fix_images.py

from typing import NamedTuple
from pathlib import Path
from argparse import ArgumentParser
from shutil import move
from imghdr import what
from hashlib import sha1
import cv2

ext_map = {
    'jpeg': 'jpg',
}

hash_map = {}


class Info(NamedTuple):
    fn: Path
    shape: tuple[int, int, int]
    size: int

    def better_than(self, other: 'Info') -> bool:
        w, h, *_ = self.shape
        ow, oh, *_ = other.shape
        if w > ow or h > oh:
            return True
        elif w < ow or h < oh:
            return False
        else:
            return self.size > other.size


def check_image(fn: Path, target_size=(320, 240)) -> None:
    image = cv2.imread(str(fn))
    if image is None:
        print('Unable to read', fn)
        return
    shape = image.shape
    image = cv2.resize(image, target_size)
    image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    image = cv2.normalize(image, None, alpha=0, beta=3, norm_type=cv2.NORM_MINMAX)
    h = sha1(image.data).hexdigest()
    file_size = fn.stat().st_size
    new_item = Info(fn, shape, file_size)
    if item := hash_map.get(h, False):
        if not new_item.better_than(item):
            new_item, item = item, new_item
        hash_map[h] = new_item
        print(f'Deleting {item.fn} as copy of {new_item.fn}')
        item.fn.unlink()
        #cv2.imshow('TEST', image * 64)
        #cv2.waitKey(-1)
    else:
        hash_map[h] = new_item


def main():
    parser = ArgumentParser('Images fixer')
    parser.add_argument('source', type=Path, help='Images directory')
    args = parser.parse_args()
    src_dir: Path = args.source
    for item in src_dir.glob('**/*'):
        if item.is_file() and (t := what(item)) is not None:
            ext = f'.{ext_map.get(t, t)}'
            if item.suffix != ext:
                new_name = item.with_suffix(ext)
                i = 1
                while new_name.is_file():
                    new_name = new_name.with_stem(f'{new_name.stem}_{i}')
                    i += 1
                print(f'Rename {item} to {new_name}')
                move(item, new_name)
                item = new_name
            check_image(item)


if __name__ == '__main__':
    main()