1 files changed, 176 insertions, 35 deletions

diff --git a/extensions/SyntaxHighlight_GeSHi/SyntaxHighlight_GeSHi.class.php b/extensions/SyntaxHighlight_GeSHi/SyntaxHighlight_GeSHi.class.php
index 186b99ac..7318574d 100644
--- a/extensions/SyntaxHighlight_GeSHi/SyntaxHighlight_GeSHi.class.php
+++ b/extensions/SyntaxHighlight_GeSHi/SyntaxHighlight_GeSHi.class.php
@@ -1,7 +1,6 @@
 <?php
 
 class SyntaxHighlight_GeSHi {
-
 	/**
 	 * Has GeSHi been initialised this session?
 	 */
@@ -9,6 +8,7 @@ class SyntaxHighlight_GeSHi {
 
 	/**
 	 * List of languages available to GeSHi
+	 * @var array
 	 */
 	private static $languages = null;
 
@@ -21,12 +21,19 @@ class SyntaxHighlight_GeSHi {
 	 * @return string
 	 */
 	public static function parserHook( $text, $args = array(), $parser ) {
-		global $wgSyntaxHighlightDefaultLang;
+		global $wgSyntaxHighlightDefaultLang, $wgUseSiteCss, $wgUseTidy;
 		wfProfileIn( __METHOD__ );
 		self::initialise();
 		$text = rtrim( $text );
 		// Don't trim leading spaces away, just the linefeeds
 		$text = preg_replace( '/^\n+/', '', $text );
+
+		if( $wgUseTidy ) {
+			// HTML Tidy will convert tabs to spaces incorrectly (bug 30930).
+			// Preemptively replace the spaces in a more controlled fashion.
+			$text = self::tabsToSpaces( $text );
+		}
+
 		// Validate language
 		if( isset( $args['lang'] ) && $args['lang'] ) {
 			$lang = $args['lang'];
@@ -62,7 +69,7 @@ class SyntaxHighlight_GeSHi {
 		// Highlighting specific lines
 		if( isset( $args['highlight'] ) ) {
 			$lines = self::parseHighlightLines( $args['highlight'] );
-			if ( count($lines) ) {
+			if ( count( $lines ) ) {
 				$geshi->highlight_lines_extra( $lines );
 			}
 		}
@@ -91,10 +98,15 @@ class SyntaxHighlight_GeSHi {
 			return $error;
 		}
 		// Armour for Parser::doBlockLevels()
-		if( $enclose === GESHI_HEADER_DIV )
+		if( $enclose === GESHI_HEADER_DIV ) {
 			$out = str_replace( "\n", '', $out );
+		}
 		// Register CSS
-		$parser->mOutput->addHeadItem( self::buildHeadItem( $geshi ), "source-{$lang}" );
+		$parser->getOutput()->addHeadItem( self::buildHeadItem( $geshi ), "source-{$lang}" );
+
+		if ( $wgUseSiteCss ) {
+			$parser->getOutput()->addModuleStyles( 'ext.geshi.local' );
+		}
 
 		$encloseTag = $enclose === GESHI_HEADER_NONE ? 'span' : 'div';
 		$attribs = Sanitizer::validateTagAttributes( $args, $encloseTag );
@@ -105,19 +117,22 @@ class SyntaxHighlight_GeSHi {
 		if ( $enclose === GESHI_HEADER_NONE ) {
 			$attribs = self::addAttribute( $attribs, 'class', 'mw-geshi ' . $lang . ' source-' . $lang );
 		} else {
-			if ( !isset( $attribs['dir'] ) ) {
-				$attribs = self::addAttribute( $attribs, 'dir', 'ltr' );
-			}
-
-			$attribs = self::addAttribute( $attribs, 'class', 'mw-geshi' );
-			$attribs = self::addAttribute( $attribs, 'style', 'text-align: left;' );
+			// Default dir="ltr" (but allow dir="rtl", although unsure if needed)
+			$attribs['dir'] = isset( $attribs['dir'] ) && $attribs['dir'] === 'rtl' ? 'rtl' : 'ltr';
+			$attribs = self::addAttribute( $attribs, 'class', 'mw-geshi mw-code mw-content-' . $attribs['dir'] );
 		}
-		$out = Xml::tags( $encloseTag, $attribs, $out );
+		$out = Html::rawElement( $encloseTag, $attribs, $out );
 
