Added support for interpreting numeric entities when the charset is i…

…so-8859-1, eg. &#9786 => ☺.

This is what browsers send when the user tries to submit characters that aren't available in the charset being used to submit the form.

lib/parse.js

@@ -16,6 +16,12 @@ var defaults = {
     strictNullHandling: false
 };
 
+var interpretNumericEntities = function (str) {
+    return str.replace(/&#(\d+);/g, function ($0, numberStr) {
+        return String.fromCharCode(parseInt(numberStr, 10));
+    });
+};
+
 var parseValues = function parseQueryStringValues(str, options) {
     var obj = {};
     var cleanStr = options.ignoreQueryPrefix ? str.replace(/^\?/, '') : str;
@@ -43,10 +49,15 @@ var parseValues = function parseQueryStringValues(str, options) {
             } else if (val === '✓') {
                 charset = 'iso-8859-1';
             }
-        } else if (has.call(obj, key)) {
-            obj[key] = [].concat(obj[key]).concat(val);
         } else {
-            obj[key] = val;
+            if (options.interpretNumericEntities && charset === 'iso-8859-1') {
+                val = interpretNumericEntities(val);
+            }
+            if (has.call(obj, key)) {
+                obj[key] = [].concat(obj[key]).concat(val);
+            } else {
+                obj[key] = val;
+            }
         }
     }
 

test/parse.js

@@ -607,5 +607,20 @@ test('parse()', function (t) {
         st.end();
     });
 
+    t.test('interprets numeric entities in iso-8859-1 when the interpretNumericEntities option is given', function (st) {
+        st.deepEqual(qs.parse('foo=%26%239786%3B', { charset: 'iso-8859-1', interpretNumericEntities: true }), { foo: '☺' });
+        st.end();
+    });
+
+    t.test('does not interpret numeric entities in iso-8859-1 when the interpretNumericEntities option is not given', function (st) {
+        st.deepEqual(qs.parse('foo=%26%239786%3B', { charset: 'iso-8859-1' }), { foo: '☺' });
+        st.end();
+    });
+
+    t.test('does not interpret numeric entities when the charset is utf-8, even when the interpretNumericEntities option is given', function (st) {
+        st.deepEqual(qs.parse('foo=%26%239786%3B', { charset: 'utf-8', interpretNumericEntities: true }), { foo: '☺' });
+        st.end();
+    });
+
     t.end();
 });