1
|
1 |
|
import json |
2
|
1 |
|
from nltk.stem.wordnet import WordNetLemmatizer |
3
|
1 |
|
from ppp_questionparsing_grammatical import Word, DependenciesTree, TreeGenerator, computeTree |
4
|
1 |
|
import data |
5
|
|
|
|
6
|
1 |
|
from unittest import TestCase |
7
|
|
|
|
8
|
1 |
|
class DependenciesTreeTests(TestCase): |
9
|
|
|
|
10
|
|
|
######## |
11
|
|
|
# Word # |
12
|
|
|
######## |
13
|
|
|
|
14
|
1 |
|
def testBasicWordConstructor1(self): |
15
|
1 |
|
w=Word('foo', 1, 'bar') |
16
|
1 |
|
self.assertEqual(w.word, 'foo') |
17
|
1 |
|
self.assertEqual(w.index, 1) |
18
|
1 |
|
self.assertEqual(w.pos, 'bar') |
19
|
1 |
|
self.assertEqual(str(w), "(foo, 1, bar)") |
20
|
1 |
|
w.append('aaa') |
21
|
1 |
|
self.assertEqual(Word('foo aaa', 1, 'bar'), w) |
22
|
|
|
|
23
|
1 |
|
def testPOS(self): |
24
|
1 |
|
for pos in {'VB', 'VBD', 'VBG', 'VBN', 'VBP', 'VBZ'}: |
25
|
1 |
|
w = Word('foo', 1, pos) |
26
|
1 |
|
self.assertTrue(w.isVerb()) |
27
|
1 |
|
self.assertFalse(w.isNoun()) |
28
|
1 |
|
for pos in {'NN', 'NNS', 'NNP', 'NNPS'}: |
29
|
1 |
|
w = Word('foo', 1, pos) |
30
|
1 |
|
self.assertFalse(w.isVerb()) |
31
|
1 |
|
self.assertTrue(w.isNoun()) |
32
|
|
|
|
33
|
|
|
################### |
34
|
|
|
# Dependency tree # |
35
|
|
|
################### |
36
|
|
|
|
37
|
1 |
|
def testBasicTreeConstructor(self): |
38
|
1 |
|
n = DependenciesTree('foo', 1) |
39
|
1 |
|
self.assertEqual(n.wordList, [Word('foo', 1)]) |
40
|
1 |
|
self.assertEqual(n.namedEntityTag, 'undef') |
41
|
1 |
|
self.assertEqual(n.dependency, 'undef') |
42
|
1 |
|
self.assertEqual(n.child, []) |
43
|
1 |
|
self.assertEqual(n.text, "") |
44
|
1 |
|
self.assertEqual(n.parent, None) |
45
|
1 |
|
self.assertEqual(n.subtreeType, 'undef') |
46
|
1 |
|
self.assertEqual(n.dfsTag, 0) |
47
|
1 |
|
self.assertFalse(n.isVerb()) |
48
|
1 |
|
self.assertFalse(n.isNoun()) |
49
|
1 |
|
n.appendWord('bar') |
50
|
1 |
|
self.assertEqual(str(DependenciesTree('foo bar', 1)), str(n)) |
51
|
|
|
|
52
|
1 |
|
def testTreePos(self): |
53
|
1 |
|
n = DependenciesTree('foo', 1) |
54
|
1 |
|
n.wordList += [Word('eat', 2, 'VB'), Word('bar', 3)] |
55
|
1 |
|
self.assertTrue(n.isVerb()) |
56
|
1 |
|
self.assertFalse(n.isNoun()) |
57
|
1 |
|
n = DependenciesTree('foo', 1) |
58
|
1 |
|
n.wordList += [Word('broomstick', 2, 'NN'), Word('bar', 3)] |
59
|
1 |
|
self.assertFalse(n.isVerb()) |
60
|
1 |
|
self.assertTrue(n.isNoun()) |
61
|
|
|
|
62
|
|
|
|
63
|
|
|
############### |
64
|
|
|
# computeTree # |
65
|
|
|
############### |
66
|
|
|
|
67
|
1 |
|
def testStr1(self): |
68
|
1 |
|
tree=computeTree(data.give_john_smith()) |
69
|
1 |
|
self.maxDiff=None |
70
|
1 |
|
tree.sort() |
71
|
1 |
|
self.assertEqual(str(tree), data.give_john_smith_string()) |
72
|
|
|
|
73
|
|
|
############### |
74
|
|
|