 		wfProfileOut( __METHOD__ );
 		return $out;
 	}
 
+	/**
+	 * @param $attribs array
+	 * @param $name string
+	 * @param $value string
+	 * @return array
+	 */
 	private static function addAttribute( $attribs, $name, $value ) {
 		if( isset( $attribs[$name] ) ) {
 			$attribs[$name] = $value . ' ' . $attribs[$name];
@@ -133,7 +148,7 @@ class SyntaxHighlight_GeSHi {
 	 *
 	 * Input is comma-separated list of lines or line ranges.
 	 *
-	 * @input string
+	 * @param $arg string
 	 * @return array of ints
 	 */
 	protected static function parseHighlightLines( $arg ) {
@@ -160,6 +175,9 @@ class SyntaxHighlight_GeSHi {
 
 	/**
 	 * Validate a provided input range
+	 * @param $start
+	 * @param $end
+	 * @return bool
 	 */
 	protected static function validHighlightRange( $start, $end ) {
 		// Since we're taking this tiny range and producing a an
@@ -176,6 +194,10 @@ class SyntaxHighlight_GeSHi {
 			$end - $start < $arbitrarilyLargeConstant;
 	}
 
+	/**
+	 * @param $args array
+	 * @return int
+	 */
 	static function getEncloseType( $args ) {
 		// Since version 1.0.8 geshi can produce valid pre, but we need to check for it
 		if ( defined('GESHI_HEADER_PRE_VALID') ) {
@@ -205,7 +227,9 @@ class SyntaxHighlight_GeSHi {
 
 	/**
 	 * Hook into Article::view() to provide syntax highlighting for
-	 * custom CSS and JavaScript pages
+	 * custom CSS and JavaScript pages.
+	 *
+	 * B/C for MW 1.20 and before. 1.21 and later use renderHook() instead.
 	 *
 	 * @param string $text
 	 * @param Title $title
@@ -213,10 +237,11 @@ class SyntaxHighlight_GeSHi {
 	 * @return bool
 	 */
 	public static function viewHook( $text, $title, $output ) {
+		global $wgUseSiteCss;
 		// Determine the language
 		$matches = array();
 		preg_match( '!\.(css|js)$!u', $title->getText(), $matches );
-		$lang = $matches[1] == 'css' ? 'css' : 'javascript';
+		$lang = isset( $matches[1] ) && $matches[1] == 'css' ? 'css' : 'javascript';
 		// Attempt to format
 		$geshi = self::prepare( $text, $lang );
 		if( $geshi instanceof GeSHi ) {
@@ -225,9 +250,68 @@ class SyntaxHighlight_GeSHi {
 				// Done
 				$output->addHeadItem( "source-$lang", self::buildHeadItem( $geshi ) );
 				$output->addHTML( "<div dir=\"ltr\">{$out}</div>" );
+				if( $wgUseSiteCss ) {
+					$output->addModuleStyles( 'ext.geshi.local' );
+				}
+				return false;
+			}
+		}
+		// Bottle out
+		return true;
+	}
+
+	/**
+	 * Hook into Content::getParserOutput to provide syntax highlighting for
+	 * script content.
+	 *
+	 * @return bool
+	 * @since MW 1.21
+	 */
+	public static function renderHook( Content $content, Title $title,
+			ParserOptions $options, $generateHtml, ParserOutput &$output
+	) {
+
+		global $wgSyntaxHighlightModels, $wgUseSiteCss;
+
+		// Determine the language
+		$model = $content->getModel();
+		if ( !isset( $wgSyntaxHighlightModels[$model] ) ) {
+			// We don't care about this model, carry on.
+			return true;
+		}
+
+		if ( !$generateHtml ) {
+			// Nothing to do.
+			return false;
+		}
+
+		// Hope that $wgSyntaxHighlightModels does not contain silly types.
+		$text = Contenthandler::getContentText( $content );
+
+		if ( $text === null || $text === false ) {
+			// Oops! Non-text content?
+			return false;
+		}
+
+		$lang = $wgSyntaxHighlightModels[$model];
+
+		// Attempt to format
+		$geshi = self::prepare( $text, $lang );
+		if( $geshi instanceof GeSHi ) {
+
+			$out = $geshi->parse_code();
+			if( !$geshi->error() ) {
+				// Done
+				$output->addHeadItem( self::buildHeadItem( $geshi ), "source-$lang" );
+				$output->setText( "<div dir=\"ltr\">{$out}</div>" );
+
+				if( $wgUseSiteCss ) {
+					$output->addModuleStyles( 'ext.geshi.local' );
+				}
 				return false;
 			}
 		}
+
 		// Bottle out
 		return true;
 	}
@@ -236,11 +320,16 @@ class SyntaxHighlight_GeSHi {
 	 * Initialise a GeSHi object to format some code, performing
 	 * common setup for all our uses of it
 	 *
+	 * @note Used only until MW 1.20
+	 *
 	 * @param string $text
 	 * @param string $lang
 	 * @return GeSHi
 	 */
-	private static function prepare( $text, $lang ) {
+	public static function prepare( $text, $lang ) {
+
+		global $wgSyntaxHighlightKeywordLinks;
+
 		self::initialise();
 		$geshi = new GeSHi( $text, $lang );
 		if( $geshi->error() == GESHI_ERROR_NO_SUCH_LANG ) {
@@ -249,7 +338,18 @@ class SyntaxHighlight_GeSHi {
 		$geshi->set_encoding( 'UTF-8' );
 		$geshi->enable_classes();
 		$geshi->set_overall_class( "source-$lang" );
-		$geshi->enable_keyword_links( false );
+		$geshi->enable_keyword_links( $wgSyntaxHighlightKeywordLinks );
+
+		// If the source code is over 100 kB, disable higlighting of symbols.
+		// If over 200 kB, disable highlighting of strings too.
+		$bytes = strlen( $text );
+		if ( $bytes > 102400 ) {
+			$geshi->set_symbols_highlighting( false );
+			if ( $bytes > 204800 ) {
+				$geshi->set_strings_highlighting( false );
+			}
+		}
+
 		return $geshi;
 	}
 
