imap-archiver/archiver.py

#!/usr/bin/env python3
import argparse
import logging
import sys
from datetime import datetime
from pathlib import Path

import humanize
import yaml

from iarchiver.database import EmailDatabase
from iarchiver.email import unix_timestamp_since_to_imap_timestamp
from iarchiver.mail_conn import MailConnection


def main(args):
    logging.basicConfig()
    logger = logging.getLogger('iarchiver')
    logger.setLevel(logging.INFO)
    with open(args.config) as f:
        config = yaml.safe_load(f)

    exclude_folders = config.get('exclude_folders', [])

    if not config.get('server') or not config.get('username') or not config.get('password') or not config.get('database_path') or not config.get('attachments_path'):
        logger.critical('Bad config file.')
        sys.exit(1)

    save_inline_attachments = config.get('save_inline_attachments', False)
    if save_inline_attachments:
        logger.info('Saving inline attachments as well')

    attachments_dir = Path(config['attachments_path'])
    attachments_dir.mkdir(parents=True, exist_ok=True)

    database = EmailDatabase(Path(config['database_path']))
    mail = MailConnection(config['server'], config['username'], config['password'], attachments_dir, save_inline_attachments=save_inline_attachments)
    mail.load_folders()

    if config['server'] == 'imap.gmail.com':
        mail.folder_structure.remove('INBOX')  # We will use "'[Gmail]/All Mail'" instead
        mail.folder_structure.remove('[Gmail]')
    num_folders_to_sync = len(mail.folder_structure)
    logger.info(f'Syncing {num_folders_to_sync} folders...')

    last_refresh = database.have_we_done_a_full_sync_at_all()
    last_refresh_imap_date = None
    if last_refresh:
        last_refresh_imap_date = unix_timestamp_since_to_imap_timestamp(last_refresh)
        logger.info(f'Syncing emails received after {last_refresh_imap_date}')

    new_emails = 0
    new_attachments = 0
    did_full_sync = False
    sync_start_time = datetime.now()

    for folder_name in mail.folder_structure:
        if folder_name in exclude_folders:
            # Exclude folder
            continue
        logger.info(folder_name)
        if last_refresh:
            search_criterion = ['(SINCE "' + last_refresh_imap_date + '")']
        else:
            did_full_sync = True
            search_criterion = ['ALL']

        for email in mail.fetch_folder(folder_name, search_criterion=search_criterion):
            timestamp, to_email, from_email, subject, raw_content, attachments = email
            is_new_email = database.insert_email(folder_name, timestamp, subject, raw_content, to_email, from_email, attachments)
            if is_new_email:
                new_emails += 1
            if len(attachments):
                new_attachments += 1

    elapsed = datetime.now() - sync_start_time
    database.finish_sync('refresh' if not did_full_sync else 'full', new_emails, new_attachments, int(elapsed.total_seconds()))
    logger.info(f'Finished email {"refresh" if not did_full_sync else "sync"} in {humanize.naturaldelta(elapsed)} and added {new_emails} new emails and {new_attachments} attachments.')


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='Sync and archive your IMAP server.')
    parser.add_argument('--config', default='config.yml', help='Path to config file.')
    args = parser.parse_args()
    main(args)