Store messages in a Maildir into a couchdb database.

maildirtocouch.py

#!/usr/bin/env python
"""Store messages in a Maildir into a couchdb database."""

import couchdb
from mailbox import Maildir
from optparse import OptionParser
import os
from pprint import pprint
import sys
from uuid import uuid4


def error(msg):
    """Insert blatantly obvious docstring here."""

    sys.stderr.write('%s\n' % msg)
    sys.exit(1)


def insertmessages(database, maildir, verbose):
    """Put each maildir message in CouchDb."""

    inserted = 0
    ignored = 0
    count = 0
    for fname, message in maildir.iteritems():
        count += 1
        # is it already in there?
        doc = database.get(fname)
        if doc:
            print '%s already in db; skipping' % fname
            ignored += 1
            continue

        (fromname, fromaddr) = message.getaddr('From')
        data = {'type': 'email',
                'fromname': fromname.decode('utf-8', 'ignore'),
                'fromaddr': fromaddr,
                'fname': fname,
                'fullpath': message.fp._file.name}
        for header in ('delivered-to', 'subject', 'date', 'status', 'sender',
                       'message-id', 'in-reply-to', 'references'):
            data[header] = message.get(header, '').decode('utf-8', 'ignore')
        for header in ('to', 'cc'):
            data[header] = message.getaddrlist(header)

        # doc_id = uuid4().hex
        doc_id = fname
        if verbose:
            print 'putting %s...' % data['fullpath']
        try:
            database[doc_id] = data
        except Exception, exc:
            pprint(data)
            error('exception for %s:, %s' % (fname, str(exc)))
        doc = database.get(doc_id)
        if doc:
            # TODO: handle failure here
            database.put_attachment(doc, message.fp, fname,
                                    content_type='message/rfc822')
        inserted += 1

        if count % 100 == 0:
            print 'messages processed: %d' % count

    print 'inserted messages: %d ignored: %d' % (inserted, ignored)
        


def main():
    """Main branching logic."""

    home = os.environ.get('HOME', os.getcwd())

    parser = OptionParser('usage: %prog -d dbname [other options]')
    parser.add_option('-v', '--verbose', dest='verbose', action='store_true',
                      help='enable loquacious mode.  like, duh.')
    parser.add_option('-d', '--database', dest='database',
                      help='mail database in CouchDb [required]')
    parser.add_option('-s', '--server', dest='server',
                      help='CouchDb URI [default: %default]',
                      default='http://localhost:5984/')
    parser.add_option('-m', '--maildir', dest='maildir',
                      help='path to the maildir to import [default: %default]',
                      default='%s/Maildir/' % home)

    (opts, _args) = parser.parse_args()

    if not opts.database:
        parser.error('a database name is required.')

    if not (os.access(opts.maildir, os.F_OK) and
            os.access(os.sep.join([opts.maildir, 'new']), os.F_OK)):
        parser.error('cannot access maildir %s' % opts.maildir)
    maildir = Maildir(opts.maildir)

    server = couchdb.Server(opts.server)
    try:
        _version = server.version
    except AttributeError, _exc:
        error('cannot connect to %s' % opts.server)

    try:
        database = server[opts.database]
    except couchdb.client.ResourceNotFound, _exc:
        error('invalid database name: %s' % opts.database)

    insertmessages(database, maildir, opts.verbose)


if __name__ == '__main__':
    main()

# eof

https://github.com/nborwankar/sproingg might be of interest to you. similar in nature to this gist

Thanks!