@adobe/helix-importer 3.1.3 → 3.2.0

package/CHANGELOG.md

@@ -1,3 +1,17 @@
+# [3.2.0](https://github.com/adobe/helix-importer/compare/v3.1.4...v3.2.0) (2024-01-29)
+
+
+### Features
+
+* manage absolute images adjustment ([#296](https://github.com/adobe/helix-importer/issues/296)) ([abd19ae](https://github.com/adobe/helix-importer/commit/abd19aeb59a8d00574762d477f012c38e996e869))
+
+## [3.1.4](https://github.com/adobe/helix-importer/compare/v3.1.3...v3.1.4) (2024-01-17)
+
+
+### Bug Fixes
+
+* import comments cleanup logic ([#290](https://github.com/adobe/helix-importer/issues/290)) ([deec675](https://github.com/adobe/helix-importer/commit/deec675c6e88f0035a3d5b3d88b4068f106a8f80))
+
 ## [3.1.3](https://github.com/adobe/helix-importer/compare/v3.1.2...v3.1.3) (2024-01-06)
 
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@adobe/helix-importer",
-  "version": "3.1.3",
+  "version": "3.2.0",
   "description": "Helix Importer tool: create md / docx from html",
   "main": "src/index.js",
   "type": "module",
@@ -13,7 +13,6 @@
     "test:web": "web-test-runner test/browser/*.test.js --node-resolve",
     "test:web:watch": "web-test-runner test/browser/*.test.js --node-resolve --watch",
     "test": "c8 mocha && npm run test:web",
-    "test-ci": "c8 mocha && npm run test:web",
     "semantic-release": "semantic-release",
     "prepare": "npx husky install"
   },
@@ -29,7 +28,7 @@
   "devDependencies": {
     "@adobe/eslint-config-helix": "2.0.5",
     "@adobe/helix-docx2md": "1.5.0",
-    "@adobe/helix-mediahandler": "2.4.4",
+    "@adobe/helix-mediahandler": "2.4.7",
     "@esm-bundle/chai": "4.3.4-fix.0",
     "@semantic-release/changelog": "6.0.3",
     "@semantic-release/exec": "6.0.3",
@@ -37,12 +36,12 @@
     "@web/test-runner": "0.18.0",
     "@web/test-runner-commands": "0.9.0",
     "@web/test-runner-mocha": "0.9.0",
-    "c8": "8.0.1",
-    "chai": "4.4.0",
+    "c8": "9.1.0",
+    "chai": "5.0.3",
     "dirname-filename-esm": "1.1.1",
     "eslint": "8.56.0",
     "husky": "8.0.3",
-    "jsdom": "23.1.0",
+    "jsdom": "24.0.0",
     "lint-staged": "15.2.0",
     "mocha": "10.2.0",
     "mocha-multi-reporters": "1.5.1",

package/src/importer/defaults/rules/adjustImageUrls.js

@@ -10,14 +10,29 @@
  * governing permissions and limitations under the License.
  */
 
-export default function adjustImageUrls(main, url) {
+export default function adjustImageUrls(main, url, current) {
   [...main.querySelectorAll('img')].forEach((img) => {
     const src = img.getAttribute('src');
-    if (src && (src.startsWith('./') || src.startsWith('/') || src.startsWith('../'))) {
+    if (src) {
       try {
-        const u = new URL(src, url);
-        // eslint-disable-next-line no-param-reassign
-        img.src = u.toString();
+        if (src.startsWith('./') || src.startsWith('/') || src.startsWith('../')) {
+          // transform relative URLs to absolute URLs
+          const targetUrl = new URL(src, url);
+          // eslint-disable-next-line no-param-reassign
+          img.src = targetUrl.toString();
+        } else if (current) {
+          // also transform absolute URLs to current host
+          const currentSrc = new URL(src);
+          const currentUrl = new URL(current);
+          if (currentSrc.host === currentUrl.host) {
+            // if current host is same than src host, switch src host with url host
+            // this is the case for absolutes URLs pointing to the same host
+            const targetUrl = new URL(url);
+            const newSrc = new URL(`${currentSrc.pathname}${currentSrc.search}${currentSrc.hash}`, `${targetUrl.protocol}//${targetUrl.host}`);
+            // eslint-disable-next-line no-param-reassign
+            img.src = newSrc.toString();
+          }
+        }
       } catch (e) {
         // eslint-disable-next-line no-console
         console.log(`Unable to adjust image URL ${img.src} - removing image`);

package/src/importer/defaults/transformDOM.js

@@ -17,7 +17,7 @@ import transformBackgroundImages from './rules/transformBackgroundImages.js';
 
 export default async function transformDOM({
   // eslint-disable-next-line no-unused-vars
-  url, document, html, params,
+  url, document, html, params = {},
 }) {
   const main = document.body;
 