# Merge # |
75
|
|
|
############### |
76
|
1 |
|
def testMerge(self): |
77
|
1 |
|
root1 = DependenciesTree('root', 1) |
78
|
1 |
|
root2 = DependenciesTree('root', 2) |
79
|
1 |
|
node1 = DependenciesTree('n', 1, 'tag1', 'stype1', 'dep1', [DependenciesTree('childn', 1)]) |
80
|
1 |
|
node1.parent = root1 |
81
|
1 |
|
root1.child += [node1] |
82
|
1 |
|
node2 = DependenciesTree('n', 2, 'tag2', 'stype2', 'dep2', [DependenciesTree('childn', 2)]) |
83
|
1 |
|
node2.parent = root2 |
84
|
1 |
|
root2.child += [node2] |
85
|
1 |
|
node1.merge(node2, True) |
86
|
1 |
|
self.assertEqual(len(root2.child), 0) |
87
|
1 |
|
self.assertEqual(len(root1.child), 1) |
88
|
1 |
|
self.assertEqual(len(node1.child), 2) |
89
|
1 |
|
self.assertEqual(node1.wordList, [Word('n', 1), Word('n', 2)]) |
90
|
1 |
|
self.assertEqual(node1.namedEntityTag, 'tag1') |
91
|
1 |
|
self.assertEqual(node1.dependency, 'dep1') |
92
|
1 |
|
self.assertEqual(node1.parent, root1) |
93
|
1 |
|
self.assertEqual(node1.subtreeType, 'stype1') |
94
|
1 |
|
self.assertEqual(node1.dfsTag, 0) |
95
|
|
|
|
96
|
|
|
############### |
97
|
|
|
# correctTree # |
98
|
|
|
############### |
99
|
|
|
|
100
|
1 |
|
def testAddNamedEntityTag1(self): |
101
|
1 |
|
foo1 = DependenciesTree('foo1', 1, namedEntityTag='42') |
102
|
1 |
|
foo2 = DependenciesTree('foo2', 3, namedEntityTag='42') |
103
|
1 |
|
bar = DependenciesTree('bar', 2, namedEntityTag='undef', dependency = 'nn', parent = foo1) |
104
|
1 |
|
generator = TreeGenerator(None) |
105
|
1 |
|
generator.nameToNodes = {('foo1',1) : foo1, ('bar',2) : bar, ('foo2', 3) : foo2} |
106
|
1 |
|
generator._correctTree(foo1) |
107
|
1 |
|
self.assertEqual(bar.namedEntityTag, '42') |
108
|
|
|
|
109
|
1 |
|
def testAddNamedEntityTag2(self): |
110
|
1 |
|
foo1 = DependenciesTree('foo1', 1, namedEntityTag='42') |
111
|
1 |
|
foo2 = DependenciesTree('foo2', 3, namedEntityTag='42') |
112
|
1 |
|
bar = DependenciesTree('bar', 2, namedEntityTag='27', dependency = 'nn', parent = foo1) |
113
|
1 |
|
generator = TreeGenerator(None) |
114
|
1 |
|
generator.nameToNodes = {('foo1',1) : foo1, ('bar',2) : bar, ('foo2', 3) : foo2} |
115
|
1 |
|
generator._correctTree(foo1) |
116
|
1 |
|
self.assertEqual(bar.namedEntityTag, '27') |
117
|
|
|
|
118
|
1 |
|
def testAddNamedEntityTag3(self): |
119
|
1 |
|
foo1 = DependenciesTree('foo1', 1, namedEntityTag='42') |
120
|
1 |
|
foo2 = DependenciesTree('foo2', 3, namedEntityTag='42') |
121
|
1 |
|
bar = DependenciesTree('bar', 2, namedEntityTag='undef', dependency = 'amod', parent = foo1) |
122
|
1 |
|
generator = TreeGenerator(None) |
123
|
1 |
|
generator.nameToNodes = {('foo1',1) : foo1, ('bar',2) : bar, ('foo2', 3) : foo2} |
124
|
1 |
|
generator._correctTree(foo1) |
125
|
|
|
self.assertEqual(bar.namedEntityTag, 'undef') |
126
|
|
|
|