Coverage for tests\sampling\test_factory_random_in_same_cluster.py: 100.00%
44 statements
« prev ^ index » next coverage.py v7.3.2, created at 2023-11-17 13:31 +0100
1# -*- coding: utf-8 -*-
3"""
4* Name: interactive-clustering/tests/sampling/test_factory_random_in_same_cluster.py
5* Description: Unittests for the `sampling.cluster_based` module, `"random_in_same_cluster"` sampler.
6* Author: Erwan SCHILD
7* Created: 17/03/2021
8* Licence: CeCILL (https://cecill.info/licences.fr.html)
9"""
11# ==============================================================================
12# IMPORT PYTHON DEPENDENCIES
13# ==============================================================================
15import pytest
17from cognitivefactory.interactive_clustering.constraints.binary import BinaryConstraintsManager
18from cognitivefactory.interactive_clustering.sampling.clusters_based import ClustersBasedConstraintsSampling
21# ==============================================================================
22# test_factory_random_in_same_cluster_sampler_for_correct_settings
23# ==============================================================================
24def test_factory_random_in_same_cluster_sampler_for_correct_settings():
25 """
26 Test that the `random_in_same_cluster sampler` works for correct settings.
27 """
29 # Check a correct initialization.
30 sampler = ClustersBasedConstraintsSampling(
31 clusters_restriction="same_cluster",
32 random_seed=1,
33 )
35 assert sampler
36 assert sampler.random_seed == 1
39# ==============================================================================
40# test_factory_random_in_same_cluster_sampler_sample_for_incorrect_constraints_manager
41# ==============================================================================
42def test_factory_random_in_same_cluster_sampler_sample_for_incorrect_constraints_manager():
43 """
44 Test that the `random_in_same_cluster sampler` sampling raises `ValueError` for incorrect `constraints_manager`.
45 """
47 # Initialize a `random_in_same sampler` instance.
48 sampler = ClustersBasedConstraintsSampling(
49 clusters_restriction="same_cluster",
50 random_seed=1,
51 )
53 # Check sample with incorrect `constraints_manager`.
54 with pytest.raises(ValueError, match="`constraints_manager`"):
55 sampler.sample(
56 constraints_manager=None,
57 nb_to_select=None,
58 )
61# ==============================================================================
62# test_factory_random_in_same_cluster_sampler_sample_for_incorrect_nb_to_select
63# ==============================================================================
64def test_factory_random_in_same_cluster_sampler_sample_for_incorrect_nb_to_select():
65 """
66 Test that the `random_in_same_cluster sampler` sampling raises `ValueError` for incorrect `nb_to_select`.
67 """
69 # Initialize a `random_in_same sampler` instance.
70 sampler = ClustersBasedConstraintsSampling(
71 clusters_restriction="same_cluster",
72 random_seed=1,
73 )
75 # Check sample with incorrect `nb_to_select`.
76 with pytest.raises(ValueError, match="`nb_to_select`"):
77 sampler.sample(
78 constraints_manager=BinaryConstraintsManager(
79 list_of_data_IDs=[
80 "bonjour",
81 "salut",
82 "coucou",
83 "au revoir",
84 "a bientôt",
85 ]
86 ),
87 nb_to_select=None,
88 )
90 # Check sample with incorrect `nb_to_select`
91 with pytest.raises(ValueError, match="`nb_to_select`"):
92 sampler.sample(
93 constraints_manager=BinaryConstraintsManager(
94 list_of_data_IDs=[
95 "bonjour",
96 "salut",
97 "coucou",
98 "au revoir",
99 "a bientôt",
100 ],
101 ),
102 nb_to_select=-99,
103 )
106# ==============================================================================
107# test_factory_random_in_same_cluster_sampler_sample_for_zero_nb_to_select
108# ==============================================================================
109def test_factory_random_in_same_cluster_sampler_sample_for_zero_nb_to_select():
110 """
111 Test that the `random_in_same_cluster sampler` sampling works for zero `nb_to_select`.
112 """
114 # Initialize a `random_in_same sampler` instance.
115 sampler = ClustersBasedConstraintsSampling(
116 clusters_restriction="same_cluster",
117 random_seed=1,
118 )
120 # Check sample with zero `nb_to_select`.
121 assert not sampler.sample(
122 constraints_manager=BinaryConstraintsManager(
123 list_of_data_IDs=[
124 "bonjour",
125 "salut",
126 "coucou",
127 "au revoir",
128 "a bientôt",
129 ],
130 ),
131 nb_to_select=0,
132 )
135# ==============================================================================
136# test_factory_random_in_same_cluster_sampler_sample_for_incorrect_clustering_result
137# ==============================================================================
138def test_factory_random_in_same_cluster_sampler_sample_for_incorrect_clustering_result():
139 """
140 Test that the `random_in_same_cluster sampler` sampling raises `ValueError` or `KeyError` for incorrect `clustering_result`.
141 """
143 # Initialize a `random_in_same sampler` instance.
144 sampler = ClustersBasedConstraintsSampling(
145 clusters_restriction="same_cluster",
146 random_seed=1,
147 )
149 # Check sample with incorrect `clustering_result`.
