[security] Sanitize paths, hosts before parsing.

index.js

@@ -20,6 +20,9 @@ var required = require('requires-port')
 var rules = [
   ['#', 'hash'],                        // Extract from the back.
   ['?', 'query'],                       // Extract from the back.
+  function sanitize(address) {          // Sanitize what is left of the address
+    return address.replace('\\', '/');
+  },
   ['/', 'pathname'],                    // Extract from the back.
   ['@', 'auth', 1],                     // Extract from the front.
   [NaN, 'host', undefined, 1, 1],       // Set left over value.
@@ -47,7 +50,7 @@ var ignore = { hash: 1, query: 1 };
  *
  * @param {Object|String} loc Optional default location object.
  * @returns {Object} lolcation object.
- * @api public
+ * @public
  */
 function lolcation(loc) {
   var location = global && global.location || {};
@@ -89,7 +92,7 @@ function lolcation(loc) {
  *
  * @param {String} address URL we want to extract from.
  * @return {ProtocolExtract} Extracted information.
- * @api private
+ * @private
  */
 function extractProtocol(address) {
   var match = protocolre.exec(address);
@@ -107,7 +110,7 @@ function extractProtocol(address) {
  * @param {String} relative Pathname of the relative URL.
  * @param {String} base Pathname of the base URL.
  * @return {String} Resolved pathname.
- * @api private
+ * @private
  */
 function resolve(relative, base) {
   var path = (base || '/').split('/').slice(0, -1).concat(relative.split('/'))
@@ -140,11 +143,14 @@ function resolve(relative, base) {
  * create an actual constructor as it's much more memory efficient and
  * faster and it pleases my OCD.
  *
+ * It is worth noting that we should not use `URL` as class name to prevent
+ * clashes with the global URL instance that got introduced in browsers.
+ *
  * @constructor
  * @param {String} address URL we want to parse.
  * @param {Object|String} location Location defaults for relative paths.
  * @param {Boolean|Function} parser Parser for the query string.
- * @api public
+ * @private
  */
 function Url(address, location, parser) {
   if (!(this instanceof Url)) {
@@ -190,10 +196,16 @@ function Url(address, location, parser) {
   // When the authority component is absent the URL starts with a path
   // component.
   //
-  if (!extracted.slashes) instructions[2] = [/(.*)/, 'pathname'];
+  if (!extracted.slashes) instructions[3] = [/(.*)/, 'pathname'];
 
   for (; i < instructions.length; i++) {
     instruction = instructions[i];
+
+    if (typeof instruction === 'function') {
+      address = instruction(address);
+      continue;
+    }
+
     parse = instruction[0];
     key = instruction[1];
 
@@ -284,8 +296,8 @@ function Url(address, location, parser) {
  *                               used to parse the query.
  *                               When setting the protocol, double slash will be
  *                               removed from the final url if it is true.
- * @returns {URL}
- * @api public
+ * @returns {URL} URL instance for chaining.
+ * @public
  */
 function set(part, value, fn) {
   var url = this;
@@ -370,8 +382,8 @@ function set(part, value, fn) {
  * Transform the properties back in to a valid and full URL string.
  *
  * @param {Function} stringify Optional query stringify function.
- * @returns {String}
- * @api public
+ * @returns {String} Compiled version of the URL.
+ * @public
  */
 function toString(stringify) {
   if (!stringify || 'function' !== typeof stringify) stringify = qs.stringify;

test/test.js

@@ -192,6 +192,28 @@ describe('url-parse', function () {
     assume(parsed.pathname).equals('/b/c');
   });
 
+  it('ignores \\ in pathnames', function () {
+    var url = 'http://google.com:80\\@yahoo.com/#what\\is going on'
+      , parsed = parse(url);
+
+    assume(parsed.port).equals('');
+    assume(parsed.username).equals('');
+    assume(parsed.password).equals('');
+    assume(parsed.hostname).equals('google.com');
+    assume(parsed.hash).equals('#what\\is going on');
+
+    parsed = parse('//\\what-is-up.com');
+    assume(parsed.pathname).equals('/what-is-up.com');
+  });
+
+  it('correctly ignores multiple slashes //', function () {
+    var url = '////what-is-up.com'
+      , parsed = parse(url);
+
+    assume(parsed.host).equals('');
+    assume(parsed.hostname).equals('');
+  });
+
   describe('origin', function () {
     it('generates an origin property', function () {
       var url = 'http://google.com:80/pathname'
@@ -252,6 +274,13 @@ describe('url-parse', function () {
       o = parse('wss://google.com:80/pathname');
       assume(o.origin).equals('wss://google.com:80');
     });
+
+    it('maintains the port number for non-default port numbers', function () {
+      var parsed = parse('http://google.com:8080/pathname');
+
+      assume(parsed.host).equals('http://google.com:8080');
+      assume(parsed.href).equals('http://google.com:8080/pathname');
+    });
   });
 
   describe('protocol', function () {