@link-assistant/web-capture 1.7.18 → 1.7.19

package/CHANGELOG.md

@@ -1,5 +1,11 @@
 # @link-assistant/web-capture
 
+## 1.7.19
+
+### Patch Changes
+
+- 3a483aa: Fix Google Docs public-export Markdown for multi-paragraph table cells.
+
 ## 1.7.18
 
 ### Patch Changes

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@link-assistant/web-capture",
-  "version": "1.7.18",
+  "version": "1.7.19",
   "description": "CLI and microservice to render web pages as HTML, Markdown, or PNG",
   "main": "src/index.js",
   "bin": {

package/src/gdocs-preprocess.js

@@ -379,7 +379,7 @@ function inlineTableCellHtml($, cell) {
         parts.push(this.data);
       }
     });
-  return parts.join('<br>');
+  return parts.join('<br><br>');
 }
 
 function hasBoldStyle(style) {

package/src/lib.js

@@ -166,11 +166,34 @@ export function convertHtmlToMarkdown(html, baseUrl) {
     style: false,
   });
   turndown.use(turndownPluginGfm.gfm);
+  preserveTableCellLineBreaks(turndown);
   // Decode HTML entities to unicode after markdown conversion
   // Preserve non-breaking spaces as   entities for clear marking
   return he.decode(turndown.turndown($.html())).replace(/\u00A0/g, ' ');
 }
 
+function preserveTableCellLineBreaks(turndown) {
+  turndown.addRule('tableCellLineBreak', {
+    filter(node) {
+      return node.nodeName === 'BR' && hasAncestorNode(node, ['TD', 'TH']);
+    },
+    replacement() {
+      return '<br>';
+    },
+  });
+}
+
+function hasAncestorNode(node, nodeNames) {
+  let current = node.parentNode;
+  while (current) {
+    if (nodeNames.includes(current.nodeName)) {
+      return true;
+    }
+    current = current.parentNode;
+  }
+  return false;
+}
+
 function selectedHtml($, selector) {
   if (!selector) {
     return null;
@@ -531,6 +554,7 @@ export function convertHtmlToMarkdownEnhanced(html, baseUrl, options = {}) {
     style: false,
   });
   turndown.use(turndownPluginGfm.gfm);
+  preserveTableCellLineBreaks(turndown);
 
   // Decode HTML entities to unicode after markdown conversion
   // Normalize non-breaking spaces to regular spaces in Markdown text

package/tests/unit/gdocs-preprocess.test.js

@@ -0,0 +1,36 @@
+import {
+  normalizeGoogleDocsExportMarkdown,
+  preprocessGoogleDocsExportHtml,
+} from '../../src/gdocs.js';
+import { convertHtmlToMarkdown } from '../../src/lib.js';
+
+describe('preprocessGoogleDocsExportHtml', () => {
+  it('keeps multi-paragraph table cells on one GFM row (issue #110)', () => {
+    const html = `
+      <table>
+        <tr><td><p>Feature</p></td><td><p>Notes</p></td></tr>
+        <tr>
+          <td>
+            <p>First paragraph.</p>
+            <p>Second paragraph.</p>
+            <p>Third paragraph.</p>
+          </td>
+          <td><p>x</p></td>
+        </tr>
+      </table>
+    `;
+
+    const preprocessed = preprocessGoogleDocsExportHtml(html);
+    const markdown = normalizeGoogleDocsExportMarkdown(
+      convertHtmlToMarkdown(preprocessed.html)
+    );
+
+    expect(preprocessed.html).toContain(
+      '<td>First paragraph.<br><br>Second paragraph.<br><br>Third paragraph.</td>'
+    );
+    expect(markdown).toContain(
+      '| First paragraph.<br><br>Second paragraph.<br><br>Third paragraph. | x |'
+    );
+    expect(markdown).not.toContain('First paragraph.  \n');
+  });
+});