backup.inc

Go to the documentation of this file.

<?php
class DumpDBZip2Output extends DumpPipeOutput {
    function __construct( $file ) {
        parent::__construct( "dbzip2", $file );
    }
}

class BackupDumper {
    var $reportingInterval = 100;
    var $reporting = true;
    var $pageCount = 0;
    var $revCount = 0;
    var $server = null; // use default
    var $pages = null; // all pages
    var $skipHeader = false; // don't output <mediawiki> and <siteinfo>
    var $skipFooter = false; // don't output </mediawiki>
    var $startId = 0;
    var $endId = 0;
    var $revStartId = 0;
    var $revEndId = 0;
    var $sink = null; // Output filters
    var $stubText = false; // include rev_text_id instead of text; for 2-pass dump
    var $dumpUploads = false;
    var $dumpUploadFileContents = false;
    var $lastTime = 0;
    var $pageCountLast = 0;
    var $revCountLast = 0;
    var $ID = 0;

    var $outputTypes = array(), $filterTypes = array();

    protected $forcedDb = null;

    protected $lb;

    function __construct( $args ) {
        $this->stderr = fopen( "php://stderr", "wt" );

        // Built-in output and filter plugins
        $this->registerOutput( 'file', 'DumpFileOutput' );
        $this->registerOutput( 'gzip', 'DumpGZipOutput' );
        $this->registerOutput( 'bzip2', 'DumpBZip2Output' );
        $this->registerOutput( 'dbzip2', 'DumpDBZip2Output' );
        $this->registerOutput( '7zip', 'Dump7ZipOutput' );

        $this->registerFilter( 'latest', 'DumpLatestFilter' );
        $this->registerFilter( 'notalk', 'DumpNotalkFilter' );
        $this->registerFilter( 'namespace', 'DumpNamespaceFilter' );

        $this->sink = $this->processArgs( $args );
    }

    function registerOutput( $name, $class ) {
        $this->outputTypes[$name] = $class;
    }

    function registerFilter( $name, $class ) {
        $this->filterTypes[$name] = $class;
    }

    function loadPlugin( $class, $file ) {
        if ( $file != '' ) {
            require_once $file;
        }
        $register = array( $class, 'register' );
        call_user_func_array( $register, array( &$this ) );
    }

    function processArgs( $args ) {
        $sink = null;
        $sinks = array();
        foreach ( $args as $arg ) {
            $matches = array();
            if ( preg_match( '/^--(.+?)(?:=(.+?)(?::(.+?))?)?$/', $arg, $matches ) ) {
                @list( /* $full */ , $opt, $val, $param ) = $matches;
                switch ( $opt ) {
                case "plugin":
                    $this->loadPlugin( $val, $param );
                    break;
                case "output":
                    if ( !is_null( $sink ) ) {
                        $sinks[] = $sink;
                    }
                    if ( !isset( $this->outputTypes[$val] ) ) {
                        $this->fatalError( "Unrecognized output sink type '$val'" );
                    }
                    $type = $this->outputTypes[$val];
                    $sink = new $type( $param );
                    break;
                case "filter":
                    if ( is_null( $sink ) ) {
                        $sink = new DumpOutput();
                    }
                    if ( !isset( $this->filterTypes[$val] ) ) {
                        $this->fatalError( "Unrecognized filter type '$val'" );
                    }
                    $type = $this->filterTypes[$val];
                    $filter = new $type( $sink, $param );

                    // references are lame in php...
                    unset( $sink );
                    $sink = $filter;

                    break;
                case "report":
                    $this->reportingInterval = intval( $val );
                    break;
                case "server":
                    $this->server = $val;
                    break;
                case "force-normal":
                    if ( !function_exists( 'utf8_normalize' ) ) {
                        $this->fatalError( "UTF-8 normalization extension not loaded. " .
                            "Install or remove --force-normal parameter to use slower code." );
                    }
                    break;
                default:
                    $this->processOption( $opt, $val, $param );
                }
            }
        }

        if ( is_null( $sink ) ) {
            $sink = new DumpOutput();
        }
        $sinks[] = $sink;

        if ( count( $sinks ) > 1 ) {
            return new DumpMultiWriter( $sinks );
        } else {
            return $sink;
        }
    }

    function processOption( $opt, $val, $param ) {
        // extension point for subclasses to add options
    }

    function dump( $history, $text = WikiExporter::TEXT ) {
        # Notice messages will foul up your XML output even if they're
        # relatively harmless.
        if ( ini_get( 'display_errors' ) ) {
            ini_set( 'display_errors', 'stderr' );
        }

        $this->initProgress( $history );

        $db = $this->backupDb();
        $exporter = new WikiExporter( $db, $history, WikiExporter::STREAM, $text );
        $exporter->dumpUploads = $this->dumpUploads;
        $exporter->dumpUploadFileContents = $this->dumpUploadFileContents;

