feat: give access to the original dom, before any preprocessing (#41)

adobe · Nov 23, 2022 · 6c394c5 · 6c394c5
1 parent fa9b4b5
commit 6c394c5
Show file tree

Hide file tree

Showing 3 changed files with 44 additions and 5 deletions.
diff --git a/package-lock.json b/package-lock.json
diff --git a/src/importer/HTML2x.js b/src/importer/HTML2x.js
@@ -21,7 +21,7 @@ import Utils from '../utils/Utils.js';
 
 // import docxStylesXML from '../resources/styles.xml';
 
-function preprocessDOM(document) {
+function setBackgroundImagesFromCSS(document) {
   const elements = document.querySelectorAll('body, header, footer, div, span, section, main');
   const getComputedStyle = document.defaultView?.getComputedStyle;
   if (getComputedStyle) {
@@ -76,8 +76,19 @@ async function html2x(
     }
   }
 
-  if (config.preprocess !== false) {
-    preprocessDOM(doc);
+  // for more advanced use cases, give access to the original dom with
+  // no preprocessing at all
+  if (transformer.preprocess) {
+    transformer.preprocess({
+      url,
+      document: doc,
+      html: doc.documentElement.outerHTML,
+      params,
+    });
+  }
+
+  if (config.setBackgroundImagesFromCSS !== false) {
+    setBackgroundImagesFromCSS(doc);
   }
 
   const html = doc.documentElement.outerHTML;

diff --git a/test/importers/HTML2x.spec.js b/test/importers/HTML2x.spec.js
@@ -72,13 +72,15 @@ describe('html2x parameters', () => {
     await html2md(URL, HTML, {
       transformDOM: testParams,
       generateDocumentPath: testParams,
+      preprocess: testParams,
     }, null, {
       originalURL: ORIGNAL_URL,
     });
 
     await html2docx(URL, HTML, {
       transformDOM: testParams,
       generateDocumentPath: testParams,
+      preprocess: testParams,
     }, null, {
       originalURL: ORIGNAL_URL,
     });
@@ -87,12 +89,14 @@ describe('html2x parameters', () => {
   it('parameters are correctly passed in multi mode', async () => {
     await html2md(URL, HTML, {
       transform: testParams,
+      preprocess: testParams,
     }, null, {
       originalURL: ORIGNAL_URL,
     });
 
     await html2docx(URL, HTML, {
       transform: testParams,
+      preprocess: testParams,
     }, null, {
       originalURL: ORIGNAL_URL,
     });
@@ -205,6 +209,30 @@ describe('html2md tests', () => {
     );
     strictEqual(out.html.trim(), '<body><img src="./image.png"></body>');
   });
+
+  it('html2md removes images with src attributes', async () => {
+    const out = await html2md('https://www.sample.com/page.html', '<html><body><img src="data:abc"></body></html>');
+    strictEqual(out.html.trim(), '<body></body>');
+    strictEqual(out.md.trim(), '');
+  });
+
+  it('html2md set image src with data-src attribute value', async () => {
+    const out = await html2md('https://www.sample.com/page.html', '<html><body><img src="data:abc" data-src="./image.jpg"></body></html>');
+    strictEqual(out.html.trim(), '<body><img src="./image.jpg" data-src="./image.jpg"></body>');
+    strictEqual(out.md.trim(), '![][image0]\n\n[image0]: ./image.jpg');
+  });
+
+  it('html2md allows to preprocess the document', async () => {
+    const out = await html2md('https://www.sample.com/page.html', '<html><body><img src="data:abc" data-fancy-src="./image.jpg"></body></html>', {
+      preprocess: ({ document }) => {
+        const img = document.querySelector('img');
+        img.setAttribute('src', img.getAttribute('data-fancy-src'));
+        img.removeAttribute('data-fancy-src');
+      },
+    });
+    strictEqual(out.html.trim(), '<body><img src="./image.jpg"></body>');
+    strictEqual(out.md.trim(), '![][image0]\n\n[image0]: ./image.jpg');
+  });
 });
 
 describe('html2docx tests', () => {