Harden link sanitization and add editor integration tests

Co-authored-by: mroxso <24775431+mroxso@users.noreply.github.com>

src/hooks/useDocumentEditor.ts

@@ -45,9 +45,9 @@ export interface DocumentEditor {
  *   extension, which is the exact seam where the Phase 2 Hocuspocus provider
  *   attaches (`Collaboration.configure({ document, provider })`). Nothing
  *   else about the app changes when that lands.
- * - Local undo/redo stays enabled: Collaboration swaps in the Yjs undo
- *   manager only when a provider connects, so solo Phase 1 editing keeps the
- *   expected Word-like per-user history.
+ * - Collaboration replaces local undo/redo with the Yjs UndoManager, which
+ *   tracks local-origin changes only — so undo behaves per-user (Word-like)
+ *   now, and stays correct when remote peers join in Phase 2.
  * - Pasted/dropped HTML passes through the allowlist sanitizer before the
  *   editor's schema parse; the schema itself drops anything else.
  */
@@ -125,12 +125,10 @@ export function useDocumentEditor({
           link: {
             openOnClick: false,
             autolink: true,
-            // Only allowlisted protocols may become links. Everything else
-            // (javascript:, data:, vbscript:) is dropped by the extension.
-            isAllowedUri: (url, ctx) => {
-              if (ctx.defaultValidate(url)) return true;
-              return sanitizeUrl(url) !== undefined;
-            },
+            // Only protocols on the shared allowlist may become links. This
+            // keeps `nostr:` mentions working while `javascript:`/`data:`
+            // hrefs are dropped.
+            isAllowedUri: (url) => sanitizeUrl(url) !== undefined,
           },
         }),
         ...(sessionDoc ? [Collaboration.configure({ document: sessionDoc })] : []),

src/lib/documents/editor.integration.test.tsx

@@ -0,0 +1,80 @@
+import { describe, expect, it } from 'vitest';
+import { Editor } from '@tiptap/core';
+import StarterKit from '@tiptap/starter-kit';
+import { TableKit } from '@tiptap/extension-table';
+import { sanitizeUrl } from '@/lib/nostrUtils';
+import { sanitizeHtml } from './sanitizeHtml';
+import { docToMarkdown } from './markdown';
+
+/**
+ * Integration tests against a real headless Tiptap editor using the same
+ * extension configuration as `useDocumentEditor`. They prove the constrained
+ * schema + sanitizer + serializer chain holds on untrusted input.
+ */
+function createEditor(): Editor {
+  return new Editor({
+    element: document.createElement('div'),
+    extensions: [
+      StarterKit.configure({
+        heading: { levels: [1, 2, 3] },
+        link: {
+          openOnClick: false,
+          autolink: true,
+          isAllowedUri: (url) => sanitizeUrl(url) !== undefined,
+        },
+      }),
+      TableKit.configure({ table: { resizable: false } }),
+    ],
+  });
+}
+
+describe('editor schema + sanitizer', () => {
+  it('rejects javascript: link hrefs end to end', () => {
+    const editor = createEditor();
+    editor.commands.setContent('<p><a href="javascript:alert(1)">click</a></p>', {
+      contentType: 'html',
+    });
+    // The unsafe href must not survive into the document.
+    expect(editor.getHTML()).not.toContain('javascript:');
+    editor.destroy();
+  });
+
+  it('keeps safe link hrefs', () => {
+    const editor = createEditor();
+    editor.commands.setContent('<p><a href="https://example.com">click</a></p>', {
+      contentType: 'html',
+    });
+    expect(editor.getHTML()).toContain('https://example.com');
+    editor.destroy();
+  });
+
+  it('drops script content pasted as HTML', () => {
+    const editor = createEditor();
+    const sanitized = sanitizeHtml('<p>hello</p><script>alert(1)</script>');
+    editor.commands.setContent(sanitized, { contentType: 'html' });
+    expect(editor.getText()).toContain('hello');
+    expect(editor.getHTML()).not.toContain('<script');
+    editor.destroy();
+  });
+
+  it('serializes typed content to portable Markdown', () => {
+    const editor = createEditor();
+    editor.commands.setContent('<h1>Title</h1><p>Some <strong>bold</strong> text</p>', {
+      contentType: 'html',
+    });
+    expect(docToMarkdown(editor.getJSON())).toBe('# Title\n\nSome **bold** text');
+    editor.destroy();
+  });
+
+  it('serializes a table to GFM', () => {
+    const editor = createEditor();
+    editor
+      .chain()
+      .focus()
+      .insertTable({ rows: 2, cols: 2, withHeaderRow: true })
+      .run();
+    const markdown = docToMarkdown(editor.getJSON());
+    expect(markdown).toContain('| --- | --- |');
+    editor.destroy();
+  });
+});