recountCategories.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
use MediaWiki\MediaWikiServices;
 
class RecountCategories extends Maintenance {
    private $mode;
 
    private $minimumId;
 
    public function __construct() {
        parent::__construct();
        $this->addDescription( <<<'TEXT'
This script refreshes the category membership counts stored in the category
table. As time passes, these counts often drift from the actual number of
category members. The script identifies rows where the value in the category
table does not match the number of categorylinks rows for that category, and
updates the category table accordingly.
 
To fully refresh the data in the category table, you need to run this script
three times: once in each mode. Alternatively, just one mode can be run if
required.
TEXT
        );
        $this->addOption(
            'mode',
            '(REQUIRED) Which category count column to recompute: "pages", "subcats" or "files".',
            true,
            true
        );
        $this->addOption(
            'begin',
            'Only recount categories with cat_id greater than the given value',
            false,
            true
        );
        $this->addOption(
            'throttle',
            'Wait this many milliseconds after each batch. Default: 0',
            false,
            true
        );
 
        $this->setBatchSize( 500 );
    }
 
    public function execute() {
        $this->mode = $this->getOption( 'mode' );
        if ( !in_array( $this->mode, [ 'pages', 'subcats', 'files' ] ) ) {
            $this->fatalError( 'Please specify a valid mode: one of "pages", "subcats" or "files".' );
        }
 
        $this->minimumId = intval( $this->getOption( 'begin', 0 ) );
 
        // do the work, batch by batch
        $affectedRows = 0;
        while ( ( $result = $this->doWork() ) !== false ) {
            $affectedRows += $result;
            usleep( $this->getOption( 'throttle', 0 ) * 1000 );
        }
 
        $this->output( "Done! Updated the {$this->mode} counts of $affectedRows categories.\n" .
            "Now run the script using the other --mode options if you haven't already.\n" );
        if ( $this->mode === 'pages' ) {
            $this->output(
                "Also run 'php cleanupEmptyCategories.php --mode remove' to remove empty,\n" .
                "nonexistent categories from the category table.\n\n" );
        }
    }
 
    protected function doWork() {
        $this->output( "Finding up to {$this->getBatchSize()} drifted rows " .
            "greater than cat_id {$this->minimumId}...\n" );
 
        $countingConds = [ 'cl_to = cat_title' ];
        if ( $this->mode === 'subcats' ) {
            $countingConds['cl_type'] = 'subcat';
        } elseif ( $this->mode === 'files' ) {
            $countingConds['cl_type'] = 'file';
        }
 
        $dbr = $this->getDB( DB_REPLICA, 'vslow' );
        $countingSubquery = $dbr->selectSQLText( 'categorylinks',
            'COUNT(*)',
            $countingConds,
            __METHOD__ );
 
        // First, let's find out which categories have drifted and need to be updated.
        // The query counts the categorylinks for each category on the replica DB,
        // but this data can't be used for updating the master, so we don't include it
        // in the results.
        $idsToUpdate = $dbr->selectFieldValues( 'category',
            'cat_id',
            [
                'cat_id > ' . (int)$this->minimumId,
                "cat_{$this->mode} != ($countingSubquery)"
            ],
            __METHOD__,
            [ 'LIMIT' => $this->getBatchSize() ]
        );
        if ( !$idsToUpdate ) {
            return false;
        }
        $this->output( "Updating cat_{$this->mode} field on " .
            count( $idsToUpdate ) . " rows...\n" );
 
        // In the next batch, start where this query left off. The rows selected
        // in this iteration shouldn't be selected again after being updated, but
        // we still keep track of where we are up to, as extra protection against
        // infinite loops.
        $this->minimumId = end( $idsToUpdate );
 
        // Now, on master, find the correct counts for these categories.
        $dbw = $this->getDB( DB_PRIMARY );
        $res = $dbw->select( 'category',
            [ 'cat_id', 'count' => "($countingSubquery)" ],
            [ 'cat_id' => $idsToUpdate ],
            __METHOD__ );
 
        // Update the category counts on the rows we just identified.
        // This logic is equivalent to Category::refreshCounts, except here, we
        // don't remove rows when cat_pages is zero and the category description page
        // doesn't exist - instead we print a suggestion to run
        // cleanupEmptyCategories.php.
        $affectedRows = 0;
        foreach ( $res as $row ) {
            $dbw->update( 'category',
                [ "cat_{$this->mode}" => $row->count ],
                [
                    'cat_id' => $row->cat_id,
                    "cat_{$this->mode} != " . (int)( $row->count ),
                ],
                __METHOD__ );
            $affectedRows += $dbw->affectedRows();
        }
 
        MediaWikiServices::getInstance()->getDBLoadBalancerFactory()->waitForReplication();
 
        return $affectedRows;
    }
}
 
$maintClass = RecountCategories::class;
require_once RUN_MAINTENANCE_IF_MAIN;