A replacement for decodeURIComponent that doesn't throw.

And add a few more tests.

lib/querystring.js

@@ -1,10 +1,10 @@
 // Query String Utilities
 
 var QueryString = exports;
+var urlDecode = process.binding('http_parser').urlDecode;
 
-QueryString.unescape = function (str, decodeSpaces) {
+// a safe fast alternative to decodeURIComponent
-  return decodeURIComponent(decodeSpaces ? str.replace(/\+/g, " ") : str);
+QueryString.unescape = urlDecode;
-};
 
 QueryString.escape = function (str) {
   return encodeURIComponent(str);
@@ -25,17 +25,17 @@ var stack = [];
  * @param name {String} (optional) Name of the current key, for handling children recursively.
  * @static
  */
-QueryString.stringify = function (obj, sep, eq, munge, name) {
+QueryString.stringify = QueryString.encode = function (obj, sep, eq, munge, name) {
   munge = typeof(munge) == "undefined" ? true : munge;
   sep = sep || "&";
   eq = eq || "=";
   if (isA(obj, null) || isA(obj, undefined) || typeof(obj) === 'function') {
-    return name ? encodeURIComponent(name) + eq : '';
+    return name ? QueryString.escape(name) + eq : '';
   }
 
   if (isBool(obj)) obj = +obj;
   if (isNumber(obj) || isString(obj)) {
-    return encodeURIComponent(name) + eq + encodeURIComponent(obj);
+    return QueryString.escape(name) + eq + QueryString.escape(obj);
   }
   if (isA(obj, [])) {
     var s = [];
@@ -71,11 +71,11 @@ QueryString.stringify = function (obj, sep, eq, munge, name) {
   return s;
 };
 
-QueryString.parseQuery = QueryString.parse = function (qs, sep, eq) {
+QueryString.parse = QueryString.decode = function (qs, sep, eq) {
   return (qs || '')
     .split(sep||"&")
     .map(pieceParser(eq||"="))
-    .reduce(mergeParams);
+    .reduce(mergeParams)
 };
 
 // Parse a key=val string.
@@ -87,26 +87,14 @@ QueryString.parseQuery = QueryString.parse = function (qs, sep, eq) {
 // return parse(foo[bar], [{bla:"baz"}])
 // return parse(foo, {bar:[{bla:"baz"}]})
 // return {foo:{bar:[{bla:"baz"}]}}
-var trimmerPattern = /^\s+|\s+$/g,
+var slicerPattern = /(.*)\[([^\]]*)\]$/;
-  slicerPattern = /(.*)\[([^\]]*)\]$/;
 var pieceParser = function (eq) {
   return function parsePiece (key, val) {
     if (arguments.length !== 2) {
       // key=val, called from the map/reduce
        key = key.split(eq);
-      return parsePiece(
+      return parsePiece(QueryString.unescape(key.shift(), true),
-        QueryString.unescape(key.shift(), true),
+                        QueryString.unescape(key.join(eq), true));
-        QueryString.unescape(key.join(eq), true)
-      );
-    }
-    key = key.replace(trimmerPattern, '');
-    if (isString(val)) {
-      val = val.replace(trimmerPattern, '');
-      // convert numerals to numbers
-      if (!isNaN(val)) {
-        var numVal = +val;
-        if (val === numVal.toString(10)) val = numVal;
-      }
     }
     var sliced = slicerPattern.exec(key);
     if (!sliced) {

src/node_http_parser.cc

@@ -315,6 +315,86 @@ class Parser : public ObjectWrap {
 };
 
