linkage 0.0.8 → 0.1.0.pre
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gitignore +1 -0
- data/.yardopts +1 -0
- data/Gemfile +1 -19
- data/Gemfile-java +3 -0
- data/README.markdown +88 -34
- data/Rakefile +16 -15
- data/TODO +4 -0
- data/lib/linkage/comparator.rb +139 -144
- data/lib/linkage/comparators/compare.rb +236 -29
- data/lib/linkage/comparators/strcompare.rb +85 -0
- data/lib/linkage/comparators/within.rb +24 -20
- data/lib/linkage/configuration.rb +44 -466
- data/lib/linkage/dataset.rb +28 -127
- data/lib/linkage/exceptions.rb +5 -0
- data/lib/linkage/field.rb +6 -37
- data/lib/linkage/field_set.rb +3 -3
- data/lib/linkage/match_recorder.rb +22 -0
- data/lib/linkage/match_set.rb +34 -0
- data/lib/linkage/match_sets/csv.rb +39 -0
- data/lib/linkage/match_sets/database.rb +45 -0
- data/lib/linkage/matcher.rb +30 -0
- data/lib/linkage/result_set.rb +25 -110
- data/lib/linkage/result_sets/csv.rb +54 -0
- data/lib/linkage/result_sets/database.rb +42 -0
- data/lib/linkage/runner.rb +57 -16
- data/lib/linkage/score_recorder.rb +30 -0
- data/lib/linkage/score_set.rb +49 -0
- data/lib/linkage/score_sets/csv.rb +64 -0
- data/lib/linkage/score_sets/database.rb +77 -0
- data/lib/linkage/version.rb +1 -1
- data/lib/linkage.rb +14 -17
- data/linkage.gemspec +13 -1
- data/linkage.gemspec-java +32 -0
- data/test/helper.rb +30 -23
- data/test/integration/test_cross_linkage.rb +46 -25
- data/test/integration/test_database_result_set.rb +55 -0
- data/test/integration/test_dual_linkage.rb +19 -94
- data/test/integration/test_self_linkage.rb +100 -203
- data/test/integration/test_within_comparator.rb +24 -77
- data/test/unit/comparators/test_compare.rb +254 -50
- data/test/unit/comparators/test_strcompare.rb +45 -0
- data/test/unit/comparators/test_within.rb +14 -26
- data/test/unit/match_sets/test_csv.rb +78 -0
- data/test/unit/match_sets/test_database.rb +63 -0
- data/test/unit/result_sets/test_csv.rb +111 -0
- data/test/unit/result_sets/test_database.rb +68 -0
- data/test/unit/score_sets/test_csv.rb +151 -0
- data/test/unit/score_sets/test_database.rb +149 -0
- data/test/unit/test_comparator.rb +46 -83
- data/test/unit/test_comparators.rb +4 -0
- data/test/unit/test_configuration.rb +99 -145
- data/test/unit/test_dataset.rb +52 -73
- data/test/unit/test_field.rb +4 -55
- data/test/unit/test_field_set.rb +6 -6
- data/test/unit/test_match_recorder.rb +23 -0
- data/test/unit/test_match_set.rb +23 -0
- data/test/unit/test_match_sets.rb +4 -0
- data/test/unit/test_matcher.rb +44 -0
- data/test/unit/test_result_set.rb +24 -223
- data/test/unit/test_result_sets.rb +4 -0
- data/test/unit/test_runner.rb +122 -17
- data/test/unit/test_runners.rb +4 -0
- data/test/unit/test_score_recorder.rb +25 -0
- data/test/unit/test_score_set.rb +37 -0
- data/test/unit/test_score_sets.rb +4 -0
- metadata +183 -90
- data/Gemfile.lock +0 -92
- data/lib/linkage/comparators/binary.rb +0 -12
- data/lib/linkage/data.rb +0 -175
- data/lib/linkage/decollation.rb +0 -93
- data/lib/linkage/expectation.rb +0 -21
- data/lib/linkage/expectations/exhaustive.rb +0 -63
- data/lib/linkage/expectations/simple.rb +0 -168
- data/lib/linkage/function.rb +0 -148
- data/lib/linkage/functions/binary.rb +0 -30
- data/lib/linkage/functions/cast.rb +0 -54
- data/lib/linkage/functions/length.rb +0 -29
- data/lib/linkage/functions/strftime.rb +0 -33
- data/lib/linkage/functions/trim.rb +0 -30
- data/lib/linkage/group.rb +0 -55
- data/lib/linkage/meta_object.rb +0 -139
- data/lib/linkage/runner/single_threaded.rb +0 -187
- data/lib/linkage/utils.rb +0 -164
- data/lib/linkage/warnings.rb +0 -5
- data/test/integration/test_collation.rb +0 -45
- data/test/integration/test_configuration.rb +0 -268
- data/test/integration/test_dataset.rb +0 -116
- data/test/integration/test_functions.rb +0 -88
- data/test/integration/test_result_set.rb +0 -85
- data/test/integration/test_scoring.rb +0 -84
- data/test/unit/expectations/test_exhaustive.rb +0 -111
- data/test/unit/expectations/test_simple.rb +0 -303
