Decode HTML entities

Author: CMTegner

index.js

@@ -2,6 +2,7 @@ var util = require('util');
 var Transform = require('stream').Transform;
 var Parser = require('htmlparser2').Parser;
 var isEmpty = require('lodash.isempty');
+var decode = (new (require('html-entities').AllHtmlEntities)()).decode;
 
 util.inherits(JSONMLParser, Transform);
 
@@ -36,7 +37,7 @@ JSONMLParser.prototype._createSourceOptions = function() {
             transform._parent = element;
         },
         ontext: function(text) {
-            transform._parent.push(text);
+            transform._parent.push(decode(text));
         },
         oncomment: function(text) {
             transform._parent.push(['#comment', text]);

package.json

@@ -28,6 +28,7 @@
   "homepage": "https://github.com/CMTegner/jsonml-parse",
   "dependencies": {
     "JSONStream": "^0.8.4",
+    "html-entities": "^1.0.10",
     "htmlparser2": "^3.7.3",
     "lodash.isempty": "^2.4.1",
     "nomnom": "^1.8.0"

test/06-html-entities/expected.json

@@ -0,0 +1,4 @@
+[
+    ["button", { "class": "close" }, "×"],
+    "\n«Yeah, ok»\næøå\n© Awesome Inc."
+]

test/06-html-entities/markup.html

@@ -0,0 +1,4 @@
+<button class="close">&times;</button>
+&laquo;Yeah, ok&raquo;
+&aelig;&oslash;&aring;
+&copy; Awesome Inc.

test/cli/expected.json

@@ -57,6 +57,12 @@ test('should parse comments', function(t) {
     verify(markup, expected, t);
 });
 
+test('should convert HTML entities into unicode characters', function(t) {
+    var markup = fs.readFileSync(__dirname + '/06-html-entities/markup.html', 'utf8');
+    var expected = require('./06-html-entities/expected.json');
+    verify(markup, expected, t);
+});
+
 test('should also offer a node-style callback API', function(t) {
     parse('<div></div>', function(err, result) {
         t.notOk(err);