def test_find_all_follow_on_links(self):
all_links = hundredgreatest.get_links_in_soup(self.soup)
follow_on_links_numbers = []
all_links_as_str = [str(l) for l in all_links]
pattern = re.compile(r"^<a.*(\S+)(\?page=)(\d)\S+</a>")
for l in all_links_as_str:
m = re.match(pattern, l)
if m:
follow_on_links_numbers.append(m.group(3))
theset = set(follow_on_links_numbers)
self.assertEqual(hundredgreatest.get_follow_on_link_numbers(self.soup), theset)
def test_get_all_links_as_strings(self):
links_as_strings = set()
for l in hundredgreatest.get_links_in_soup(self.soup):
links_as_strings.add(str(l))
self.assertEqual(hundredgreatest.all_links_as_strings(self.soup), links_as_strings)
def test_get_links_in_soup(self):
links = self.soup.find_all("a")
self.assertEqual(hundredgreatest.get_links_in_soup(self.soup), links)