phpBB/develop/repair_bots.php

   1 <?php
   2 /**
   3 * Rebuild BOTS
   4 *
   5 * You should make a backup from your whole database. Things can and will go wrong.
   6 * This will only work if no BOTs were added.
   7 *
   8 */
   9 die("Please read the first lines of this script for instructions on how to enable it");
  10
  11 set_time_limit(0);
  12
  13 define('IN_PHPBB', true);
  14 define('PHPBB_ROOT_PATH', './../');
  15 define('PHP_EXT', substr(strrchr(__FILE__, '.'), 1));
  16 include(PHPBB_ROOT_PATH . 'common.' . PHP_EXT);
  17 include(PHPBB_ROOT_PATH . 'includes/functions_user.' . PHP_EXT);
  18
  19
  20 // Start session management
  21 $user->session_begin();
  22 $auth->acl($user->data);
  23 $user->setup();
  24
  25 $bots = array(
  26         'AdsBot [Google]'                       => array('AdsBot-Google', ''),
  27         'Alexa [Bot]'                           => array('ia_archiver', ''),
  28         'Alta Vista [Bot]'                      => array('Scooter/', ''),
  29         'Ask Jeeves [Bot]'                      => array('Ask Jeeves', ''),
  30         'Baidu [Spider]'                        => array('Baiduspider+(', ''),
  31         'Exabot [Bot]'                          => array('Exabot/', ''),
  32         'FAST Enterprise [Crawler]'     => array('FAST Enterprise Crawler', ''),
  33         'FAST WebCrawler [Crawler]'     => array('FAST-WebCrawler/', ''),
  34         'Francis [Bot]'                         => array('http://www.neomo.de/', ''),
  35         'Gigabot [Bot]'                         => array('Gigabot/', ''),
  36         'Google Adsense [Bot]'          => array('Mediapartners-Google', ''),
  37         'Google Desktop'                        => array('Google Desktop', ''),
  38         'Google Feedfetcher'            => array('Feedfetcher-Google', ''),
  39         'Google [Bot]'                          => array('Googlebot', ''),
  40         'Heise IT-Markt [Crawler]'      => array('heise-IT-Markt-Crawler', ''),
  41         'Heritrix [Crawler]'            => array('heritrix/1.', ''),
  42         'IBM Research [Bot]'            => array('ibm.com/cs/crawler', ''),
  43         'ICCrawler - ICjobs'            => array('ICCrawler - ICjobs', ''),
  44         'ichiro [Crawler]'                      => array('ichiro/2', ''),
  45         'Majestic-12 [Bot]'                     => array('MJ12bot/', ''),
  46         'Metager [Bot]'                         => array('MetagerBot/', ''),
  47         'MSN NewsBlogs'                         => array('msnbot-NewsBlogs/', ''),
  48         'MSN [Bot]'                                     => array('msnbot/', ''),
  49         'MSNbot Media'                          => array('msnbot-media/', ''),
  50         'NG-Search [Bot]'                       => array('NG-Search/', ''),
  51         'Nutch [Bot]'                           => array('http://lucene.apache.org/nutch/', ''),
  52         'Nutch/CVS [Bot]'                       => array('NutchCVS/', ''),
  53         'OmniExplorer [Bot]'            => array('OmniExplorer_Bot/', ''),
  54         'Online link [Validator]'       => array('online link validator', ''),
  55         'psbot [Picsearch]'                     => array('psbot/0', ''),
  56         'Seekport [Bot]'                        => array('Seekbot/', ''),
  57         'Sensis [Crawler]'                      => array('Sensis Web Crawler', ''),
  58         'SEO Crawler'                           => array('SEO search Crawler/', ''),
  59         'Seoma [Crawler]'                       => array('Seoma [SEO Crawler]', ''),
  60         'SEOSearch [Crawler]'           => array('SEOsearch/', ''),
  61         'Snappy [Bot]'                          => array('Snappy/1.1 ( http://www.urltrends.com/ )', ''),
  62         'Steeler [Crawler]'                     => array('http://www.tkl.iis.u-tokyo.ac.jp/~crawler/', ''),
  63         'Synoo [Bot]'                           => array('SynooBot/', ''),
  64         'Telekom [Bot]'                         => array('crawleradmin.t-info@telekom.de', ''),
