maintenance/cleanupSpam.php

   1 <?php
   2 /**
   3  * Cleanup all spam from a given hostname
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Maintenance
  22  */
  23
  24 require_once( dirname( __FILE__ ) . '/Maintenance.php' );
  25
  26 class CleanupSpam extends Maintenance {
  27         public function __construct() {
  28                 parent::__construct();
  29                 $this->mDescription = "Cleanup all spam from a given hostname";
  30                 $this->addOption( 'all', 'Check all wikis in $wgLocalDatabases' );
  31                 $this->addArg( 'hostname', 'Hostname that was spamming' );
  32         }
  33
  34         public function execute() {
  35                 global $wgLocalDatabases, $wgUser;
  36
  37                 $username = wfMsg( 'spambot_username' );
  38                 $wgUser = User::newFromName( $username );
  39                 // Create the user if necessary
  40                 if ( !$wgUser->getId() ) {
  41                         $wgUser->addToDatabase();
  42                 }
  43                 $spec = $this->getArg();
  44                 $like = LinkFilter::makeLikeArray( $spec );
  45                 if ( !$like ) {
  46                         $this->error( "Not a valid hostname specification: $spec", true );
  47                 }
  48
  49                 if ( $this->hasOption( 'all' ) ) {
  50                         // Clean up spam on all wikis
  51                         $this->output( "Finding spam on " . count( $wgLocalDatabases ) . " wikis\n" );
  52                         $found = false;
  53                         foreach ( $wgLocalDatabases as $wikiID ) {
  54                                 $dbr = wfGetDB( DB_SLAVE, array(), $wikiID );
  55
  56                                 $count = $dbr->selectField( 'externallinks', 'COUNT(*)',
  57                                         array( 'el_index' . $dbr->buildLike( $like ) ), __METHOD__ );
  58                                 if ( $count ) {
  59                                         $found = true;
  60                                         passthru( "php cleanupSpam.php --wiki='$wikiID' $spec | sed 's/^/$wikiID:  /'" );
  61                                 }
  62                         }
  63                         if ( $found ) {
  64                                 $this->output( "All done\n" );
  65                         } else {
  66                                 $this->output( "None found\n" );
  67                         }
  68                 } else {
  69                         // Clean up spam on this wiki
  70
  71                         $dbr = wfGetDB( DB_SLAVE );
  72                         $res = $dbr->select( 'externallinks', array( 'DISTINCT el_from' ),
  73                                 array( 'el_index' . $dbr->buildLike( $like ) ), __METHOD__ );
  74                         $count = $dbr->numRows( $res );
  75                         $this->output( "Found $count articles containing $spec\n" );
  76                         foreach ( $res as $row ) {
  77                                 $this->cleanupArticle( $row->el_from, $spec );
  78                         }
  79                         if ( $count ) {
  80                                 $this->output( "Done\n" );
  81                         }
  82                 }
  83         }
  84
  85         private function cleanupArticle( $id, $domain ) {
  86                 $title = Title::newFromID( $id );
  87                 if ( !$title ) {
  88                         $this->error( "Internal error: no page for ID $id" );
  89                         return;
  90                 }
  91
  92                 $this->output( $title->getPrefixedDBkey() . " ..." );
  93                 $rev = Revision::newFromTitle( $title );
  94                 $revId = $rev->getId();
  95                 $currentRevId = $revId;
  96
  97                 while ( $rev && LinkFilter::matchEntry( $rev->getText() , $domain ) ) {
  98                         # Revision::getPrevious can't be used in this way before MW 1.6 (Revision.php 1.26)
  99                         # $rev = $rev->getPrevious();
 100                         $revId = $title->getPreviousRevisionID( $revId );
 101                         if ( $revId ) {
 102                                 $rev = Revision::newFromTitle( $title, $revId );
 103                         } else {
 104                                 $rev = false;
 105                         }
 106                 }
 107                 if ( $revId == $currentRevId ) {
 108                         // The regex didn't match the current article text
 109                         // This happens e.g. when a link comes from a template rather than the page itself
 110                         $this->output( "False match\n" );
 111                 } else {
 112                         $dbw = wfGetDB( DB_MASTER );
 113                         $dbw->begin();
 114                         if ( !$rev ) {
 115                                 // Didn't find a non-spammy revision, blank the page
 116                                 $this->output( "blanking\n" );
 117                                 $article = new Article( $title );
 118                                 $article->doEdit( '', wfMsg( 'spam_blanking', $domain ) );
 119                         } else {
 120                                 // Revert to this revision
 121                                 $this->output( "reverting\n" );
 122                                 $article = new Article( $title );
 123                                 $article->doEdit( $rev->getText(), wfMsg( 'spam_reverting', $domain ), EDIT_UPDATE );
 124                         }
 125                         $dbw->commit();
 126                         wfDoUpdates();
 127                 }
 128         }
 129 }
 130
 131 $maintClass = "CleanupSpam";
 132 require_once( RUN_MAINTENANCE_IF_MAIN );