From d5b526914b9e5b0e97a68130545d5b0b33d7d7ad Mon Sep 17 00:00:00 2001
From: Jermolene <jeremy@osmosoft.com>
Date: Wed, 12 Feb 2014 15:24:03 +0000
Subject: [PATCH] Simplify the regular expression for HTML comments
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

We were getting catastrophic backtracking in Chrome for some input
texts where the closing “—>” of the comment was omitted.
---
 .../parsers/wikiparser/rules/commentblock.js  | 21 ++++++++++++++++---
 .../parsers/wikiparser/rules/commentinline.js | 21 ++++++++++++++++---
 2 files changed, 36 insertions(+), 6 deletions(-)

diff --git a/core/modules/parsers/wikiparser/rules/commentblock.js b/core/modules/parsers/wikiparser/rules/commentblock.js
index fe712f03a..cb1bd1a93 100644
--- a/core/modules/parsers/wikiparser/rules/commentblock.js
+++ b/core/modules/parsers/wikiparser/rules/commentblock.js
@@ -9,6 +9,8 @@ Wiki text block rule for HTML comments. For example:
 <!-- This is a comment -->
 ```
 
+Note that the syntax for comments is simplified to an opening "<!--" sequence and a closing "-->" sequence -- HTML itself implements a more complex format (see http://ostermiller.org/findhtmlcomment.html)
+
 \*/
 (function(){
 
@@ -21,13 +23,26 @@ exports.types = {block: true};
 
 exports.init = function(parser) {
 	this.parser = parser;
-	// Regexp to match - HTML comment regexp by Stephen Ostermiller, http://ostermiller.org/findhtmlcomment.html
-	this.matchRegExp = /\<![ \r\n\t]*(?:--(?:[^\-]|[\r\n]|-[^\-])*--[ \r\n\t]*)\>\r?\n/mg;
+	this.matchRegExp = /\<!--/mg;
+	this.endMatchRegExp = /--\>/mg;
+};
+
+exports.findNextMatch = function(startPos) {
+	this.matchRegExp.lastIndex = startPos;
+	this.match = this.matchRegExp.exec(this.parser.source);
+	if(this.match) {
+		this.endMatchRegExp.lastIndex = startPos + this.match[0].length;
+		this.endMatch = this.endMatchRegExp.exec(this.parser.source);
+		if(this.endMatch) {
+			return this.match.index;
+		}
+	}
+	return undefined;
 };
 
 exports.parse = function() {
 	// Move past the match
-	this.parser.pos = this.matchRegExp.lastIndex;
+	this.parser.pos = this.endMatchRegExp.lastIndex;
 	// Don't return any elements
 	return [];
 };
diff --git a/core/modules/parsers/wikiparser/rules/commentinline.js b/core/modules/parsers/wikiparser/rules/commentinline.js
index e10ecedd0..fd9c6a10e 100644
--- a/core/modules/parsers/wikiparser/rules/commentinline.js
+++ b/core/modules/parsers/wikiparser/rules/commentinline.js
@@ -9,6 +9,8 @@ Wiki text inline rule for HTML comments. For example:
 <!-- This is a comment -->
 ```
 
+Note that the syntax for comments is simplified to an opening "<!--" sequence and a closing "-->" sequence -- HTML itself implements a more complex format (see http://ostermiller.org/findhtmlcomment.html)
+
 \*/
 (function(){
 
@@ -21,13 +23,26 @@ exports.types = {inline: true};
 
 exports.init = function(parser) {
 	this.parser = parser;
-	// Regexp to match - HTML comment regexp by Stephen Ostermiller, http://ostermiller.org/findhtmlcomment.html
-	this.matchRegExp = /\<![ \r\n\t]*(?:--(?:[^\-]|[\r\n]|-[^\-])*--[ \r\n\t]*)\>/mg;
+	this.matchRegExp = /\<!--/mg;
+	this.endMatchRegExp = /--\>/mg;
+};
+
+exports.findNextMatch = function(startPos) {
+	this.matchRegExp.lastIndex = startPos;
+	this.match = this.matchRegExp.exec(this.parser.source);
+	if(this.match) {
+		this.endMatchRegExp.lastIndex = startPos + this.match[0].length;
+		this.endMatch = this.endMatchRegExp.exec(this.parser.source);
+		if(this.endMatch) {
+			return this.match.index;
+		}
+	}
+	return undefined;
 };
 
 exports.parse = function() {
 	// Move past the match
-	this.parser.pos = this.matchRegExp.lastIndex;
+	this.parser.pos = this.endMatchRegExp.lastIndex;
 	// Don't return any elements
 	return [];
 };