import { NodeHtmlMarkdown } from 'node-html-markdown';

// MIT-licensed HTML→Markdown rendering for the web_fetch tool. Output feeds an
// LLM, so structural fidelity matters more than exact whitespace.
const OPTIONS = {
  // GFM-style emphasis markers (matches what most models expect).
  emDelimiter: '*',
  strongDelimiter: '**',
  bulletMarker: '*',
  codeFence: '```',
  codeBlockStyle: 'fenced' as const,
  // Always use []() syntax for links rather than <url> autolinks.
  useInlineLinks: false,
  // Collapse runs of blank lines to a single separator.
  maxConsecutiveNewlines: 1,
  // Strip non-content elements entirely (script/style are skipped by default,
  // but listing them here is explicit; head/nav/footer/etc. drop their text).
  ignore: ['script', 'style', 'head', 'noscript', 'svg', 'math', 'nav', 'footer'],
};

export function htmlToMarkdown(sourceHtml: string): string {
  if (!sourceHtml) return '';
  return NodeHtmlMarkdown.translate(sourceHtml, OPTIONS).trim();
}