fix: preserve table header row in markdown export (#2532)

apps/server/src/collaboration/collaboration.util.spec.ts

@@ -0,0 +1,54 @@
+import { jsonToMarkdown } from './collaboration.util';
+
+const cell = (type: 'tableHeader' | 'tableCell', text: string) => ({
+  type,
+  content: [
+    {
+      type: 'paragraph',
+      content: text ? [{ type: 'text', text }] : [],
+    },
+  ],
+});
+
+const row = (type: 'tableHeader' | 'tableCell', texts: string[]) => ({
+  type: 'tableRow',
+  content: texts.map((text) => cell(type, text)),
+});
+
+const tableDoc = (rows: ReturnType<typeof row>[]) => ({
+  type: 'doc',
+  content: [{ type: 'table', content: rows }],
+});
+
+describe('jsonToMarkdown', () => {
+  it('uses the table header row as the markdown header', () => {
+    const markdown = jsonToMarkdown(
+      tableDoc([
+        row('tableHeader', ['Name', 'Role']),
+        row('tableCell', ['Ada', 'Engineer']),
+      ]),
+    );
+
+    expect(markdown.trim().split('\n')).toEqual([
+      '| Name | Role |',
+      '| --- | --- |',
+      '| Ada | Engineer |',
+    ]);
+  });
+
+  it('adds an empty header when the table has no header row', () => {
+    const markdown = jsonToMarkdown(
+      tableDoc([
+        row('tableCell', ['Ada', 'Engineer']),
+        row('tableCell', ['Alan', 'Mathematician']),
+      ]),
+    );
+
+    expect(markdown.trim().split('\n')).toEqual([
+      '|     |     |',
+      '| --- | --- |',
+      '| Ada | Engineer |',
+      '| Alan | Mathematician |',
+    ]);
+  });
+});

packages/editor-ext/src/lib/markdown/utils/turndown.utils.ts

@@ -37,7 +37,11 @@ export function htmlToMarkdown(html: string): string {
     footnoteRef,
     footnotesList,
   ]);
-  return turndownService.turndown(html).replaceAll('<br>', ' ');
+  const htmlWithoutColgroups = html.replace(
+    /<colgroup\b[^>]*>[\s\S]*?<\/colgroup>/gi,
+    '',
+  );
+  return turndownService.turndown(htmlWithoutColgroups).replaceAll('<br>', ' ');
 }
 
 function listParagraph(turndownService: _TurndownService) {