SyntaxHighlight.class.php

Go to the documentation of this file.

<?php
use MediaWiki\Shell\Shell;
use Symfony\Component\Process\ProcessBuilder;
 
// @codingStandardsIgnoreStart
class SyntaxHighlight {
// @codingStandardsIgnoreEnd
 
    const HIGHLIGHT_MAX_LINES = 1000;
 
    const HIGHLIGHT_MAX_BYTES = 102400;
 
    const HIGHLIGHT_CSS_CLASS = 'mw-highlight';
 
    const CACHE_VERSION = 2;
 
    private static $mimeLexers = [
        'text/javascript'  => 'javascript',
        'application/json' => 'javascript',
        'text/xml'         => 'xml',
    ];
 
    private static function getLexer( $lang ) {
        static $lexers = null;
 
        if ( $lang === null ) {
            return null;
        }
 
        if ( !$lexers ) {
            $lexers = require __DIR__ . '/SyntaxHighlight.lexers.php';
        }
 
        $lexer = strtolower( $lang );
 
        if ( in_array( $lexer, $lexers ) ) {
            return $lexer;
        }
 
        $geshi2pygments = SyntaxHighlightGeSHiCompat::getGeSHiToPygmentsMap();
 
        // Check if this is a GeSHi lexer name for which there exists
        // a compatible Pygments lexer with a different name.
        if ( isset( $geshi2pygments[$lexer] ) ) {
            $lexer = $geshi2pygments[$lexer];
            if ( in_array( $lexer, $lexers ) ) {
                return $lexer;
            }
        }
 
        return null;
    }
 
    public static function onParserFirstCallInit( Parser &$parser ) {
        foreach ( [ 'source', 'syntaxhighlight' ] as $tag ) {
            $parser->setHook( $tag, [ 'SyntaxHighlight_GeSHi', 'parserHook' ] );
        }
    }
 
    public static function parserHook( $text, $args = [], $parser ) {
        global $wgUseTidy;
 
        // Replace strip markers (For e.g. {{#tag:syntaxhighlight|<nowiki>...}})
        $out = $parser->mStripState->unstripNoWiki( $text );
 
        // Don't trim leading spaces away, just the linefeeds
        $out = preg_replace( '/^\n+/', '', rtrim( $out ) );
 
        // Convert deprecated attributes
        if ( isset( $args['enclose'] ) ) {
            if ( $args['enclose'] === 'none' ) {
                $args['inline'] = true;
            }
            unset( $args['enclose'] );
        }
 
        $lexer = isset( $args['lang'] ) ? $args['lang'] : '';
 
        $result = self::highlight( $out, $lexer, $args );
        if ( !$result->isGood() ) {
            $parser->addTrackingCategory( 'syntaxhighlight-error-category' );
        }
        $out = $result->getValue();
 
        // HTML Tidy will convert tabs to spaces incorrectly (bug 30930).
        // But the conversion from tab to space occurs while reading the input,
        // before the conversion from &#9; to tab, so we can armor it that way.
        if ( $wgUseTidy ) {
            $out = str_replace( "\t", '&#9;', $out );
        }
 
        // Allow certain HTML attributes
        $htmlAttribs = Sanitizer::validateAttributes( $args, [ 'style', 'class', 'id', 'dir' ] );
        if ( !isset( $htmlAttribs['class'] ) ) {
            $htmlAttribs['class'] = self::HIGHLIGHT_CSS_CLASS;
        } else {
            $htmlAttribs['class'] .= ' ' . self::HIGHLIGHT_CSS_CLASS;
        }
        if ( !( isset( $htmlAttribs['dir'] ) && $htmlAttribs['dir'] === 'rtl' ) ) {
            $htmlAttribs['dir'] = 'ltr';
        }
 
        if ( isset( $args['inline'] ) ) {
            // Enforce inlineness. Stray newlines may result in unexpected list and paragraph processing
            // (also known as doBlockLevels()).
            $out = str_replace( "\n", ' ', $out );
            $out = Html::rawElement( 'code', $htmlAttribs, $out );
 
        } else {
            // Not entirely sure what benefit this provides, but it was here already
            $htmlAttribs['class'] .= ' ' . 'mw-content-' . $htmlAttribs['dir'];
 