- data/test/unit/functions/test_binary.rb +0 -54
- data/test/unit/functions/test_cast.rb +0 -98
- data/test/unit/functions/test_length.rb +0 -52
- data/test/unit/functions/test_strftime.rb +0 -60
- data/test/unit/functions/test_trim.rb +0 -43
- data/test/unit/runner/test_single_threaded.rb +0 -12
- data/test/unit/test_data.rb +0 -445
- data/test/unit/test_decollation.rb +0 -201
- data/test/unit/test_function.rb +0 -233
- data/test/unit/test_group.rb +0 -38
- data/test/unit/test_meta_object.rb +0 -208
- data/test/unit/test_utils.rb +0 -341
@@ -1,6 +1,6 @@
|
|
1
|
-
require
|
1
|
+
require File.expand_path("../../test_comparators", __FILE__)
|
2
2
|
|
3
|
-
class UnitTests::TestCompare < Test::Unit::TestCase
|
3
|
+
class UnitTests::TestComparators::TestCompare < Test::Unit::TestCase
|
4
4
|
def self.const_missing(name)
|
5
5
|
if Linkage::Comparators.const_defined?(name)
|
6
6
|
Linkage::Comparators.const_get(name)
|
@@ -9,97 +9,301 @@ class UnitTests::TestCompare < Test::Unit::TestCase
|
|
9
9
|
end
|
10
10
|
end
|
11
11
|
|
12
|
-
test "subclass of
|
13
|
-
assert_equal
|
14
|
-
end
|
15
|
-
|
16
|
-
test "valid parameters" do
|
17
|
-
meta_object_1 = stub('meta object', :name => :foo, :side => :lhs, :ruby_type => { :type => Integer }, :static? => false)
|
18
|
-
meta_object_2 = stub('meta object', :object => '>', :ruby_type => { :type => String }, :static? => true, :raw? => true)
|
19
|
-
meta_object_3 = stub('meta object', :name => :bar, :side => :rhs, :ruby_type => { :type => Integer }, :static? => false)
|
20
|
-
assert_nothing_raised do
|
21
|
-
Compare.new(meta_object_1, meta_object_2, meta_object_3)
|
22
|
-
end
|
12
|
+
test "subclass of Comparator" do
|
13
|
+
assert_equal Linkage::Comparator, Compare.superclass
|
23
14
|
end
|
24
15
|
|
25
16
|
test "score for not equal to" do
|
26
|
-
|
27
|
-
|
28
|
-
|
29
|
-
|
17
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
18
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
19
|
+
comp = Compare.new([field_1], [field_2], :not_equal)
|
20
|
+
assert_equal :simple, comp.type
|
30
21
|
assert_equal 1, comp.score({:foo => 10}, {:bar => 5})
|
31
22
|
assert_equal 0, comp.score({:foo => 5}, {:bar => 5})
|
32
23
|
assert_equal 1, comp.score({:foo => 0}, {:bar => 5})
|
33
24
|
end
|
34
25
|
|
26
|
+
test "score for not equal to with multiple fields" do
|
27
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
28
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
29
|
+
field_3 = stub('baz field', :name => :baz, :ruby_type => { :type => Integer })
|
30
|
+
field_4 = stub('qux field', :name => :qux, :ruby_type => { :type => Integer })
|
31
|
+
comp = Compare.new([field_1, field_2], [field_3, field_4], :not_equal)
|
32
|
+
assert_equal :simple, comp.type
|
33
|
+
assert_equal 1, comp.score({:foo => 10, :bar => 5}, {:baz => 5, :qux => 10})
|
34
|
+
assert_equal 0, comp.score({:foo => 10, :bar => 5}, {:baz => 10, :qux => 5})
|
35
|
+
assert_equal 0, comp.score({:foo => 10, :bar => 5}, {:baz => 10, :qux => 7})
|
36
|
+
assert_equal 0, comp.score({:foo => 10, :bar => 5}, {:baz => 7, :qux => 5})
|
37
|
+
end
|
38
|
+
|
35
39
|
test "score for greater than" do
|
36
|
-
|
37
|
-
|
38
|
-
|
39
|
-
|
40
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
41
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
42
|
+
comp = Compare.new([field_1], [field_2], :greater_than)
|
43
|
+
assert_equal :simple, comp.type
|
40
44
|
assert_equal 1, comp.score({:foo => 10}, {:bar => 5})
|
41
45
|
assert_equal 0, comp.score({:foo => 5}, {:bar => 5})
|
42
46
|
assert_equal 0, comp.score({:foo => 0}, {:bar => 5})
|
43
47
|
end
|
44
48
|
|
49
|
+
test "score for greater than with multiple fields" do
|
50
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
51
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
52
|
+
field_3 = stub('baz field', :name => :baz, :ruby_type => { :type => Integer })
|
53
|
+
field_4 = stub('qux field', :name => :qux, :ruby_type => { :type => Integer })
|
54
|
+
