orphans.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
use MediaWiki\MediaWikiServices;
use Wikimedia\Rdbms\IMaintainableDatabase;
 
class Orphans extends Maintenance {
    public function __construct() {
        parent::__construct();
        $this->addDescription( "Look for 'orphan' revisions hooked to pages which don't exist\n" .
            "and 'childless' pages with no revisions\n" .
            "Then, kill the poor widows and orphans\n" .
            "Man this is depressing"
        );
        $this->addOption( 'fix', 'Actually fix broken entries' );
    }
 
    public function execute() {
        $this->checkOrphans( $this->hasOption( 'fix' ) );
        $this->checkSeparation( $this->hasOption( 'fix' ) );
        # Does not work yet, do not use
        # $this->checkWidows( $this->hasOption( 'fix' ) );
    }
 
    private function lockTables( $db, $extraTable = [] ) {
        $tbls = [ 'page', 'revision', 'redirect' ];
        if ( $extraTable ) {
            $tbls = array_merge( $tbls, $extraTable );
        }
        $db->lockTables( [], $tbls, __METHOD__, false );
    }
 
    private function checkOrphans( $fix ) {
        $dbw = $this->getDB( DB_MASTER );
        $commentStore = CommentStore::getStore();
 
        if ( $fix ) {
            $this->lockTables( $dbw );
        }
 
        $commentQuery = $commentStore->getJoin( 'rev_comment' );
        $actorQuery = ActorMigration::newMigration()->getJoin( 'rev_user' );
 
        $this->output( "Checking for orphan revision table entries... "
            . "(this may take a while on a large wiki)\n" );
        $result = $dbw->select(
            [ 'revision', 'page' ] + $commentQuery['tables'] + $actorQuery['tables'],
            [ 'rev_id', 'rev_page', 'rev_timestamp' ] + $commentQuery['fields'] + $actorQuery['fields'],
            [ 'page_id' => null ],
            __METHOD__,
            [],
            [ 'page' => [ 'LEFT JOIN', [ 'rev_page=page_id' ] ] ] + $commentQuery['joins']
                + $actorQuery['joins']
        );
        $orphans = $result->numRows();
        if ( $orphans > 0 ) {
            $this->output( "$orphans orphan revisions...\n" );
            $this->output( sprintf(
                "%10s %10s %14s %20s %s\n",
                'rev_id', 'rev_page', 'rev_timestamp', 'rev_user_text', 'rev_comment'
            ) );
 
            $contLang = MediaWikiServices::getInstance()->getContentLanguage();
            foreach ( $result as $row ) {
                $comment = $commentStore->getComment( 'rev_comment', $row )->text;
                if ( $comment !== '' ) {
                    $comment = '(' . $contLang->truncateForVisual( $comment, 40 ) . ')';
                }
                $rev_user_text = $contLang->truncateForVisual( $row->rev_user_text, 20 );
                # pad $rev_user_text to 20 characters.  Note that this may
                # yield poor results if $rev_user_text contains combining
                # or half-width characters.  Alas.
                if ( mb_strlen( $rev_user_text ) < 20 ) {
                    $rev_user_text = str_repeat( ' ', 20 - mb_strlen( $rev_user_text ) );
                }
                $this->output( sprintf( "%10d %10d %14s %s %s\n",
                    $row->rev_id,
                    $row->rev_page,
                    $row->rev_timestamp,
                    $rev_user_text,
                    $comment ) );
                if ( $fix ) {
                    $dbw->delete( 'revision', [ 'rev_id' => $row->rev_id ] );
                }
            }
            if ( !$fix ) {
                $this->output( "Run again with --fix to remove these entries automatically.\n" );
            }
        } else {
            $this->output( "No orphans! Yay!\n" );
        }
 
        if ( $fix ) {
            $dbw->unlockTables( __METHOD__ );
        }
    }
 
    private function checkWidows( $fix ) {
        $dbw = $this->getDB( DB_MASTER );
        $page = $dbw->tableName( 'page' );
        $revision = $dbw->tableName( 'revision' );
 
