1.33.0/php/renderDump_8php_source.html

<?php

require_once __DIR__ . '/Maintenance.php';


class DumpRenderer extends Maintenance {


    private $count = 0;

    private $outputDirectory, $startTime;


    public function __construct() {

        parent::__construct();

        $this->addDescription(

            'Take page text out of an XML dump file and render basic HTML out to files' );

        $this->addOption( 'output-dir', 'The directory to output the HTML files to', true, true );

        $this->addOption( 'prefix', 'Prefix for the rendered files (defaults to wiki)', false, true );

        $this->addOption( 'parser', 'Use an alternative parser class', false, true );

    }


    public function execute() {

        $this->outputDirectory = $this->getOption( 'output-dir' );

        $this->prefix = $this->getOption( 'prefix', 'wiki' );

        $this->startTime = microtime( true );


        if ( $this->hasOption( 'parser' ) ) {

            global $wgParserConf;

            $wgParserConf['class'] = $this->getOption( 'parser' );

            $this->prefix .= "-{$wgParserConf['class']}";

        }


        $source = new ImportStreamSource( $this->getStdin() );

        $importer = new WikiImporter( $source, $this->getConfig() );


        $importer->setRevisionCallback(

            [ $this, 'handleRevision' ] );

        $importer->setNoticeCallback( function ( $msg, $params ) {

            echo wfMessage( $msg, $params )->text() . "\n";

        } );


        $importer->doImport();


        $delta = microtime( true ) - $this->startTime;

        $this->error( "Rendered {$this->count} pages in " . round( $delta, 2 ) . " seconds " );

        if ( $delta > 0 ) {

            $this->error( round( $this->count / $delta, 2 ) . " pages/sec" );

        }

        $this->error( "\n" );

    }


    public function handleRevision( $rev ) {

        $title = $rev->getTitle();

        if ( !$title ) {

            $this->error( "Got bogus revision with null title!" );


            return;

        }

        $display = $title->getPrefixedText();


        $this->count++;


        $sanitized = rawurlencode( $display );

        $filename = sprintf( "%s/%s-%07d-%s.html",

            $this->outputDirectory,

            $this->prefix,

            $this->count,

            $sanitized );

        $this->output( sprintf( "%s\n", $filename, $display ) );


        $user = new User();

        $options = ParserOptions::newFromUser( $user );


        $content = $rev->getContent();

        $output = $content->getParserOutput( $title, null, $options );


        file_put_contents( $filename,

            "<!DOCTYPE html>\n" .

            "<html lang=\"en\" dir=\"ltr\">\n" .

            "<head>\n" .

            "<meta charset=\"UTF-8\" />\n" .

            "<title>" . htmlspecialchars( $display ) . "</title>\n" .

            "</head>\n" .

            "<body>\n" .

            $output->getText() .

            "</body>\n" .

            "</html>" );

    }

}


$maintClass = DumpRenderer::class;

require_once RUN_MAINTENANCE_IF_MAIN;