comp = Compare.new([field_1, field_2], [field_3, field_4], :greater_than)
|
55
|
+
assert_equal :simple, comp.type
|
56
|
+
assert_equal 1, comp.score({:foo => 10, :bar => 5}, {:baz => 5, :qux => 0})
|
57
|
+
assert_equal 0, comp.score({:foo => 10, :bar => 5}, {:baz => 15, :qux => 10})
|
58
|
+
assert_equal 0, comp.score({:foo => 10, :bar => 5}, {:baz => 15, :qux => 0})
|
59
|
+
assert_equal 0, comp.score({:foo => 10, :bar => 5}, {:baz => 5, :qux => 10})
|
60
|
+
end
|
61
|
+
|
45
62
|
test "score for greater than or equal to" do
|
46
|
-
|
47
|
-
|
48
|
-
|
49
|
-
|
63
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
64
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
65
|
+
comp = Compare.new([field_1], [field_2], :greater_than_or_equal)
|
66
|
+
assert_equal :simple, comp.type
|
50
67
|
assert_equal 1, comp.score({:foo => 10}, {:bar => 5})
|
51
68
|
assert_equal 1, comp.score({:foo => 5}, {:bar => 5})
|
52
69
|
assert_equal 0, comp.score({:foo => 0}, {:bar => 5})
|
53
70
|
end
|
54
71
|
|
72
|
+
test "score for greater than or equal to with multiple fields" do
|
73
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
74
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
75
|
+
field_3 = stub('baz field', :name => :baz, :ruby_type => { :type => Integer })
|
76
|
+
field_4 = stub('qux field', :name => :qux, :ruby_type => { :type => Integer })
|
77
|
+
comp = Compare.new([field_1, field_2], [field_3, field_4], :greater_than_or_equal)
|
78
|
+
assert_equal :simple, comp.type
|
79
|
+
assert_equal 1, comp.score({:foo => 10, :bar => 5}, {:baz => 5, :qux => 0})
|
80
|
+
assert_equal 1, comp.score({:foo => 10, :bar => 5}, {:baz => 10, :qux => 5})
|
81
|
+
assert_equal 0, comp.score({:foo => 10, :bar => 5}, {:baz => 5, :qux => 10})
|
82
|
+
assert_equal 0, comp.score({:foo => 10, :bar => 5}, {:baz => 15, :qux => 0})
|
83
|
+
end
|
84
|
+
|
55
85
|
test "score for less than or equal to" do
|
56
|
-
|
57
|
-
|
58
|
-
|
59
|
-
|
60
|
-
assert_equal 0, comp.score({:foo => 10}, {:bar => 5})
|
86
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
87
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
88
|
+
comp = Compare.new([field_1], [field_2], :less_than_or_equal)
|
89
|
+
assert_equal :simple, comp.type
|
61
90
|
assert_equal 1, comp.score({:foo => 5}, {:bar => 5})
|
62
91
|
assert_equal 1, comp.score({:foo => 0}, {:bar => 5})
|
92
|
+
assert_equal 0, comp.score({:foo => 10}, {:bar => 5})
|
93
|
+
end
|
94
|
+
|
95
|
+
test "score for less than or equal to with multiple fields" do
|
96
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
97
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
98
|
+
field_3 = stub('baz field', :name => :baz, :ruby_type => { :type => Integer })
|
99
|
+
field_4 = stub('qux field', :name => :qux, :ruby_type => { :type => Integer })
|
100
|
+
comp = Compare.new([field_1, field_2], [field_3, field_4], :less_than_or_equal)
|
101
|
+
assert_equal :simple, comp.type
|
102
|
+
assert_equal 1, comp.score({:foo => 5, :bar => 0}, {:baz => 10, :qux => 5})
|
103
|
+
assert_equal 1, comp.score({:foo => 5, :bar => 0}, {:baz => 5, :qux => 0})
|
104
|
+
assert_equal 0, comp.score({:foo => 5, :bar => 0}, {:baz => 0, :qux => 5})
|
105
|
+
assert_equal 0, comp.score({:foo => 5, :bar => 0}, {:baz => 10, :qux => -5})
|
63
106
|
end
|
64
107
|
|
65
108
|
test "score for less than" do
|
66
|
-
|
67
|
-
|
68
|
-
|
69
|
-
|
109
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
110
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
111
|
+
comp = Compare.new([field_1], [field_2], :less_than)
|
112
|
+
assert_equal :simple, comp.type
|
113
|
+
assert_equal 1, comp.score({:foo => 0}, {:bar => 5})
|
70
114
|
assert_equal 0, comp.score({:foo => 10}, {:bar => 5})
|
71
115
|
assert_equal 0, comp.score({:foo => 5}, {:bar => 5})
|
72
|
-
|
116
|
+
end
|
117
|
+
|
118
|
+
test "score for less than with multiple fields" do
|
119
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
120
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
