populateImageSha1.php

Go to the documentation of this file.

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

class PopulateImageSha1 extends LoggedUpdateMaintenance {
    public function __construct() {
        parent::__construct();
        $this->addDescription( 'Populate the img_sha1 field' );
        $this->addOption( 'force', "Recalculate sha1 for rows that already have a value" );
        $this->addOption( 'multiversiononly', "Calculate only for files with several versions" );
        $this->addOption( 'method', "Use 'pipe' to pipe to mysql command line,\n" .
            "\t\tdefault uses Database class", false, true );
        $this->addOption(
            'file',
            'Fix for a specific file, without File: namespace prefixed',
            false,
            true
        );
    }

    protected function getUpdateKey() {
        return 'populate img_sha1';
    }

    protected function updateSkippedMessage() {
        return 'img_sha1 column of image table already populated.';
    }

    public function execute() {
        if ( $this->getOption( 'file' ) || $this->hasOption( 'multiversiononly' ) ) {
            $this->doDBUpdates(); // skip update log checks/saves
        } else {
            parent::execute();
        }
    }

    public function doDBUpdates() {
        $method = $this->getOption( 'method', 'normal' );
        $file = $this->getOption( 'file', '' );
        $force = $this->getOption( 'force' );
        $isRegen = ( $force || $file != '' ); // forced recalculation?

        $t = -microtime( true );
        $dbw = $this->getDB( DB_MASTER );
        if ( $file != '' ) {
            $res = $dbw->select(
                'image',
                [ 'img_name' ],
                [ 'img_name' => $file ],
                __METHOD__
            );
            if ( !$res ) {
                $this->error( "No such file: $file", true );

                return false;
            }
            $this->output( "Populating img_sha1 field for specified files\n" );
        } else {
            if ( $this->hasOption( 'multiversiononly' ) ) {
                $conds = [];
                $this->output( "Populating and recalculating img_sha1 field for versioned files\n" );
            } elseif ( $force ) {
                $conds = [];
                $this->output( "Populating and recalculating img_sha1 field\n" );
            } else {
                $conds = [ 'img_sha1' => '' ];
                $this->output( "Populating img_sha1 field\n" );
            }
            if ( $this->hasOption( 'multiversiononly' ) ) {
                $res = $dbw->select( 'oldimage',
                    [ 'img_name' => 'DISTINCT(oi_name)' ], $conds, __METHOD__ );
            } else {
                $res = $dbw->select( 'image', [ 'img_name' ], $conds, __METHOD__ );
            }
        }

        $imageTable = $dbw->tableName( 'image' );
        $oldImageTable = $dbw->tableName( 'oldimage' );

        if ( $method == 'pipe' ) {
            // Opening a pipe allows the SHA-1 operation to be done in parallel
            // with the database write operation, because the writes are queued
            // in the pipe buffer. This can improve performance by up to a
            // factor of 2.
            global $wgDBuser, $wgDBserver, $wgDBpassword, $wgDBname;
            $cmd = 'mysql -u' . wfEscapeShellArg( $wgDBuser ) .
                ' -h' . wfEscapeShellArg( $wgDBserver ) .
                ' -p' . wfEscapeShellArg( $wgDBpassword, $wgDBname );
            $this->output( "Using pipe method\n" );
            $pipe = popen( $cmd, 'w' );
        }

        $numRows = $res->numRows();
        $i = 0;
        foreach ( $res as $row ) {
            if ( $i % $this->mBatchSize == 0 ) {
                $this->output( sprintf(
                    "Done %d of %d, %5.3f%%  \r", $i, $numRows, $i / $numRows * 100 ) );
                wfWaitForSlaves();
            }

            $file = wfLocalFile( $row->img_name );
            if ( !$file ) {
                continue;
            }

            // Upgrade the current file version...
            $sha1 = $file->getRepo()->getFileSha1( $file->getPath() );
            if ( strval( $sha1 ) !== '' ) { // file on disk and hashed properly
                if ( $isRegen && $file->getSha1() !== $sha1 ) {
                    // The population was probably done already. If the old SHA1
                    // does not match, then both fix the SHA1 and the metadata.
                    $file->upgradeRow();
                } else {
                    $sql = "UPDATE $imageTable SET img_sha1=" . $dbw->addQuotes( $sha1 ) .
                        " WHERE img_name=" . $dbw->addQuotes( $file->getName() );
                    if ( $method == 'pipe' ) {
                        fwrite( $pipe, "$sql;\n" );
                    } else {
                        $dbw->query( $sql, __METHOD__ );
                    }
                }
            }
            // Upgrade the old file versions...
            foreach ( $file->getHistory() as $oldFile ) {
                $sha1 = $oldFile->getRepo()->getFileSha1( $oldFile->getPath() );
                if ( strval( $sha1 ) !== '' ) { // file on disk and hashed properly
                    if ( $isRegen && $oldFile->getSha1() !== $sha1 ) {
                        // The population was probably done already. If the old SHA1
                        // does not match, then both fix the SHA1 and the metadata.
                        $oldFile->upgradeRow();
                    } else {
                        $sql = "UPDATE $oldImageTable SET oi_sha1=" . $dbw->addQuotes( $sha1 ) .
                            " WHERE (oi_name=" . $dbw->addQuotes( $oldFile->getName() ) . " AND" .
                            " oi_archive_name=" . $dbw->addQuotes( $oldFile->getArchiveName() ) . ")";
                        if ( $method == 'pipe' ) {
                            fwrite( $pipe, "$sql;\n" );
                        } else {
                            $dbw->query( $sql, __METHOD__ );
                        }
                    }
                }
            }
            $i++;
        }
        if ( $method == 'pipe' ) {
            fflush( $pipe );
            pclose( $pipe );
        }
        $t += microtime( true );
        $this->output( sprintf( "\nDone %d files in %.1f seconds\n", $numRows, $t ) );

        return !$file; // we only updated *some* files, don't log
    }
}

$maintClass = "PopulateImageSha1";
require_once RUN_MAINTENANCE_IF_MAIN;