def testBasicNamedEntityChildParent(self):
tagList = ['LOCATION', 'PERSON', 'NUMBER', 'MONEY', 'MISC']
for tag in tagList:
parent = DependenciesTree('parent', 1, namedEntityTag = tag)
child = DependenciesTree('child', 2, parent = parent, namedEntityTag = tag)
parent.child.append(child)
child.dependency = 'conj_and'
NamedEntityMerging(parent).merge()
self.assertEqual(parent.wordList, [Word('parent', 1)])
self.assertEqual(parent.child, [child])
self.assertEqual(child.parent, parent)
child.dependency = 'foo'
NamedEntityMerging(parent).merge()
self.assertIn(Word('parent', 1), parent.wordList)
self.assertIn(Word('child', 2), parent.wordList)
self.assertEqual(parent.child, [])
for (tag1, tag2) in itertools.permutations(tagList, 2):
parent = DependenciesTree('parent', 1, namedEntityTag = tag1)
child = DependenciesTree('child', 2, parent = parent, namedEntityTag = tag2)
parent.child.append(child)
child.dependency = 'conj_and'
NamedEntityMerging(parent).merge()
self.assertEqual(parent.wordList, [Word('parent', 1)])
self.assertEqual(parent.child, [child])
self.assertEqual(child.parent, parent)
child.dependency = 'foo'
NamedEntityMerging(parent).merge()
self.assertEqual(parent.wordList, [Word('parent', 1)])
self.assertEqual(parent.child, [child])
self.assertEqual(child.parent, parent)
def tripleProductionData():
'''
Return data corresponding to a tree (root-0)--dep-->(child-1)
'''
root = DependenciesTree("root-0")
child = DependenciesTree("child-1",dependency="dep",parent=root)
root.child = [child]
nodeToID = {root:0, child:1}
bt = TriplesBucket()
return (root,nodeToID,bt)
def testBasicTreeConstructor(self):
n = DependenciesTree('foo', 1)
self.assertEqual(n.wordList, [Word('foo', 1)])
self.assertEqual(n.namedEntityTag, 'undef')
self.assertEqual(n.dependency, 'undef')
self.assertEqual(n.child, [])
self.assertEqual(n.text, "")
self.assertEqual(n.parent, None)
self.assertEqual(n.subtreeType, 'undef')
self.assertEqual(n.dfsTag, 0)
self.assertFalse(n.isVerb())
self.assertFalse(n.isNoun())
n.appendWord('bar')
self.assertEqual(str(DependenciesTree('foo bar', 1)), str(n))
def testAddNamedEntityTag3(self):
foo1 = DependenciesTree('foo1', 1, namedEntityTag='42')
foo2 = DependenciesTree('foo2', 3, namedEntityTag='42')
bar = DependenciesTree('bar',
2,
namedEntityTag='undef',
dependency='amod',
parent=foo1)
generator = TreeGenerator(None)
generator.nameToNodes = {
('foo1', 1): foo1,
('bar', 2): bar,
('foo2', 3): foo2
}
generator._correctTree(foo1)
self.assertEqual(bar.namedEntityTag, 'undef')
def testTreePos(self):
n = DependenciesTree('foo', 1)
n.wordList += [Word('eat', 2, 'VB'), Word('bar', 3)]
self.assertTrue(n.isVerb())
self.assertFalse(n.isNoun())
n = DependenciesTree('foo', 1)
n.wordList += [Word('broomstick', 2, 'NN'), Word('bar', 3)]
self.assertFalse(n.isVerb())
self.assertTrue(n.isNoun())
def testBasicPrepositionEdge(self):
for prep in ['in', 'of', 'with', 'by']:
parent = DependenciesTree('parent', 1)
parent.wordList[0].pos = 'VB'
child = DependenciesTree('child', 2, parent = parent, dependency = 'prep_'+prep)
parent.child.append(child)
PrepositionMerging(parent).merge()
self.assertEqual(parent.wordList, [Word('parent '+prep, 1, 'VB')])
self.assertEqual(parent.child, [child])
self.assertEqual(child.dependency, 'prep')
parent = DependenciesTree('parent', 1)
parent.wordList[0].pos = 'VB'
child = DependenciesTree('child', 2, parent = parent, dependency = 'agent')
parent.child.append(child)
PrepositionMerging(parent).merge()
self.assertEqual(parent.wordList, [Word('parent by', 1, 'VB')])
self.assertEqual(parent.child, [child])
def testTreePos(self):
n = DependenciesTree('foo', 1)
n.wordList += [Word('eat', 2, 'VB'), Word('bar', 3)]
self.assertTrue(n.isVerb())
self.assertFalse(n.isNoun())
n = DependenciesTree('foo', 1)
n.wordList += [Word('broomstick', 2, 'NN'), Word('bar', 3)]
self.assertFalse(n.isVerb())
self.assertTrue(n.isNoun())
def testBasicTreeConstructor(self):
n = DependenciesTree('foo', 1)
self.assertEqual(n.wordList, [Word('foo', 1)])
self.assertEqual(n.namedEntityTag, 'undef')
self.assertEqual(n.dependency, 'undef')
self.assertEqual(n.child, [])
self.assertEqual(n.text, "")
self.assertEqual(n.parent, None)
self.assertEqual(n.subtreeType, 'undef')
self.assertEqual(n.dfsTag, 0)