121
|
+
field_3 = stub('baz field', :name => :baz, :ruby_type => { :type => Integer })
|
122
|
+
field_4 = stub('qux field', :name => :qux, :ruby_type => { :type => Integer })
|
123
|
+
comp = Compare.new([field_1, field_2], [field_3, field_4], :less_than)
|
124
|
+
assert_equal :simple, comp.type
|
125
|
+
assert_equal 1, comp.score({:foo => 5, :bar => 0}, {:baz => 10, :qux => 5})
|
126
|
+
assert_equal 0, comp.score({:foo => 5, :bar => 0}, {:baz => 5, :qux => 0})
|
127
|
+
assert_equal 0, comp.score({:foo => 5, :bar => 0}, {:baz => 0, :qux => 5})
|
128
|
+
assert_equal 0, comp.score({:foo => 5, :bar => 0}, {:baz => 10, :qux => -5})
|
129
|
+
end
|
130
|
+
|
131
|
+
test "score_datasets with one field equality" do
|
132
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
133
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
134
|
+
comp = Compare.new([field_1], [field_2], :equal)
|
135
|
+
assert_equal :advanced, comp.type
|
136
|
+
observer = stub('observer', :update => nil)
|
137
|
+
comp.add_observer(observer)
|
138
|
+
|
139
|
+
dataset_1 = stub('dataset 1')
|
140
|
+
dataset_1.expects(:order).with(:foo).returns(dataset_1)
|
141
|
+
dataset_1.expects(:each).multiple_yields(
|
142
|
+
[(record_1_1 = {:id => 1, :foo => 123})],
|
143
|
+
[(record_1_2 = {:id => 2, :foo => 456})],
|
144
|
+
[(record_1_3 = {:id => 3, :foo => 456})]
|
145
|
+
)
|
146
|
+
|
147
|
+
dataset_2 = stub('dataset 2')
|
148
|
+
dataset_2.expects(:order).with(:bar).returns(dataset_2)
|
149
|
+
dataset_2.expects(:each).multiple_yields(
|
150
|
+
[(record_2_1 = {:id => 100, :bar => 123})],
|
151
|
+
[(record_2_2 = {:id => 101, :bar => 456})],
|
152
|
+
[(record_2_3 = {:id => 102, :bar => 456})]
|
153
|
+
)
|
154
|
+
|
155
|
+
observer.expects(:update).with(comp, record_1_1, record_2_1, 1)
|
156
|
+
observer.expects(:update).with(comp, record_1_2, record_2_2, 1)
|
157
|
+
observer.expects(:update).with(comp, record_1_2, record_2_3, 1)
|
158
|
+
observer.expects(:update).with(comp, record_1_3, record_2_2, 1)
|
159
|
+
observer.expects(:update).with(comp, record_1_3, record_2_3, 1)
|
160
|
+
|
161
|
+
comp.score_datasets(dataset_1, dataset_2)
|
162
|
+
end
|
163
|
+
|
164
|
+
test "score_datasets with multiple field equality" do
|
165
|
+
field_1_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
166
|
+
field_1_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
167
|
+
field_2_1 = stub('baz field', :name => :baz, :ruby_type => { :type => Integer })
|
168
|
+
field_2_2 = stub('qux field', :name => :qux, :ruby_type => { :type => Integer })
|
169
|
+
comp = Compare.new([field_1_1, field_1_2], [field_2_1, field_2_2], :equal)
|
170
|
+
assert_equal :advanced, comp.type
|
171
|
+
observer = stub('observer', :update => nil)
|
172
|
+
comp.add_observer(observer)
|
173
|
+
|
174
|
+
dataset_1 = stub('dataset 1')
|
175
|
+
dataset_1.expects(:order).with(:foo, :bar).returns(dataset_1)
|
176
|
+
dataset_1.expects(:each).multiple_yields(
|
177
|
+
[(record_1_1 = {:id => 1, :foo => 123, :bar => 123})],
|
178
|
+
[(record_1_2 = {:id => 2, :foo => 123, :bar => 456})],
|
179
|
+
[(record_1_3 = {:id => 3, :foo => 123, :bar => 789})],
|
180
|
+
[(record_1_4 = {:id => 4, :foo => 456, :bar => 123})]
|
181
|
+
)
|
182
|
+
|
183
|
+
dataset_2 = stub('dataset 2')
|
184
|
+
dataset_2.expects(:order).with(:baz, :qux).returns(dataset_2)
|
185
|
+
dataset_2.expects(:each).multiple_yields(
|
186
|
+
[(record_2_1 = {:id => 100, :baz => 123, :qux => 123})],
|
187
|
+
[(record_2_2 = {:id => 101, :baz => 123, :qux => 123})],
|
188
|
+
[(record_2_3 = {:id => 102, :baz => 123, :qux => 789})],
|
189
|
+
[(record_2_4 = {:id => 103, :baz => 456, :qux => 123})],
|
190
|
+
[(record_2_5 = {:id => 104, :baz => 456, :qux => 456})]
|
191
|
+
)
|
192
|
+
|
193
|
+
observer.expects(:update).with(comp, record_1_1, record_2_1, 1)
|
194
|
+
observer.expects(:update).with(comp, record_1_1, record_2_2, 1)
|
195
|
+
observer.expects(:update).with(comp, record_1_3, record_2_3, 1)
|
196