            // Unwrap Pygments output to provide our own wrapper. We can't just always use the 'nowrap'
            // option (pass 'inline'), since it disables other useful things like line highlighting.
            // Tolerate absence of quotes for Html::element() and wgWellFormedXml=false.
            $m = [];
            if ( preg_match( '/^<div class="?mw-highlight"?>(.*)<\/div>$/s', trim( $out ), $m ) ) {
                $out = trim( $m[1] );
            } else {
                throw new MWException( 'Unexpected output from Pygments encountered' );
            }
 
            // Use 'nowiki' strip marker to prevent list processing (also known as doBlockLevels()).
            // However, leave the wrapping <div/> outside to prevent <p/>-wrapping.
            $marker = $parser::MARKER_PREFIX . '-syntaxhighlightinner-' .
                sprintf( '%08X', $parser->mMarkerIndex++ ) . $parser::MARKER_SUFFIX;
            $parser->mStripState->addNoWiki( $marker, $out );
 
            $out = Html::openElement( 'div', $htmlAttribs ) .
                $marker .
                Html::closeElement( 'div' );
        }
 
        // Register CSS
        $parser->getOutput()->addModuleStyles( 'ext.pygments' );
 
        return $out;
    }
 
    public static function getPygmentizePath() {
        global $wgPygmentizePath;
 
        // If $wgPygmentizePath is unset, use the bundled copy.
        if ( $wgPygmentizePath === false ) {
            $wgPygmentizePath = __DIR__ . '/pygments/pygmentize';
        }
 
        return $wgPygmentizePath;
    }
 
    public static function highlight( $code, $lang = null, $args = [] ) {
        $status = new Status;
 
        $lexer = self::getLexer( $lang );
        if ( $lexer === null && $lang !== null ) {
            $status->warning( 'syntaxhighlight-error-unknown-language', $lang );
        }
 
        $length = strlen( $code );
        if ( strlen( $code ) > self::HIGHLIGHT_MAX_BYTES ) {
            $status->warning( 'syntaxhighlight-error-exceeds-size-limit',
                $length, self::HIGHLIGHT_MAX_BYTES );
            $lexer = null;
        }
 
        if ( Shell::isDisabled() ) {
            $status->warning( 'syntaxhighlight-error-pygments-invocation-failure' );
            wfWarn(
                'MediaWiki determined that it cannot invoke Pygments. ' .
                'As a result, SyntaxHighlight_GeSHi will not perform any syntax highlighting. ' .
                'See the debug log for details: ' .
                'https://www.mediawiki.org/wiki/Manual:$wgDebugLogFile'
            );
            $lexer = null;
        }
 
        $inline = isset( $args['inline'] );
 
        if ( $lexer === null ) {
            if ( $inline ) {
                $status->value = htmlspecialchars( trim( $code ), ENT_NOQUOTES );
            } else {
                $pre = Html::element( 'pre', [], $code );
                $status->value = Html::rawElement(
                    'div',
                    [ 'class' => self::HIGHLIGHT_CSS_CLASS ],
                    $pre
                );
            }
            return $status;
        }
 
        $options = [
            'cssclass' => self::HIGHLIGHT_CSS_CLASS,
            'encoding' => 'utf-8',
        ];
 
        // Line numbers
        if ( isset( $args['line'] ) ) {
            $options['linenos'] = 'inline';
        }
 
        if ( $lexer === 'php' && strpos( $code, '<?php' ) === false ) {
            $options['startinline'] = 1;
        }
 
        // Highlight specific lines
        if ( isset( $args['highlight'] ) ) {
            $lines = self::parseHighlightLines( $args['highlight'] );
            if ( count( $lines ) ) {
                $options['hl_lines'] = implode( ' ', $lines );
            }
        }
 
        // Starting line number
        if ( isset( $args['start'] ) && ctype_digit( $args['start'] ) ) {
            $options['linenostart'] = (int)$args['start'];
        }
 
        if ( $inline ) {
            $options['nowrap'] = 1;
        }
 
        $cache = ObjectCache::getMainWANInstance();
        $cacheKey = self::makeCacheKey( $code, $lexer, $options );
        $output = $cache->get( $cacheKey );
 
        if ( $output === false ) {
            $optionPairs = [];
            foreach ( $options as $k => $v ) {
                $optionPairs[] = "{$k}={$v}";
            }
            $builder = new ProcessBuilder();
            $builder->setPrefix( self::getPygmentizePath() );
            $process = $builder
                ->add( '-l' )->add( $lexer )
                ->add( '-f' )->add( 'html' )
                ->add( '-O' )->add( implode( ',', $optionPairs ) )
                ->getProcess();
 
