Full document, spreadsheet, slideshow, and diagram tooling
fix: prefer markdown conversion when clipboard HTML contains raw syntax
When copying from apps like GitHub or text editors, the clipboard often
includes both text/html and text/plain. The HTML may wrap raw markdown
text in basic tags without converting [text](url) to <a> links.
Previously the paste handler always deferred to TipTap for HTML
clipboard content. Now it checks whether the HTML contains unrendered
markdown link syntax or task list patterns. If so, the plain text
markdown conversion path is used instead, producing proper TipTap
link marks and task items.
+57
-4
+8
-3
src/docs/main.ts
+8
-3
src/docs/main.ts
···
52
52
import { exportDocx } from './docx-export.js';
53
53
import { importDocx, isValidDocx } from './docx-import.js';
54
54
import { markdownToHtml } from './markdown-parser.js';
55
-
import { looksLikeMarkdown } from './markdown-paste.js';
55
+
import { looksLikeMarkdown, htmlContainsRawMarkdown } from './markdown-paste.js';
56
56
import { htmlToMarkdown as turndownHtmlToMarkdown } from './markdown-export.js';
57
57
import { createMarkdownToggle, TOGGLE_MODE } from './markdown-toggle.js';
58
58
import { VersionManager, computeWordCount } from '../lib/version-history.js';
···
273
273
handlePaste: (_view, event) => {
274
274
const clip = event.clipboardData;
275
275
if (!clip) return false;
276
-
// If clipboard has HTML, let TipTap handle it natively
277
-
if (clip.types.includes('text/html')) return false;
278
276
const text = clip.getData('text/plain');
279
277
if (!text || !looksLikeMarkdown(text)) return false;
278
+
// If clipboard has HTML, only override when the HTML contains
279
+
// unrendered markdown syntax (e.g. literal [text](url) not in <a> tags).
280
+
// Well-rendered HTML (from rich text editors) is left to TipTap.
281
+
if (clip.types.includes('text/html')) {
282
+
const clipHtml = clip.getData('text/html');
283
+
if (!htmlContainsRawMarkdown(clipHtml)) return false;
284
+
}
280
285
const html = markdownToHtml(text);
281
286
if (!html) return false;
282
287
event.preventDefault();
+20
src/docs/markdown-paste.ts
+20
src/docs/markdown-paste.ts
···
4
4
* Conservatively detects whether pasted plain text is markdown.
5
5
* Requires 2+ signals or 1 strong signal (heading, code block, table)
6
6
* to avoid false positives on normal text.
7
+
*
8
+
* Also detects when clipboard HTML contains unrendered markdown link
9
+
* syntax, indicating the source didn't convert markdown to rich text.
7
10
*/
8
11
9
12
const PATTERNS: readonly RegExp[] = [
···
45
48
46
49
return false;
47
50
}
51
+
52
+
/**
53
+
* Check whether clipboard HTML contains unrendered markdown link syntax.
54
+
* When apps copy content, they may include both text/html and text/plain.
55
+
* If the HTML contains literal `[text](url)` strings (not rendered as <a> tags),
56
+
* the plain text markdown path should be preferred.
57
+
*/
58
+
// Looser task list pattern for HTML context (no start-of-line anchor)
59
+
const RAW_TASK_LIST_IN_HTML = /- \[([ xX])\]\s/;
60
+
61
+
export function htmlContainsRawMarkdown(html: string): boolean {
62
+
if (!html) return false;
63
+
// Strip actual <a> tags and their content to avoid false positives,
64
+
// then check if raw markdown link syntax remains in the text content.
65
+
const stripped = html.replace(/<a\b[^>]*>.*?<\/a>/gi, '');
66
+
return MARKDOWN_LINK_RE.test(stripped) || RAW_TASK_LIST_IN_HTML.test(stripped);
67
+
}
+29
-1
tests/markdown-paste.test.ts
+29
-1
tests/markdown-paste.test.ts
···
1
1
import { describe, it, expect } from 'vitest';
2
-
import { looksLikeMarkdown } from '../src/docs/markdown-paste.js';
2
+
import { looksLikeMarkdown, htmlContainsRawMarkdown } from '../src/docs/markdown-paste.js';
3
3
4
4
describe('looksLikeMarkdown', () => {
5
5
// --- Strong solo signals (should return true with just 1) ---
···
108
108
expect(looksLikeMarkdown(' - Indented item\n - Another\n> Quote')).toBe(true);
109
109
});
110
110
});
111
+
112
+
describe('htmlContainsRawMarkdown', () => {
113
+
it('detects raw markdown links in HTML', () => {
114
+
expect(htmlContainsRawMarkdown('<p>[Example](https://example.com)</p>')).toBe(true);
115
+
});
116
+
117
+
it('detects raw task lists in HTML', () => {
118
+
expect(htmlContainsRawMarkdown('<li>- [ ] Buy milk</li>')).toBe(true);
119
+
});
120
+
121
+
it('ignores properly rendered <a> tags', () => {
122
+
expect(htmlContainsRawMarkdown('<p>Check <a href="https://example.com">Example</a></p>')).toBe(false);
123
+
});
124
+
125
+
it('detects raw links alongside rendered links', () => {
126
+
expect(htmlContainsRawMarkdown(
127
+
'<p><a href="https://a.com">A</a> and [B](https://b.com)</p>'
128
+
)).toBe(true);
129
+
});
130
+
131
+
it('returns false for empty input', () => {
132
+
expect(htmlContainsRawMarkdown('')).toBe(false);
133
+
});
134
+
135
+
it('returns false for plain HTML without markdown', () => {
136
+
expect(htmlContainsRawMarkdown('<p>Hello world</p>')).toBe(false);
137
+
});
138
+
});