Script to extract all email addresses from all emails in all folders of an IMAP account.

email_addresses.py

"""Create a connection to an IMAP server and find ALL email addresses
Original script by abought: https://gist.github.com/abought/15a1e08705b121c1b7bd
References:
http://www.voidynullness.net/blog/2013/07/25/gmail-email-with-python-via-imap/
and
https://yuji.wordpress.com/2011/06/22/python-imaplib-imap-example-with-gmail/
"""
__author__ = 'jmb'

import email
import imaplib
import getpass
import sys
import re
import os


# EDIT these as required:
FILENAME = 'out.txt'
DEFAULT_MAIL_SERVER = 'imap.server'

# No user parameters below this line
# ADDR_PATTERN = re.compile('<(.*?)>')  # Finds email as <nospam@nospam.com>
# Find ALL email address in all fields:
ADDR_PATTERN = re.compile(r"(^[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+$)")


def connect(user, pwd, server=DEFAULT_MAIL_SERVER):
    """Connect to [the specified] mail server. Return an open connection"""
    conn = imaplib.IMAP4_SSL(server)
    try:
        conn.login(user, pwd)
    except imaplib.IMAP4.error:
        print "Failed to login"
        sys.exit(1)
    return conn


def print_folders(conn):
    """Print a list of open mailbox folders"""
    for f in conn.list():
        print " ", f


def get_folder(conn, folder_name):
    """Fetch a specific folder (or label) from server"""

    if conn.state == "SELECTED":
        # Explicitly close any previously opened folders; may not be necessary
        conn.close()

    rv, data = conn.select(folder_name)
    if rv != 'OK':
        print "Could not open specified folder. Known labels:"
        print_folders(conn)
    return conn


def get_email_ids(conn, query='ALL'):
    """Get the numeric IDs for all emails in a given folder"""
    if conn.state != "SELECTED":
        raise imaplib.IMAP4.error("Cannot search without selecting a folder")

    rv, data = conn.uid('search', None, query)
    if rv != 'OK':
        print "Could not fetch email ids"  # for some reason...
        return []

    return data[0].split()


def fetch_message(conn, msg_uid):
    """
    Fetch a specific message uid (not sequential id!) from the given folder;
    return the parsed message. User must ensure that specified
    message ID exists in that folder.
    """
    # TODO: Could we fetch just the envelope of the response to save bandwidth?
    rv, data = conn.uid('fetch', msg_uid, "(RFC822)")
    if rv != 'OK':
        print "ERROR fetching message #", msg_uid
        return {}

    return email.message_from_string(data[0][1])  # dict-like object


def get_recipients(msg_parsed):
    """Given a parsed message, extract and return recipient list"""
    recipients = []
    addr_fields = ['From', 'To', 'Cc', 'Bcc']

    for f in addr_fields:
        rfield = msg_parsed.get(f, "")  # Empty string if field not present
        rlist = re.findall(ADDR_PATTERN, rfield)
        recipients.extend(rlist)

    return recipients


if __name__ == "__main__":
    username = raw_input("Full email address: ")
    password = getpass.getpass()

    # Connect
    mail_conn = connect(username, password)

    # Open output file
    file = open(FILENAME, "a")

    # Go through each folder
    for f in mail_conn.list()[1]:
        folder = f.split()[2].strip('"')
        if folder == ".":
            continue
        mail_conn = get_folder(mail_conn, folder)
        msg_uid_list = get_email_ids(mail_conn)
        print "Scanning folder: ", folder, " with ", len(msg_uid_list), " messages"

        # Fetch a list of recipients
        all_recipients = []
        for msg_uid in msg_uid_list:
            msg = fetch_message(mail_conn, msg_uid)
            recip_list = get_recipients(msg)
            all_recipients.extend(recip_list)

        print "Writing", len(set(all_recipients)), "email addresses to file", file.name
        output = ""
        for address in set(all_recipients):
            output = output + address + "\n"
        file.write(output)
        file.flush()
        os.fsync(file.fileno())

    file.close()
    print("\nWritten to file: " + FILENAME)

    try:
        mail_conn.close()  # Close currently selected folder (if any)
    finally:
        mail_conn.logout()