@@ -260,8 +360,20 @@ class SyntaxHighlight_GeSHi {
 	 * @param GeSHi $geshi
 	 * @return string
 	 */
-	private static function buildHeadItem( $geshi ) {
-		global $wgUseSiteCss, $wgSquidMaxage;
+	public static function buildHeadItem( $geshi ) {
+		/**
+		 * Geshi comes by default with a font-family set to monospace which
+		 * ends ultimately ends up causing the font-size to be smaller than
+		 * one would expect (causing bug 26204).
+		 * We append to the default geshi style a CSS hack which is to specify
+		 * monospace twice which "reset" the browser font-size specified for monospace.
+		 *
+		 * The hack is documented in MediaWiki core under
+		 * docs/uidesign/monospace.html and in bug 33496.
+		 */
+		$geshi->set_code_style( 'font-family: monospace, monospace;',
+			/** preserve defaults */ true );
+
 		$lang = $geshi->language;
 		$css = array();
 		$css[] = '<style type="text/css">/*<![CDATA[*/';
@@ -272,14 +384,6 @@ class SyntaxHighlight_GeSHi {
 		$css[] = $geshi->get_stylesheet( false );
 		$css[] = '/*]]>*/';
 		$css[] = '</style>';
-		if( $wgUseSiteCss ) {
-			$title = Title::makeTitle( NS_MEDIAWIKI, 'Geshi.css' );
-			$q = "usemsgcache=yes&action=raw&ctype=text/css&smaxage={$wgSquidMaxage}";
-			$css[] = '<style type="text/css">/*<![CDATA[*/';
-			$css[] = '@import "' . $title->getLocalUrl( $q ) . '";';
-			$css[] = '/*]]>*/';
-			$css[] = '</style>';
-		}
 		return implode( "\n", $css );
 	}
 
