Remove all binary attachments from email messages

mailbinpurge.py

#!/usr/bin/python
# vim: set fileencoding=utf-8:
# Remove all binary attachments from email messages
#
# This is a standard UNIX filter; it reads a message or mailbox from standard
# input, and outputs it again on standard output, with all binary attachments
# removed (with message/external-body).
#
# Written by Frédéric Brière <[email protected]>.  Copy at will.

import email
import email.message
import optparse
import sys
import time

# Adapted from <http://code.activestate.com/recipes/576553/>, originally
# written by Romain Dartigues.
def messages_list(input):
    """Extract a list of messages from a mailbox.

    This parses a mbox mailbox, and returns a list of messages, as
    email.message.Message objects.  Unlike the mailbox module, this allows
    reading from standard input.

    """
    lines = []

    while True:
        line = input.readline()

        if line[:5] == 'From ' or line == '':
            if lines:
                yield email.message_from_string(''.join(lines))
                lines = []

        if line == '':
            return
        else:   
            lines.append(line)

def gut_message(message):
    """Remove body from a message, and wrap in a message/external-body."""
    wrapper = email.message.Message()
    wrapper.add_header('Content-Type', 'message/external-body',
                       access_type='x-spam-deleted',
                       expiration=time.strftime("%a, %d %b %Y %H:%M:%S %z"),
                       size=str(len(message.get_payload())))

    message.set_payload('')
    wrapper.set_payload([message])

    return wrapper

def message_is_binary(message):
    """Determine if a non-multipart message is of binary type."""
    return message.get_content_maintype() not in set(['text', 'message'])

def clean_message(message):
    """Clean a message of all its binary parts.
    
    This guts all binary attachments, and returns the message itself for
    convenience.

    """
    if message.is_multipart():
        # Don't recurse in already-deleted attachments
        if message.get_content_type() != 'message/external-body':
            parts = message.get_payload()
            parts[:] = map(clean_message, parts)
    elif message_is_binary(message):
        # TODO: Don't gut if this is the topmost message
        message = gut_message(message)

    return message


def main():
    usage = 'Usage: %prog [ --mbox ]'
    parser = optparse.OptionParser(usage=usage)

    parser.add_option('--mbox', action='store_true', default=False,
                      help='Input is in mbox format')

    (options, args) = parser.parse_args()

    if options.mbox:
        messages = messages_list(sys.stdin)
    else:
        messages = [email.message_from_string(''.join(sys.stdin.readlines()))]

    for message in messages:
        print clean_message(message).as_string(unixfrom=True)


if __name__ == '__main__':
    main()