cleanupTable.inc

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';

class TableCleanup extends Maintenance {
    protected $defaultParams = array(
        'table' => 'page',
        'conds' => array(),
        'index' => 'page_id',
        'callback' => 'processRow',
    );

    protected $dryrun = false;
    protected $maxLag = 10; # if slaves are lagged more than 10 secs, wait
    public $batchSize = 100;
    public $reportInterval = 100;

    protected $processed, $updated, $count, $startTime, $table;

    public function __construct() {
        parent::__construct();
        $this->addOption( 'dry-run', 'Perform a dry run' );
    }

    public function execute() {
        global $wgUser;
        $wgUser = User::newFromName( 'Conversion script' );
        $this->dryrun = $this->hasOption( 'dry-run' );
        if ( $this->dryrun ) {
            $this->output( "Checking for bad titles...\n" );
        } else {
            $this->output( "Checking and fixing bad titles...\n" );
        }
        $this->runTable( $this->defaultParams );
    }

    protected function init( $count, $table ) {
        $this->processed = 0;
        $this->updated = 0;
        $this->count = $count;
        $this->startTime = microtime( true );
        $this->table = $table;
    }

    protected function progress( $updated ) {
        $this->updated += $updated;
        $this->processed++;
        if ( $this->processed % $this->reportInterval != 0 ) {
            return;
        }
        $portion = $this->processed / $this->count;
        $updateRate = $this->updated / $this->processed;

        $now = microtime( true );
        $delta = $now - $this->startTime;
        $estimatedTotalTime = $delta / $portion;
        $eta = $this->startTime + $estimatedTotalTime;

        $this->output(
            sprintf( "%s %s: %6.2f%% done on %s; ETA %s [%d/%d] %.2f/sec <%.2f%% updated>\n",
                wfWikiID(),
                wfTimestamp( TS_DB, intval( $now ) ),
                $portion * 100.0,
                $this->table,
                wfTimestamp( TS_DB, intval( $eta ) ),
                $this->processed,
                $this->count,
                $this->processed / $delta,
                $updateRate * 100.0
            )
        );
        flush();
    }

    public function runTable( $params ) {
        $dbr = wfGetDB( DB_SLAVE );

        if ( array_diff( array_keys( $params ),
            array( 'table', 'conds', 'index', 'callback' ) )
        ) {
            throw new MWException( __METHOD__ . ': Missing parameter ' . implode( ', ', $params ) );
        }

        $table = $params['table'];
        // count(*) would melt the DB for huge tables, we can estimate here
        $count = $dbr->estimateRowCount( $table, '*', '', __METHOD__ );
        $this->init( $count, $table );
        $this->output( "Processing $table...\n" );

        $index = (array)$params['index'];
        $indexConds = array();
        $options = array(
            'ORDER BY' => implode( ',', $index ),
            'LIMIT' => $this->batchSize
        );
        $callback = array( $this, $params['callback'] );

        while ( true ) {
            $conds = array_merge( $params['conds'], $indexConds );
            $res = $dbr->select( $table, '*', $conds, __METHOD__, $options );
            if ( !$res->numRows() ) {
                // Done
                break;
            }

            foreach ( $res as $row ) {
                call_user_func( $callback, $row );
            }

            if ( $res->numRows() < $this->batchSize ) {
                // Done
                break;
            }

            // Update the conditions to select the next batch.
            // Construct a condition string by starting with the least significant part
            // of the index, and adding more significant parts progressively to the left
            // of the string.
            $nextCond = '';
            foreach ( array_reverse( $index ) as $field ) {
                $encValue = $dbr->addQuotes( $row->$field );
                if ( $nextCond === '' ) {
                    $nextCond = "$field > $encValue";
                } else {
                    $nextCond = "$field > $encValue OR ($field = $encValue AND ($nextCond))";
                }
            }
            $indexConds = array( $nextCond );
        }

        $this->output( "Finished $table... $this->updated of $this->processed rows updated\n" );
    }

    protected function hexChar( $matches ) {
        return sprintf( "\\x%02x", ord( $matches[1] ) );
    }
}

class TableCleanupTest extends TableCleanup {
    function processRow( $row ) {
        $this->progress( mt_rand( 0, 1 ) );
    }
}