From 2afb966fbfc17c9a084402f5401d0a7b17ad64c3 Mon Sep 17 00:00:00 2001
From: Pavol Bielik <pbielik@acm.org>
Date: Sun, 25 Jan 2015 16:37:14 +0100
Subject: [PATCH] Update encoding. Note that using some unicode characters is
 broken due to UglifyJS.

---
 lib/feature_extractor.js | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/lib/feature_extractor.js b/lib/feature_extractor.js
index 4aa2467f..4174d2b9 100644
--- a/lib/feature_extractor.js
+++ b/lib/feature_extractor.js
@@ -558,9 +558,11 @@ StringMap.prototype.getId = function(input){
 /* ------------------------ */
 
 function escapeString(input){
-	//escape method escapes certain characters that encodeURIComponent cannot process
-	//unescape is used to make the resulting string smaller
-	return unescape(encodeURIComponent(escape(input)));
+	try {
+		return encodeURIComponent(input);
+	} catch (ex){
+		throw new Parse_Error(" unable to encode '" + input.replace(/^!(String|Number|RegExp|Atom|Bool')!/, '') + "'");
+	}
 }
 
 function parseFile(code, file) {