        if ( $fix ) {
            $this->lockTables( $dbw );
        }
 
        $this->output( "\nChecking for childless page table entries... "
            . "(this may take a while on a large wiki)\n" );
        $result = $dbw->query( "
            SELECT *
            FROM $page LEFT OUTER JOIN $revision ON page_latest=rev_id
            WHERE rev_id IS NULL
        " );
        $widows = $result->numRows();
        if ( $widows > 0 ) {
            $this->output( "$widows childless pages...\n" );
            $this->output( sprintf( "%10s %11s %2s %s\n", 'page_id', 'page_latest', 'ns', 'page_title' ) );
            foreach ( $result as $row ) {
                printf( "%10d %11d %2d %s\n",
                    $row->page_id,
                    $row->page_latest,
                    $row->page_namespace,
                    $row->page_title );
                if ( $fix ) {
                    $dbw->delete( 'page', [ 'page_id' => $row->page_id ] );
                }
            }
            if ( !$fix ) {
                $this->output( "Run again with --fix to remove these entries automatically.\n" );
            }
        } else {
            $this->output( "No childless pages! Yay!\n" );
        }
 
        if ( $fix ) {
            $dbw->unlockTables( __METHOD__ );
        }
    }
 
    private function checkSeparation( $fix ) {
        $dbw = $this->getDB( DB_MASTER );
        $page = $dbw->tableName( 'page' );
        $revision = $dbw->tableName( 'revision' );
 
        if ( $fix ) {
            $this->lockTables( $dbw, [ 'user', 'text' ] );
        }
 
        $this->output( "\nChecking for pages whose page_latest links are incorrect... "
            . "(this may take a while on a large wiki)\n" );
        $result = $dbw->query( "
            SELECT *
            FROM $page LEFT OUTER JOIN $revision ON page_latest=rev_id
        " );
        $found = 0;
        foreach ( $result as $row ) {
            $result2 = $dbw->query( "
                SELECT MAX(rev_timestamp) as max_timestamp
                FROM $revision
                WHERE rev_page=" . (int)( $row->page_id )
            );
            $row2 = $dbw->fetchObject( $result2 );
            if ( $row2 ) {
                if ( $row->rev_timestamp != $row2->max_timestamp ) {
                    if ( $found == 0 ) {
                        $this->output( sprintf( "%10s %10s %14s %14s\n",
                            'page_id', 'rev_id', 'timestamp', 'max timestamp' ) );
                    }
                    ++$found;
                    $this->output( sprintf( "%10d %10d %14s %14s\n",
                        $row->page_id,
                        $row->page_latest,
                        $row->rev_timestamp,
                        $row2->max_timestamp ) );
                    if ( $fix ) {
                        # ...
                        $maxId = $dbw->selectField(
                            'revision',
                            'rev_id',
                            [
                                'rev_page' => $row->page_id,
                                'rev_timestamp' => $row2->max_timestamp ] );
                        $this->output( "... updating to revision $maxId\n" );
                        $maxRev = Revision::newFromId( $maxId );
                        $title = Title::makeTitle( $row->page_namespace, $row->page_title );
                        $article = WikiPage::factory( $title );
                        $article->updateRevisionOn( $dbw, $maxRev );
                    }
                }
            } else {
                $this->output( "wtf\n" );
            }
        }
 
        if ( $found ) {
            $this->output( "Found $found pages with incorrect latest revision.\n" );
        } else {
            $this->output( "No pages with incorrect latest revision. Yay!\n" );
        }
        if ( !$fix && $found > 0 ) {
            $this->output( "Run again with --fix to remove these entries automatically.\n" );
        }
 
        if ( $fix ) {
            $dbw->unlockTables( __METHOD__ );
        }
    }
}
 
$maintClass = Orphans::class;
require_once RUN_MAINTENANCE_IF_MAIN;