compareParsers.php

Go to the documentation of this file.

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

class CompareParsers extends DumpIterator {

    private $count = 0;
    private $saveFailed;
    private $stripParametersEnabled;
    private $showParsedOutput;
    private $showDiff;
    private $options;
    private $failed;

    public function __construct() {
        parent::__construct();
        $this->saveFailed = false;
        $this->addDescription( 'Run a file or dump with several parsers' );
        $this->addOption( 'parser1', 'The first parser to compare.', true, true );
        $this->addOption( 'parser2', 'The second parser to compare.', true, true );
        $this->addOption( 'tidy', 'Run tidy on the articles.', false, false );
        $this->addOption(
            'save-failed',
            'Folder in which articles which differ will be stored.',
            false,
            true
        );
        $this->addOption( 'show-diff', 'Show a diff of the two renderings.', false, false );
        $this->addOption(
            'diff-bin',
            'Binary to use for diffing (can also be provided by DIFF env var).',
            false,
            false
        );
        $this->addOption(
            'strip-parameters',
            'Remove parameters of html tags to increase readability.',
            false,
            false
        );
        $this->addOption(
            'show-parsed-output',
            'Show the parsed html if both Parsers give the same output.',
            false,
            false
        );
    }

    public function checkOptions() {
        if ( $this->hasOption( 'save-failed' ) ) {
            $this->saveFailed = $this->getOption( 'save-failed' );
        }

        $this->stripParametersEnabled = $this->hasOption( 'strip-parameters' );
        $this->showParsedOutput = $this->hasOption( 'show-parsed-output' );

        $this->showDiff = $this->hasOption( 'show-diff' );
        if ( $this->showDiff ) {
            $bin = $this->getOption( 'diff-bin', getenv( 'DIFF' ) );
            if ( $bin != '' ) {
                global $wgDiff;
                $wgDiff = $bin;
            }
        }

        $user = new User();
        $this->options = ParserOptions::newFromUser( $user );

        if ( $this->hasOption( 'tidy' ) ) {
            if ( !MWTidy::isEnabled() ) {
                $this->fatalError( 'Tidy was requested but $wgTidyConfig is not set in LocalSettings.php' );
            }
            $this->options->setTidy( true );
        }

        $this->failed = 0;
    }

    public function conclusions() {
        $this->error( "{$this->failed} failed revisions out of {$this->count}" );
        if ( $this->count > 0 ) {
            $this->output( " (" . ( $this->failed / $this->count ) . "%)\n" );
        }
    }

    function stripParameters( $text ) {
        if ( !$this->stripParametersEnabled ) {
            return $text;
        }

        return preg_replace( '/(<a) [^>]+>/', '$1>', $text );
    }

    public function processRevision( $rev ) {
        $title = $rev->getTitle();

        $parser1Name = $this->getOption( 'parser1' );
        $parser2Name = $this->getOption( 'parser2' );

        self::checkParserLocally( $parser1Name );
        self::checkParserLocally( $parser2Name );

        $parser1 = new $parser1Name();
        $parser2 = new $parser2Name();

        $content = $rev->getContent();

        if ( $content->getModel() !== CONTENT_MODEL_WIKITEXT ) {
            $this->error( "Page {$title->getPrefixedText()} does not contain wikitext "
                . "but {$content->getModel()}\n" );

            return;
        }

        '@phan-var WikitextContent $content';
        $text = strval( $content->getText() );

        $output1 = $parser1->parse( $text, $title, $this->options );
        $output2 = $parser2->parse( $text, $title, $this->options );

        if ( $output1->getText() != $output2->getText() ) {
            $this->failed++;
            $this->error( "Parsing for {$title->getPrefixedText()} differs\n" );

            if ( $this->saveFailed ) {
                file_put_contents(
                    $this->saveFailed . '/' . rawurlencode( $title->getPrefixedText() ) . ".txt",
                    $text
                );
            }
            if ( $this->showDiff ) {
                $this->output( wfDiff(
                    $this->stripParameters( $output1->getText() ),
                    $this->stripParameters( $output2->getText() ),
                    ''
                ) );
            }
        } else {
            $this->output( $title->getPrefixedText() . "\tOK\n" );

            if ( $this->showParsedOutput ) {
                $this->output( $this->stripParameters( $output1->getText() ) );
            }
        }
    }

    private static function checkParserLocally( $parserName ) {
        /* Look for the parser in a file appropiately named in the current folder */
        if ( !class_exists( $parserName ) && file_exists( "$parserName.php" ) ) {
            global $wgAutoloadClasses;
            $wgAutoloadClasses[$parserName] = realpath( '.' ) . "/$parserName.php";
        }
    }
}

$maintClass = CompareParsers::class;
require_once RUN_MAINTENANCE_IF_MAIN;