  65         'TurnitinBot [Bot]'                     => array('TurnitinBot/', ''),
  66         'Voyager [Bot]'                         => array('voyager/1.0', ''),
  67         'W3 [Sitesearch]'                       => array('W3 SiteSearch Crawler', ''),
  68         'W3C [Linkcheck]'                       => array('W3C-checklink/', ''),
  69         'W3C [Validator]'                       => array('W3C_*Validator', ''),
  70         'WiseNut [Bot]'                         => array('http://www.WISEnutbot.com', ''),
  71         'YaCy [Bot]'                            => array('yacybot', ''),
  72         'Yahoo MMCrawler [Bot]'         => array('Yahoo-MMCrawler/', ''),
  73         'Yahoo Slurp [Bot]'                     => array('Yahoo! DE Slurp', ''),
  74         'Yahoo [Bot]'                           => array('Yahoo! Slurp', ''),
  75         'YahooSeeker [Bot]'                     => array('YahooSeeker/', ''),
  76 );
  77
  78 $bot_ids = array();
  79 user_get_id_name($bot_ids, array_keys($bots), phpbb::USER_IGNORE);
  80 foreach($bot_ids as $bot)
  81 {
  82         user_delete('remove', $bot);
  83 }
  84 // Done
  85 add_bots($bots);
  86 echo 'done';
  87
  88
  89 /**
  90 * Add the search bots into the database
  91 * This code should be used in execute_last if the source database did not have bots
  92 * If you are converting bots this function should not be called
  93 * @todo We might want to look at sharing the bot list between the install code and this code for consistency
  94 */
  95 function add_bots($bots)
  96 {
  97         global $db, $config;
  98
  99         $sql = 'SELECT group_id FROM ' . GROUPS_TABLE . " WHERE group_name = 'BOTS'";
 100         $result = $db->sql_query($sql);
 101         $group_id = (int) $db->sql_fetchfield('group_id', $result);
 102         $db->sql_freeresult($result);
 103         $db->sql_query('TRUNCATE TABLE ' . BOTS_TABLE);
 104
 105         if (!$group_id)
 106         {
 107                 add_default_groups();
 108
 109                 $sql = 'SELECT group_id FROM ' . GROUPS_TABLE . " WHERE group_name = 'BOTS'";
 110                 $result = $db->sql_query($sql);
 111                 $group_id = (int) $db->sql_fetchfield('group_id', $result);
 112                 $db->sql_freeresult($result);
 113
 114         }
 115
 116
 117
 118
 119         foreach ($bots as $bot_name => $bot_ary)
 120         {
 121                 $user_row = array(
 122                         'user_type'                             => phpbb::USER_IGNORE,
 123                         'group_id'                              => $group_id,
 124                         'username'                              => $bot_name,
 125                         'user_regdate'                  => time(),
 126                         'user_password'                 => '',
 127                         'user_colour'                   => '9E8DA7',
 128                         'user_email'                    => '',
 129                         'user_lang'                             => $config['default_lang'],
 130                         'user_style'                    => 1,
 131                         'user_timezone'                 => 0,
 132                         'user_allow_massemail'  => 0,
 133                 );
 134
 135                 $user_id = user_add($user_row);
 136
 137                 if ($user_id)
 138                 {
 139                         $sql = 'INSERT INTO ' . BOTS_TABLE . ' ' . $db->sql_build_array('INSERT', array(
 140                                 'bot_active'    => 1,
 141                                 'bot_name'              => $bot_name,
 142                                 'user_id'               => $user_id,
 143                                 'bot_agent'             => $bot_ary[0],
 144                                 'bot_ip'                => $bot_ary[1])
 145                         );
 146                         $db->sql_query($sql);
 147                 }
 148         }
 149 }
 150
 151 ?>