        $wrapper = new ExportProgressFilter( $this->sink, $this );
        $exporter->setOutputSink( $wrapper );

        if ( !$this->skipHeader ) {
            $exporter->openStream();
        }
        # Log item dumps: all or by range
        if ( $history & WikiExporter::LOGS ) {
            if ( $this->startId || $this->endId ) {
                $exporter->logsByRange( $this->startId, $this->endId );
            } else {
                $exporter->allLogs();
            }
        # Page dumps: all or by page ID range
        } elseif ( is_null( $this->pages ) ) {
            if ( $this->startId || $this->endId ) {
                $exporter->pagesByRange( $this->startId, $this->endId );
            } elseif ( $this->revStartId || $this->revEndId ) {
                $exporter->revsByRange( $this->revStartId, $this->revEndId );
            } else {
                $exporter->allPages();
            }
        # Dump of specific pages
        } else {
            $exporter->pagesByName( $this->pages );
        }

        if ( !$this->skipFooter ) {
            $exporter->closeStream();
        }

        $this->report( true );
    }

    function initProgress( $history = WikiExporter::FULL ) {
        $table = ( $history == WikiExporter::CURRENT ) ? 'page' : 'revision';
        $field = ( $history == WikiExporter::CURRENT ) ? 'page_id' : 'rev_id';

        $dbr = $this->forcedDb;
        if ( $this->forcedDb === null ) {
            $dbr = wfGetDB( DB_SLAVE );
        }
        $this->maxCount = $dbr->selectField( $table, "MAX($field)", '', __METHOD__ );
        $this->startTime = microtime( true );
        $this->lastTime = $this->startTime;
        $this->ID = getmypid();
    }

    function backupDb() {
        if ( $this->forcedDb !== null ) {
            return $this->forcedDb;
        }

        $this->lb = wfGetLBFactory()->newMainLB();
        $db = $this->lb->getConnection( DB_SLAVE, 'backup' );

        // Discourage the server from disconnecting us if it takes a long time
        // to read out the big ol' batch query.
        $db->setSessionOptions( array( 'connTimeout' => 3600 * 24 ) );

        return $db;
    }

    function setDb( DatabaseBase $db = null ) {
        $this->forcedDb = $db;
    }

    function __destruct() {
        if ( isset( $this->lb ) ) {
            $this->lb->closeAll();
        }
    }

    function backupServer() {
        global $wgDBserver;
        return $this->server
            ? $this->server
            : $wgDBserver;
    }

    function reportPage() {
        $this->pageCount++;
    }

    function revCount() {
        $this->revCount++;
        $this->report();
    }

    function report( $final = false ) {
        if ( $final xor ( $this->revCount % $this->reportingInterval == 0 ) ) {
            $this->showReport();
        }
    }

    function showReport() {
        if ( $this->reporting ) {
            $now = wfTimestamp( TS_DB );
            $nowts = microtime( true );
            $deltaAll = $nowts - $this->startTime;
            $deltaPart = $nowts - $this->lastTime;
            $this->pageCountPart = $this->pageCount - $this->pageCountLast;
            $this->revCountPart = $this->revCount - $this->revCountLast;

            if ( $deltaAll ) {
                $portion = $this->revCount / $this->maxCount;
                $eta = $this->startTime + $deltaAll / $portion;
                $etats = wfTimestamp( TS_DB, intval( $eta ) );
                $pageRate = $this->pageCount / $deltaAll;
                $revRate = $this->revCount / $deltaAll;
            } else {
                $pageRate = '-';
                $revRate = '-';
                $etats = '-';
            }
            if ( $deltaPart ) {
                $pageRatePart = $this->pageCountPart / $deltaPart;
                $revRatePart = $this->revCountPart / $deltaPart;
            } else {
                $pageRatePart = '-';
                $revRatePart = '-';
            }
            $this->progress( sprintf( "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), %d revs (%0.1f|%0.1f/sec all|curr), ETA %s [max %d]",
                    $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate, $revRatePart, $etats, $this->maxCount ) );
            $this->lastTime = $nowts;
            $this->revCountLast = $this->revCount;
        }
    }

    function progress( $string ) {
        fwrite( $this->stderr, $string . "\n" );
    }

    function fatalError( $msg ) {
        $this->progress( "$msg\n" );
        die( 1 );
    }
}

class ExportProgressFilter extends DumpFilter {
    function __construct( &$sink, &$progress ) {
        parent::__construct( $sink );
        $this->progress = $progress;
    }

    function writeClosePage( $string ) {
        parent::writeClosePage( $string );
        $this->progress->reportPage();
    }

    function writeRevision( $rev, $string ) {
        parent::writeRevision( $rev, $string );
        $this->progress->revCount();
    }
}