fixDefaultJsonContentPages.php

Go to the documentation of this file.

<?php
use MediaWiki\MediaWikiServices;
use MediaWiki\Revision\RevisionRecord;
use MediaWiki\Revision\SlotRecord;
use MediaWiki\Title\Title;
 
require_once __DIR__ . '/Maintenance.php';
 
class FixDefaultJsonContentPages extends LoggedUpdateMaintenance {
    public function __construct() {
        parent::__construct();
        $this->addDescription(
            'Fix instances of JSON pages prior to them being the ContentHandler default' );
        $this->setBatchSize( 100 );
    }
 
    protected function getUpdateKey() {
        return __CLASS__;
    }
 
    protected function doDBUpdates() {
        $dbr = $this->getDB( DB_REPLICA );
        $namespaces = [
            NS_MEDIAWIKI => $dbr->buildLike( $dbr->anyString(), '.json' ),
            NS_USER => $dbr->buildLike( $dbr->anyString(), '/', $dbr->anyString(), '.json' ),
        ];
        foreach ( $namespaces as $ns => $like ) {
            $lastPage = 0;
            do {
                $rows = $dbr->select(
                    'page',
                    [ 'page_id', 'page_title', 'page_namespace', 'page_content_model' ],
                    [
                        'page_namespace' => $ns,
                        'page_title ' . $like,
                        'page_id > ' . $dbr->addQuotes( $lastPage )
                    ],
                    __METHOD__,
                    [ 'ORDER BY' => 'page_id', 'LIMIT' => $this->getBatchSize() ]
                );
                foreach ( $rows as $row ) {
                    $this->handleRow( $row );
                    $lastPage = $row->page_id;
                }
            } while ( $rows->numRows() >= $this->getBatchSize() );
        }
 
        return true;
    }
 
    protected function handleRow( stdClass $row ) {
        $title = Title::makeTitle( $row->page_namespace, $row->page_title );
        $this->output( "Processing {$title} ({$row->page_id})...\n" );
        $rev = MediaWikiServices::getInstance()
            ->getRevisionLookup()
            ->getRevisionByTitle( $title );
        $content = $rev->getContent( SlotRecord::MAIN, RevisionRecord::RAW );
        $dbw = $this->getDB( DB_PRIMARY );
        if ( $content instanceof JsonContent ) {
            if ( $content->isValid() ) {
                // Yay, actually JSON. We need to just change the
                // page_content_model because revision will automatically
                // use the default, which is *now* JSON.
                $this->output( "Setting page_content_model to json..." );
                $dbw->update(
                    'page',
                    [ 'page_content_model' => CONTENT_MODEL_JSON ],
                    [ 'page_id' => $row->page_id ],
                    __METHOD__
                );
                $this->output( "done.\n" );
                $this->waitForReplication();
            } else {
                // Not JSON...force it to wikitext. We need to update the
                // revision table so that these revisions are always processed
                // as wikitext in the future. page_content_model is already
                // set to "wikitext".
                $this->output( "Setting rev_content_model to wikitext..." );
                // Grab all the ids for batching
                $ids = $dbw->selectFieldValues(
                    'revision',
                    'rev_id',
                    [ 'rev_page' => $row->page_id ],
                    __METHOD__
                );
                foreach ( array_chunk( $ids, 50 ) as $chunk ) {
                    $dbw->update(
                        'revision',
                        [ 'rev_content_model' => CONTENT_MODEL_WIKITEXT ],
                        [ 'rev_page' => $row->page_id, 'rev_id' => $chunk ],
                        __METHOD__
                    );
                    $this->waitForReplication();
                }
                $this->output( "done.\n" );
            }
        } else {
            $this->output( "not a JSON page? Skipping\n" );
        }
    }
}
 
$maintClass = FixDefaultJsonContentPages::class;
require_once RUN_MAINTENANCE_IF_MAIN;