def test_extract_links(self):
response = '<a>http://tomato.org/test</a><b>http://tomato.org/test</b><a>http://tomato.com/test</a>'
links = extract_links(response)
self.assertEquals(links, ['http://tomato.org/test', 'http://tomato.com/test'])
def test_extract_links_with_blacklist_extensions(self):
response = '<a>http://tomato.org/test</a><b>http://tomato.org/test</b><a>http://tomato.com/test</a>'
links = extract_links(response, blacklist_extensions=['.org'])
self.assertEquals(links, ['http://tomato.com/test'])
def test_extract_links_with_regex(self):
response = '<a>http://tomato.org/test</a><b>http://tomato.org/test</b><a>http://tomato.com/test</a>'
links = extract_links(response, regex='tomato\.org')
self.assertEquals(links, ['http://tomato.org/test'])
def test_extract_links_with_whitelist_domains(self):
response = '<a>http://tomato.org/test</a><b>http://tomato.org/test</b><a>http://tomato.com/test</a>'
links = extract_links(response, whitelist_domains=['tomato.org'])
self.assertEquals(links, ['http://tomato.org/test'])
def test_extract_links_common_white_and_black_lists(self):
response = '<a>http://tomato.org/test</a><b>http://tomato.org/test</b><a>http://tomato.com/test</a>'
with self.assertRaises(LinkExtractorException):
extract_links(response, whitelist_domains=['test'], blacklist_domains=['test'])
