backupPrefetch.inc

Go to the documentation of this file.

<?php
class BaseDump {
        var $reader = null;
        var $atEnd = false;
        var $atPageEnd = false;
        var $lastPage = 0;
        var $lastRev = 0;
        var $infiles = null;

        function BaseDump( $infile ) {
                $this->infiles = explode(';',$infile);
                $this->reader = new XMLReader();
                $infile = array_shift($this->infiles);
                if (defined( 'LIBXML_PARSEHUGE' ) ) {
                        $this->reader->open( $infile, null, LIBXML_PARSEHUGE );
                }
                else {
                        $this->reader->open( $infile );
                }
        }

        function prefetch( $page, $rev ) {
                $page = intval( $page );
                $rev = intval( $rev );
                while ( $this->lastPage < $page && !$this->atEnd ) {
                        $this->debug( "BaseDump::prefetch at page $this->lastPage, looking for $page" );
                        $this->nextPage();
                }
                if ( $this->lastPage > $page || $this->atEnd ) {
                        $this->debug( "BaseDump::prefetch already past page $page looking for rev $rev  [$this->lastPage, $this->lastRev]" );
                        return null;
                }
                while ( $this->lastRev < $rev && !$this->atEnd && !$this->atPageEnd ) {
                        $this->debug( "BaseDump::prefetch at page $this->lastPage, rev $this->lastRev, looking for $page, $rev" );
                        $this->nextRev();
                }
                if ( $this->lastRev == $rev && !$this->atEnd ) {
                        $this->debug( "BaseDump::prefetch hit on $page, $rev [$this->lastPage, $this->lastRev]" );
                        return $this->nextText();
                } else {
                        $this->debug( "BaseDump::prefetch already past rev $rev on page $page  [$this->lastPage, $this->lastRev]" );
                        return null;
                }
        }

        function debug( $str ) {
                wfDebug( $str . "\n" );
                // global $dumper;
                // $dumper->progress( $str );
        }

        function nextPage() {
                if ( $this->skipTo( 'page', 'mediawiki' ) ) {
                        if ( $this->skipTo( 'id' ) ) {
                                $this->lastPage = intval( $this->nodeContents() );
                                $this->lastRev = 0;
                                $this->atPageEnd = false;
                        }
                } else {
                        $this->close();
                        if (count($this->infiles)) {
                                $infile = array_shift($this->infiles);
                                $this->reader->open( $infile );
                                $this->atEnd = false;
                        }
                }
        }

        function nextRev() {
                if ( $this->skipTo( 'revision' ) ) {
                        if ( $this->skipTo( 'id' ) ) {
                                $this->lastRev = intval( $this->nodeContents() );
                        }
                } else {
                        $this->atPageEnd = true;
                }
        }

        function nextText() {
                $this->skipTo( 'text' );
                return strval( $this->nodeContents() );
        }

        function skipTo( $name, $parent = 'page' ) {
                if ( $this->atEnd ) {
                        return false;
                }
                while ( $this->reader->read() ) {
                        if ( $this->reader->nodeType == XMLReader::ELEMENT &&
                                $this->reader->name == $name ) {
                                return true;
                        }
                        if ( $this->reader->nodeType == XMLReader::END_ELEMENT &&
                                $this->reader->name == $parent ) {
                                $this->debug( "BaseDump::skipTo found </$parent> searching for <$name>" );
                                return false;
                        }
                }
                return $this->close();
        }

        function nodeContents() {
                if ( $this->atEnd ) {
                        return null;
                }
                if ( $this->reader->isEmptyElement ) {
                        return "";
                }
                $buffer = "";
                while ( $this->reader->read() ) {
                        switch( $this->reader->nodeType ) {
                        case XMLReader::TEXT:
//                      case XMLReader::WHITESPACE:
                        case XMLReader::SIGNIFICANT_WHITESPACE:
                                $buffer .= $this->reader->value;
                                break;
                        case XMLReader::END_ELEMENT:
                                return $buffer;
                        }
                }
                return $this->close();
        }

        function close() {
                $this->reader->close();
                $this->atEnd = true;
                return null;
        }
}