This is a script I used to clean up the havoc Thunderbird caused by copying thousands of messages from INBOX as five copies to a folder

imapdedup.py

"""Helpers for removing duplicate messages in an IMAP mailbox

Dependency: IMAPClient (easy_install IMAPClient)

Author: Antti Kaihola <[email protected]>

License: New BSD license
"""


import collections
import imapclient


class IMAPDeDupper(object):
    def __init__(self, client):
        self.client = client

    def get_all_messages(self):
        uids = self.client.search()
        return self.client.fetch(uids, ('internaldate', 'envelope',))

    def make_fingerprint(self, msg):
        return msg['INTERNALDATE'], msg['ENVELOPE'][:9]

    def group_messages_by_fingerprint(self):
        """Returns a dict grouping duplicate messages in current folder

        Message fingerprints as returned by ``make_fingerprint`` are
        used as dict keys, and lists of message UIDs as dict values.
        """
        msgs = self.get_all_messages()
        grouped = collections.defaultdict(list)
        for uid, msg in msgs.iteritems():
            fingerprint = self.make_fingerprint(msg)
            grouped[fingerprint].append(uid)
        return grouped

    def get_duplicates(self):
        """Returns UIDs of duplicate and unique messages in the current folder

        Return value: a 2-tuple of lists ``([<duplicates>], [<uniques>])``

        For sets of duplicate messages, the first message is not
        included in the lists, and following messages are added to
        the ``duplicates`` list.
        """
        grouped = self.group_messages_by_fingerprint()
        uniques = []
        duplicates = []
        for fingerprint, uids in grouped.iteritems():
            if len(uids) > 1:
                duplicates.extend(uids[1:])
            else:
                uniques.append(uids[0])
        return duplicates, uniques

    def get_matching(self, fingerprints):
        """Deletes matching messages in the current folder

        Given a list (or dict with keys) of message fingerprints,
        searches the current folder for messages matching any of the
        fingerprints.

        Return value: a 2-tuple of UID lists
        ``([<matching>], [<nonmatching>])``
        """
        grouped = self.group_messages_by_fingerprint()
        matching = []
        nonmatching = []
        for fingerprint, uids in grouped.iteritems():
            if fingerprint in fingerprints:
                matching.extend(uids)
            else:
                nonmatching.extend(uids)
        return matching, nonmatching


def example():
    c = imapclient.IMAPClient('imap.mydomain.com')
    c.login('login', 'password')
    d = IMAPDeDupper(c)

    # delete duplicate messages in folder ``2009``
    c.select_folder('2009')
    dups, uniques = d.get_duplicates()
    print 'In folder "2009", deleting message UIDS:\n%s' % repr(dups)
    c.delete_messages(dups)
    c.expunge()

    # delete inbox messages which exist in folder ``2009``
    c.select_folder('2009')
    twoten = d.group_messages_by_fingerprint()
    c.select_folder('INBOX')
    dups, uniqs = d.get_matching(twoten)
    print 'In the inbox, deleting message UIDS:\n%s' % repr(dups)
    c.delete_messages(dups)
    c.expunge()



if __name__ == '__main__':
    pass
    # TODO: implement command line interface