Leon4gr45/builder

1import { Readability } from '@mozilla/readability';2import TurndownService from 'turndown';3 4let turndown: TurndownService | null = null;5function getTurndown(): TurndownService {6  if (!turndown) turndown = new TurndownService({ headingStyle: 'atx', codeBlockStyle: 'fenced' });7  return turndown;8}9 10// Minimum readable-text length before we trust Readability's extraction over11// the raw document body. Short fixtures/pages get low-quality Readability12// output (heading demotion, URL normalization), so fall back to the raw body.13const MIN_READABLE_LENGTH = 200;14 15export function htmlToMarkdown(html: string, baseUrl: string): string {16  try {17    const doc = new DOMParser().parseFromString(html, 'text/html');18    // Resolve relative links against the source URL so markdown links work.19    const base = doc.createElement('base');20    base.href = baseUrl;21    doc.head?.appendChild(base);22    const article = new Readability(doc).parse();23    const readableText = article?.textContent?.trim() ?? '';24    const contentHtml =25      article?.content && readableText.length >= MIN_READABLE_LENGTH26        ? article.content27        : doc.body?.innerHTML || html;28    const md = getTurndown().turndown(contentHtml).trim();29    return md || (doc.body?.textContent || '').trim();30  } catch {31    return html;32  }33}34