trim leading/trailing whitespace from sections that can occur when pa…

…rsing indented HTML

src/js/parsers/dom.js

@@ -36,6 +36,14 @@ export function transformHTMLText(textContent) {
   return text;
 }
 
+export function trimSectionText(section) {
+  if (section.isMarkerable && section.markers.length) {
+    let { head, tail } = section.markers;
+    head.value = head.value.replace(/^\s+/, '');
+    tail.value = tail.value.replace(/\s+$/, '');
+  }
+}
+
 function isGoogleDocsContainer(element) {
   return !isTextNode(element) &&
          !isCommentNode(element) &&
@@ -108,6 +116,10 @@ class DOMParser {
       this.appendSections(post, sections);
     });
 
+    // trim leading/trailing whitespace of markerable sections to avoid
+    // unnessary whitespace from indented HTML input
+    forEach(post.sections, section => trimSectionText(section));
+
     return post;
   }
 

src/js/parsers/section.js

@@ -35,7 +35,8 @@ import {
 } from 'mobiledoc-kit/utils/array-utils';
 
 import {
-  transformHTMLText
+  transformHTMLText,
+  trimSectionText
 } from '../parsers/dom';
 
 import assert from '../utils/assert';
@@ -283,6 +284,7 @@ class SectionParser {
 
     // push listItems onto the listSection or add a new section
     if (state.section.isListItem && lastSection && lastSection.isListSection) {
+      trimSectionText(state.section);
       lastSection.items.append(state.section);
     } else {
       // remove empty list sections before creating a new section

tests/unit/parsers/dom-test.js

@@ -361,6 +361,24 @@ test('singly-nested ol lis are parsed correctly', (assert) => {
   assert.equal(section.items.objectAt(1).text, 'second element');
 });
 
+test('nested html doesn\'t create unneccessary whitespace', (assert) => {
+  let element = buildDOM(`
+    <div>
+      <p>
+        One
+      <p>
+      <p>
+        Two
+      </p>
+    </div>
+  `);
+  const post = parser.parse(element);
+
+  assert.equal(post.sections.length, 2, '2 sections');
+  assert.equal(post.sections.objectAt(0).text, 'One');
+  assert.equal(post.sections.objectAt(1).text, 'Two');
+});
+
 /*
  * FIXME: Google docs nests uls like this
 test('lis in nested uls are flattened (when ul is child of ul)', (assert) => {

tests/unit/parsers/section-test.js

@@ -322,7 +322,7 @@ test('#parse handles insignificant whitespace', (assert) => {
   let [list] = sections;
   assert.equal(list.type, 'list-section');
   assert.equal(list.items.length, 1, '1 list item');
-  assert.equal(list.items.objectAt(0).text, '         One            ');
+  assert.equal(list.items.objectAt(0).text, 'One');
 });
 
 test('#parse avoids empty paragraph around wrapped list', (assert) => {