make-work-email-mbox.py

#!/usr/bin/python

import mailbox
import os
import re
from subprocess import check_output
import sys

top_level_mail_directory = os.path.join(os.environ['HOME'], 'Mail')

if len(sys.argv) != 2:
    print "Usage: OUTPUT-MBOX-FILENAME"
    sys.exit(1)

output_filename = sys.argv[1]

def handle_mail(filename):
    # header_name = 'Envelope-to'
    # work_header_value = 'mark-mysocietyaddress@longair.net'
    # header_name = 'Delivered-To'
    # work_header_value = 'mark@mysociety.org'
    header_name = 'Envelope-to'
    work_header_value = 'mark-mysocietycron@longair.net'
    with open(filename, 'rb') as file_input:
        header_value = check_output(
            ['formail', '-x', header_name],
            stdin=file_input).strip()
    if header_value != work_header_value:
        return
    print filename
    with open(output_filename, 'ab') as mbox_output:
        with open(filename, 'rb') as file_input:
            mbox_output.write("From IGNORETHIS\n")
            mbox_output.write(file_input.read())
            mbox_output.write("\n\n")

for folder_name in sorted(os.listdir(top_level_mail_directory)):
    nnml_directory = os.path.join(top_level_mail_directory, folder_name)
    if not os.path.isdir(nnml_directory):
        continue
    if folder_name in ('spam', 'temporary'):
        continue
    for numbered_filename in os.listdir(nnml_directory):
        if not re.search(r'^\d+$', numbered_filename):
            continue
        full_mail_filename = os.path.join(nnml_directory, numbered_filename)
        handle_mail(full_mail_filename)