self.assertFalse(n.isVerb())
self.assertFalse(n.isNoun())
n.appendWord('bar')
self.assertEqual(str(DependenciesTree('foo bar', 1)), str(n))
def testBasicPrepositionNode(self):
parent = DependenciesTree('parent', 1)
child = DependenciesTree('child', 2, parent = parent, dependency = 'foo')
parent.child.append(child)
PrepositionMerging(parent).merge()
self.assertEqual(parent.wordList, [Word('parent', 1)])
self.assertEqual(parent.child, [child])
self.assertEqual(child.parent, parent)
for prep in PrepositionMerging.prepositionSet:
parent = DependenciesTree('parent', 1)
child = DependenciesTree(prep, 2, parent = parent, dependency = 'foo')
parent.child.append(child)
child.dependency = 'conj_and'
PrepositionMerging(parent).merge()
self.assertIn(Word('parent', 1), parent.wordList)
self.assertIn(Word(prep, 2), parent.wordList)
self.assertEqual(parent.child, [])
def testMerge(self):
root1 = DependenciesTree('root', 1)
root2 = DependenciesTree('root', 2)
node1 = DependenciesTree('n', 1, 'tag1', 'stype1', 'dep1', [DependenciesTree('childn', 1)])
node1.parent = root1
root1.child += [node1]
node2 = DependenciesTree('n', 2, 'tag2', 'stype2', 'dep2', [DependenciesTree('childn', 2)])
node2.parent = root2
root2.child += [node2]
node1.merge(node2, True)
self.assertEqual(len(root2.child), 0)
self.assertEqual(len(root1.child), 1)
self.assertEqual(len(node1.child), 2)
self.assertEqual(node1.wordList, [Word('n', 1), Word('n', 2)])
self.assertEqual(node1.namedEntityTag, 'tag1')
self.assertEqual(node1.dependency, 'dep1')
self.assertEqual(node1.parent, root1)
self.assertEqual(node1.subtreeType, 'stype1')
self.assertEqual(node1.dfsTag, 0)
def testBasicNamedEntitySisterBrother(self):
tagList = ['LOCATION', 'PERSON', 'NUMBER', 'MONEY', 'MISC']
for tag in tagList:
parent = DependenciesTree('parent', 1, namedEntityTag = 'undef')
child1 = DependenciesTree('child1', 2, parent = parent, dependency = 'conj_and', namedEntityTag = tag)
child2 = DependenciesTree('child2', 3, parent = parent, dependency = 'conj_and', namedEntityTag = tag)
parent.child += [child1, child2]
NamedEntityMerging(parent).merge()
self.assertEqual(parent.wordList, [Word('parent', 1)])
self.assertEqual(parent.child, [child1, child2])
self.assertEqual(child1.parent, parent)
self.assertEqual(child2.parent, parent)
child1.dependency = 'foo'
child2.dependency = 'foo'
NamedEntityMerging(parent).merge()
self.assertEqual(parent.wordList, [Word('parent', 1)])
self.assertEqual(len(parent.child), 1)
self.assertIn(Word('child1', 2), parent.child[0].wordList)
self.assertIn(Word('child2', 3), parent.child[0].wordList)
self.assertEqual(parent.child[0].parent, parent)
for (tag1, tag2) in itertools.permutations(tagList, 2):
parent = DependenciesTree('parent', 1, namedEntityTag = 'undef')
child1 = DependenciesTree('child1', 2, parent = parent, dependency = 'conj_and', namedEntityTag = tag1)
child2 = DependenciesTree('child2', 3, parent = parent, dependency = 'conj_and', namedEntityTag = tag2)
parent.child += [child1, child2]
NamedEntityMerging(parent).merge()
self.assertEqual(parent.wordList, [Word('parent', 1)])
self.assertEqual(parent.child, [child1, child2])
self.assertEqual(child1.parent, parent)
self.assertEqual(child2.parent, parent)
child1.dependency = 'foo'
child2.dependency = 'foo'
NamedEntityMerging(parent).merge()
self.assertEqual(parent.wordList, [Word('parent', 1)])
self.assertEqual(parent.child, [child1, child2])
self.assertEqual(child1.parent, parent)
self.assertEqual(child2.parent, parent)
def testMerge(self):
root1 = DependenciesTree('root', 1)
root2 = DependenciesTree('root', 2)
node1 = DependenciesTree('n', 1, 'tag1', 'stype1', 'dep1',
[DependenciesTree('childn', 1)])
node1.parent = root1
root1.child += [node1]
node2 = DependenciesTree('n', 2, 'tag2', 'stype2', 'dep2',
[DependenciesTree('childn', 2)])
node2.parent = root2
root2.child += [node2]
node1.merge(node2, True)
self.assertEqual(len(root2.child), 0)
self.assertEqual(len(root1.child), 1)
self.assertEqual(len(node1.child), 2)
self.assertEqual(node1.wordList, [Word('n', 1), Word('n', 2)])
self.assertEqual(node1.namedEntityTag, 'tag1')
self.assertEqual(node1.dependency, 'dep1')
self.assertEqual(node1.parent, root1)
self.assertEqual(node1.subtreeType, 'stype1')
self.assertEqual(node1.dfsTag, 0)