add regex for all wide chars.

lib/unicode.js

@@ -549,6 +549,16 @@ exports.wideChars = new RegExp('(['
   + '\\uffe0-\\uffe6'
   + '])', 'g');
 
+exports.allWide = new RegExp('('
+  // 0x20000 - 0x2fffd:
+  + '[\\ud840-\\ud87f][\\udc00-\\udffd]'
+  + '|'
+  // 0x30000 - 0x3fffd:
+  + '[\\ud880-\\ud8bf][\\udc00-\\udffd]'
+  + '|'
+  + exports.wideChars.source.slice(1, -1)
+  + ')', 'g');
+
 // Regex to detect a surrogate pair.
 exports.surrogate = /[\ud800-\udbff][\udc00-\udfff]/g;
 

lib/widget.js

@@ -2427,7 +2427,7 @@ Element.prototype.parseContent = function(noTags) {
       content = content.replace(unicode.wideChars, '$1_');
     } else {
       // no double-width: replace them with question-marks.
-      content = content.replace(unicode.wideChars, '??');
+      content = content.replace(unicode.allWide, '??');
       // delete combining characters since they're 0-width anyway.
       // NOTE: We could drop this, the non-surrogates would get changed to ? by
       // the unicode filter, and surrogates changed to ? by the surrogate