示例#1
0
    def test_load_whitelist(self):
        d = {"generic": {"path": "conf/whitelists/generic.example.yml"}}
        results = utils.load_whitelist(d)
        self.assertIsInstance(results, set)
        self.assertIn("google.com", results)
        self.assertIn("amazon.com", results)
        self.assertIn("facebook.com", results)

        d = {
            "generic": {
                "path": "conf/whitelists/generic.example.yml",
                "expiry": None
            }
        }
        results = utils.load_whitelist(d)
        self.assertIsInstance(results, set)
        self.assertIn("google.com", results)
        self.assertIn("amazon.com", results)
        self.assertIn("facebook.com", results)

        d = {
            "generic": {
                "path": "conf/whitelists/generic.example.yml",
                "expiry": "2016-06-28T12:33:00.000Z"
            }
        }
        results = utils.load_whitelist(d)
        self.assertIsInstance(results, set)
        self.assertEqual(len(results), 0)
示例#2
0
    def test_reformat_urls(self):

        body = """
        bla bla https://tweetdeck.twitter.com/random bla bla
        http://kafka.apache.org/documentation.html
        http://kafka.apache.org/documentation1.html
        bla bla bla https://docs.python.org/2/library/re.html bla bla
        bla bla bla https://docs.python.org/2/library/re_2.html> bla bla
        <p>https://tweetdeck.twitter.com/random</p> bla bla
        <p>https://tweetdeck.twitter.com/random_2</p>
        """

        d = {"generic": {"path": "conf/whitelists/generic.example.yml"}}
        whitelist = utils.load_whitelist(d)
        urls = utils.text2urls_whitelisted(body, whitelist, self.faup)
        self.assertIsInstance(urls, dict)

        urls = utils.reformat_urls(urls)
        self.assertIsInstance(urls, list)

        with self.assertRaises(TypeError):
            utils.reformat_urls(dict)
示例#3
0
    def test_text2urls_whitelisted(self):

        body = """
        bla bla https://tweetdeck.twitter.com/random bla bla
        http://kafka.apache.org/documentation.html
        http://kafka.apache.org/documentation1.html
        bla bla bla https://docs.python.org/2/library/re.html bla bla
        bla bla bla https://docs.python.org/2/library/re_2.html> bla bla
        <p>https://tweetdeck.twitter.com/random</p> bla bla
        <p>https://tweetdeck.twitter.com/random_2</p>
        """

        d = {"generic": {"path": "conf/whitelists/generic.example.yml"}}
        whitelist = utils.load_whitelist(d)
        urls = utils.text2urls_whitelisted(body, whitelist, self.faup)

        self.assertIsInstance(urls, dict)
        self.assertNotIn("apache.org", urls)
        self.assertIn("python.org", urls)
        self.assertIsInstance(urls["python.org"], list)
        self.assertIn("twitter.com", urls)
        self.assertIsInstance(urls["twitter.com"], list)