@adobe/helix-importer 2.9.0 → 2.9.2

package/CHANGELOG.md

@@ -1,3 +1,17 @@
+## [2.9.2](https://github.com/adobe/helix-importer/compare/v2.9.1...v2.9.2) (2023-05-04)
+
+
+### Bug Fixes
+
+* replace nbsp with space ([#142](https://github.com/adobe/helix-importer/issues/142)) ([f093c74](https://github.com/adobe/helix-importer/commit/f093c749be0dcd75ec3a2b44313346fac2f3ede4))
+
+## [2.9.1](https://github.com/adobe/helix-importer/compare/v2.9.0...v2.9.1) (2023-04-22)
+
+
+### Bug Fixes
+
+* **deps:** update adobe fixes ([#139](https://github.com/adobe/helix-importer/issues/139)) ([7cf1d8c](https://github.com/adobe/helix-importer/commit/7cf1d8cfc72f854b1c1268a4360592ae783ada1e))
+
 # [2.9.0](https://github.com/adobe/helix-importer/compare/v2.8.13...v2.9.0) (2023-04-20)
 
 

package/README.md

@@ -34,7 +34,7 @@ async function main() {
 }
 ```
 
-In this example, the [WPContentPager](./src/wp/explorers/WPContentPager.ts) extends the [PagignExplorer](src/explorer/PagingExplorer.ts) which implements the 2 methods:
+In this example, the [WPContentPager](./src/wp/explorers/WPContentPager.ts) extends the [PagingExplorer](src/explorer/PagingExplorer.ts) which implements the 2 methods:
 - `fetch` which defines how to fetch one page on results
 - `explore` which extracts the list of urls present on that page
 
@@ -71,4 +71,4 @@ TODO: publish npm module
 
 ```js
 import { ... } from '@adobe/helix-importer';
-```
+```

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/helix-importer",
-  "version": "2.9.0",
+  "version": "2.9.2",
   "description": "Helix Importer tool: create md / docx from html",
   "main": "src/index.js",
   "type": "module",
@@ -27,24 +27,24 @@
   "devDependencies": {
     "@adobe/eslint-config-helix": "2.0.2",
     "@adobe/helix-docx2md": "1.4.1",
-    "@adobe/helix-mediahandler": "2.1.5",
+    "@adobe/helix-mediahandler": "2.1.7",
     "@semantic-release/changelog": "6.0.3",
     "@semantic-release/exec": "6.0.3",
     "@semantic-release/git": "10.0.1",
     "c8": "7.13.0",
     "dirname-filename-esm": "1.1.1",
-    "eslint": "8.38.0",
+    "eslint": "8.39.0",
     "husky": "8.0.3",
-    "lint-staged": "13.2.1",
+    "lint-staged": "13.2.2",
     "mocha": "10.2.0",
     "mocha-multi-reporters": "1.5.1",
     "mock-fs": "5.2.0",
-    "semantic-release": "21.0.1"
+    "semantic-release": "21.0.2"
   },
   "license": "Apache-2.0",
   "dependencies": {
     "@adobe/helix-markdown-support": "6.1.1",
-    "@adobe/helix-md2docx": "2.0.50",
+    "@adobe/helix-md2docx": "2.0.51",
     "@adobe/mdast-util-gridtables": "1.0.6",
     "@adobe/remark-gridtables": "1.0.2",
     "form-data": "4.0.0",

package/src/utils/DOMUtils.js

@@ -103,10 +103,12 @@ export default class DOMUtils {
         (tag.textContent === ''
           || tag.textContent === ' '
           || tag.textContent === ' '
-          || tag.textContent.charCodeAt(0) === 160)
+          || (tag.textContent.charCodeAt(0) === 160 && tag.textContent.length === 1))
         && !tag.querySelector(DOMUtils.EMPTY_TAGS_TO_PRESERVE.join(','))
       ) {
         tag.remove();
+      } else {
+        tag.innerHTML = tag.innerHTML.replace(/ /gm, ' ');
       }
     }
   }
@@ -145,7 +147,7 @@ export default class DOMUtils {
     // remove spans
     document.querySelectorAll('span').forEach((span) => {
       // do not touch spans with images and span with a css class or an id
-      if (!span.querySelector('img') && span.classList.length === 0 && !span.id) {
+      if (!span.querySelector('img') && span.classList.length === 0 && !span.id && !span.getAttribute('style')) {
         if (span.textContent === '') {
           span.remove();
         } else {

package/test/importers/fixtures/space.spec.html

@@ -4,10 +4,11 @@
     <p>A simple paragraph</p>
     <p>A paragraph with a br inside.<br> This should be next line.</p>
     <p>A paragraph with a br at the end.<br></p>
+    <p>A paragraph with a br at the end and &amp;nbsp; "&nbsp;".<br> &nbsp;</p>
     <p>A paragraph followed by a br</p>
     <br>
     <p>A paragraph after the br</p>
-    &nbsp;
+    <p>&nbsp;</p>
     <p>A paragraph after the nbsp;</p>
   </body>
 </html>

package/test/importers/fixtures/space.spec.md

@@ -7,6 +7,8 @@ This should be next line.
 
 A paragraph with a br at the end.
 
+A paragraph with a br at the end and \  " ".
+
 A paragraph followed by a br
 
 \
@@ -14,6 +16,4 @@ A paragraph followed by a br
 
 A paragraph after the br
 
- 
-
 A paragraph after the nbsp;

package/test/utils/DOMUtils.spec.js

@@ -95,6 +95,11 @@ describe('DOMUtils#reviewParagraphs tests', () => {
     test('<p><video width="320" height="240" controls=""><source src="movie.mp4" type="video/mp4"></video></p>', '<p><video width="320" height="240" controls=""><source src="movie.mp4" type="video/mp4"></video></p>');
     test('<p><iframe src="www.iframe.com"></iframe></p>', '<p><iframe src="www.iframe.com"></iframe></p>');
   });
+
+  it('reviewParagraphs replaces &nbsp; with spaces', () => {
+    test('<p>usefull with space&nbsp;</p>', '<p>usefull with space </p>');
+    test('<p>&nbsp;more&nbsp;spaces&nbsp;<br> &nbsp;</p>', '<p> more spaces <br>  </p>');
+  });
 });
 
 describe('DOMUtils#reviewHeadings tests', () => {
@@ -193,6 +198,9 @@ describe('DOMUtils#removeSpans tests', () => {
     test('<div>Spans potentially used to do layouting: <span class="tab1">tab1</span><span class="tab2">tab2</span></div>', '<div>Spans potentially used to do layouting: <span class="tab1">tab1</span><span class="tab2">tab2</span></div>');
     test('<div>Spans potentially used to do layouting: <span id="tab1">tab1</span><span id="tab2">tab2</span></div>', '<div>Spans potentially used to do layouting: <span id="tab1">tab1</span><span id="tab2">tab2</span></div>');
   });
+  it('keeps styled spans', () => {
+    test('<p><span style="text-decoration: underline;">This should be underlined.</span></p>', '<p><span style="text-decoration: underline;">This should be underlined.</span></p>');
+  });
 });
 
 describe('DOMUtils#removeNoscripts tests', () => {