refreshImageMetadata.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
use MediaWiki\MediaWikiServices;
use Wikimedia\Rdbms\IDatabase;
use Wikimedia\Rdbms\IMaintainableDatabase;
 
class RefreshImageMetadata extends Maintenance {
 
    protected $dbw;
 
    public function __construct() {
        parent::__construct();
 
        $this->addDescription( 'Script to update image metadata records' );
        $this->setBatchSize( 200 );
 
        $this->addOption(
            'force',
            'Reload metadata from file even if the metadata looks ok',
            false,
            false,
            'f'
        );
        $this->addOption(
            'broken-only',
            'Only fix really broken records, leave old but still compatible records alone.'
        );
        $this->addOption(
            'convert-to-json',
            'Fix records with an out of date serialization format.'
        );
        $this->addOption(
            'split',
            'Enable splitting out large metadata items to the text table. Implies --convert-to-json.'
        );
        $this->addOption(
            'verbose',
            'Output extra information about each upgraded/non-upgraded file.',
            false,
            false,
            'v'
        );
        $this->addOption( 'start', 'Name of file to start with', false, true );
        $this->addOption( 'end', 'Name of file to end with', false, true );
 
        $this->addOption(
            'mediatype',
            'Only refresh files with this media type, e.g. BITMAP, UNKNOWN etc.',
            false,
            true
        );
        $this->addOption(
            'mime',
            "Only refresh files with this MIME type. Can accept wild-card 'image/*'. "
                . "Potentially inefficient unless 'mediatype' is also specified",
            false,
            true
        );
        $this->addOption(
            'metadata-contains',
            '(Inefficient!) Only refresh files where the img_metadata field '
                . 'contains this string. Can be used if its known a specific '
                . 'property was being extracted incorrectly.',
            false,
            true
        );
        $this->addOption(
            'sleep',
            'Time to sleep between each batch (in seconds). Default: 0',
            false,
            true
        );
    }
 
    public function execute() {
        $force = $this->hasOption( 'force' );
        $brokenOnly = $this->hasOption( 'broken-only' );
        $verbose = $this->hasOption( 'verbose' );
        $start = $this->getOption( 'start', false );
        $split = $this->hasOption( 'split' );
        $sleep = (int)$this->getOption( 'sleep', 0 );
        $reserialize = $this->hasOption( 'convert-to-json' );
 
        $upgraded = 0;
        $leftAlone = 0;
        $error = 0;
 
        $dbw = $this->getDB( DB_PRIMARY );
        $batchSize = $this->getBatchSize();
        if ( $batchSize <= 0 ) {
            $this->fatalError( "Batch size is too low...", 12 );
        }
 
        $repo = $this->newLocalRepo( $force, $brokenOnly, $reserialize, $split );
        $conds = $this->getConditions( $dbw );
 
        // For the WHERE img_name > 'foo' condition that comes after doing a batch
        $conds2 = [];
        if ( $start !== false ) {
            $conds2[] = 'img_name >= ' . $dbw->addQuotes( $start );
        }
 
        $options = [
            'LIMIT' => $batchSize,
            'ORDER BY' => 'img_name ASC',
        ];
 
        $fileQuery = LocalFile::getQueryInfo();
        $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
 
        do {
            $res = $dbw->select(
                $fileQuery['tables'],
                $fileQuery['fields'],
                array_merge( $conds, $conds2 ),
                __METHOD__,
                $options,
                $fileQuery['joins']
            );
 
            if ( $res->numRows() > 0 ) {
                $row1 = $res->current();
                $this->output( "Processing next {$res->numRows()} row(s) starting with {$row1->img_name}.\n" );
                $res->rewind();
            }
 
            foreach ( $res as $row ) {
                try {
                    // LocalFile will upgrade immediately here if obsolete
                    $file = $repo->newFileFromRow( $row );
                    $file->maybeUpgradeRow();
                    if ( $file->getUpgraded() ) {
                        // File was upgraded.
                        $upgraded++;
                        $this->output( "Refreshed File:{$row->img_name}.\n" );
                    } else {
                        $leftAlone++;
                        if ( $force ) {
                            $file->upgradeRow();
                            if ( $verbose ) {
                                $this->output( "Forcibly refreshed File:{$row->img_name}.\n" );
                            }
                        } else {
                            if ( $verbose ) {
                                $this->output( "Skipping File:{$row->img_name}.\n" );
                            }
                        }
                    }
                } catch ( Exception $e ) {
                    $this->output( "{$row->img_name} failed. {$e->getMessage()}\n" );
                }
            }
            $conds2 = [ 'img_name > ' . $dbw->addQuotes( $row->img_name ) ];
            $lbFactory->waitForReplication();
            if ( $sleep ) {
                sleep( $sleep );
            }
        } while ( $res->numRows() === $batchSize );
 
        $total = $upgraded + $leftAlone;
        if ( $force ) {
            $this->output( "\nFinished refreshing file metadata for $total files. "
                . "$upgraded needed to be refreshed, $leftAlone did not need to "
                . "be but were refreshed anyways, and $error refreshes were suspicious.\n" );
        } else {
            $this->output( "\nFinished refreshing file metadata for $total files. "
                . "$upgraded were refreshed, $leftAlone were already up to date, "
                . "and $error refreshes were suspicious.\n" );
        }
    }
 
    private function getConditions( $dbw ) {
        $conds = [];
 
        $end = $this->getOption( 'end', false );
        $mime = $this->getOption( 'mime', false );
        $mediatype = $this->getOption( 'mediatype', false );
        $like = $this->getOption( 'metadata-contains', false );
 
        if ( $end !== false ) {
            $conds[] = 'img_name <= ' . $dbw->addQuotes( $end );
        }
        if ( $mime !== false ) {
            list( $major, $minor ) = File::splitMime( $mime );
            $conds['img_major_mime'] = $major;
            if ( $minor !== '*' ) {
                $conds['img_minor_mime'] = $minor;
            }
        }
        if ( $mediatype !== false ) {
            $conds['img_media_type'] = $mediatype;
        }
        if ( $like ) {
            $conds[] = 'img_metadata ' . $dbw->buildLike( $dbw->anyString(), $like, $dbw->anyString() );
        }
 
        return $conds;
    }
 
    private function newLocalRepo( $force, $brokenOnly, $reserialize, $split ): LocalRepo {
        if ( $brokenOnly && $force ) {
            $this->fatalError( 'Cannot use --broken-only and --force together. ', 2 );
        }
        $reserialize = $reserialize || $split;
        if ( $brokenOnly && $reserialize ) {
            $this->fatalError( 'Cannot use --broken-only with --convert-to-json or --split. ',
                2 );
        }
 
        $overrides = [
            'updateCompatibleMetadata' => !$brokenOnly,
        ];
        if ( $reserialize ) {
            $overrides['reserializeMetadata'] = true;
            $overrides['useJsonMetadata'] = true;
        }
        if ( $split ) {
            $overrides['useSplitMetadata'] = true;
        }
 
        return MediaWikiServices::getInstance()->getRepoGroup()
            ->newCustomLocalRepo( $overrides );
    }
}
 
$maintClass = RefreshImageMetadata::class;
require_once RUN_MAINTENANCE_IF_MAIN;