developer.intra2net.com Git - imap-fix-internaldate/blob - src/fix_imap_internaldate.py

   1 '''
   2 fix_imap_internaldate.py - Fix the INTERNALDATE field on IMAP servers
   3
   4 Copyright (c) 2012 Intra2net AG
   5 Author: Plamen Dimitrov
   6
   7 This program is free software: you can redistribute it and/or modify
   8 it under the terms of the GNU General Public License as published by
   9 the Free Software Foundation, either version 3 of the License, or
  10 (at your option) any later version.
  11
  12 This program is distributed in the hope that it will be useful,
  13 but WITHOUT ANY WARRANTY; without even the implied warranty of
  14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15 GNU General Public License for more details.
  16 '''
  17
  18 import sys
  19 import csv
  20 import argparse
  21 # python version handling
  22 try:
  23     import configparser
  24 except ImportError:
  25     print("This module needs python version 3 or later.")
  26     sys.exit()
  27 import logging
  28 from mail_date_parser import MailDateParser
  29 from mail_iterator import MailIterator
  30 from caching_data import CachingData
  31
  32 CONFIG_FILENAME = "fix_imap_internaldate.cfg"
  33 LOG_FILENAME = "fix_imap_internaldate.log"
  34 CSV_FILENAME = "userdata.csv"
  35
  36 def main():
  37     """Interprets command arguments and initializes configuration and logger.
  38         Then begins mail synchronization."""
  39
  40     # parse arguments
  41     parser = argparse.ArgumentParser(description="Fix the INTERNALDATE field on IMAP servers. "
  42                                                 "Small tool to fix the IMAP internaldate "
  43                                                 "in case it's too much off compared to the last date "
  44                                                 "stored in the received lines.")
  45     parser.add_argument('-u', '--update', dest='test_mode', action='store_false',
  46                         default=True, help='update all e-mails and exit test mode')
  47
  48     # config and logging setup
  49     config = load_configuration()
  50     prepare_logger(config)
  51     args = parser.parse_args()
  52     if(args.test_mode):
  53         logging.info("Testing mode initiated. No message will be modified on the server.")
  54     else:
  55         logging.info("Update mode initiated. Messages will be modified.")
  56
  57     # proceed to main functionality
  58     try:
  59         synchronize_csv(config, args.test_mode)
  60     except KeyboardInterrupt:
  61         logging.info("Script was interrupted by the user.")
  62
  63     logging.info("All done. Exiting.")
  64     return
  65
  66 def load_configuration():
  67     """Loads the script configuration from a file or creates such."""
  68     config = configparser.RawConfigParser()
  69     success = config.read(CONFIG_FILENAME)
  70     if(len(success)==0):
  71         config.add_section('basic_settings')
  72         config.set('basic_settings', 'file_log_level', logging.INFO)
  73         config.set('basic_settings', 'console_log_level', logging.INFO)
  74         config.set('basic_settings', 'imap_server', 'imap.company.com')
  75         config.set('basic_settings', 'tolerance_mins', 30)
  76         config.set('basic_settings', 'skip_shared_folders', "ON")
  77         config.set('basic_settings', 'fallback_to_date_header', "OFF")
  78         with open(CONFIG_FILENAME, 'w') as configfile:
  79             config.write(configfile)
  80             configfile.write("# 0 NOTSET, 10 DEBUG, 20 INFO, 30 WARNING, 40 ERROR, 50 CRITICAL")
  81     return config
  82
  83 def prepare_logger(config):
  84     """Sets up the logging functionality"""
  85
  86     # reset the log
  87     with open(LOG_FILENAME, 'w'):
  88         pass
  89
  90     # add basic configuration
  91     logging.basicConfig(filename=LOG_FILENAME,
  92                         format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
  93                         level=config.getint('basic_settings', 'file_log_level'))
  94
  95     # add a handler for a console output
  96     console = logging.StreamHandler()
  97     console.setLevel(config.getint('basic_settings', 'console_log_level'))
  98     formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
  99     console.setFormatter(formatter)
 100     logging.getLogger('').addHandler(console)
 101     return
 102
 103 def synchronize_csv(config, test_mode):
 104     """Iterates through csv list of users and synchronizes their messages."""
 105
 106     # initialize loop permanent data
 107     caching_data = CachingData()
 108     date_parser = MailDateParser()
 109     server = config.get('basic_settings', 'imap_server')
 110     tolerance = config.getint('basic_settings', 'tolerance_mins') * 60
 111
 112     # iterate through the users in the csv data
 113     user_reader = csv.DictReader(open(CSV_FILENAME, "r"), delimiter=',')
 114     for user in user_reader:
 115         try:
 116             session = MailIterator(server, user['username'], user['password'])
 117         except UserWarning as ex:
 118             logging.error(ex)
 119             continue
 120         for mailbox in session:
 121             try:
 122                 box = caching_data.retrieve_cached_mailbox(mailbox[0], mailbox[1], user['username'])
 123                 mail_ids = session.fetch_messages()
 124                 new_ids = box.synchronize(mail_ids, tolerance)
 125                 logging.info("%s new messages out of %s found in %s.", len(new_ids), len(mail_ids), box.name)
 126             except UserWarning as ex:
 127                 logging.error(ex)
 128                 continue
 129             for mid in new_ids:
 130                 try:
 131                     fetched_internal_date = session.fetch_internal_date(mid)
 132                     internal_date = date_parser.extract_internal_date(fetched_internal_date)
 133                     fetched_correct_date = session.fetch_received_date(mid)
 134                     correct_date = date_parser.extract_received_date(fetched_correct_date)
 135                     # check for empty received headers
 136                     if(correct_date == ""):
 137                         logging.debug("No received date could be found in message uid: %s - mailbox: %s - user: %s.",
 138                                         mid.decode('iso-8859-1'), box.name, box.owner)
 139                         box.no_received_field += 1
 140                         # correct these messages if required and override received_date from basic date
 141                         if(config.get('basic_settings', 'fallback_to_date_header') == "ON"):
 142                             fetched_correct_date = session.fetch_basic_date(mid)
 143                             correct_date = date_parser.extract_received_date(fetched_correct_date)
 144                         else:
 145                             # skip synchronization for this message
 146                             continue
 147                     else:
 148                         # preserve only the first received line as fetched if everything is ok
 149                         fetched_correct_date = fetched_correct_date.split("Received:")[1]
 150                 except UserWarning as ex:
 151                     logging.error(ex)
 152                     continue
 153                 if(date_parser.compare_dates(correct_date, internal_date, tolerance)):
 154                     logging.warning("Date conflict found in message uid: %s - mailbox: %s - user: %s.\nInternal date %s is different from extracted date %s from header:\n%s.",
 155                                     mid.decode('iso-8859-1'), box.name, box.owner,
 156                                     internal_date.strftime("%d %b %Y %H:%M:%S"),
 157                                     correct_date.strftime("%d %b %Y %H:%M:%S"),
 158                                     fetched_correct_date)
 159                     if(not test_mode):
 160                         try:
 161                             session.update_message(mid, box.name, correct_date)
 162                         except UserWarning as ex:
 163                             logging.error(ex)
 164                             continue
 165
 166                     # count total emails for every user and mailbox
 167                     box.date_conflicts += 1
 168
 169             # if all messages were successfully fixed confirm caching
 170             if(not test_mode):
 171                 box.confirm_change()
 172
 173     # final report on date conflicts
 174     caching_data.report_conflicts()
 175     return
 176
 177 if(__name__ == "__main__"):
 178     main()