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         $reverted = false;
  16         $revId = $rev->getId();
  17         $currentRevId = $revId;
  18         $regex = LinkFilter::makeRegex( $domain );
  19
  20         while ( $rev && preg_match( $regex, $rev->getText() ) ) {
  21                 # Revision::getPrevious can't be used in this way before MW 1.6 (Revision.php 1.26)
  22                 #$rev = $rev->getPrevious();
  23                 $revId = $title->getPreviousRevisionID( $revId );
  24                 if ( $revId ) {
  25                         $rev = Revision::newFromTitle( $title, $revId );
  26                 } else {
  27                         $rev = false;
  28                 }
  29         }
  30         if ( $revId == $currentRevId ) {
  31                 // The regex didn't match the current article text
  32                 // This happens e.g. when a link comes from a template rather than the page itself
  33                 print "False match\n";
  34         } else {
  35                 $dbw =& wfGetDB( DB_MASTER );
  36                 $dbw->immediateBegin();
  37                 if ( !$rev ) {
  38                         // Didn't find a non-spammy revision, blank the page
  39                         print "blanking\n";
  40                         $article = new Article( $title );
  41                         $article->updateArticle( '', wfMsg( 'spam_blanking', $domain ),
  42                                 false, false );
  43
  44                 } else {
  45                         // Revert to this revision
  46                         print "reverting\n";
  47                         $article = new Article( $title );
  48                         $article->updateArticle( $rev->getText(), wfMsg( 'spam_reverting', $domain ), false, false );
  49                 }
  50                 $dbw->immediateCommit();
  51                 wfDoUpdates();
  52         }
  53 }
  54 //------------------------------------------------------------------------------
  55
  56 $username = wfMsg( 'spambot_username' );
  57 $fname = $username;
  58 $wgUser = User::newFromName( $username );
  59 // Create the user if necessary
  60 if ( !$wgUser->getID() ) {
  61         $wgUser->addToDatabase();
  62 }
  63
  64 if ( !isset( $args[0] ) ) {
  65         print "Usage: php cleanupSpam.php <hostname>\n";
  66         exit(1);
  67 }
  68 $spec = $args[0];
  69 $like = LinkFilter::makeLike( $spec );
  70 if ( !$like ) {
  71         print "Not a valid hostname specification: $spec\n";
  72         exit(1);
  73 }
  74
  75 $dbr =& wfGetDB( DB_SLAVE );
  76
  77 $res = $dbr->select( 'externallinks', array( 'el_from' ),
  78         array( 'el_index LIKE ' . $dbr->addQuotes( $like ) ), $fname );
  79 $count = $dbr->numRows( $res );
  80 print "Found $count articles containing $spec\n";
  81 while ( $row = $dbr->fetchObject( $res ) ) {
  82         cleanupArticle( $row->el_from, $spec );
  83 }
  84 if ( $count ) {
  85         print "Done\n";
  86 }
  87
  88 ?>