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     try:
  71             config.get('basic_settings', 'file_log_level')
  72             config.get('basic_settings', 'console_log_level')
  73             config.get('basic_settings', 'imap_server')
  74             config.getint('basic_settings', 'tolerance_mins')
  75             config.get('basic_settings', 'skip_shared_folders')
  76             config.get('basic_settings', 'fallback_to_date_header')
  77     except configparser.NoOptionError:
  78         success = []
  79     except ValueError:
  80         success = []
  81
  82     # if corrupted settings save file and load default
  83     if(len(success)==0):
  84         if(not config.has_section('basic_settings')):
  85             config.add_section('basic_settings')
  86         config.set('basic_settings', 'file_log_level', logging.INFO)
  87         config.set('basic_settings', 'console_log_level', logging.INFO)
  88         config.set('basic_settings', 'imap_server', 'imap.company.com')
  89         config.set('basic_settings', 'tolerance_mins', 30)
  90         config.set('basic_settings', 'skip_shared_folders', "ON")
  91         config.set('basic_settings', 'fallback_to_date_header', "OFF")
  92         with open(CONFIG_FILENAME, 'w') as configfile:
  93             config.write(configfile)
  94             configfile.write("# 0 NOTSET, 10 DEBUG, 20 INFO, 30 WARNING, 40 ERROR, 50 CRITICAL")
  95
  96     return config
  97
  98 def prepare_logger(config):
  99     """Sets up the logging functionality"""
 100
 101     # reset the log
 102     with open(LOG_FILENAME, 'w'):
 103         pass
 104
 105     # add basic configuration
 106     logging.basicConfig(filename=LOG_FILENAME,
 107                         format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
 108                         level=config.getint('basic_settings', 'file_log_level'))
 109
 110     # add a handler for a console output
 111     console = logging.StreamHandler()
 112     console.setLevel(config.getint('basic_settings', 'console_log_level'))
 113     formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
 114     console.setFormatter(formatter)
 115     logging.getLogger('').addHandler(console)
 116     return
 117
 118 def synchronize_csv(config, test_mode):
 119     """Iterates through csv list of users and synchronizes their messages."""
 120
 121     # initialize loop permanent data
 122     caching_data = CachingData()
 123     date_parser = MailDateParser()
 124     server = config.get('basic_settings', 'imap_server')
 125     tolerance = config.getint('basic_settings', 'tolerance_mins') * 60
 126
 127     # iterate through the users in the csv data
 128     user_reader = csv.DictReader(open(CSV_FILENAME, "r"), delimiter=',')
 129     for user in user_reader:
 130         try:
 131             session = MailIterator(server, user['username'], user['password'],
 132                                    config.get('basic_settings', 'skip_shared_folders')=="ON")
 133         except UserWarning as ex:
 134             logging.error(ex)
 135             continue
 136         for mailbox in session:
 137             try:
 138                 box = caching_data.retrieve_cached_mailbox(mailbox[0], mailbox[1], user['username'])
 139                 mail_ids = session.fetch_messages()
 140                 new_ids = box.synchronize(mail_ids, tolerance)
 141                 logging.info("%s new messages out of %s found in %s.", len(new_ids), len(mail_ids), box.name)
 142             except UserWarning as ex:
 143                 logging.error(ex)
 144                 continue
 145             for mid in new_ids:
 146                 try:
 147                     fetched_internal_date = session.fetch_internal_date(mid)
 148                     internal_date = date_parser.extract_internal_date(fetched_internal_date)
 149                     fetched_correct_date = session.fetch_received_date(mid)
 150                     correct_date = date_parser.extract_received_date(fetched_correct_date)
 151                     # check for empty received headers
 152                     if(correct_date == ""):
 153                         logging.debug("No received date could be found in message uid: %s - mailbox: %s - user: %s.",
 154                                         mid.decode('iso-8859-1'), box.name, box.owner)
 155                         box.no_received_field += 1
 156                         # correct these messages if required and override received_date from basic date
 157                         if(config.get('basic_settings', 'fallback_to_date_header') == "ON"):
 158                             fetched_correct_date = session.fetch_basic_date(mid)
 159                             correct_date = date_parser.extract_received_date(fetched_correct_date)
 160                         else:
 161                             # skip synchronization for this message
 162                             continue
 163                     else:
 164                         # preserve only the first received line as fetched if everything is ok
 165                         fetched_correct_date = fetched_correct_date.split("Received:")[1]
 166                 except UserWarning as ex:
 167                     logging.error(ex)
 168                     continue
 169                 if(date_parser.compare_dates(correct_date, internal_date, tolerance)):
 170                     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.",
 171                                     mid.decode('iso-8859-1'), box.name, box.owner,
 172                                     internal_date.strftime("%d %b %Y %H:%M:%S"),
 173                                     correct_date.strftime("%d %b %Y %H:%M:%S"),
 174                                     fetched_correct_date)
 175                     if(not test_mode):
 176                         try:
 177                             session.update_message(mid, box.name, correct_date)
 178                         except UserWarning as ex:
 179                             logging.error(ex)
 180                             continue
 181
 182                     # count total emails for every user and mailbox
 183                     box.date_conflicts += 1
 184
 185             # if all messages were successfully fixed confirm caching
 186             if(not test_mode):
 187                 box.confirm_change()
 188
 189     # final report on date conflicts
 190     caching_data.report_conflicts()
 191     return
 192
 193 if(__name__ == "__main__"):
 194     main()