            $process->setInput( $code );
 
            /* Workaround for T151523 (buggy $process->getOutput()).
                If/when this issue is fixed in HHVM or Symfony,
                replace this with "$process->run(); $output = $process->getOutput();"
            */
            $output = '';
            $process->run( function ( $type, $capturedOutput ) use ( &$output ) {
                $output .= $capturedOutput;
            } );
 
            if ( !$process->isSuccessful() ) {
                $status->warning( 'syntaxhighlight-error-pygments-invocation-failure' );
                wfWarn( 'Failed to invoke Pygments: ' . $process->getErrorOutput() );
                $status->value = self::highlight( $code, null, $args )->getValue();
                return $status;
            }
 
            $cache->set( $cacheKey, $output );
        }
 
        if ( $inline ) {
            $output = trim( $output );
        }
 
        $status->value = $output;
        return $status;
    }
 
    private static function makeCacheKey( $code, $lexer, $options ) {
        $optionString = FormatJson::encode( $options, false, FormatJson::ALL_OK );
        $hash = md5( "{$code}|{$lexer}|{$optionString}|" . self::CACHE_VERSION );
        if ( function_exists( 'wfGlobalCacheKey' ) ) {
            return wfGlobalCacheKey( 'highlight', $hash );
        } else {
            return 'highlight:' . $hash;
        }
    }
 
    protected static function parseHighlightLines( $lineSpec ) {
        $lines = [];
        $values = array_map( 'trim', explode( ',', $lineSpec ) );
        foreach ( $values as $value ) {
            if ( ctype_digit( $value ) ) {
                $lines[] = (int)$value;
            } elseif ( strpos( $value, '-' ) !== false ) {
                list( $start, $end ) = array_map( 'trim', explode( '-', $value ) );
                if ( self::validHighlightRange( $start, $end ) ) {
                    for ( $i = intval( $start ); $i <= $end; $i++ ) {
                        $lines[] = $i;
                    }
                }
            }
            if ( count( $lines ) > self::HIGHLIGHT_MAX_LINES ) {
                $lines = array_slice( $lines, 0, self::HIGHLIGHT_MAX_LINES );
                break;
            }
        }
        return $lines;
    }
 
    protected static function validHighlightRange( $start, $end ) {
        // Since we're taking this tiny range and producing a an
        // array of every integer between them, it would be trivial
        // to DoS the system by asking for a huge range.
        // Impose an arbitrary limit on the number of lines in a
        // given range to reduce the impact.
        return
            ctype_digit( $start ) &&
            ctype_digit( $end ) &&
            $start > 0 &&
            $start < $end &&
            $end - $start < self::HIGHLIGHT_MAX_LINES;
    }
 
    public static function onContentGetParserOutput( Content $content, Title $title,
        $revId, ParserOptions $options, $generateHtml, ParserOutput &$output
    ) {
        global $wgParser, $wgTextModelsToParse;
 
        if ( !$generateHtml ) {
            // Nothing special for us to do, let MediaWiki handle this.
            return true;
        }
 
        // Determine the language
        $extension = ExtensionRegistry::getInstance();
        $models = $extension->getAttribute( 'SyntaxHighlightModels' );
        $model = $content->getModel();
        if ( !isset( $models[$model] ) ) {
            // We don't care about this model, carry on.
            return true;
        }
        $lexer = $models[$model];
 
        // Hope that $wgSyntaxHighlightModels does not contain silly types.
        $text = ContentHandler::getContentText( $content );
        if ( !$text ) {
            // Oops! Non-text content? Let MediaWiki handle this.
            return true;
        }
 
        // Parse using the standard parser to get links etc. into the database, HTML is replaced below.
        // We could do this using $content->fillParserOutput(), but alas it is 'protected'.
        if ( $content instanceof TextContent && in_array( $model, $wgTextModelsToParse ) ) {
            $output = $wgParser->parse( $text, $title, $options, true, true, $revId );
        }
 
