kepano/defuddle 0.13.0

on GitHub

Breaking changes

• defuddle/node now accepts any DOM Document (linkedom, happy-dom, JSDOM, etc.), not just JSDOM.
• JSDOM is no longer a peer dependency. linkedom is now the recommended DOM parser
• Passing a raw HTML string or JSDOM instance to defuddle/node is deprecated and will be removed in the next major version.

Recommended usage

import { parseHTML } from 'linkedom';
import { Defuddle } from 'defuddle/node';

const { document } = parseHTML(html);
const result = await Defuddle(document, 'https://example.com/article');

Passing a JSDOM instance still works but is deprecated:

// @deprecated — pass dom.window.document directly instead
const result = await Defuddle(dom, url);
// Preferred
const result = await Defuddle(dom.window.document, url);

Improvements

• Generic document support for non-HTML content (#166)
• YouTube: Use existing page transcript before fetching via API
• YouTube: Improved transcript grouping, sentence merging, and cross-environment support
• YouTube: Fix diarization stripping - speaker markers from auto-captions
• Add .post-body entry point for Ghost CMS sites
• Smarter retry for hidden content (#163)
• CJK word count support (#158)
• Precompile partial selector regex for faster parsing (#157)

Fixes

• Fix syntax highlighting for Lean (#159, #160)
• Fix filenames on Windows (#155)
• Fix spacing between exclamation and image in markdown
• Fix newlines in Verso