|
+
observer.expects(:update).with(comp, record_1_4, record_2_4, 1)
|
197
|
+
|
198
|
+
comp.score_datasets(dataset_1, dataset_2)
|
199
|
+
end
|
200
|
+
|
201
|
+
test "score_dataset with same single field equality" do
|
202
|
+
field = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
203
|
+
comp = Compare.new([field], [field], :equal)
|
204
|
+
assert_equal :advanced, comp.type
|
205
|
+
observer = stub('observer', :update => nil)
|
206
|
+
comp.add_observer(observer)
|
207
|
+
|
208
|
+
dataset = stub('dataset')
|
209
|
+
dataset.expects(:order).with(:foo).returns(dataset)
|
210
|
+
dataset.expects(:each).multiple_yields(
|
211
|
+
[(record_1 = {:id => 1, :foo => 123})],
|
212
|
+
[(record_2 = {:id => 2, :foo => 456})],
|
213
|
+
[(record_3 = {:id => 3, :foo => 456})]
|
214
|
+
)
|
215
|
+
|
216
|
+
observer.expects(:update).with(comp, record_2, record_3, 1)
|
217
|
+
comp.score_dataset(dataset)
|
218
|
+
end
|
219
|
+
|
220
|
+
test "score_dataset with same multiple field equality" do
|
221
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
222
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
223
|
+
comp = Compare.new([field_1, field_2], [field_1, field_2], :equal)
|
224
|
+
assert_equal :advanced, comp.type
|
225
|
+
observer = stub('observer', :update => nil)
|
226
|
+
comp.add_observer(observer)
|
227
|
+
|
228
|
+
dataset = stub('dataset')
|
229
|
+
dataset.expects(:order).with(:foo, :bar).returns(dataset)
|
230
|
+
dataset.expects(:each).multiple_yields(
|
231
|
+
[(record_1 = {:id => 1, :foo => 123, :bar => 123})],
|
232
|
+
[(record_2 = {:id => 2, :foo => 123, :bar => 123})],
|
233
|
+
[(record_3 = {:id => 3, :foo => 123, :bar => 123})],
|
234
|
+
[(record_4 = {:id => 4, :foo => 123, :bar => 456})],
|
235
|
+
[(record_5 = {:id => 5, :foo => 456, :bar => 123})],
|
236
|
+
[(record_6 = {:id => 6, :foo => 456, :bar => 456})],
|
237
|
+
[(record_7 = {:id => 7, :foo => 456, :bar => 456})]
|
238
|
+
)
|
239
|
+
|
240
|
+
observer.expects(:update).with(comp, record_1, record_2, 1)
|
241
|
+
observer.expects(:update).with(comp, record_1, record_3, 1)
|
242
|
+
observer.expects(:update).with(comp, record_2, record_3, 1)
|
243
|
+
observer.expects(:update).with(comp, record_6, record_7, 1)
|
244
|
+
|
245
|
+
comp.score_dataset(dataset)
|
246
|
+
end
|
247
|
+
|
248
|
+
test "score_dataset with different single field equality" do
|
249
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
250
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
251
|
+
comp = Compare.new([field_1], [field_2], :equal)
|
252
|
+
assert_equal :advanced, comp.type
|
253
|
+
observer = stub('observer', :update => nil)
|
254
|
+
comp.add_observer(observer)
|
255
|
+
|
256
|
+
dataset = stub('dataset')
|
257
|
+
dataset_1 = stub('dataset 1')
|
258
|
+
dataset.expects(:order).with(:foo).returns(dataset_1)
|
259
|
+
dataset_1.expects(:each).multiple_yields(
|
260
|
+
[(record_1 = {:id => 1, :foo => 123, :bar => 456})],
|
261
|
+
[(record_2 = {:id => 2, :foo => 456, :bar => 456})],
|
262
|
+
[(record_3 = {:id => 3, :foo => 456, :bar => 123})]
|
263
|
+
)
|
264
|
+
|
265
|
+
dataset_2 = stub('dataset 2')
|
266
|
+
dataset.expects(:order).with(:bar).returns(dataset_2)
|
267
|
+
dataset_2.expects(:each).multiple_yields(
|
268
|
+
[record_3],
|
269
|
+
[record_1],
|
270
|
+
[record_2]
|
271
|
+
)
|
272
|
+
|
273
|
+
observer.expects(:update).with(comp, record_1, record_3, 1)
|
274
|
+
observer.expects(:update).with(comp, record_2, record_1, 1)
|
275
|
+
observer.expects(:update).with(comp, record_2, record_2, 1)
|
276
|
+
observer.expects(:update).with(comp, record_3, record_1, 1)
|
277
|
+
observer.expects(:update).with(comp, record_3, record_2, 1)
|
278
|
+
|
279
|
+
comp.score_dataset(dataset)
|
73
280
|
end
|
74
281
|
|
75
282
|
test "registers itself" do
|
76
283
|
assert_equal Compare, Linkage::Comparator['compare']
|
77
284
|
end
|
78
285
|
|
79
|
-
test "requires
|
80
|
-
|
81
|
-
|
82
|
-
