Update created and updated date from data from Google's takeout from json file and delete json file.

update_date.py

"""
Update created and updated system date from data from Google's takeout from json file 
and delete json file.

Change the PATH_TO_SCAN and execute this script.
Sometimes we have missing files from takeout (see error logs)
"""
import glob
import os
import json
import subprocess
from datetime import datetime

PATH_TO_SCAN = "/Users/david/Documents/migration/tak/"


def search_pattern_files(path):
    txtfiles = []
    for file in glob.glob(path):
        txtfiles.append(file)
    return txtfiles


def clean_title(image_name):
    # replace ' by _
    return image_name.replace('\u0027', '_').replace('%', '_')


def split_title(image_name):
    # Split and cut
    name_splits = image_name.split('.')
    extension = name_splits[len(name_splits)-1:][0]
    name = ".".join(name_splits[:len(name_splits)-1])
    name = name[:47]
    return f"{name}.{extension}"


def update_image_date_system(dir_path, image_name, date):
    image_name = clean_title(image_name)
    date = datetime.fromtimestamp(int(date))

    print(f"Directory: {dir_path}, image: {image_name}, for date: {date}")

    try:
        subprocess.check_call(["ExifTool", "-r",
                               f"-FileCreateDate={date}",
                               f"-FileModifyDate={date}",
                               f"{dir_path}{image_name}"])

    except subprocess.CalledProcessError:
        # google have 47 chars max image name
        image_clean_name = split_title(image_name)
        print(f"### RETRY ### Directory: {dir_path}, image: {image_clean_name}, for date: {date}")
        subprocess.check_call(["ExifTool", "-r",
                               f"-FileCreateDate={date}",
                               f"-FileModifyDate={date}",
                               f"{dir_path}{image_clean_name}"])


if __name__ == '__main__':
    errors = []
    count_success = 0
    for root, dirs, files in os.walk(PATH_TO_SCAN):
        print(f"Root: {root} with {len(dirs)} folders")
        for _dir in dirs:
            path_dir = f'{root}{_dir}/'
            json_files_path = search_pattern_files(f"{path_dir}*.json")
            for json_path in json_files_path:
                with open(json_path) as json_file:
                    data = json.load(json_file)
                print(f"Json PATH: {json_path}")
                photo_taken_time = data.get('photoTakenTime') or None
                if not photo_taken_time:
                    print('Json invalid - stop')
                    print(f'****** DELETE ****** {json_path}')
                    os.remove(json_path)
                    continue
                image_name_to_update = data['title']
                timestamp_to_replace = photo_taken_time['timestamp']
                try:
                    update_image_date_system(path_dir,
                                             image_name_to_update,
                                             timestamp_to_replace)
                except subprocess.CalledProcessError as e:
                    errors.append(f"ERROR {e} - "
                                  f"json file: {json_path}")
                else:
                    count_success += 1
                    print(f'****** DELETE ****** {json_path}')
                    os.remove(json_path)
    print("###################")
    print("-------------------")
    for er in errors:
        print(er)
    print("-------------------")
    print(f'SUCCESS: {count_success} ERROR: {len(errors)}')
    print("###################")