maintenance/cleanupSpam.php

   1 <?php
   2 /**
   3  * @file
   4  * @ingroup Maintenance
   5  */
   6
   7 require_once( 'commandLine.inc' );
   8 require_once( "$IP/includes/LinkFilter.php" );
   9
  10 function cleanupArticle( $id, $domain ) {
  11         $title = Title::newFromID( $id );
  12         if ( !$title ) {
  13                 print "Internal error: no page for ID $id\n";
  14                 return;
  15         }
  16
  17         print $title->getPrefixedDBkey() . " ...";
  18         $rev = Revision::newFromTitle( $title );
  19         $revId = $rev->getId();
  20         $currentRevId = $revId;
  21
  22         while ( $rev && LinkFilter::matchEntry( $rev->getText() , $domain ) ) {
  23                 # Revision::getPrevious can't be used in this way before MW 1.6 (Revision.php 1.26)
  24                 #$rev = $rev->getPrevious();
  25                 $revId = $title->getPreviousRevisionID( $revId );
  26                 if ( $revId ) {
  27                         $rev = Revision::newFromTitle( $title, $revId );
  28                 } else {
  29                         $rev = false;
  30                 }
  31         }
  32         if ( $revId == $currentRevId ) {
  33                 // The regex didn't match the current article text
  34                 // This happens e.g. when a link comes from a template rather than the page itself
  35                 print "False match\n";
  36         } else {
  37                 $dbw = wfGetDB( DB_MASTER );
  38                 $dbw->immediateBegin();
  39                 if ( !$rev ) {
  40                         // Didn't find a non-spammy revision, blank the page
  41                         print "blanking\n";
  42                         $article = new Article( $title );
  43                         $article->updateArticle( '', wfMsg( 'spam_blanking', $domain ),
  44                                 false, false );
  45
  46                 } else {
  47                         // Revert to this revision
  48                         print "reverting\n";
  49                         $article = new Article( $title );
  50                         $article->updateArticle( $rev->getText(), wfMsg( 'spam_reverting', $domain ), false, false );
  51                 }
  52                 $dbw->immediateCommit();
  53                 wfDoUpdates();
  54         }
  55 }
  56 //------------------------------------------------------------------------------
  57
  58
  59
  60
  61 $username = wfMsg( 'spambot_username' );
  62 $fname = $username;
  63 $wgUser = User::newFromName( $username );
  64 // Create the user if necessary
  65 if ( !$wgUser->getId() ) {
  66         $wgUser->addToDatabase();
  67 }
  68
  69 if ( !isset( $args[0] ) ) {
  70         print "Usage: php cleanupSpam.php <hostname>\n";
  71         exit(1);
  72 }
  73 $spec = $args[0];
  74 $like = LinkFilter::makeLike( $spec );
  75 if ( !$like ) {
  76         print "Not a valid hostname specification: $spec\n";
  77         exit(1);
  78 }
  79
  80 $dbr = wfGetDB( DB_SLAVE );
  81
  82 if ( isset($options['all']) ) {
  83         // Clean up spam on all wikis
  84         $dbr = wfGetDB( DB_SLAVE );
  85         print "Finding spam on " . count($wgLocalDatabases) . " wikis\n";
  86         $found = false;
  87         foreach ( $wgLocalDatabases as $db ) {
  88                 $count = $dbr->selectField( "`$db`.externallinks", 'COUNT(*)',
  89                         array( 'el_index LIKE ' . $dbr->addQuotes( $like ) ), $fname );
  90                 if ( $count ) {
  91                         $found = true;
  92                         passthru( "php cleanupSpam.php $db $spec | sed s/^/$db:  /" );
  93                 }
  94         }
  95         if ( $found ) {
  96                 print "All done\n";
  97         } else {
  98                 print "None found\n";
  99         }
 100 } else {
 101         // Clean up spam on this wiki
 102         $res = $dbr->select( 'externallinks', array( 'DISTINCT el_from' ),
 103                 array( 'el_index LIKE ' . $dbr->addQuotes( $like ) ), $fname );
 104         $count = $dbr->numRows( $res );
 105         print "Found $count articles containing $spec\n";
 106         while ( $row = $dbr->fetchObject( $res ) ) {
 107                 cleanupArticle( $row->el_from, $spec );
 108         }
 109         if ( $count ) {
 110                 print "Done\n";
 111         }
 112 }
 113
 114