Fixes for parsing top-level and trailing text nodes. (#12)

* When there are top-level text nodes, add them to the results array instead of one level up when at level 0.

* Make sure we calculate the correct end of the text node when it's the last node in the html string.

* Add tests to verify parser behavior on top-level and trailing text nodes.

* We ought to drop text nodes that are nothing but whitespace since not useful in HTML.

* Update tests for dropping whitespace-only nodes.

Author: rayd

lib/parse.js

@@ -63,10 +63,21 @@ module.exports = function parse(html, options) {
             level--;
             if (!inComponent && nextChar !== '<' && nextChar) {
                 // trailing text node
-                arr[level].children.push({
-                    type: 'text',
-                    content: html.slice(start, html.indexOf('<', start))
-                });
+                // if we're at the root, push a base text node. otherwise add as
+                // a child to the current node.
+                parent = level === -1 ? result : arr[level].children;
+
+                // calculate correct end of the content slice in case there's
+                // no tag after the text node.
+                var end = html.indexOf('<', start);
+                var content = html.slice(start, end === -1 ? undefined : end);
+                // if a node is nothing but whitespace, no need to add it.
+                if (!/^\s*$/.test(content)) {
+                    parent.push({
+                        type: 'text',
+                        content: content
+                    });
+                }
             }
         }
     });

test/parse.js

@@ -292,6 +292,80 @@ test('parse', function (t) {
             { type: 'text', content: ' 10 ' },
         ]
     }], 'should not give voidElements children');
+
+    html = '<div></div>\n';
+    parsed = HTML.parse(html);
+    t.deepEqual(parsed, [{
+        type: 'tag',
+        name: 'div',
+        attrs: {},
+        voidElement: false,
+        children: []
+    }], 'should not explode on trailing whitespace');
+
+    html = '<div>Hi</div> There ';
+    parsed = HTML.parse(html);
+    t.deepEqual(parsed, [{
+        type: 'tag',
+        name: 'div',
+        attrs: {},
+        voidElement: false,
+        children: [
+            { type: 'text', content: 'Hi' }
+        ]
+    },{
+        type: 'text', content: ' There '
+    }], 'should handle trailing text nodes at the top-level');
+
+    html = '<div>Hi</div> There <span>something</span> <a></a>else ';
+    parsed = HTML.parse(html);
+    t.deepEqual(parsed, [{
+        type: 'tag',
+        name: 'div',
+        attrs: {},
+        voidElement: false,
+        children: [
+            { type: 'text', content: 'Hi' }
+        ]
+    },{
+        type: 'text', content: ' There '
+    },{
+        type: 'tag',
+        name: 'span',
+        attrs: {},
+        voidElement: false,
+        children: [
+            { type: 'text', content: 'something' }
+        ]
+    },{
+        type: 'tag',
+        name: 'a',
+        attrs: {},
+        voidElement: false,
+        children: []
+    },{
+        type: 'text', content: 'else '
+    }], 'should handle text nodes in the middle of tags at the top-level');
+
+    html = '<div>Hi</div>\n\n <span>There</span> \t ';
+    parsed = HTML.parse(html);
+    t.deepEqual(parsed, [{
+        type: 'tag',
+        name: 'div',
+        attrs: {},
+        voidElement: false,
+        children: [
+            { type: 'text', content: 'Hi' }
+        ]
+    },{
+        type: 'tag',
+        name: 'span',
+        attrs: {},
+        voidElement: false,
+        children: [
+            { type: 'text', content: 'There' }
+        ]
+    }], 'should remove text nodes that are nothing but whitespace');
     t.end();
 });