MAJOR: respect whitespaces

src/parse.js

@@ -102,9 +102,17 @@ export default function parse(html, options) {
         // calculate correct end of the content slice in case there's
         // no tag after the text node.
         const end = html.indexOf('<', start)
-        const content = html.slice(start, end === -1 ? undefined : end)
-        // if a node is nothing but whitespace, no need to add it.
-        if (!/^\s*$/.test(content)) {
+        let content = html.slice(start, end === -1 ? undefined : end)
+        // if a node is nothing but whitespace, collapse it as the spec states:
+        // https://www.w3.org/TR/html4/struct/text.html#h-9.1
+        if (/^\s*$/.test(content)) {
+          content = ' ';
+        }
+        // don't add whitespace-only text nodes if they would be trailing text nodes
+        // or if they would be leading whitespace-only text nodes:
+        //  * end > -1 indicates this is not a trailing text node
+        //  * leading node is when level is -1 and parent has length 0
+        if ((end > -1 && level + parent.length >= 0) || content !== ' ') {
           parent.push({
             type: 'text',
             content: content,

test/parse.js

@@ -659,6 +659,10 @@ test('parse', function (t) {
         voidElement: false,
         children: [{ type: 'text', content: 'something' }],
       },
+      {
+        type: 'text',
+        content: ' ',
+      },
       {
         type: 'tag',
         name: 'a',
@@ -686,6 +690,7 @@ test('parse', function (t) {
         voidElement: false,
         children: [{ type: 'text', content: 'Hi' }],
       },
+      { type: 'text', content: ' ' },
       {
         type: 'tag',
         name: 'span',
@@ -881,3 +886,53 @@ test('ReDoS vulnerability reported by Sam Sanoop of Snyk', function (t) {
   t.ok(duration < 100, 'should not hang')
   t.end()
 })
+
+test('whitespace', function (t) {
+  let html = '<div></div>\n'
+  let 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>\n\n <span>There</span> \t <div> </div>'
+  parsed = HTML.parse(html)
+  t.deepEqual(parsed, [{
+    type: 'tag',
+    name: 'div',
+    attrs: {},
+    voidElement: false,
+    children: [
+      { type: 'text', content: 'Hi' }
+    ]
+  },{
+    type: 'text',
+    content: ' '
+  },
+  {
+    type: 'tag',
+    name: 'span',
+    attrs: {},
+    voidElement: false,
+    children: [
+      { type: 'text', content: 'There' }
+    ]
+  },{
+    type: 'text',
+    content: ' '
+  },{
+    type: 'tag',
+    name: 'div',
+    attrs: {},
+    voidElement: false,
+    children: [
+      { type: 'text', content: ' ' }
+    ]
+  }], 'should collapse whitespace')
+  // See https://www.w3.org/TR/html4/struct/text.html#h-9.1
+
+  t.end()
+})