From: Brad Jorsch <bjorsch@wikimedia.org>
Date: Tue, 13 Jun 2017 16:03:56 +0000 (-0400)
Subject: Hide <style> tags from Tidy
X-Git-Tag: 1.31.0-rc.0~2988^2
X-Git-Url: http://git.cyclocoop.org/%22%20.%20generer_url_ecrire%28%22auteur_infos%22%2C%20%22id_auteur=%24id%22%29%20.%20%22?a=commitdiff_plain;h=83b798bbabbe048552fba735834c773ac2015aed;p=lhc%2Fweb%2Fwiklou.git

Hide <style> tags from Tidy

Some versions of html-tidy (e.g. the one currently in use on WMF wikis)
will try to move all <style> tags in the body into the head, effectively
removing them for our purposes. We need to avoid that for TemplateStyles.

Bug: T167349
Change-Id: I133776d16f366cad73ed30af0e5a665fdf9f5ed9
---

diff --git a/includes/tidy/RaggettWrapper.php b/includes/tidy/RaggettWrapper.php
index 9f6feb8e55..b793a58af7 100644
--- a/includes/tidy/RaggettWrapper.php
+++ b/includes/tidy/RaggettWrapper.php
@@ -48,6 +48,12 @@ class RaggettWrapper {
 		// Modify inline Microdata <link> and <meta> elements so they say <html-link> and <html-meta> so
 		// we can trick Tidy into not stripping them out by including them in tidy's new-empty-tags config
 		$wrappedtext = preg_replace( '!<(link|meta)([^>]*?)(/{0,1}>)!', '<html-$1$2$3', $wrappedtext );
+		// Similar for inline <style> tags, but those aren't empty.
+		$wrappedtext = preg_replace_callback( '!<style([^>]*)>(.*?)</style>!s', function ( $m ) {
+			return '<html-style' . $m[1] . '>'
+				. $this->replaceCallback( [ $m[2] ] )
+				. '</html-style>';
+		}, $wrappedtext );
 
 		// Preserve empty li elements (T49673) by abusing Tidy's datafld hack
 		// The whitespace class is as in TY_(InitMap)
@@ -78,8 +84,9 @@ class RaggettWrapper {
 	 * @return string
 	 */
 	public function postprocess( $text ) {
-		// Revert <html-{link,meta}> back to <{link,meta}>
+		// Revert <html-{link,meta,style}> back to <{link,meta,style}>
 		$text = preg_replace( '!<html-(link|meta)([^>]*?)(/{0,1}>)!', '<$1$2$3', $text );
+		$text = preg_replace( '!<(/?)html-(style)([^>]*)>!', '<$1$2$3>', $text );
 
 		// Remove datafld
 		$text = str_replace( '<li datafld=""', '<li', $text );
diff --git a/includes/tidy/tidy.conf b/includes/tidy/tidy.conf
index d93ce2251b..d4a3199367 100644
--- a/includes/tidy/tidy.conf
+++ b/includes/tidy/tidy.conf
@@ -20,3 +20,5 @@ fix-uri: no
 # html-{meta,link} is a hack we use to prevent Tidy from stripping <meta> and <link> used in the body for Microdata
 new-empty-tags: html-meta, html-link, wbr, source, track
 new-inline-tags: video, audio, bdi, data, time, mark
+# html-style is a hack we use to prevent pre-HTML5 versions of Tidy from stripping <style> used in the body for TemplateStyles
+new-blocklevel-tags: html-style
diff --git a/tests/parser/parserTests.txt b/tests/parser/parserTests.txt
index d973c3e9df..44bcdffeb4 100644
--- a/tests/parser/parserTests.txt
+++ b/tests/parser/parserTests.txt
@@ -28315,3 +28315,20 @@ unclosed internal link XSS (T137264)
 !! html/parsoid
 <p>[[#%3Cscript%3Ealert(1)%3C/script%3E|</p>
 !! end
+
+!! test
+Validating that <style> isn't eaten by tidy (T167349)
+!! options
+# Use $wgRawHtml to inject a <style> tag, since you normally can't in wikitext
+wgRawHtml=1
+!! wikitext
+<div class="foo">
+<html><style>.foo::before { content: "<foo>"; }</style></html>
+<html><style data-mw-foobar="baz">.foo::after { content: "<bar>"; }</style></html>
+</div>
+!! html+tidy
+<div class="foo">
+<style>.foo::before { content: "<foo>"; }</style>
+<style data-mw-foobar="baz">.foo::after { content: "<bar>"; }</style>
+</div>
+!! end