meta_object_3 = stub('meta object', :name => :bar, :side => :lhs, :ruby_type => { :type => Integer }, :static? => false)
|
286
|
+
test "requires equal size sets" do
|
287
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
288
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
83
289
|
assert_raises do
|
84
|
-
Compare.new(
|
290
|
+
Compare.new([field_1, field_2], [], :greater_than_or_equal)
|
85
291
|
end
|
86
292
|
end
|
87
293
|
|
88
|
-
test "requires that
|
89
|
-
|
90
|
-
|
91
|
-
meta_object_3 = stub('meta object', :name => :bar, :side => :rhs, :ruby_type => { :type => Date }, :static? => false)
|
294
|
+
test "requires that sets have values with alike types" do
|
295
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
296
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Date })
|
92
297
|
assert_raises do
|
93
|
-
Compare.new(
|
298
|
+
Compare.new([field_1], [field_2], :greater_than_or_equal)
|
94
299
|
end
|
95
300
|
end
|
96
301
|
|
97
|
-
test "requires
|
98
|
-
|
99
|
-
|
100
|
-
meta_object_3 = stub('meta object', :name => :bar, :side => :rhs, :ruby_type => { :type => Integer }, :static? => false)
|
302
|
+
test "requires valid operation" do
|
303
|
+
field_1 = stub('foo field', :name => :foo, :ruby_type => { :type => Integer })
|
304
|
+
field_2 = stub('bar field', :name => :bar, :ruby_type => { :type => Integer })
|
101
305
|
assert_raises do
|
102
|
-
Compare.new(
|
306
|
+
Compare.new([field_1], [field_2], 'foo')
|
103
307
|
end
|
104
308
|
end
|
105
309
|
end
|
@@ -0,0 +1,45 @@
|
|
1
|
+
require File.expand_path("../../test_comparators", __FILE__)
|
2
|
+
|
3
|
+
class UnitTests::TestComparators::TestStrcompare < Test::Unit::TestCase
|
4
|
+
def self.const_missing(name)
|
5
|
+
if Linkage::Comparators.const_defined?(name)
|
6
|
+
Linkage::Comparators.const_get(name)
|
7
|
+
else
|
8
|
+
super
|
9
|
+
end
|
10
|
+
end
|
11
|
+
|
12
|
+
test "subclass of Comparator" do
|
13
|
+
assert_equal Linkage::Comparator, Strcompare.superclass
|
14
|
+
end
|
15
|
+
|
16
|
+
test "requires string types" do
|
17
|
+
field_1 = stub('field 1', :name => :foo, :ruby_type => { :type => Integer })
|
18
|
+
field_2 = stub('field 2', :name => :bar, :ruby_type => { :type => Integer })
|
19
|
+
assert_raises do
|
20
|
+
Strcompare.new(field_1, field_2, :jarowinkler)
|
21
|
+
end
|
22
|
+
end
|
23
|
+
|
24
|
+
test "requires valid operator" do
|
25
|
+
field_1 = stub('field 1', :name => :foo, :ruby_type => { :type => String })
|
26
|
+
field_2 = stub('field 2', :name => :bar, :ruby_type => { :type => String })
|
27
|
+
assert_raises do
|
28
|
+
Strcompare.new(field_1, field_2, 'foo')
|
29
|
+
end
|
30
|
+
end
|
31
|
+
|
32
|
+
test "score for jarowinkler" do
|
33
|
+
field_1 = stub('field 1', :name => :foo, :ruby_type => { :type => String })
|
34
|
+
field_2 = stub('field 2', :name => :bar, :ruby_type => { :type => String })
|
35
|
+
comp = Strcompare.new(field_1, field_2, :jarowinkler)
|
36
|
+
assert_equal 0.961, comp.score({:foo => 'martha'}, {:bar => 'marhta'})
|
37
|
+
assert_equal 0.840, comp.score({:foo => 'dwayne'}, {:bar => 'duane'})
|
38
|
+
assert_equal 0.813, comp.score({:foo => 'dixon'}, {:bar => 'dicksonx'})
|
39
|
+
assert_equal 0, comp.score({:foo => 'cat'}, {:bar => 'dog'})
|
40
|
+
end
|
41
|
+
|
42
|
+
test "registers itself" do
|
43
|
+
assert_equal Strcompare, Linkage::Comparator['strcompare']
|
44
|
+
end
|
45
|
+
end
|
@@ -1,6 +1,6 @@
|
|
1
|
-
require
|
1
|
+
require File.expand_path("../../test_comparators", __FILE__)
|
2
2
|
|
3
|
-
class UnitTests::TestWithin < Test::Unit::TestCase
|
3
|
+
class UnitTests::TestComparators::TestWithin < Test::Unit::TestCase
|
4
4
|
def self.const_missing(name)
|
5
5
|
if Linkage::Comparators.const_defined?(name)
|
6
6
|
Linkage::Comparators.const_get(name)
|
@@ -9,24 +9,22 @@ class UnitTests::TestWithin < Test::Unit::TestCase
|
|
9
9
|
end
|
10
10
|
end
|
11
11
|
|
12
|
-
test "subclass of
|
13
|
-
assert_equal Linkage::
|
12
|
+
test "subclass of Comparator" do
|
13
|
+
assert_equal Linkage::Comparator, Within.superclass
