image_crawl/postprocess_image_names.py

#!/usr/bin/env python
if __name__ != '__main__': raise Exception("Do not import me!")

import os
import re

from cvargparse import Arg
from os.path import isfile, join, basename, splitext
from tqdm import tqdm

from utils import parser
from utils import imaging

def main(args):

	# class_names = imaging.get_classnames(args.classnames)

	content = imaging.get_content(args.folder, args.extensions)

	for root, fnames in tqdm(content):
		folder = basename(root)
		if not re.match(r"^\d+\..*$", folder): continue
		fnames = sorted(fnames, key=lambda name: int(re.match(r"^(\d+)\..*$", name).group(1)))
		for i, fname in enumerate(fnames):
			ext = splitext(fname)[1][1:].strip().lower()

			if ext == "jpeg":
				ext = "jpg"

			new_name = f"{folder}_{i:04d}.{ext}"
			new_path = join(root, new_name)
			assert not isfile(new_path), f"File exists: {new_path}"

			os.rename(join(root, fname), join(root, new_name))

main(parser.parse_args([
])