def test_should_treat_relative_links_properly(self):
html = '''
<a href="/relative">Relative</a>
<a href="http://test.com/absolule">Absolute</a>
'''
soup = Soup(markup=html)
solution_one.discover_links(url='', soup=soup)
self.assertEquals(2, solution_one.queue.qsize())
self.assertEquals(2, len(solution_one.discovered))
def test_should_add_allowed_urls_to_queue_and_discovered(self):
html = '''
<a href="/valid">Valid</a>
<a href="#invalid">Invalid</a>
<a href="mailto://[email protected]">Invalid Too</a>
<a href="/also-valid">Also Valid</a>
'''
soup = Soup(markup=html)
solution_one.discover_links(url='', soup=soup)
self.assertEquals(2, solution_one.queue.qsize())
self.assertEquals(2, len(solution_one.discovered))
def test_should_ignore_links_starting_with_hashtag(self):
html = '''<a href="#top">Top</a>'''
soup = Soup(markup=html)
solution_one.discover_links(url='', soup=soup)
self.assertEquals(0, solution_one.queue.qsize())
self.assertEquals(0, len(solution_one.discovered))
def test_should_ignore_links_without_href(self):
html = '''<a class='href'>no href</a>'''
soup = Soup(markup=html)
solution_one.discover_links(url='', soup=soup)
self.assertEquals(0, solution_one.queue.qsize())
self.assertEquals(0, len(solution_one.discovered))