150 with pytest.raises(ValueError, match="`clustering_result`"):
151 sampler.sample(
152 constraints_manager=BinaryConstraintsManager(
153 list_of_data_IDs=[
154 "bonjour",
155 "salut",
156 "coucou",
157 "au revoir",
158 "a bientôt",
159 ],
160 ),
161 nb_to_select=3,
162 clustering_result="unknown",
163 )
165 # Check sample with incorrect `clustering_result`.
166 with pytest.raises(KeyError, match="'a bientôt'|'au revoir'|'bonjour'|'coucou'|'salut'"):
167 sampler.sample(
168 constraints_manager=BinaryConstraintsManager(
169 list_of_data_IDs=[
170 "bonjour",
171 "salut",
172 "coucou",
173 "au revoir",
174 "a bientôt",
175 ],
176 ),
177 nb_to_select=3,
178 clustering_result={
179 "first": 1,
180 "second": 2,
181 },
182 )
185# ==============================================================================
186# test_factory_random_in_same_cluster_sampler_sample_for_empty_constraints_manager
187# ==============================================================================
188def test_factory_random_in_same_cluster_sampler_sample_for_empty_constraints_manager():
189 """
190 Test that the `random_in_same_cluster sampler` sampling works for empty `constraints_manager`.
191 """
193 # Initialize a `random_in_same sampler` instance.
194 sampler = ClustersBasedConstraintsSampling(
195 clusters_restriction="same_cluster",
196 random_seed=1,
197 )
199 # Check sample with empty `constraints_manager`.
200 assert sampler.sample(
201 constraints_manager=BinaryConstraintsManager(
202 list_of_data_IDs=[
203 "bonjour",
204 "salut",
205 "coucou",
206 "au revoir",
207 "a bientôt",
208 ],
209 ),
210 nb_to_select=3,
211 clustering_result={
212 "bonjour": 0,
213 "salut": 0,
214 "coucou": 0,
215 "au revoir": 1,
216 "a bientôt": 1,
217 },
218 ) == [("bonjour", "coucou"), ("coucou", "salut"), ("bonjour", "salut")]
221# ==============================================================================
222# test_factory_random_in_same_cluster_sampler_sample_for_correct_constraints_manager
223# ==============================================================================
224def test_factory_random_in_same_cluster_sampler_sample_for_correct_constraints_manager():
225 """
226 Test that the `random_in_same_cluster sampler` sampling works for correct `constraints_manager`.
227 """
229 # Initialize a `random_in_same sampler` instance.
230 sampler = ClustersBasedConstraintsSampling(
231 clusters_restriction="same_cluster",
232 random_seed=1,
233 )
235 # Initialize a `BinaryConstraintsManager` instance.
236 constraints_manager = BinaryConstraintsManager(
237 list_of_data_IDs=[
238 "bonjour",
239 "salut",
240 "coucou",
241 "au revoir",
242 "a bientôt",
243 ]
244 )
245 constraints_manager.add_constraint(data_ID1="bonjour", data_ID2="salut", constraint_type="MUST_LINK")
246 constraints_manager.add_constraint(data_ID1="au revoir", data_ID2="a bientôt", constraint_type="MUST_LINK")
248 # Check sample with correct `constraints_manager`.
249 assert sampler.sample(
250 constraints_manager=constraints_manager,
251 nb_to_select=3,
252 clustering_result={
253 "bonjour": 0,
254 "salut": 0,
255 "coucou": 0,
256 "au revoir": 1,
257 "a bientôt": 1,
258 },
259 ) == [("bonjour", "coucou"), ("coucou", "salut")]
262# ==============================================================================
263# test_factory_random_in_same_cluster_sampler_sample_for_full_annotated_constraints_manager
264# ==============================================================================
265def test_factory_random_in_same_cluster_sampler_sample_for_full_annotated_constraints_manager():
266 """
267 Test that the `random_in_same_cluster sampler` sampling works for full annotated `constraints_manager`.
268 """
270 # Initialize a `random_in_same sampler` instance.
271 sampler = ClustersBasedConstraintsSampling(
272 clusters_restriction="same_cluster",
273 random_seed=1,
274 )
276 # Initialize a `BinaryConstraintsManager` instance.
277 constraints_manager = BinaryConstraintsManager(
278 list_of_data_IDs=[
279 "bonjour",
280 "salut",
281 "coucou",
282 "au revoir",
283 "a bientôt",
284 ]
285 )
286 constraints_manager.add_constraint(data_ID1="bonjour", data_ID2="salut", constraint_type="MUST_LINK")
287 constraints_manager.add_constraint(data_ID1="bonjour", data_ID2="coucou", constraint_type="MUST_LINK")
288 constraints_manager.add_constraint(data_ID1="bonjour", data_ID2="au revoir", constraint_type="CANNOT_LINK")
289 constraints_manager.add_constraint(data_ID1="au revoir", data_ID2="a bientôt", constraint_type="MUST_LINK")
291 # Check sample for full annotated `constraints_manager`.
292 assert not sampler.sample(
293 constraints_manager=constraints_manager,
294 nb_to_select=3,
295 clustering_result={
296 "bonjour": 0,
297 "salut": 0,
298 "coucou": 0,
299 "au revoir": 1,
300 "a bientôt": 1,
301 },
302 )