def test_nested_with_text(self): body = '<html><a href="/abc">foo<div>bar</div></a></html>' url = URL('http://www.w3af.com/') headers = Headers() headers['content-type'] = 'text/html' resp = HTTPResponse(200, body, headers, url, url, charset='utf-8') p = SGMLParser(resp) tags = p.get_tags_by_filter(('a', 'b'), yield_text=True) tags = list(tags) self.assertEqual([Tag('a', {'href': '/abc'}, 'foo')], tags)
def test_none(self): body = '<html><a href="/abc">foo<div>bar</div></a></html>' url = URL('http://www.w3af.com/') headers = Headers() headers['content-type'] = 'text/html' resp = HTTPResponse(200, body, headers, url, url, charset='utf-8') p = SGMLParser(resp) tags = p.get_tags_by_filter(None) tags = list(tags) tag_names = [tag.name for tag in tags] self.assertEqual(tag_names, ['html', 'body', 'a', 'div'])