|
14
14
|
end
|
15
15
|
|
16
16
|
test "valid parameters" do
|
17
|
-
|
18
|
-
|
19
|
-
meta_object_3 = stub('meta object', :name => :bar, :side => :rhs, :ruby_type => { :type => Integer }, :static? => false)
|
17
|
+
field_1 = stub('field 1', :name => :foo, :ruby_type => { :type => Integer })
|
18
|
+
field_2 = stub('field 2', :name => :bar, :ruby_type => { :type => Integer })
|
20
19
|
assert_nothing_raised do
|
21
|
-
Within.new(
|
20
|
+
Within.new(field_1, field_2, 123)
|
22
21
|
end
|
23
22
|
end
|
24
23
|
|
25
24
|
test "score" do
|
26
|
-
|
27
|
-
|
28
|
-
|
29
|
-
comp = Within.new(meta_object_1, meta_object_2, meta_object_3)
|
25
|
+
field_1 = stub('field 1', :name => :foo, :ruby_type => { :type => Integer })
|
26
|
+
field_2 = stub('field 2', :name => :bar, :ruby_type => { :type => Integer })
|
27
|
+
comp = Within.new(field_1, field_2, 123)
|
30
28
|
assert_equal 1, comp.score({:foo => 123}, {:bar => 124})
|
31
29
|
assert_equal 1, comp.score({:foo => 124}, {:bar => 123})
|
32
30
|
assert_equal 1, comp.score({:foo => 0}, {:bar => 123})
|
@@ -37,21 +35,11 @@ class UnitTests::TestWithin < Test::Unit::TestCase
|
|
37
35
|
assert_equal Within, Linkage::Comparator['within']
|
38
36
|
end
|
39
37
|
|
40
|
-
test "requires argument
|
41
|
-
|
42
|
-
|
43
|
-
meta_object_3 = stub('meta object', :name => :bar, :side => :lhs, :ruby_type => { :type => Integer }, :static? => false)
|
38
|
+
test "requires that second argument has the same type as the first argument" do
|
39
|
+
field_1 = stub('field 1', :name => :foo, :ruby_type => { :type => Integer })
|
40
|
+
field_2 = stub('field 2', :name => :bar, :ruby_type => { :type => Date })
|
44
41
|
assert_raises do
|
45
|
-
Within.new(
|
46
|
-
end
|
47
|
-
end
|
48
|
-
|
49
|
-
test "requires that 3rd argument has the same type as the first argument" do
|
50
|
-
meta_object_1 = stub('meta object', :name => :foo, :side => :lhs, :ruby_type => { :type => Integer }, :static? => false)
|
51
|
-
meta_object_2 = stub('meta object', :object => 123, :ruby_type => { :type => Fixnum }, :static? => true, :object => 123)
|
52
|
-
meta_object_3 = stub('meta object', :name => :bar, :side => :rhs, :ruby_type => { :type => Date }, :static? => false)
|
53
|
-
assert_raises do
|
54
|
-
Within.new(meta_object_1, meta_object_2, meta_object_3)
|
42
|
+
Within.new(field_1, field_2, 123)
|
55
43
|
end
|
56
44
|
end
|
57
45
|
end
|
@@ -0,0 +1,78 @@
|
|
1
|
+
require File.expand_path("../../test_match_sets", __FILE__)
|
2
|
+
|
3
|
+
class UnitTests::TestMatchSets::TestCSV < Test::Unit::TestCase
|
4
|
+
test "open_for_writing" do
|
5
|
+
match_set = Linkage::MatchSets::CSV.new('foo.csv')
|
6
|
+
csv = stub('csv')
|
7
|
+
CSV.expects(:open).with('foo.csv', 'wb').returns(csv)
|
8
|
+
csv.expects(:<<).with(%w{id_1 id_2 score})
|
9
|
+
match_set.open_for_writing
|
10
|
+
end
|
11
|
+
|
12
|
+
test "open_for_writing when already open" do
|
13
|
+
match_set = Linkage::MatchSets::CSV.new('foo.csv')
|
14
|
+
csv = stub('csv')
|
15
|
+
CSV.expects(:open).once.with('foo.csv', 'wb').returns(csv)
|
16
|
+
csv.expects(:<<).once.with(%w{id_1 id_2 score})
|
17
|
+
match_set.open_for_writing
|
18
|
+
match_set.open_for_writing
|
19
|
+
end
|
20
|
+
|
21
|
+
test "open_for_writing when file exists" do
|
22
|
+
match_set = Linkage::MatchSets::CSV.new('foo.csv')
|
23
|
+
File.expects(:exist?).with('foo.csv').returns(true)
|
24
|
+
assert_raises(Linkage::ExistsError) do
|
25
|
+
match_set.open_for_writing
|
26
|
+
end
|
27
|
+
end
|
28
|
+
|
29
|
+
test "open_for_writing when file exists and forcing overwrite" do
|
30
|
+
match_set = Linkage::MatchSets::CSV.new('foo.csv', :overwrite => true)
|
31
|
+
File.stubs(:exist?).with('foo.csv').returns(true)
|
32
|
+
assert_nothing_raised do
|
33
|
+
csv = stub('csv')
|
34
|
+
CSV.expects(:open).with('foo.csv', 'wb').returns(csv)
|
35
|
+
csv.expects(:<<).with(%w{id_1 id_2 score})
|
36
|
+
match_set.open_for_writing
|
37
|
+
end
|
38
|
+
end
|
39
|
+
|
40
|
+
test "add_match when unopened raises exception" do
