cleanupSpam.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
use MediaWiki\Revision\RevisionRecord;
use MediaWiki\Revision\SlotRecord;
 
require_once __DIR__ . '/Maintenance.php';
 
class CleanupSpam extends Maintenance {
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Cleanup all spam from a given hostname' );
        $this->addOption( 'all', 'Check all wikis in $wgLocalDatabases' );
        $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 $IP, $wgLocalDatabases, $wgUser;
 
        $username = wfMessage( 'spambot_username' )->text();
        $user = User::newSystemUser( $username );
        if ( !$user ) {
            $this->fatalError( "Invalid username specified in 'spambot_username' message: $username" );
        }
        // Hack: Grant bot rights so we don't flood RecentChanges
        $user->addGroup( 'bot' );
        $wgUser = $user;
 
        $spec = $this->getArg( 0 );
 
        $protConds = [];
        foreach ( [ 'http://', 'https://' ] as $prot ) {
            $conds = LinkFilter::getQueryConditions( $spec, [ 'protocol' => $prot ] );
            if ( !$conds ) {
                $this->fatalError( "Not a valid hostname specification: $spec" );
            }
            $protConds[$prot] = $conds;
        }
 
        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 = $this->getDB( DB_REPLICA, [], $wikiId );
 
                foreach ( $protConds as $conds ) {
                    $count = $dbr->selectField(
                        'externallinks',
                        'COUNT(*)',
                        $conds,
                        __METHOD__
                    );
                    if ( $count ) {
                        $found = true;
                        $cmd = wfShellWikiCmd(
                            "$IP/maintenance/cleanupSpam.php",
                            [ '--wiki', $wikiId, $spec ]
                        );
                        passthru( "$cmd | sed 's/^/$wikiId:  /'" );
                    }
                }
            }
            if ( $found ) {
                $this->output( "All done\n" );
            } else {
                $this->output( "None found\n" );
            }
        } else {
            // Clean up spam on this wiki
 
            $count = 0;
            $dbr = $this->getDB( DB_REPLICA );
            foreach ( $protConds as $prot => $conds ) {
                $res = $dbr->select(
                    'externallinks',
                    [ 'DISTINCT el_from' ],
                    $conds,
                    __METHOD__
                );
                $count = $dbr->numRows( $res );
                $this->output( "Found $count articles containing $spec\n" );
                foreach ( $res as $row ) {
                    $this->cleanupArticle(
                        $row->el_from,
                        $spec,
                        $prot,
                        $wgUser
                    );
                }
            }
            if ( $count ) {
                $this->output( "Done\n" );
            }
        }
    }
 
    private function cleanupArticle( $id, $domain, $protocol, User $deleter ) {
        $title = Title::newFromID( $id );
        if ( !$title ) {
            $this->error( "Internal error: no page for ID $id" );
 
            return;
        }
 
        $this->output( $title->getPrefixedDBkey() . " ..." );
 
        $revLookup = MediaWikiServices::getInstance()->getRevisionLookup();
        $rev = $revLookup->getRevisionByTitle( $title );
        $currentRevId = $rev->getId();
 
        while ( $rev && ( $rev->isDeleted( RevisionRecord::DELETED_TEXT ) ||
            LinkFilter::matchEntry(
                $rev->getContent( SlotRecord::MAIN, RevisionRecord::RAW ),
                $domain,
                $protocol
            ) )
        ) {
            $rev = $revLookup->getPreviousRevision( $rev );
        }
 
        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 = $this->getDB( DB_MASTER );
            $this->beginTransaction( $dbw, __METHOD__ );
            $page = WikiPage::factory( $title );
            if ( $rev ) {
                // Revert to this revision
                $content = $rev->getContent( SlotRecord::MAIN, RevisionRecord::RAW );
 
                $this->output( "reverting\n" );
                $page->doEditContent(
                    $content,
                    wfMessage( 'spam_reverting', $domain )->inContentLanguage()->text(),
                    EDIT_UPDATE | EDIT_FORCE_BOT,
                    $rev->getId()
                );
            } elseif ( $this->hasOption( 'delete' ) ) {
                // Didn't find a non-spammy revision, blank the page
                $this->output( "deleting\n" );
                $page->doDeleteArticleReal(
                    wfMessage( 'spam_deleting', $domain )->inContentLanguage()->text(),
                    $deleter
                );
            } else {
                // Didn't find a non-spammy revision, blank the page
                $handler = MediaWikiServices::getInstance()
                    ->getContentHandlerFactory()
                    ->getContentHandler( $title->getContentModel() );
                $content = $handler->makeEmptyContent();
 
                $this->output( "blanking\n" );
                $page->doEditContent(
                    $content,
                    wfMessage( 'spam_blanking', $domain )->inContentLanguage()->text(),
                    EDIT_UPDATE | EDIT_FORCE_BOT
                );
            }
            $this->commitTransaction( $dbw, __METHOD__ );
        }
    }
}
 
$maintClass = CleanupSpam::class;
require_once RUN_MAINTENANCE_IF_MAIN;