striptags.py

import re
import unittest
from unittest import TestCase


def striptags(html, *tags):
    result = html
    for tag in tags:
        result = re.sub(r'<{tag}>.+?</{tag}>'.format(tag=tag), r'', result)
    return result


def strip_headers(html):
    return striptags(html, 'h1', 'h2', 'h3', 'h4', 'h5', 'h6')


class TempTestCase(TestCase):
    def test_striptags(self):
        self.assertEqual('a', striptags('<x>b</x>a', 'x'))

    def test_strip_headers(self):
        self.assertEqual('\nxxx', strip_headers('<h1>header</h1>\nxxx'))
        self.assertEqual('aaa\nxxx', strip_headers('aaa<h2>header</h2>\nxxx'))
        self.assertEqual('aaa\nxxx', strip_headers('<h1>hea1</h1>aaa<h2>header</h2>\nxxx'))


if __name__ == '__main__':
    unittest.main()

Не работает со сложными заголовками