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         $regex = LinkFilter::makeRegex( $domain );
  18
  19         while ( $rev && preg_match( $regex, $rev->getText() ) ) {
  20                 # Revision::getPrevious can't be used in this way before MW 1.6 (Revision.php 1.26)
  21                 #$rev = $rev->getPrevious();
  22                 $revId = $title->getPreviousRevisionID( $revId );
  23                 if ( $revId ) {
  24                         $rev = Revision::newFromTitle( $title, $revId );
  25                 } else {
  26                         $rev = false;
  27                 }
  28         }
  29         if ( $revId == $currentRevId ) {
  30                 // The regex didn't match the current article text
  31                 // This happens e.g. when a link comes from a template rather than the page itself
  32                 print "False match\n";
  33         } else {
  34                 $dbw = wfGetDB( DB_MASTER );
  35                 $dbw->immediateBegin();
  36                 if ( !$rev ) {
  37                         // Didn't find a non-spammy revision, blank the page
  38                         print "blanking\n";
  39                         $article = new Article( $title );
  40                         $article->updateArticle( '', wfMsg( 'spam_blanking', $domain ),
  41                                 false, false );
  42
  43                 } else {
  44                         // Revert to this revision
  45                         print "reverting\n";
  46                         $article = new Article( $title );
  47                         $article->updateArticle( $rev->getText(), wfMsg( 'spam_reverting', $domain ), false, false );
  48                 }
  49                 $dbw->immediateCommit();
  50                 wfDoUpdates();
  51         }
  52 }
  53 //------------------------------------------------------------------------------
  54
  55
  56
  57
  58 $username = wfMsg( 'spambot_username' );
  59 $fname = $username;
  60 $wgUser = User::newFromName( $username );
  61 // Create the user if necessary
  62 if ( !$wgUser->getID() ) {
  63         $wgUser->addToDatabase();
  64 }
  65
  66 if ( !isset( $args[0] ) ) {
  67         print "Usage: php cleanupSpam.php <hostname>\n";
  68         exit(1);
  69 }
  70 $spec = $args[0];
  71 $like = LinkFilter::makeLike( $spec );
  72 if ( !$like ) {
  73         print "Not a valid hostname specification: $spec\n";
  74         exit(1);
  75 }
  76
  77 $dbr = wfGetDB( DB_SLAVE );
  78
  79 if ( isset($options['all']) ) {
  80         // Clean up spam on all wikis
  81         $dbr = wfGetDB( DB_SLAVE );
  82         print "Finding spam on " . count($wgLocalDatabases) . " wikis\n";
  83         $found = false;
  84         foreach ( $wgLocalDatabases as $db ) {
  85                 $count = $dbr->selectField( "`$db`.externallinks", 'COUNT(*)',
  86                         array( 'el_index LIKE ' . $dbr->addQuotes( $like ) ), $fname );
  87                 if ( $count ) {
  88                         $found = true;
  89                         passthru( "php cleanupSpam.php $db $spec | sed s/^/$db:  /" );
  90                 }
  91         }
  92         if ( $found ) {
  93                 print "All done\n";
  94         } else {
  95                 print "None found\n";
  96         }
  97 } else {
  98         // Clean up spam on this wiki
  99         $res = $dbr->select( 'externallinks', array( 'DISTINCT el_from' ),
 100                 array( 'el_index LIKE ' . $dbr->addQuotes( $like ) ), $fname );
 101         $count = $dbr->numRows( $res );
 102         print "Found $count articles containing $spec\n";
 103         while ( $row = $dbr->fetchObject( $res ) ) {
 104                 cleanupArticle( $row->el_from, $spec );
 105         }
 106         if ( $count ) {
 107                 print "Done\n";
 108         }
 109 }
 110
 111