def test_crawl_limit(self):
c = Crawler("http://a.com")
c.SLEEP_TIME = 0
def side_effect():
c.process_q.pop(0)
c._process_next_url = mock.Mock(side_effect=side_effect)
c.render_sitemap = mock.Mock()
c.URL_LIMIT = 10
c.process_q = ["test"] * 5
c.crawl()
self.assertEqual(c._process_next_url.call_count, 5)
c._process_next_url.call_count = 0
c.process_q = ["test"] * 10
c.URL_LIMIT = 5
c.crawl()
self.assertEqual(c._process_next_url.call_count, 5)
c._process_next_url.call_count = 0
c.process_q = ["test"] * 10
c.URL_LIMIT = float("inf")
c.crawl()
self.assertEqual(c._process_next_url.call_count, 10)
def test__process_next_url_blacklist(self):
c = Crawler("http://a.com")
c.bad_urls = {"http://a.com/a/b/c/": True}
c.process_q.append("http://a.com/a/b/c/")
c._make_request = mock.Mock(return_value=None)
c._process_html = mock.Mock()
c._process_next_url()
self.assertEqual(len(c.process_q), 1)
self.assertEqual(len(c.bad_urls), 2)
c._process_next_url()
self.assertEqual(len(c.process_q), 0)
self.assertEqual(len(c.bad_urls), 2)
self.assertEqual(c._process_html.call_count, 0)
