MediaWiki
REL1_22
|
00001 <?php 00024 require_once __DIR__ . '/Maintenance.php'; 00025 00031 class CleanupSpam extends Maintenance { 00032 00033 public function __construct() { 00034 parent::__construct(); 00035 $this->mDescription = "Cleanup all spam from a given hostname"; 00036 $this->addOption( 'all', 'Check all wikis in $wgLocalDatabases' ); 00037 $this->addOption( 'delete', 'Delete pages containing only spam instead of blanking them' ); 00038 $this->addArg( 'hostname', 'Hostname that was spamming, single * wildcard in the beginning allowed' ); 00039 } 00040 00041 public function execute() { 00042 global $wgLocalDatabases, $wgUser; 00043 00044 $username = wfMessage( 'spambot_username' )->text(); 00045 $wgUser = User::newFromName( $username ); 00046 if ( !$wgUser ) { 00047 $this->error( "Invalid username specified in 'spambot_username' message: $username", true ); 00048 } 00049 // Create the user if necessary 00050 if ( !$wgUser->getId() ) { 00051 $wgUser->addToDatabase(); 00052 } 00053 $spec = $this->getArg(); 00054 $like = LinkFilter::makeLikeArray( $spec ); 00055 if ( !$like ) { 00056 $this->error( "Not a valid hostname specification: $spec", true ); 00057 } 00058 00059 if ( $this->hasOption( 'all' ) ) { 00060 // Clean up spam on all wikis 00061 $this->output( "Finding spam on " . count( $wgLocalDatabases ) . " wikis\n" ); 00062 $found = false; 00063 foreach ( $wgLocalDatabases as $wikiID ) { 00064 $dbr = wfGetDB( DB_SLAVE, array(), $wikiID ); 00065 00066 $count = $dbr->selectField( 'externallinks', 'COUNT(*)', 00067 array( 'el_index' . $dbr->buildLike( $like ) ), __METHOD__ ); 00068 if ( $count ) { 00069 $found = true; 00070 passthru( "php cleanupSpam.php --wiki='$wikiID' $spec | sed 's/^/$wikiID: /'" ); 00071 } 00072 } 00073 if ( $found ) { 00074 $this->output( "All done\n" ); 00075 } else { 00076 $this->output( "None found\n" ); 00077 } 00078 } else { 00079 // Clean up spam on this wiki 00080 00081 $dbr = wfGetDB( DB_SLAVE ); 00082 $res = $dbr->select( 'externallinks', array( 'DISTINCT el_from' ), 00083 array( 'el_index' . $dbr->buildLike( $like ) ), __METHOD__ ); 00084 $count = $dbr->numRows( $res ); 00085 $this->output( "Found $count articles containing $spec\n" ); 00086 foreach ( $res as $row ) { 00087 $this->cleanupArticle( $row->el_from, $spec ); 00088 } 00089 if ( $count ) { 00090 $this->output( "Done\n" ); 00091 } 00092 } 00093 } 00094 00095 private function cleanupArticle( $id, $domain ) { 00096 $title = Title::newFromID( $id ); 00097 if ( !$title ) { 00098 $this->error( "Internal error: no page for ID $id" ); 00099 return; 00100 } 00101 00102 $this->output( $title->getPrefixedDBkey() . " ..." ); 00103 $rev = Revision::newFromTitle( $title ); 00104 $currentRevId = $rev->getId(); 00105 00106 while ( $rev && ( $rev->isDeleted( Revision::DELETED_TEXT ) 00107 || LinkFilter::matchEntry( $rev->getContent( Revision::RAW ), $domain ) ) ) { 00108 $rev = $rev->getPrevious(); 00109 } 00110 00111 if ( $rev && $rev->getId() == $currentRevId ) { 00112 // The regex didn't match the current article text 00113 // This happens e.g. when a link comes from a template rather than the page itself 00114 $this->output( "False match\n" ); 00115 } else { 00116 $dbw = wfGetDB( DB_MASTER ); 00117 $dbw->begin( __METHOD__ ); 00118 $page = WikiPage::factory( $title ); 00119 if ( $rev ) { 00120 // Revert to this revision 00121 $content = $rev->getContent( Revision::RAW ); 00122 00123 $this->output( "reverting\n" ); 00124 $page->doEditContent( $content, wfMessage( 'spam_reverting', $domain )->inContentLanguage()->text(), 00125 EDIT_UPDATE, $rev->getId() ); 00126 } elseif ( $this->hasOption( 'delete' ) ) { 00127 // Didn't find a non-spammy revision, blank the page 00128 $this->output( "deleting\n" ); 00129 $page->doDeleteArticle( wfMessage( 'spam_deleting', $domain )->inContentLanguage()->text() ); 00130 } else { 00131 // Didn't find a non-spammy revision, blank the page 00132 $handler = ContentHandler::getForTitle( $title ); 00133 $content = $handler->makeEmptyContent(); 00134 00135 $this->output( "blanking\n" ); 00136 $page->doEditContent( $content, wfMessage( 'spam_blanking', $domain )->inContentLanguage()->text() ); 00137 } 00138 $dbw->commit( __METHOD__ ); 00139 } 00140 } 00141 } 00142 00143 $maintClass = "CleanupSpam"; 00144 require_once RUN_MAINTENANCE_IF_MAIN;