        $status = self::highlight( $text, $lexer );
        if ( !$status->isOK() ) {
            return true;
        }
        $out = $status->getValue();
 
        $output->addModuleStyles( 'ext.pygments' );
        $output->setText( '<div dir="ltr">' . $out . '</div>' );
 
        // Inform MediaWiki that we have parsed this page and it shouldn't mess with it.
        return false;
    }
 
    public static function onApiFormatHighlight( IContextSource $context, $text, $mime, $format ) {
        if ( !isset( self::$mimeLexers[$mime] ) ) {
            return true;
        }
 
        $lexer = self::$mimeLexers[$mime];
        $status = self::highlight( $text, $lexer );
        if ( !$status->isOK() ) {
            return true;
        }
 
        $out = $status->getValue();
        if ( preg_match( '/^<pre([^>]*)>/i', $out, $m ) ) {
            $attrs = Sanitizer::decodeTagAttributes( $m[1] );
            $attrs['class'] .= ' api-pretty-content';
            $encodedAttrs = Sanitizer::safeEncodeTagAttributes( $attrs );
            $out = '<pre' . $encodedAttrs. '>' .  substr( $out, strlen( $m[0] ) );
        }
        $output = $context->getOutput();
        $output->addModuleStyles( 'ext.pygments' );
        $output->addHTML( '<div dir="ltr">' . $out . '</div>' );
 
        // Inform MediaWiki that we have parsed this page and it shouldn't mess with it.
        return false;
    }
 
    public static function onResourceLoaderRegisterModules( &$resourceLoader ) {
        if ( ! ExtensionRegistry::getInstance()->isLoaded( 'VisualEditor' ) ) {
            return;
        }
 
        $resourceLoader->register( 'ext.geshi.visualEditor', [
            'class' => 'ResourceLoaderSyntaxHighlightVisualEditorModule',
            'localBasePath' => __DIR__ . DIRECTORY_SEPARATOR . 'modules',
            'remoteExtPath' => 'SyntaxHighlight_GeSHi/modules',
            'scripts' => [
                've-syntaxhighlight/ve.dm.MWSyntaxHighlightNode.js',
                've-syntaxhighlight/ve.ce.MWSyntaxHighlightNode.js',
                've-syntaxhighlight/ve.ui.MWSyntaxHighlightWindow.js',
                've-syntaxhighlight/ve.ui.MWSyntaxHighlightDialog.js',
                've-syntaxhighlight/ve.ui.MWSyntaxHighlightDialogTool.js',
                've-syntaxhighlight/ve.ui.MWSyntaxHighlightInspector.js',
                've-syntaxhighlight/ve.ui.MWSyntaxHighlightInspectorTool.js',
            ],
            'styles' => [
                've-syntaxhighlight/ve.ce.MWSyntaxHighlightNode.css',
                've-syntaxhighlight/ve.ui.MWSyntaxHighlightDialog.css',
                've-syntaxhighlight/ve.ui.MWSyntaxHighlightInspector.css',
            ],
            'dependencies' => [
                'ext.visualEditor.mwcore',
            ],
            'messages' => [
                'syntaxhighlight-visualeditor-mwsyntaxhighlightinspector-code',
                'syntaxhighlight-visualeditor-mwsyntaxhighlightinspector-language',
                'syntaxhighlight-visualeditor-mwsyntaxhighlightinspector-none',
                'syntaxhighlight-visualeditor-mwsyntaxhighlightinspector-showlines',
                'syntaxhighlight-visualeditor-mwsyntaxhighlightinspector-startingline',
                'syntaxhighlight-visualeditor-mwsyntaxhighlightinspector-title',
            ],
            'targets' => [ 'desktop', 'mobile' ],
        ] );
    }
 
    public static function prepare( $text, $lang ) {
        wfDeprecated( __METHOD__ );
        return new GeSHi( self::highlight( $text, $lang )->getValue() );
    }
 
    public static function buildHeadItem( $geshi ) {
        wfDeprecated( __METHOD__ );
        $geshi->parse_code();
        return '';
    }
}
class_alias( SyntaxHighlight::class, 'SyntaxHighlight_GeSHi' );