Clean html string - remove css, script and other unnecessary content. keep html tags, json and json schema only. Useful to feed the html content to LLM for analysis

cleanHtml.js

// Clean html - remove css, script, keep html tags, json and json schema only.
function cleanHtml(html) {
  // First remove HTML comments
  let cleaned = html.replace(/<!--[\s\S]*?-->/g, '');

  // Remove unwanted tags, but preserve script tags with type="application/json" or type="application/ld+json"
  cleaned = cleaned.replace(
    /<script(?![^>]*type="application\/(json|ld\+json)")[^>]*>[\s\S]*?<\/script>|<style[^>]*>[\s\S]*?<\/style>|<(link|svg|img|image)[^>]*?>[\s\S]*?<\/\1>|<(link|svg|img|image)([^>]*?)\/?>|<\/(link|svg|img|image)>/gi,
    ''
  );

  // Replace any tag's attributes, keeping href and specified script types if they exist
  cleaned = cleaned.replace(/<([a-z][a-z0-9]*)(?:[^>]*href="([^"]*)")?(?:[^>]*type="(application\/(?:json|ld\+json))")?[^>]*?>/gi, (match, tag, href, type) => {
    const attributes = [];
    if (href) attributes.push(`href="${href}"`);
    if (type && tag.toLowerCase() === 'script') attributes.push(`type="${type}"`);
    return `<${tag}${attributes.length ? ` ${attributes.join(' ')}` : ''}>`;
  });

  // Remove empty tags (can need multiple passes for nested empty tags)
  // But don't remove script tags with allowed types even if empty
  let prevHtml;
  do {
    prevHtml = cleaned;
    cleaned = cleaned.replace(/<(?!script\s+type="application\/(json|ld\+json)")([a-z][a-z0-9]*)>\s*<\/\2>/gi, '');
  } while (cleaned !== prevHtml);

  return cleaned;
}