[fix] Sanitize strings by removing lone surrogates

lib/browser.js

@@ -2,11 +2,11 @@
  * Module dependencies.
  */
 
-var keys = require('./keys');
+var utils = require('./utils');
 var hasBinary = require('has-binary');
 var sliceBuffer = require('arraybuffer.slice');
 var after = require('after');
-var utf8 = require('wtf-8');
+var utf8 = require('utf8');
 
 var base64encoder;
 if (global.ArrayBuffer) {
@@ -56,7 +56,7 @@ var packets = exports.packets = {
   , noop:     6
 };
 
-var packetslist = keys(packets);
+var packetslist = utils.keys(packets);
 
 /**
  * Premade error packet.
@@ -117,7 +117,7 @@ exports.encodePacket = function (packet, supportsBinary, utf8encode, callback) {
 
   // data fragment is optional
   if (undefined !== packet.data) {
-    encoded += utf8encode ? utf8.encode(String(packet.data)) : String(packet.data);
+    encoded += utf8encode ? utf8.encode(utils.sanitizeString(String(packet.data))) : String(packet.data);
   }
 
   return callback('' + encoded);

lib/index.js

@@ -2,9 +2,9 @@
  * Module dependencies.
  */
 
-var utf8 = require('wtf-8');
+var utf8 = require('utf8');
 var after = require('after');
-var keys = require('./keys');
+var utils = require('./utils');
 
 /**
  * Current protocol version.
@@ -25,7 +25,7 @@ var packets = exports.packets = {
   , noop:     6
 };
 
-var packetslist = keys(packets);
+var packetslist = utils.keys(packets);
 
 /**
  * Premade error packet.
@@ -72,7 +72,7 @@ exports.encodePacket = function (packet, supportsBinary, utf8encode, callback) {
 
   // data fragment is optional
   if (undefined !== packet.data) {
-    encoded += utf8encode ? utf8.encode(String(packet.data)) : String(packet.data);
+    encoded += utf8encode ? utf8.encode(utils.sanitizeString(String(packet.data))) : String(packet.data);
   }
 
   return callback('' + encoded);

lib/utils.js

@@ -0,0 +1,38 @@
+
+/**
+ * Gets the keys for an object.
+ *
+ * @return {Array} keys
+ * @api private
+ */
+
+var keys = Object.keys || function keys (obj) {
+  var arr = [];
+  var has = Object.prototype.hasOwnProperty;
+
+  for (var i in obj) {
+    if (has.call(obj, i)) {
+      arr.push(i);
+    }
+  }
+  return arr;
+};
+
+// from https://gist.github.com/mathiasbynens/bbe7f870208abcfec860
+var loneSurrogatesRegex = /[\uD800-\uDBFF](?![\uDC00-\uDFFF])|(?:[^\uD800-\uDBFF]|^)[\uDC00-\uDFFF]/g;
+
+/**
+ * Sanitize a WTF-8 string, replacing lone surrogates with
+ * U+FFFD 'REPLACEMENT CHARACTER'
+ *
+ * @return {String} str
+ * @api private
+ */
+var sanitizeString = function (str) {
+  return str.replace(loneSurrogatesRegex, '\uFFFD');
+};
+
+module.exports = {
+  keys: keys,
+  sanitizeString: sanitizeString
+};

package.json

@@ -16,7 +16,7 @@
     "base64-arraybuffer": "0.1.5",
     "blob": "0.0.4",
     "has-binary": "0.1.6",
-    "wtf-8": "1.0.0"
+    "utf8": "2.1.0"
   },
   "scripts": {
     "test": "make test"

test/parser.js

@@ -124,6 +124,15 @@ module.exports = function(parser) {
             expect(data).to.match(/^[0-9]$/);
          });
         });
+
+        it('should encode a string message with lone surrogates replaced', function(done) {
+          var data = '\uDC00 a\uDC00 \uDBFF\uDC00 \uDBFFb \uDBFF';
+          encode({ type: 'message', data: data }, null, true, function(encoded) {
+           expect(decode(encoded, null, true)).to.eql({ type: 'message', data: '\uFFFD \uFFFD \uDBFF\uDC00 \uFFFDb \uFFFD' });
+           done();
+          });
+        });
+
       });
 
       describe('decoding error handing', function () {