Ejemplo n.º 1
0
 def test_next_url_new_domain_block(self):
     my_example = "https://my.example.com/"
     www_example = "https://www.example.com/"
     schedule = Scheduler(my_example) # Constructor calls upon schedule_url
     schedule.schedule_url(www_example)
     self.assertEqual(schedule.next_url(), my_example)
     self.assertEqual(schedule.next_url(), www_example)
Ejemplo n.º 2
0
 def test_schedule_url_already_crawled(self):
     url = "https://my.example.com/path/to/location"
     schedule = Scheduler(url) # Constructor calls upon schedule_url
     schedule.next_url()
     self.assertFalse(schedule.schedule_url(url))
Ejemplo n.º 3
0
 def test_next_url_extensions(self):
     schedule = Scheduler("https://my.example.com/path/to/location") # Constructor calls upon schedule_url
     schedule.schedule_url("https://my.example.com/path/to/location?key=val;word=bird#frag")
     correct_output = "https://my.example.com/path/to/location?key=val;word=bird#frag" 
     self.assertEqual(schedule.next_url(), correct_output)
Ejemplo n.º 4
0
 def test_next_url_empty_queue(self):
     schedule = Scheduler("https://my.example.com/path/to/location") # Constructor calls upon schedule_url
     schedule.next_url()
     self.assertFalse(schedule.next_url())
Ejemplo n.º 5
0
 def test_next_url(self):
     schedule = Scheduler("https://my.example.com/path/to/location") # Constructor calls upon schedule_url
     correct_output = "https://my.example.com/path/to/location"
     self.assertEqual(schedule.next_url(), correct_output)