Merge "Clean up applyPatch() usage in the installer."

[lhc/web/wiklou.git] / maintenance / cleanupSpam.php

diff --git a/maintenance/cleanupSpam.php b/maintenance/cleanupSpam.php
index d1927f9..2051c89 100644 (file)
--- a/maintenance/cleanupSpam.php
+++ b/maintenance/cleanupSpam.php
@@ -1,6 +1,6 @@
 <?php
 /**
- * Cleanup all spam from a given hostname
+ * Cleanup all spam from a given hostname.
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -17,24 +17,35 @@
  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  * http://www.gnu.org/copyleft/gpl.html
  *
+ * @file
  * @ingroup Maintenance
  */
 
-require_once( dirname(__FILE__) . '/Maintenance.php' );
+require_once( dirname( __FILE__ ) . '/Maintenance.php' );
 
+/**
+ * Maintenance script to cleanup all spam from a given hostname.
+ *
+ * @ingroup Maintenance
+ */
 class CleanupSpam extends Maintenance {
+
        public function __construct() {
                parent::__construct();
                $this->mDescription = "Cleanup all spam from a given hostname";
                $this->addOption( 'all', 'Check all wikis in $wgLocalDatabases' );
-               $this->addArg( 'hostname', 'Hostname that was spamming' );
+               $this->addOption( 'delete', 'Delete pages containing only spam instead of blanking them' );
+               $this->addArg( 'hostname', 'Hostname that was spamming, single * wildcard in the beginning allowed' );
        }
 
        public function execute() {
-               global $wgLocalDatabases;
+               global $wgLocalDatabases, $wgUser;
 
-               $username = wfMsg( 'spambot_username' );
+               $username = wfMessage( 'spambot_username' )->text();
                $wgUser = User::newFromName( $username );
+               if ( !$wgUser ) {
+                       $this->error( "Invalid username", true );
+               }
                // Create the user if necessary
                if ( !$wgUser->getId() ) {
                        $wgUser->addToDatabase();
@@ -44,15 +55,15 @@ class CleanupSpam extends Maintenance {
                if ( !$like ) {
                        $this->error( "Not a valid hostname specification: $spec", true );
                }
-       
-               if ( $this->hasOption('all') ) {
+
+               if ( $this->hasOption( 'all' ) ) {
                        // Clean up spam on all wikis
                        $this->output( "Finding spam on " . count( $wgLocalDatabases ) . " wikis\n" );
                        $found = false;
                        foreach ( $wgLocalDatabases as $wikiID ) {
                                $dbr = wfGetDB( DB_SLAVE, array(), $wikiID );
 
-                               $count = $dbr->selectField( 'externallinks', 'COUNT(*)', 
+                               $count = $dbr->selectField( 'externallinks', 'COUNT(*)',
                                        array( 'el_index' . $dbr->buildLike( $like ) ), __METHOD__ );
                                if ( $count ) {
                                        $found = true;
@@ -68,7 +79,7 @@ class CleanupSpam extends Maintenance {
                        // Clean up spam on this wiki
 
                        $dbr = wfGetDB( DB_SLAVE );
-                       $res = $dbr->select( 'externallinks', array( 'DISTINCT el_from' ), 
+                       $res = $dbr->select( 'externallinks', array( 'DISTINCT el_from' ),
                                array( 'el_index' . $dbr->buildLike( $like ) ), __METHOD__ );
                        $count = $dbr->numRows( $res );
                        $this->output( "Found $count articles containing $spec\n" );
@@ -87,47 +98,41 @@ class CleanupSpam extends Maintenance {
                        $this->error( "Internal error: no page for ID $id" );
                        return;
                }
-       
+
                $this->output( $title->getPrefixedDBkey() . " ..." );
                $rev = Revision::newFromTitle( $title );
-               $revId = $rev->getId();
-               $currentRevId = $revId;
-       
-               while ( $rev && LinkFilter::matchEntry( $rev->getText() , $domain ) ) {
-                       # Revision::getPrevious can't be used in this way before MW 1.6 (Revision.php 1.26)
-                       #$rev = $rev->getPrevious();
-                       $revId = $title->getPreviousRevisionID( $revId );
-                       if ( $revId ) {
-                               $rev = Revision::newFromTitle( $title, $revId );
-                       } else {
-                               $rev = false;
-                       }
+               $currentRevId = $rev->getId();
+
+               while ( $rev && ( $rev->isDeleted( Revision::DELETED_TEXT ) || LinkFilter::matchEntry( $rev->getText() , $domain ) ) ) {
+                       $rev = $rev->getPrevious();
                }
-               if ( $revId == $currentRevId ) {
+
+               if ( $rev && $rev->getId() == $currentRevId ) {
                        // The regex didn't match the current article text
                        // This happens e.g. when a link comes from a template rather than the page itself
                        $this->output( "False match\n" );
                } else {
                        $dbw = wfGetDB( DB_MASTER );
-                       $dbw->immediateBegin();
-                       if ( !$rev ) {
-                               // Didn't find a non-spammy revision, blank the page
-                               $this->output( "blanking\n" );
-                               $article = new Article( $title );
-                               $article->updateArticle( '', wfMsg( 'spam_blanking', $domain ),
-                                       false, false );
-       
-                       } else {
+                       $dbw->begin( __METHOD__ );
+                       $page = WikiPage::factory( $title );
+                       if ( $rev ) {
                                // Revert to this revision
                                $this->output( "reverting\n" );
-                               $article = new Article( $title );
-                               $article->updateArticle( $rev->getText(), wfMsg( 'spam_reverting', $domain ), false, false );
+                               $page->doEdit( $rev->getText(), wfMessage( 'spam_reverting', $domain )->inContentLanguage()->text(),
+                                       EDIT_UPDATE, $rev->getId() );
+                       } elseif ( $this->hasOption( 'delete' ) ) {
+                               // Didn't find a non-spammy revision, blank the page
+                               $this->output( "deleting\n" );
+                               $page->doDeleteArticle( wfMessage( 'spam_deleting', $domain )->inContentLanguage()->text() );
+                       } else {
+                               // Didn't find a non-spammy revision, blank the page
+                               $this->output( "blanking\n" );
+                               $page->doEdit( '', wfMessage( 'spam_blanking', $domain )->inContentLanguage()->text() );
                        }
-                       $dbw->immediateCommit();
-                       wfDoUpdates();
+                       $dbw->commit( __METHOD__ );
                }
        }
 }
 
 $maintClass = "CleanupSpam";
-require_once( DO_MAINTENANCE );
+require_once( RUN_MAINTENANCE_IF_MAIN );