def test_html_cleaner_strings(self):
r = HtmlCleaner()
docs = list(r.process([self.d1["body"]]))
words = docs[0]
self.assertEqual(len(words), 53)
self.assertEqual(words[33], ".")
self.assertEqual(words[34], "\n")
self.assertEqual(words[35], " ")
def test_html_cleaner_docs(self):
r = HtmlCleaner()
docs = list(r.process(self.docs))
words = list(docs[0].words())
self.assertEqual(len(words), 11)
self.assertEqual(words[6], "pipeline")
self.assertEqual(words[7], ".")
self.assertEqual(words[8], "another")
