generate_askfm_assets.py

#!/usr/bin/env python3
import sys
import re

def main():
    if len(sys.argv) < 2:
        print("Usage: generate_askfm_assets.py filename1 [filename2 ...]")
        sys.exit(1)

    input_filenames = sys.argv[1:]

    small_thumbnails = set()
    big_thumbnails = set()
    videos = set()

    for input_filename in input_filenames:
        try:
            with open(input_filename, 'r') as f:
                for line in f:
                    filename = line.strip()
                    if not filename:
                        continue
                    # Match the filename pattern and extract the ID
                    match = re.match(r'p_video_answer_(\d+)(_big)?\.(jpg|mp4)', filename)
                    if match:
                        id_num = match.group(1)
                        # Pad the ID to at least 9 digits
                        id_padded = id_num.zfill(9)
                        # Split into three groups of three digits
                        groups = [id_padded[i:i+3] for i in range(0, 9, 3)]
                        # Build the path
                        path = '/'.join(groups)
                        # Build the base URL
                        base_url = f"asset:cvad.ask.fm/{path}/p_video_answer_{id_num}"
                        # Determine which type of file it is
                        if match.group(2) == '_big':
                            # It's a big thumbnail
                            big_thumbnail_url = f"{base_url}_big.jpg"
                            big_thumbnails.add(big_thumbnail_url)
                            # Generate small thumbnail and video URLs
                            small_thumbnail_url = f"{base_url}.jpg"
                            video_url = f"{base_url}.mp4"
                            small_thumbnails.add(small_thumbnail_url)
                            videos.add(video_url)
                        else:
                            # It's a small thumbnail or video
                            extension = match.group(3)
                            if extension == 'jpg':
                                small_thumbnail_url = f"{base_url}.jpg"
                                small_thumbnails.add(small_thumbnail_url)
                                # Generate big thumbnail and video URLs
                                big_thumbnail_url = f"{base_url}_big.jpg"
                                video_url = f"{base_url}.mp4"
                                big_thumbnails.add(big_thumbnail_url)
                                videos.add(video_url)
                            elif extension == 'mp4':
                                video_url = f"{base_url}.mp4"
                                videos.add(video_url)
                                # Generate small and big thumbnail URLs
                                small_thumbnail_url = f"{base_url}.jpg"
                                big_thumbnail_url = f"{base_url}_big.jpg"
                                small_thumbnails.add(small_thumbnail_url)
                                big_thumbnails.add(big_thumbnail_url)
                    else:
                        print(f"Invalid filename format in {input_filename}: {filename}", file=sys.stderr)
        except FileNotFoundError:
            print(f"File not found: {input_filename}", file=sys.stderr)
            continue

    # Write to output files
    with open('small_thumbnails.txt', 'w') as f:
        for url in sorted(small_thumbnails):
            f.write(url + '\n')
    with open('big_thumbnails.txt', 'w') as f:
        for url in sorted(big_thumbnails):
            f.write(url + '\n')
    with open('videos.txt', 'w') as f:
        for url in sorted(videos):
            f.write(url + '\n')

if __name__ == "__main__":
    main()