 
+static Handle<Value> UrlDecode (const Arguments& args) {
+  HandleScope scope;
+
+  if (!args[0]->IsString()) {
+    return ThrowException(Exception::TypeError(
+          String::New("First arg must be a string")));
+  }
+
+  bool decode_spaces = args[1]->IsTrue();
+
+  String::Utf8Value in_v(args[0]->ToString());
+  size_t l = in_v.length();
+  char* out = strdup(*in_v);
+
+  enum { CHAR, HEX0, HEX1 } state = CHAR;
+
+  int n, m, hexchar; 
+  size_t in_index = 0, out_index = 0;
+  char c;
+  for (; in_index <= l; in_index++) {
+    c = out[in_index];
+    switch (state) {
+      case CHAR:
+        switch (c) {
+          case '%':
+            n = 0;
+            m = 0;
+            state = HEX0;
+            break;
+          case '+':
+            if (decode_spaces) c = ' ';
+            // pass thru
+          default:
+            out[out_index++] = c;
+            break;
+        }
+        break;
+
+      case HEX0:
+        state = HEX1;
+        hexchar = c;
+        if ('0' <= c && c <= '9') {
+          n = c - '0';
+        } else if ('a' <= c && c <= 'f') {
+          n = c - 'a' + 10;
+        } else if ('A' <= c && c <= 'F') {
+          n = c - 'A' + 10;
+        } else {
+          out[out_index++] = '%';
+          out[out_index++] = c;
+          state = CHAR;
+          break;
+        }
+        break;
+
+      case HEX1:
+        state = CHAR;
+        if ('0' <= c && c <= '9') {
+          m = c - '0';
+        } else if ('a' <= c && c <= 'f') {
+          m = c - 'a' + 10;
+        } else if ('A' <= c && c <= 'F') {
+          m = c - 'A' + 10;
+        } else {
+          out[out_index++] = '%';
+          out[out_index++] = hexchar;
+          out[out_index++] = c;
+          break;
+        }
+        out[out_index++] = 16*n + m;
+        break;
+    }
+  }
+
+  Local<String> out_v = String::New(out, out_index-1);
+  free(out);
+  return scope.Close(out_v);
+}
+
+
 void InitHttpParser(Handle<Object> target) {
   HandleScope scope;
 
@@ -327,6 +407,7 @@ void InitHttpParser(Handle<Object> target) {
   NODE_SET_PROTOTYPE_METHOD(t, "reinitialize", Parser::Reinitialize);
 
   target->Set(String::NewSymbol("HTTPParser"), t->GetFunction());
+  NODE_SET_METHOD(target, "urlDecode", UrlDecode);
 
   on_message_begin_sym    = NODE_PSYMBOL("onMessageBegin");
   on_path_sym             = NODE_PSYMBOL("onPath");

test/simple/test-querystring.js

@@ -29,7 +29,9 @@ var qsTestCases = [
   ["foo[bar][bla]=baz&foo[bar][bla]=blo", "foo%5Bbar%5D%5Bbla%5D%5B%5D=baz&foo%5Bbar%5D%5Bbla%5D%5B%5D=blo", {"foo":{"bar":{"bla":["baz","blo"]}}}],
   ["foo[bar][][bla]=baz&foo[bar][][bla]=blo", "foo%5Bbar%5D%5B%5D%5Bbla%5D=baz&foo%5Bbar%5D%5B%5D%5Bbla%5D=blo", {"foo":{"bar":[{"bla":"baz"},{"bla":"blo"}]}}],
   ["foo[bar][bla][]=baz&foo[bar][bla][]=blo", "foo%5Bbar%5D%5Bbla%5D%5B%5D=baz&foo%5Bbar%5D%5Bbla%5D%5B%5D=blo", {"foo":{"bar":{"bla":["baz","blo"]}}}],
-  [" foo = bar ", "foo=bar", {"foo":"bar"}]
+  [" foo = bar ", "%20foo%20=%20bar%20", {" foo ":" bar "}],
+  ["foo=%zx", "foo=%25zx", {"foo":"%zx"}],
+  ["foo=%EF%BF%BD", "foo=%EF%BF%BD", {"foo" : "\ufffd" }]
 ];
 
 // [ wonkyQS, canonicalQS, obj ]