps a filter — but note that reading the images to get their sizes might increase build time a lot (especially if you try to do that with remote images.)

[aledeg]

Actually, I know the image size but not the orientation. I am trying to add that on my thumbnails.
I guess I will have to do it by hand.
Thank you!

[Kwpolska]

Figuring out the orientation would still require reading the images.

[felixfontein]

For most sane images formats, it suffices to read the header / metadata to get that information. For example, Pillow's Image.open() only reads metadata (until you explicitly tell the resulting object to load the pixel data), and you can use the width and height properties of that object without loading the pixel data. That's pretty quick even if you have quite a few images.

For example, I ran

import os
from PIL import Image

data = []
Image.MAX_IMAGE_PIXELS = None
for p, _, fns in os.walk('.'):
    for fn in fns:
        if not fn.endswith(('.png', '.jpg', '.jpeg', '.JPG', '.JPEG')):
            continue
        fn = os.path.join(p, fn)
        try:
            with Image.open(fn) as i:
                data.append((fn, i.width, i.height))
        except Exception as e:
            print('{0}: {1}'.format(fn, e))
print(len(data))

in a folder with ~5500 files, it took 0.56 seconds.

When I ran it in a larger photo collection (~750 GB, ~140'000 files), it took 5.5 minutes. So for a certain amount of images, caching makes sense :) But still, it is a lot faster than actually reading all that image data :)