@@ -35,7 +35,7 @@ export default async function transformDOM({
 
   createMetadata(main, document);
   transformBackgroundImages(main, document);
-  adjustImageUrls(main, url);
+  adjustImageUrls(main, url, params.originalURL);
   convertIcons(main, document);
 
   return main;

package/src/utils/DOMUtils.js

@@ -139,7 +139,8 @@ export default class DOMUtils {
     // eslint-disable-next-line no-param-reassign
     document.body.innerHTML = document.body.innerHTML
       // remove html comments
-      .replace(/<!--(?!>)[\S\s]*?-->/gm, '');
+      .replace(/><!--(?!>)[\S\s]*?-->/gm, '>')
+      .replace(/\n<!--(?!>)[\S\s]*?-->/gm, '\n');
   }
 
   static removeSpans(document) {

package/test/importers/defaults/fixtures/adjust-image-urls.expected.html

@@ -1,7 +1,13 @@
 <body>
   <h1>Hello World</h1>
   <img src="https://wwww.sample.com/image1.png">
-  <img src="https://wwww.sample.com/path/image2.png">
-  <img src="https://wwww.sample.com/image3.png">
-  <img src="https://wwww.anotherhost.com/image4.png">
+  <img src="https://wwww.sample.com/image2.png?p=v&amp;p1=v1">
+  <img src="https://wwww.sample.com/path/image3.png">
+  <img src="https://wwww.sample.com/path/image4.png?p=v&amp;p1=v1">
+  <img src="https://wwww.sample.com/go/to/image5.png">
+  <img src="https://wwww.sample.com/path/go/to/image6.png">
+  <img src="https://wwww.sample.com/image7.png">
+  <img src="https://wwww.sample.com/image8.png?p=v&amp;p1=v1">
+  <img src="https://wwww.sample.com/go/to/image9.png">
+  <img src="https://wwww.anotherhost.com/image10.png">
 </body>

package/test/importers/defaults/fixtures/adjust-image-urls.input.html

@@ -2,9 +2,15 @@
   <body>
     <h1>Hello World</h1>
     <img src="/image1.png">
-    <img src="./image2.png">
-    <img src="https://wwww.sample.com/image3.png">
-    <img src="https://wwww.anotherhost.com/image4.png">
+    <img src="/image2.png?p=v&p1=v1">
+    <img src="./image3.png">
+    <img src="./image4.png?p=v&p1=v1">
+    <img src="/go/to/image5.png">
+    <img src="./go/to/image6.png">
+    <img src="https://wwww.currenthost.com/image7.png">
+    <img src="https://wwww.currenthost.com/image8.png?p=v&p1=v1">
+    <img src="https://wwww.currenthost.com/go/to/image9.png">
+    <img src="https://wwww.anotherhost.com/image10.png">
     <img src="/\/: #brokenlink#?!$$">
   </body>
 </html>

package/test/importers/defaults/transformDOM.spec.js

@@ -77,6 +77,9 @@ describe('defaultTransformDOM tests', () => {
   it('default transformation adjusts image urls', async () => {
     await runTest('adjust-image-urls', {
       url: 'https://wwww.sample.com/path/page.html',
+      params: {
+        originalURL: 'https://wwww.currenthost.com/path/page.html',
+      },
     });
   });
 

package/test/utils/DOMUtils.spec.js

@@ -182,6 +182,8 @@ describe('DOMUtils#removeCommments tests', () => {
     test('<p><!-- useless comment \n multiline --></p>', '<p></p>');
     test('<p><!-- useless comment \n multiline \n multiline --></p>', '<p></p>');
     test('<!-- useless comment --><p>The content stays</p><!-- another useless comment with \n line break -->', '<p>The content stays</p>');
+    test('<p>This is a paragraph.</p>\n\x3C!--\n<p>Look at this cool image:</p>\n<img border="0" src="pic_trulli.jpg" alt="Trulli">\n-->\n<p>This is a paragraph too.</p>\x3C!-- same line -->\n\x3C!-- single line -->', '<p>This is a paragraph.</p>\n\n<p>This is a paragraph too.</p>\n');
+    test('<div some-crazy-attribute="" <!--=""><!-- useless comment --></div>', '<div some-crazy-attribute="" <!--=""></div>');
   });
 });