feat: manage absolute images adjustment

adobe · Jan 26, 2024 · afa47d9 · afa47d9
1 parent 079249a
commit afa47d9
Show file tree

Hide file tree

Showing 5 changed files with 54 additions and 18 deletions.
diff --git a/src/importer/defaults/rules/adjustImageUrls.js b/src/importer/defaults/rules/adjustImageUrls.js
@@ -10,18 +10,39 @@
  * 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('../'))) {
-      try {
-        const u = new URL(src, url);
-        // eslint-disable-next-line no-param-reassign
-        img.src = u.toString();
-      } catch (e) {
-        // eslint-disable-next-line no-console
-        console.log(`Unable to adjust image URL ${img.src} - removing image`);
-        img.remove();
+    if (src) {
+      if (src.startsWith('./') || src.startsWith('/') || src.startsWith('../')) {
+        // transform relative URLs to absolute URLs
+        try {
+          const targetUrl = new URL(src, url);
+          // eslint-disable-next-line no-param-reassign
+          img.src = targetUrl.toString();
+        } catch (e) {
+          // eslint-disable-next-line no-console
+          console.log(`Unable to adjust image URL ${img.src} - removing image`);
+          img.remove();
+        }
+      } else if (current) {
+        // also transform absolute URLs to current host
+        try {
+          const currentSrc = new URL(src);
+          const currentUrl = new URL(current);
+          if (currentSrc.host === currentUrl.host) {
+            // if current host is same than src host, switch src host with url host
+            // this is the case for absolutes URLs pointing to the same host
+            const targetUrl = new URL(url);
+            const newSrc = new URL(`${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`);
+          img.remove();
+        }
       }
     }
   });

diff --git a/src/importer/defaults/transformDOM.js b/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;

diff --git a/test/importers/defaults/fixtures/adjust-image-urls.expected.html b/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>
diff --git a/test/importers/defaults/fixtures/adjust-image-urls.input.html b/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>
diff --git a/test/importers/defaults/transformDOM.spec.js b/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',
+      },
     });
   });