Symbiosis email archiver

archive_emails.py

#!/usr/bin/env python
import os
import os.path
import time
import shutil
import glob
import email


def get_maildirs():
    ignored_domains = []
    ignored_users = {}

    domains = []
    for mbox in glob.glob('/srv/*/mailboxes/*/Maildir'):
        parts = mbox.split('/')
        # Ignored domains
        if parts[2] in ignored_domains:
            continue
        # Ignored users
        if parts[2] in ignored_users.keys() and \
           parts[4] in ignored_users[parts[2]]:
            continue
        domains.append(mbox)

    return domains


def archive_emails(mbox):
    if not os.path.isdir(mbox):
        return False

    archive_limit = 31536000  # 365 days
    for root, dirs, files in os.walk(mbox, topdown=False):
        for f in files:
            path = os.path.join(root, f)

            try:
                fh = open(path, 'r')
                mail = email.message_from_file(fh)
                date = email.utils.parsedate(mail['date'])
                if not date:
                    continue
                year = time.strftime("%Y", date)
                ctime = time.mktime(date)
            except:
                continue

            if (time.time() - ctime) > archive_limit:
                parts = path.split('/')
                msg_file = parts[-1]
                folder_dir = parts[-2]
                folder = parts[-3]

                if folder_dir in ['tmp', 'cur', 'new']:
                    if folder.startswith('.Archives.'):
                        lf = '.'.join(folder.split('.')[3:])
                    else:
                        if folder == 'Maildir':
                            lf = 'Inbox'
                        else:
                            lf = folder

                    archive_folder = '.Archives.%s.%s' % (year, lf.lstrip('.'))
                    dir_path = os.path.join(mbox, archive_folder, folder_dir)
                    archive_path = os.path.join(dir_path, msg_file)

                    if not os.path.isdir(dir_path):
                        os.makedirs(dir_path)

                    if path != archive_path:
                        print 'mv %s to %s' % (path, dir_path)
                        shutil.move(path, archive_path)

if __name__ == '__main__':
    for mbox in get_maildirs():
        archive_emails(mbox)