def test_unpaired_urls_link_after_table(self): fp = '../tests/test-posts/link_after_table.htm' text = read_file(fp) elements = detect_pcpp_html_elements(text) urls = elements['anon'] # No identifiable links unpaired_urls = get_urls_with_no_table(urls, elements['tables']) self.assertEqual(0, len(unpaired_urls))
def test_unpaired_urls_iden_anon_same_euro(self): fp = '../tests/test-posts/same_anon_iden_links_euro.htm' text = read_file(fp) elements = detect_pcpp_html_elements(text) all_anon_urls, iden_anon_urls = combine_iden_anon_urls( elements['anon'], elements['identifiable'], self.pcpp) self.assertEqual(1, len(all_anon_urls)) self.assertEqual('https://dk.pcpartpicker.com/list/ZqWwj2', all_anon_urls[0])
def test_unpaired_urls_table_euro(self): fp = '../tests/test-posts/table_euro.htm' text = read_file(fp) elements = detect_pcpp_html_elements(text) all_anon_urls, iden_anon_urls = combine_iden_anon_urls( elements['anon'], elements['identifiable'], self.pcpp) remaining_urls = get_urls_with_no_table(all_anon_urls, elements['tables']) self.assertEqual(0, len(remaining_urls))
def test_detect_pcpp_html_iden_view_no_table(self): fp = '../tests/test-posts/iden_link_view.htm' text = read_file(fp) elements = detect_pcpp_html_elements(text) self.assertEqual(1, len(elements['identifiable'])) expected_iden = 'https://pcpartpicker.com/user/haydenholton/saved/szvVWZ' self.assertEqual(expected_iden, elements['identifiable'][0]) self.assertEqual(0, len(elements['anon'])) self.assertEqual(0, len(elements['tables']))
def test_detect_pcpp_html_table_broken(self): fp = '../tests/test-posts/table_broken.htm' text = read_file(fp) elements = detect_pcpp_html_elements(text) self.assertEqual(0, len(elements['identifiable'])) self.assertEqual(1, len(elements['anon'])) expected_anon = 'https://pcpartpicker.com/list/9qMcNP' self.assertEqual(expected_anon, elements['anon'][0]) self.assertEqual(0, len(elements['tables']))
def test_detect_pcpp_html_table_euro(self): fp = '../tests/test-posts/table_euro.htm' text = read_file(fp) elements = detect_pcpp_html_elements(text) self.assertEqual(0, len(elements['identifiable'])) self.assertEqual(1, len(elements['anon'])) expected_anon = 'https://be.pcpartpicker.com/list/KytcBc' self.assertEqual(expected_anon, elements['anon'][0]) self.assertEqual(1, len(elements['tables'])) self.assertTrue(elements['tables'][0].is_valid())
def test_unpaired_urls_unpaired_url_before_after(self): fp = '../tests/test-posts/unpaired_link_before_after.htm' text = read_file(fp) elements = detect_pcpp_html_elements(text) urls = elements['anon'] # No identifiable links self.assertEqual(2, len(urls)) unpaired_urls = get_urls_with_no_table(urls, elements['tables']) expected_url = "https://mx.pcpartpicker.com/list/ZGFq7X" self.assertEqual(1, len(unpaired_urls)) self.assertEqual(expected_url, unpaired_urls[0])
def test_unpaired_urls_table_iden_only(self): fp = '../tests/test-posts/table_iden_only.htm' text = read_file(fp) elements = detect_pcpp_html_elements(text) anon_urls = elements['anon'] iden_urls = elements['identifiable'] all_urls, iden_anon_urls = combine_iden_anon_urls( anon_urls, iden_urls, self.pcpp) self.assertEqual(1, len(all_urls)) rem_urls = get_urls_with_no_table(all_urls, elements['tables']) self.assertEqual(0, len(rem_urls))