Generate HTML Table of Contents

gentoc.py

def gen_toc(html, ignore_h1_in_toc=False):
  if ignore_h1_in_toc:
    first_level = 2
  else:
    first_level = 1

  soup = BeautifulSoup.BeautifulSoup(html)
  headers = soup.findAll(re.compile('^h[%d-6]' % first_level))
  output = []
  for header in headers:
    hid = header.get('id')
    title = header.text
    level = int(header.name[-1])
    if hid:
      output.append('%s<a href="#%s">%s</a>' % (
            (level - first_level) * 4 * '&nbsp;', hid, title))
    else:
      output.append('%s%s' % ((level - first_level) * 4 * '&nbsp;', title))
  return '<br/>\n'.join(output)