def test_accessors(self):
t = PrefTree()
urls = ['arxiv.org/abs/1410.1454','arxiv.org/pdf/1410.1454v2']
for u in urls:
t.add_url(u, True)
self.assertEqual(t['arxiv.org/'], t[hashable_list('arxiv.org/')])
t['biorxiv.org/'] = PrefTree()
t[hashable_list('sciencedirect.com/')] = PrefTree()
del t[hashable_list('biorxiv.org/')]
del t['sciencedirect.com/']
def test_create(self):
t = PrefTree()
urls = ['aaba','cadb','abdc','abcd','afgh','abec']
for u in urls:
t.add_url(u)
self.assertTrue(t.check_sanity())
self.assertFalse(t.has_wildcard())
t.print_as_tree()
self.assertEqual(sorted([flatten(u) for u, c, s in t.urls()]), sorted(urls))
for u in urls:
self.assertEqual(t.match(u), (1,0))
self.assertEqual(t.match('bac'), (0,0))
def test_match(self):
t = PrefTree()
urls = [
('aab/ced',True),
('aab/ru',False),
('bcadr/ste',False),
('aab/t/est',True),
('aab/t/le',True),
('aab/stts',False),
]
for u, s in urls:
t.add_url(u, s)
t, pruned = t.prune(min_rate=0.9, min_children=2,min_urls=1)
c, s, b = t.match_with_branch('aab/t/lu')
self.assertEqual((c,s),(2,2))
self.assertEqual(''.join(b), 'aab/t/*')
c, s, b = t.match_with_branch('aab/xx')
self.assertEqual((c,s),(0,0))
self.assertEqual(''.join(b), 'aab/<unk>')
c, s, b = t.match_with_branch('aab/ced')
self.assertEqual((c,s),(1,1))
self.assertEqual(''.join(b), 'aab/ced')
def test_prune_with_reverse(self):
t = PrefTree()
for url, success in [
('researchgate.net/publication/233865122_uriset', False),
('researchgate.net/publication/143874230_albtedru', False),
('researchgate.net/publication/320748374_kelbcad', False),
('researchgate.net/publication/233865122_uriset.pdf', True),
('researchgate.net/publication/143874230_albtedru.pdf', True),
('researchgate.net/publication/320748374_kelbcad.pdf', True),
('onlinelibrary.wiley.com/wol1/doi/10.1002/anie.200800037.abstract', False),
('onlinelibrary.wiley.com/wol1/doi/10.1002/anie.200800037.pdf', False)]:
t.add_url(url, success)
t.print_as_tree()
t, pruned = t.prune(reverse=True)
t.print_as_tree()
self.assertTrue(t.check_sanity())
for u, c, s in t.urls():
print flatten(u), c, s
def test_check_sanity(self):
t = PrefTree()
t['abc'] = PrefTree()
t['aa'] = PrefTree()
self.assertFalse(t.check_sanity())
t = PrefTree(url_count=-1)
self.assertFalse(t.check_sanity())
t = PrefTree()
t['abc'] = PrefTree()
t.is_wildcard = True
self.assertFalse(t.check_sanity())
t2 = PrefTree()
t2['def'] = t
self.assertFalse(t2.check_sanity())
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()
def test_prune_empty_tree(self):
t = PrefTree()
t.prune() # should not raise anything
def test_with_tokenization(self):
t = PrefTree()
t.add_url(prepare_url('eprint.iacr.org/2016/093'), False)
t.add_url(prepare_url('eprint.iacr.org/2016/093.pdf'), True)
t.add_url(prepare_url('eprint.iacr.org/2015/1248.pdf'), True)
t.add_url(prepare_url('eprint.iacr.org/2015/1248'), False)
t.print_as_tree()
self.assertEqual(t.match(prepare_url('eprint.iacr.org/2014/528.pdf')), (2,2))
self.assertEqual(t.match(prepare_url('eprint.iacr.org/2014/528')), (2,0))
def test_wildcard(self):
t = PrefTree()
t.add_url('arxiv.org/pdf/', True)
t['arxiv.org/pdf/'].is_wildcard = True
self.assertTrue(t.check_sanity())
self.assertTrue(t.has_wildcard())
self.assertEqual(t.match('arxiv.org/pdf/1410.1454v2'), (1,1))
t.add_url('arxiv.org/pdf/1412.8548v1', True)
self.assertEqual(t.match('arxiv.org/pdf/1410.1454v2'), (2,2))
t.print_as_tree()
self.assertEqual(len(t.urls()), 1)
def test_urls(self):
t = PrefTree()
urls = ['aa','aa.pdf','bb','bb.pdf']
for u in urls:
t.add_url(u)
self.assertEqual(len(t.urls()), 4)
def test_empty(self):
t = PrefTree()
self.assertTrue(t.check_sanity())