示例#1
0
    def test__process_html_link(self):
        c = Crawler("http://test.com")
        soup = BeautifulSoup(self.html_test_string)

        for link in soup.find_all("a"):
            c._process_html_link(link, "/")

        self.assertEqual(len(c.sitemap.nodes()), 3)
        self.assertEqual(len(c.sitemap.edges()), 2)
        self.assertEqual(len(c.process_q), 3)
示例#2
0
    def test__process_html(self):
        soup = BeautifulSoup(self.html_test_string)
        c = Crawler("http://test.com")
        c._process_html_asset = mock.Mock()
        c._process_html_link = mock.Mock()

        c._process_html(soup)
        self.assertEqual(c._process_html_asset.call_count, 3)
        self.assertEqual(c._process_html_link.call_count, 4)