def test_prune_failures(self):
t = PrefTree()
for url, success in [
('sciencedirect.com/paper1.pdf', False),
('sciencedirect.com/paper2.pdf', False),
('sciencedirect.com/paper3.pdf', False),
('mysciencework.com/paper9.pdf', True),
('mysciencework.com/paper8.pdf', False)]:
t.add_url(url, success)
t, pruned = t.prune(min_rate=0.95, min_children=2, min_urls=2)
self.assertEqual(t.match('sciencedirect.com/paper4.pdf'), (3,0))
self.assertFalse(t.predict_success('sciencedirect.com/paper4.pdf'))
def test_prune(self):
t = PrefTree()
with self.assertRaises(ValueError):
t.prune(min_urls=0)
for url, success in [
('arxiv.org/pdf/1410.1234', True),
('arxiv.org/pdf/1409.1094', True),
('arxiv.org/pdf/1201.5480', True),
('arxiv.org/pdf/1601.01234', True),
('arxiv.org/pdf/1602.01i34', False), # oops
]:
t.add_url(url, success)
t, pruned = t.prune(min_rate=0.75,min_children=2,min_urls=1)
self.assertEqual(len(t.urls()), 1)
self.assertTrue(t.has_wildcard())
self.assertEqual(t.match('arxiv.org/pdf/1784.1920'), (5,4))
self.assertEqual(t.match('arxiv.org/pdf/2340.0124'), (0,0))
self.assertTrue(t.predict_success('arxiv.org/pdf/1784.1920', threshold=0.6, min_urls=3))
t.print_as_tree()
