Run this script on a Mac to find all email addresses of persons who wrote emails to a specific email address.

find_emails.py

#!/usr/bin/python
# -*- coding: iso-8859-1 -*-
__version__ = "$Revision: 0.1 $"
__author__ = "Pierrick Terrettaz"
__date__ = "2008-05-08"


import popen2
import re
import sys
import os

if __name__ == '__main__':
    prog = os.path.basename(sys.argv[0])

    if len(sys.argv) < 2:
        print '== Find all sender addresses in your mailboxes =='
        print 'Usage: ' + prog + ' emailTo [-onlyin path]'
        print '       emailTo : your email address'
        print '       -onlyin : find onlyin a certain folder scope'
        print '                 (for more information see mdfind help)'
        print 'Requires : Mac OSX platform with mdfind, mdls programs'
        print '\n       Example : '+prog+' [email protected] -onlyin /Users/pik/Library/Mail/Mailboxes'
        sys.exit(1)
    
    onlyin = ''
    if len(sys.argv) >= 4 and sys.argv[2] == '-onlyin':
        onlyin = '-onlyin "%s"' % sys.argv[3]

    re_email = re.compile('.*"(.+)".*')
    
    addresses = {}
    
    mdfind_cmd = 'mdfind '+onlyin+' "(kMDItemRecipientEmailAddresses == \''+sys.argv[1]+'\')"'
    
    (stdout, stdin) = popen2.popen2(mdfind_cmd)
    for path in stdout.xreadlines():
        path = path[:-1]
            
        (stdout2, stdin2) = popen2.popen2('mdls -name kMDItemAuthorEmailAddresses -raw "'+path+'"')
        for line in stdout2.xreadlines():
            line = line[:-1]
            res = re_email.findall(line)
            if len(res) == 1:
                address = res[0]
                if address not in addresses:
                    addresses[address] = 0
                addresses[address] += 1
    
    rev_items = [(v, k) for k, v in addresses.items()]
    rev_items.sort()
    
    for v, k in rev_items:
        print "%d\t: %s" % (v, k)