@@ -291,7 +395,8 @@ class SyntaxHighlight_GeSHi {
 	 * @return string HTML fragment
 	 */
 	private static function formatLanguageError( $text ) {
-		$error = self::formatError( htmlspecialchars( wfMsgForContent( 'syntaxhighlight-err-language' ) ), $text );
+		$msg = wfMessage( 'syntaxhighlight-err-language' )->inContentLanguage()->escaped();
+		$error = self::formatError( $msg, $text );
 		return $error . '<pre>' . htmlspecialchars( $text ) . '</pre>';
 	}
 
@@ -306,10 +411,10 @@ class SyntaxHighlight_GeSHi {
 		if( $error ) {
 			$html .= "<p>{$error}</p>";
 		}
-		$html .= '<p>' . htmlspecialchars( wfMsgForContent( 'syntaxhighlight-specify' ) )
+		$html .= '<p>' . wfMessage( 'syntaxhighlight-specify')->inContentLanguage()->escaped()
 			. ' <samp>&lt;source lang=&quot;html4strict&quot;&gt;...&lt;/source&gt;</samp></p>'
-			. '<p>' . htmlspecialchars( wfMsgForContent( 'syntaxhighlight-supported' ) ) . '</p>'
-			. self::formatLanguages();
+			. '<p>' . wfMessage( 'syntaxhighlight-supported' )->inContentLanguage()->escaped()
+			. '</p>' . self::formatLanguages();
 		return "<div style=\"border: solid red 1px; padding: .5em;\">{$html}</div>";
 	}
 
@@ -327,7 +432,7 @@ class SyntaxHighlight_GeSHi {
 			}
 			return '<p class="mw-collapsible mw-collapsed" style="padding: 0em 1em;">' . implode( ', ', $list ) . '</p><br style="clear: all"/>';
 		} else {
-			return '<p>' . htmlspecialchars( wfMsgForContent( 'syntaxhighlight-err-loading' ) ) . '</p>';
+			return '<p>' . wfMessage( 'syntaxhighlight-err-loading' )->inContentLanguage()->escaped() . '</p>';
 		}
 	}
 
@@ -350,11 +455,12 @@ class SyntaxHighlight_GeSHi {
 
 	/**
 	 * Initialise messages and ensure the GeSHi class is loaded
+	 * @return bool
 	 */
 	private static function initialise() {
 		if( !self::$initialised ) {
 			if( !class_exists( 'GeSHi' ) ) {
-				require( 'geshi/geshi.php' );
+				require( dirname( __FILE__ ) . '/geshi/geshi.php' );
 			}
 			self::$initialised = true;
 		}
@@ -363,6 +469,8 @@ class SyntaxHighlight_GeSHi {
 
 	/**
 	 * Get the GeSHI's version information while Special:Version is read.
+	 * @param $extensionTypes
+	 * @return bool
 	 */
 	public static function hSpecialVersion_GeSHi( &$extensionTypes ) {
 		global $wgExtensionCredits;
@@ -373,9 +481,42 @@ class SyntaxHighlight_GeSHi {
 
 	/**
 	 * @see SyntaxHighlight_GeSHi::hSpecialVersion_GeSHi
+	 * @param $sp
+	 * @param $extensionTypes
+	 * @return bool
 	 */
 	public static function hOldSpecialVersion_GeSHi( &$sp, &$extensionTypes ) {
 		return self::hSpecialVersion_GeSHi( $extensionTypes );
 	}
-	
-}
\ No newline at end of file
+
+	/**
+	 * Convert tabs to spaces
+	 *
+	 * @param string $text
+	 * @return string
+	 */
+	private static function tabsToSpaces( $text ) {
+		$lines = explode( "\n", $text );
+		$lines = array_map( array( __CLASS__, 'tabsToSpacesLine' ), $lines );
+		return implode( "\n", $lines );
+	}
+
+	/**
+	 * Convert tabs to spaces for a single line
+	 *
+	 * @param $line
+	 * @internal param string $text
+	 * @return string
+	 */
+	private static function tabsToSpacesLine( $line ) {
+		$parts = explode( "\t", $line );
+		$width = 8; // To match tidy's config & typical browser defaults
+		$out = $parts[0];
+		foreach( array_slice( $parts, 1 ) as $chunk ) {
+			$spaces = $width - (strlen( $out ) % $width);
+			$out .= str_repeat( ' ', $spaces );
+			$out .= $chunk;
+		}
+		return $out;
+	}
+}