populateRevisionSha1.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
use MediaWiki\MediaWikiServices;
 
class PopulateRevisionSha1 extends LoggedUpdateMaintenance {
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Populates the rev_sha1 and ar_sha1 fields' );
        $this->setBatchSize( 200 );
    }
 
    protected function getUpdateKey() {
        return 'populate rev_sha1';
    }
 
    protected function doDBUpdates() {
        $db = $this->getDB( DB_PRIMARY );
 
        if ( !$db->tableExists( 'revision', __METHOD__ ) ) {
            $this->fatalError( "revision table does not exist" );
        } elseif ( !$db->tableExists( 'archive', __METHOD__ ) ) {
            $this->fatalError( "archive table does not exist" );
        } elseif ( !$db->fieldExists( 'revision', 'rev_sha1', __METHOD__ ) ) {
            $this->output( "rev_sha1 column does not exist\n\n", true );
            return false;
        }
 
        $revStore = MediaWikiServices::getInstance()->getRevisionStore();
 
        $this->output( "Populating rev_sha1 column\n" );
        $rc = $this->doSha1Updates( $revStore, 'revision', 'rev_id',
            $revStore->getQueryInfo(), 'rev'
        );
 
        $this->output( "Populating ar_sha1 column\n" );
        $ac = $this->doSha1Updates( $revStore, 'archive', 'ar_rev_id',
            $revStore->getArchiveQueryInfo(), 'ar'
        );
 
        $this->output( "rev_sha1 and ar_sha1 population complete "
            . "[$rc revision rows, $ac archive rows].\n" );
 
        return true;
    }
 
    protected function doSha1Updates( $revStore, $table, $idCol, $queryInfo, $prefix ) {
        $db = $this->getDB( DB_PRIMARY );
        $batchSize = $this->getBatchSize();
        $start = $db->selectField( $table, "MIN($idCol)", '', __METHOD__ );
        $end = $db->selectField( $table, "MAX($idCol)", '', __METHOD__ );
        if ( !$start || !$end ) {
            $this->output( "...$table table seems to be empty.\n" );
 
            return 0;
        }
 
        $count = 0;
        # Do remaining chunk
        $end += $batchSize - 1;
        $blockStart = $start;
        $blockEnd = $start + $batchSize - 1;
        while ( $blockEnd <= $end ) {
            $this->output( "...doing $idCol from $blockStart to $blockEnd\n" );
 
            $cond = "$idCol BETWEEN " . (int)$blockStart . " AND " . (int)$blockEnd .
                " AND $idCol IS NOT NULL AND {$prefix}_sha1 = ''";
            $res = $db->select(
                $queryInfo['tables'], $queryInfo['fields'], $cond, __METHOD__, [], $queryInfo['joins']
            );
 
            $this->beginTransaction( $db, __METHOD__ );
            foreach ( $res as $row ) {
                if ( $this->upgradeRow( $revStore, $row, $table, $idCol, $prefix ) ) {
                    $count++;
                }
            }
            $this->commitTransaction( $db, __METHOD__ );
 
            $blockStart += $batchSize;
            $blockEnd += $batchSize;
        }
 
        return $count;
    }
 
    protected function upgradeRow( $revStore, $row, $table, $idCol, $prefix ) {
        $db = $this->getDB( DB_PRIMARY );
 
        // Create a revision and use it to get the sha1 from the content table, if possible.
        try {
            $rev = ( $table === 'archive' )
                ? $revStore->newRevisionFromArchiveRow( $row )
                : $revStore->newRevisionFromRow( $row );
            $sha1 = $rev->getSha1();
        } catch ( Exception $e ) {
            $this->output( "Data of revision with {$idCol}={$row->$idCol} unavailable!\n" );
            return false; // T24624? T22757?
        }
 
        $db->update( $table,
            [ "{$prefix}_sha1" => $sha1 ],
            [ $idCol => $row->$idCol ],
            __METHOD__
        );
 
        return true;
    }
}
 
$maintClass = PopulateRevisionSha1::class;
require_once RUN_MAINTENANCE_IF_MAIN;