maintenance/cleanupSpam.php

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