populateRevisionLength.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
class PopulateRevisionLength extends LoggedUpdateMaintenance {
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Populates the rev_len and ar_len fields' );
        $this->setBatchSize( 200 );
    }
 
    protected function getUpdateKey() {
        return 'populate rev_len and ar_len';
    }
 
    public function doDBUpdates() {
        $dbw = $this->getDB( DB_MASTER );
        if ( !$dbw->tableExists( 'revision' ) ) {
            $this->error( "revision table does not exist", true );
        } elseif ( !$dbw->tableExists( 'archive' ) ) {
            $this->error( "archive table does not exist", true );
        } elseif ( !$dbw->fieldExists( 'revision', 'rev_len', __METHOD__ ) ) {
            $this->output( "rev_len column does not exist\n\n", true );
 
            return false;
        }
 
        $this->output( "Populating rev_len column\n" );
        $rev = $this->doLenUpdates( 'revision', 'rev_id', 'rev', Revision::selectFields() );
 
        $this->output( "Populating ar_len column\n" );
        $ar = $this->doLenUpdates( 'archive', 'ar_id', 'ar', Revision::selectArchiveFields() );
 
        $this->output( "rev_len and ar_len population complete "
            . "[$rev revision rows, $ar archive rows].\n" );
 
        return true;
    }
 
    protected function doLenUpdates( $table, $idCol, $prefix, $fields ) {
        $dbr = $this->getDB( DB_REPLICA );
        $dbw = $this->getDB( DB_MASTER );
        $start = $dbw->selectField( $table, "MIN($idCol)", false, __METHOD__ );
        $end = $dbw->selectField( $table, "MAX($idCol)", false, __METHOD__ );
        if ( !$start || !$end ) {
            $this->output( "...$table table seems to be empty.\n" );
 
            return 0;
        }
 
        # Do remaining chunks
        $blockStart = intval( $start );
        $blockEnd = intval( $start ) + $this->mBatchSize - 1;
        $count = 0;
 
        while ( $blockStart <= $end ) {
            $this->output( "...doing $idCol from $blockStart to $blockEnd\n" );
            $res = $dbr->select(
                $table,
                $fields,
                [
                    "$idCol >= $blockStart",
                    "$idCol <= $blockEnd",
                    "{$prefix}_len IS NULL"
                ],
                __METHOD__
            );
 
            if ( $res->numRows() > 0 ) {
                $this->beginTransaction( $dbw, __METHOD__ );
                # Go through and update rev_len from these rows.
                foreach ( $res as $row ) {
                    if ( $this->upgradeRow( $row, $table, $idCol, $prefix ) ) {
                        $count++;
                    }
                }
                $this->commitTransaction( $dbw, __METHOD__ );
            }
 
            $blockStart += $this->mBatchSize;
            $blockEnd += $this->mBatchSize;
            wfWaitForSlaves();
        }
 
        return $count;
    }
 
    protected function upgradeRow( $row, $table, $idCol, $prefix ) {
        $dbw = $this->getDB( DB_MASTER );
 
        $rev = ( $table === 'archive' )
            ? Revision::newFromArchiveRow( $row )
            : new Revision( $row );
 
        $content = $rev->getContent();
        if ( !$content ) {
            # This should not happen, but sometimes does (T22757)
            $id = $row->$idCol;
            $this->output( "Content of $table $id unavailable!\n" );
 
            return false;
        }
 
        # Update the row...
        $dbw->update( $table,
            [ "{$prefix}_len" => $content->getSize() ],
            [ $idCol => $row->$idCol ],
            __METHOD__
        );
 
        return true;
    }
}
 
$maintClass = "PopulateRevisionLength";
require_once RUN_MAINTENANCE_IF_MAIN;