mailbounce.py

#!/usr/bin/python
# -*- coding: utf-8 -*-

from email.parser import Parser
import os, re, csv, argparse, sys

class EMailBounce(object):
    def __init__(self, filename):
        self.Delivery_status = {}
        self.Recipient = ''
        self.Reason = ''

        self.Filename = filename
        self.parse()
        self.get_delivery_status()
        self.get_infos()

    def parse(self):
        fp = open(self.Filename, mode='r', encoding='iso-8859-15')
        self.msg = Parser().parse(fp)
        fp.close()

    def get_delivery_status(self):
        self.Delivery_status = {}
        if self.msg.is_multipart() and self.msg.get_content_type() == 'multipart/report':
            for submsg in self.msg.get_payload():
                if submsg.get_content_type() == 'message/delivery-status':
                    for subsubmsg in submsg.get_payload():
                        for item in subsubmsg.items():
                            self.Delivery_status[item[0].lower()] = item[1]

    def get_infos(self):
        if 'original-recipient' in self.Delivery_status:
            self.Recipient = self.Delivery_status['original-recipient']
        elif 'final-recipient' in self.Delivery_status:
            self.Recipient = self.Delivery_status['final-recipient']

        self.Recipient = self.Recipient.replace('rfc822;', '').strip()

        if 'diagnostic-code' in self.Delivery_status:
            self.Reason = self.Delivery_status['diagnostic-code'].strip()
            self.Reason = re.sub('^.+;', '', self.Reason)
            self.Reason = re.sub('\s*\r?\n\s*', ' ', self.Reason)

parser = argparse.ArgumentParser(description='Geht die E-Mail Dateien im angegebenen Ordner durch, holt die E-Mail Adressen raus und speichert diese in eine CSV Datei.')
parser.add_argument('dir', help='Ordner mit den E-Mail Dateien.')
args = parser.parse_args()

rows = []

if not os.path.exists(args.dir):
    sys.stderr.write('Ordner existiert nicht\n')
    exit(1)

for root, dirs, files in os.walk(args.dir):
    for f in files:
        em = EMailBounce(os.path.join(root, f))
        if em.Recipient.strip() == '':
            print('failed: {}'.format(f))
        else:
            rows.append({'email': em.Recipient, 'reason': em.Reason})

csv.register_dialect('christophsdialect', delimiter=';', quotechar='"', lineterminator='\n', quoting=csv.QUOTE_MINIMAL)

with open('bounces.csv', 'w', newline='') as csvfile:
    writer = csv.DictWriter(csvfile, ['email', 'reason'], dialect='christophsdialect')
    writer.writerows(rows)