def test_strip_tags(self): # Assert HTML parser and tag stripper. for html, plain in ( (u"<b>ünîcøde</b>", u"ünîcøde"), ( "<img src=""/>", ""), ( "<p>text</p>", "text\n\n"), ( "<li>text</li>", "* text\n"), ( "<td>text</td>", "text\t"), ( "<br /><br/><br>", "\n\n\n")): self.assertEqual(web.strip_tags(html), plain) # Assert exclude tags and attributes v = web.strip_tags("<a href=\"\" onclick=\"\">text</a>", exclude={"a": ["href"]}) self.assertEqual(v, "<a href=\"\">text</a>") print "pattern.web.strip_tags()"