adversarial-mailman2to3/export_mailman2_settings.py

import argparse
import json
import sys

from bs4 import BeautifulSoup

from mailman.interfaces.archiver import ArchivePolicy
from mailman.interfaces.action import Action
from mailman.utilities.importer import NAME_MAPPINGS, member_moderation_action_mapping, dmarc_action_mapping

KEYFILTER = ('submit')

EXCLUDES = set((
    'accept_these_nonmembers',
    'delivery_status',
    'digest_members',
    'discard_these_nonmembers',
    'hold_these_nonmembers',
    'members',
    'reject_these_nonmembers',
    'user_options',
) + tuple({
    # from `convert_to_uri` in mailman's `importer.py`
    'goodbye_msg': 'list:user:notice:goodbye',
    'msg_header': 'list:member:regular:header',
    'msg_footer': 'list:member:regular:footer',
    'digest_header': 'list:member:digest:header',
    'digest_footer': 'list:member:digest:footer',
}.keys()) + (
    # some manual settings which don't seem to be handled by the conversion
    # script
    'archive_volume_frequency',
    'autoresponse_postings_text_upload',
    'autoresponse_admin_text_upload',
    'autoresponse_request_text_upload',
    'bounce_notify_owner_on_bounce_increment',
    'digestable',
    'digest_is_default',
    'mime_is_default_digest',
    'digest_size_threshhold',
    '_new_volume',
    '_send_digest_now',
    'umbrella_list',
    'umbrella_member_suffix'
))


def msg(*args, **kwargs):
    print(*args, file=sys.stderr, **kwargs)


def get_form_data(htmlfile):
    data_clean = {}

    soup = BeautifulSoup(htmlfile.read(), 'html.parser')

    for field in soup.find_all('textarea'):
        if field['name'] in EXCLUDES:
            continue

        name = NAME_MAPPINGS.get(field['name'], field['name'])

        if 'msg' in name:
            continue

        data_clean[name] = [line for line in field.get_text().split('\n') if
                            line != ""]

    for field in soup.find_all('input'):
        if field['name'] in EXCLUDES:
            continue

        if field['type'].lower() in ('hidden', 'submit'):
            continue

        if field['type'].lower() == 'radio':
            if 'checked' not in field.attrs:
                continue

        name = NAME_MAPPINGS.get(field['name'], field['name'])

        if 'msg' in name:
            continue

        try:
            value = field['value']
            try:
                value = int(value)
            except ValueError:
                ...
        except KeyError:
            value = ''

        data_clean[name] = value

    return data_clean


if __name__ == "__main__":
    parser = argparse.ArgumentParser(
        prog=sys.argv[0], description="Munge Mailman config data"
    )
    parser.add_argument("config", help="A list of files (named specifically) to import as configurations",
                        nargs='+',
                        action="append")
    args = parser.parse_args()

    globalconfig = {}
    for page in args.config[0]:
        msg(page)
        with open(page) as inf:
            result = get_form_data(inf)
        for key, value in result.items():
            if key in globalconfig:
                msg(f"warning - duplicate key {key}")
            else:
                globalconfig[key] = value

    # Handle the moderation policy.
    #
    # The mlist.default_member_action and mlist.default_nonmember_action enum
    # values are different in Mailman 2.1, because they have been merged into a
    # single enum in Mailman 3.
    #
    # Unmoderated lists used to have default_member_moderation set to a false
    # value; this translates to the Defer default action.  Moderated lists with
    # the default_member_moderation set to a true value used to store the
    # action in the member_moderation_action flag, the values were: 0==Hold,
    # 1=Reject, 2==Discard
    if bool(globalconfig.get('default_member_moderation', 0)):
        globalconfig['default_member_action'] = member_moderation_action_mapping(
            globalconfig.get('member_moderation_action')).name
        del globalconfig['default_member_moderation']
    else:
        globalconfig['default_member_action'] = Action.defer.name
    # Handle DMARC mitigations.
    # This would be straightforward except for from_is_list.  The issue
    # is in MM 2.1 the from_is_list action applies if dmarc_moderation_action
    # doesn't apply and they can be different.
    # We will map as follows:
    # from_is_list > dmarc_moderation_action
    #    dmarc_mitigate_action = from_is_list action
    #    dmarc_mitigate_unconditionally = True
    # from_is_list <= dmarc_moderation_action
    #    dmarc_mitigate_action = dmarc_moderation_action
    #    dmarc_mitigate_unconditionally = False
    # The text attributes are handled above.
    if (globalconfig.get('from_is_list', 0) >
            globalconfig.get('dmarc_moderation_action', 0)):
        globalconfig['dmarc_mitigate_action'] = dmarc_action_mapping(
            globalconfig.get('from_is_list', 0)).name
        globalconfig['dmarc_mitigate_unconditionally'] = True
        del globalconfig['from_is_list']
    else:
        globalconfig['dmarc_mitigate_action'] = dmarc_action_mapping(
            globalconfig.get('dmarc_moderation_action', 0)).name
        globalconfig['dmarc_mitigate_unconditionally'] = False
    # Handle the archiving policy.  In MM2.1 there were two boolean options
    # but only three of the four possible states were valid.  Now there's just
    # an enum.
    if globalconfig.get('archive'):
        # For maximum safety, if for some strange reason there's no
        # archive_private key, treat the list as having private archives.
        if globalconfig.get('archive_private', True):
            globalconfig['archive_policy'] = ArchivePolicy.private.name
            del globalconfig['archive_private']
        else:
            globalconfig['archive_policy'] = ArchivePolicy.public.name
        del globalconfig['archive']
    else:
        globalconfig['archive_policy'] = ArchivePolicy.never.name

    print(json.dumps(globalconfig))