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 and Thomas Jarosch
   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
  22 try:
  23     import configparser
  24 except ImportError:
  25     import ConfigParser as configparser
  26
  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
  71     # if no file is found create a default one
  72     if(len(success)==0):
  73         if(not config.has_section('basic_settings')):
  74             config.add_section('basic_settings')
  75         config.set('basic_settings', 'file_log_level', logging.INFO)
  76         config.set('basic_settings', 'console_log_level', logging.INFO)
  77         config.set('basic_settings', 'imap_server', 'imap.company.com')
  78         config.set('basic_settings', 'tolerance_mins', 30)
  79         config.set('basic_settings', 'skip_shared_folders', "True")
  80         config.set('basic_settings', 'fallback_to_date_header', "False")
  81         with open(CONFIG_FILENAME, 'w') as configfile:
  82             config.write(configfile)
  83             configfile.write("# 0 NOTSET, 10 DEBUG, 20 INFO, 30 WARNING, 40 ERROR, 50 CRITICAL\n")
  84             print("Created initial config %s" % CONFIG_FILENAME)
  85
  86     try:
  87         config.get('basic_settings', 'file_log_level')
  88         config.get('basic_settings', 'console_log_level')
  89         config.get('basic_settings', 'imap_server')
  90         config.getint('basic_settings', 'tolerance_mins')
  91         config.getboolean('basic_settings', 'skip_shared_folders')
  92         config.getboolean('basic_settings', 'fallback_to_date_header')
  93     except (configparser.NoSectionError, configparser.NoOptionError, ValueError) as ex:
  94         print("Could not read config file '%s': %s." % (CONFIG_FILENAME, ex))
  95         print("Please change or remove the config file.")
  96         sys.exit()
  97
  98     return config
  99
 100 def prepare_logger(config):
 101     """Sets up the logging functionality"""
 102
 103     # reset the log
 104     with open(LOG_FILENAME, 'w'):
 105         pass
 106
 107     # add basic configuration
 108     logging.basicConfig(filename=LOG_FILENAME,
 109                         format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
 110                         level=config.getint('basic_settings', 'file_log_level'))
 111
 112     # add a handler for a console output
 113     console = logging.StreamHandler()
 114     console.setLevel(config.getint('basic_settings', 'console_log_level'))
 115     formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
 116     console.setFormatter(formatter)
 117     logging.getLogger('').addHandler(console)
 118     return
 119
 120 def synchronize_csv(config, test_mode):
 121     """Iterates through csv list of users and synchronizes their messages."""
 122
 123     # initialize loop permanent data
 124     caching_data = CachingData(config.getboolean('basic_settings', 'fallback_to_date_header'))
 125     date_parser = MailDateParser()
 126     server = config.get('basic_settings', 'imap_server')
 127     tolerance = config.getint('basic_settings', 'tolerance_mins') * 60
 128     skip_shared_folders = config.getboolean('basic_settings', 'skip_shared_folders')
 129
 130     # iterate through the users in the csv data
 131     user_reader = csv.DictReader(open(CSV_FILENAME, "r"), delimiter=',')
 132     for user in user_reader:
 133         try:
 134             session = MailIterator(server, user['username'], user['password'], skip_shared_folders)
 135         except UserWarning as ex:
 136             logging.error(ex)
 137             continue
 138         for mailbox in session:
 139             try:
 140                 box = caching_data.retrieve_cached_mailbox(mailbox[0], mailbox[1], user['username'])
 141                 mail_ids = session.fetch_messages()
 142                 new_ids = box.synchronize(mail_ids, tolerance)
 143                 logging.info("%s new messages out of %s found in %s.", len(new_ids), len(mail_ids), box.name)
 144             except UserWarning as ex:
 145                 logging.error(ex)
 146                 continue
 147             for mid in new_ids:
 148                 try:
 149                     fetched_internal_date = session.fetch_internal_date(mid)
 150                     internal_date = date_parser.extract_internal_date(fetched_internal_date)
 151                     fetched_correct_date = session.fetch_received_date(mid)
 152                     correct_date = date_parser.extract_received_date(fetched_correct_date)
 153                     # check for empty received headers
 154                     if(correct_date == ""):
 155                         logging.debug("No received date could be found in message uid: %s - mailbox: %s - user: %s.",
 156                                         mid.decode('iso-8859-1'), box.name, box.owner)
 157                         box.no_received_field += 1
 158                         # correct these messages if required and override received_date from basic date
 159                         if(config.getboolean('basic_settings', 'fallback_to_date_header')):
 160                             fetched_correct_date = session.fetch_basic_date(mid)
 161                             correct_date = date_parser.extract_received_date(fetched_correct_date)
 162                             if(correct_date == ""):
 163                                 logging.debug("No fallback date header could be found in message uid: %s - mailbox: %s - user: %s.",
 164                                                 mid.decode('iso-8859-1'), box.name, box.owner)
 165
 166                         if(correct_date == ""):
 167                             # skip synchronization for this message
 168                             continue
 169                     else:
 170                         # preserve only the first received line as fetched if everything is ok
 171                         fetched_correct_date = fetched_correct_date.split("Received:")[1]
 172                 except UserWarning as ex:
 173                     logging.error(ex)
 174                     continue
 175                 if(date_parser.compare_dates(correct_date, internal_date, tolerance)):
 176                     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.",
 177                                     mid.decode('iso-8859-1'), box.name, box.owner,
 178                                     internal_date.strftime("%d %b %Y %H:%M:%S"),
 179                                     correct_date.strftime("%d %b %Y %H:%M:%S"),
 180                                     fetched_correct_date)
 181                     if(not test_mode):
 182                         try:
 183                             session.update_message(mid, box.name, correct_date)
 184                         except UserWarning as ex:
 185                             logging.error(ex)
 186                             continue
 187
 188                     # count total emails for every user and mailbox
 189                     box.date_conflicts += 1
 190
 191             # if all messages were successfully fixed confirm caching
 192             if(not test_mode):
 193                 box.confirm_change()
 194
 195     # final report on date conflicts
 196     caching_data.report_conflicts()
 197     return
 198
 199 if(__name__ == "__main__"):
 200     main()