|
41
|
+
match_set = Linkage::MatchSets::CSV.new('foo.csv')
|
42
|
+
assert_raises { match_set.add_match(1, 2, 1) }
|
43
|
+
end
|
44
|
+
|
45
|
+
test "add_match" do
|
46
|
+
tempfile = Tempfile.new('linkage')
|
47
|
+
tempfile.close
|
48
|
+
match_set = Linkage::MatchSets::CSV.new(tempfile.path, :overwrite => true)
|
49
|
+
match_set.open_for_writing
|
50
|
+
match_set.add_match(1, 2, 1)
|
51
|
+
match_set.close
|
52
|
+
|
53
|
+
expected = "id_1,id_2,score\n1,2,1\n"
|
54
|
+
assert_equal expected, File.read(tempfile.path)
|
55
|
+
end
|
56
|
+
|
57
|
+
test "add_match removes extra decimals" do
|
58
|
+
match_set = Linkage::MatchSets::CSV.new('foo.csv')
|
59
|
+
csv = stub('csv')
|
60
|
+
CSV.stubs(:open).with('foo.csv', 'wb').returns(csv)
|
61
|
+
csv.stubs(:<<).with(%w{id_1 id_2 score})
|
62
|
+
match_set.open_for_writing
|
63
|
+
|
64
|
+
csv.expects(:<<).with do |(id_1, id_2, score)|
|
65
|
+
id_1 == 1 && id_2 == 2 && score.equal?(1)
|
66
|
+
end
|
67
|
+
match_set.add_match(1, 2, 1.0)
|
68
|
+
|
69
|
+
csv.expects(:<<).with do |(id_1, id_2, score)|
|
70
|
+
id_1 == 1 && id_2 == 2 && score.equal?(0)
|
71
|
+
end
|
72
|
+
match_set.add_match(1, 2, 0.0)
|
73
|
+
end
|
74
|
+
|
75
|
+
test "registers itself" do
|
76
|
+
assert_equal Linkage::MatchSets::CSV, Linkage::MatchSet['csv']
|
77
|
+
end
|
78
|
+
end
|
@@ -0,0 +1,63 @@
|
|
1
|
+
require File.expand_path("../../test_match_sets", __FILE__)
|
2
|
+
|
3
|
+
class UnitTests::TestMatchSets::TestDatabase < Test::Unit::TestCase
|
4
|
+
def setup
|
5
|
+
@dataset = stub('dataset')
|
6
|
+
@database = stub('database', :[] => @dataset)
|
7
|
+
end
|
8
|
+
|
9
|
+
test "open_for_writing for database with no matches table" do
|
10
|
+
match_set = Linkage::MatchSets::Database.new(@database)
|
11
|
+
@database.stubs(:table_exists?).with(:matches).returns(false)
|
12
|
+
@database.expects(:create_table).with(:matches)
|
13
|
+
@database.expects(:[]).with(:matches).returns(@dataset)
|
14
|
+
match_set.open_for_writing
|
15
|
+
end
|
16
|
+
|
17
|
+
test "open_for_writing when already open" do
|
18
|
+
match_set = Linkage::MatchSets::Database.new(@database)
|
19
|
+
@database.stubs(:table_exists?).with(:matches).returns(false)
|
20
|
+
@database.expects(:create_table).with(:matches)
|
21
|
+
@database.expects(:[]).with(:matches).returns(@dataset)
|
22
|
+
match_set.open_for_writing
|
23
|
+
match_set.open_for_writing
|
24
|
+
end
|
25
|
+
|
26
|
+
test "open_for_writing when matches table already exists" do
|
27
|
+
match_set = Linkage::MatchSets::Database.new(@database)
|
28
|
+
@database.expects(:table_exists?).with(:matches).returns(true)
|
29
|
+
@database.expects(:create_table).with(:matches).never
|
30
|
+
assert_raises(Linkage::ExistsError) do
|
31
|
+
match_set.open_for_writing
|
32
|
+
end
|
33
|
+
end
|
34
|
+
|
35
|
+
test "open_for_writing when matches table already exists and in overwrite mode" do
|
36
|
+
match_set = Linkage::MatchSets::Database.new(@database, :overwrite => true)
|
37
|
+
@database.expects(:drop_table?).with(:matches)
|
38
|
+
@database.expects(:create_table).with(:matches)
|
39
|
+
@database.expects(:[]).with(:matches).returns(@dataset)
|
40
|
+
match_set.open_for_writing
|
41
|
+
end
|
42
|
+
|
43
|
+
test "add_match when unopened raises exception" do
|
44
|
+
match_set = Linkage::MatchSets::Database.new(@database)
|
45
|
+
assert_raises { match_set.add_match(1, 2, 1) }
|
46
|
+
end
|
47
|
+
|
48
|
+
test "add_match" do
|
49
|
+
match_set = Linkage::MatchSets::Database.new(@database)
|
50
|
+
@database.stubs(:table_exists?).with(:matches).returns(false)
|
51
|
+
@database.stubs(:create_table)
|
52
|
+
match_set.open_for_writing
|
53
|
+
|
54
|
+
@dataset.expects(:insert).with({
|
55
|
+
:id_1 => 1, :id_2 => 2, :score => 1
|
56
|
+
})
|
57
|
+
match_set.add_match(1, 2, 1)
|
58
|
+
end
|
59
|
+
|
60
|
+
test "registers itself" do
|
61
|
+
assert_equal Linkage::MatchSets::Database, Linkage::MatchSet['database']
|
62
|
+
end
|
63
|
+
end
|