BaseDump.php

Go to the documentation of this file.

<?php
use MediaWiki\Revision\SlotRecord;
 
class BaseDump {
    protected $reader = null;
    protected $atEnd = false;
    protected $atPageEnd = false;
    protected $lastPage = 0;
    protected $lastRev = 0;
    protected $infiles = null;
 
    public function __construct( $infile ) {
        $this->infiles = explode( ';', $infile );
        $this->reader = new XMLReader();
        $infile = array_shift( $this->infiles );
        if ( !$this->reader->open( $infile, null, LIBXML_PARSEHUGE ) ) {
            $this->debug( __METHOD__ . ' was unable to open xml' );
            $this->atEnd = true;
        }
    }
 
    public function prefetch( $page, $rev, $slot = SlotRecord::MAIN ) {
        $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 or failed to open/read input file, "
                . "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]" );
 
            if ( $slot !== SlotRecord::MAIN ) {
                $lastSlot = SlotRecord::MAIN;
                while ( $lastSlot !== $slot ) {
                    if ( !$this->skipTo( 'content', 'revision' ) ||
                        !$this->skipTo( 'role', 'revision' )
                    ) {
                        return null;
                    }
                    $lastSlot = $this->nodeContents();
                }
            }
 
            return $this->nextText();
        } else {
            $this->debug( "BaseDump::prefetch already past rev $rev on page $page "
                . "[$this->lastPage, $this->lastRev]" );
 
            return null;
        }
    }
 
    protected function debug( $str ) {
        wfDebug( $str );
        // global $dumper;
        // $dumper->progress( $str );
    }
 
    private 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( $this->infiles );
                if ( !$this->reader->open( $infile, null, LIBXML_PARSEHUGE ) ) {
                    $this->debug( __METHOD__ . ' was unable to open xml' );
                    $this->atEnd = true;
                } else {
                    $this->atEnd = false;
                }
            }
        }
    }
 
    private function nextRev() {
        if ( $this->skipTo( 'revision' ) ) {
            if ( $this->skipTo( 'id' ) ) {
                $this->lastRev = intval( $this->nodeContents() );
            }
        } else {
            $this->atPageEnd = true;
        }
    }
 
    private function nextText() {
        if ( !$this->skipTo( 'text', 'revision' ) ) {
            return null;
        }
 
        return strval( $this->nodeContents() );
    }
 
    private 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();
    }
 
    private 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();
    }
 
    public function close() {
        $this->reader->close();
        $this->atEnd = true;
 
        return null;
    }
}