consolidate_takeout.py

#! /usr/bin/env python3

import os
import json
import re
import argparse
import shutil
import logging
import sys

# Usage:
# First generate a file_list.txt
# cd "Google Photos"
# find . -type f > file_list.txt
# Then run the script
# consolidate_takeout.py . file_list.txt album_name

def parse_arguments():
    parser = argparse.ArgumentParser(description='Process a specific album in a takeout directory and a file list.')
    parser.add_argument('takeout_dir', type=str, help='The directory containing the takeout files.')
    parser.add_argument('file_list', type=str, help='The path to the file list (file_list.txt).')
    parser.add_argument('album_name', type=str, help='The name of the album to process.')
    return parser.parse_args()

def consolidate_album(album_dir, file_list):
    # Set up logging
    logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(message)s')

    # Read the file list into a set for quick lookup
    # file_list has full paths relative to the album directory
    with open(file_list, 'r') as f:
        file_set = set(line.strip() for line in f)

    # Check if the folder name starts with 20XX, if so get the year.
    # We'll use it to restrict the search for the same picture name as
    # we often have multiple pictures with the same name in different years.
    album_year = re.match(r'2\d{3}', os.path.basename(album_dir)).group() if re.match(r'2\d{3}', os.path.basename(album_dir)) else None
    print (f"Album year: {album_year}")

    # Iterate over files in the album directory
    for filename in os.listdir(album_dir):
        if filename.endswith('.supplemental-metadata.json'):
            image_file = filename.replace('.supplemental-metadata.json', '')
            image_path = os.path.join(album_dir, image_file)

            # Check if the image file exists
            if not os.path.exists(image_path):
                # Check for a unique match in the file list
                matches = [f for f in file_set if os.path.basename(f) == image_file and (album_year is None or album_year in f)]
                if len(matches) == 1:
                    # If there is a single match, it must be the right one, just copy it and we're done.
                    shutil.copy(matches[0], album_dir)
                    logging.info(f"[SUCCESS] copied {matches[0]} to {album_dir}")
                elif len(matches) > 1:
                    # Multiple matches, we'll need to manually review which ones correspond to the
                    # album.
                    logging.error(f"Multiple matches found for {image_file} in file list: {matches}")
                else:
                    logging.error(f"No match found for {image_file} in file list.")

def main():
    args = parse_arguments()
    takeout_dir = args.takeout_dir
    album_name = args.album_name
    file_list = args.file_list
    
    # Process the specified album directory
    album_dir = os.path.join(takeout_dir, album_name)
    if os.path.isdir(album_dir):
        consolidate_album(album_dir, file_list)
    else:
        logging.error(f"The specified album directory {album_dir} does not exist.")

if __name__ == "__main__":
    main()