forked from mir-evaluation/mir_eval
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_util.py
339 lines (278 loc) · 10.8 KB
/
test_util.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
''' Unit tests for utils
'''
import collections
import numpy as np
import nose.tools
import mir_eval
from mir_eval import util
def test_interpolate_intervals():
"""Check that an interval set is interpolated properly, with boundaries
conditions and out-of-range values.
"""
labels = list('abc')
intervals = np.array([(n, n + 1.0) for n in range(len(labels))])
time_points = [-1.0, 0.1, 0.9, 1.0, 2.3, 4.0]
expected_ans = ['N', 'a', 'a', 'b', 'c', 'N']
assert (util.interpolate_intervals(intervals, labels, time_points, 'N') ==
expected_ans)
def test_interpolate_intervals_gap():
"""Check that an interval set is interpolated properly, with gaps."""
labels = list('abc')
intervals = np.array([[0.5, 1.0], [1.5, 2.0], [2.5, 3.0]])
time_points = [0.0, 0.75, 1.25, 1.75, 2.25, 2.75, 3.5]
expected_ans = ['N', 'a', 'N', 'b', 'N', 'c', 'N']
assert (util.interpolate_intervals(intervals, labels, time_points, 'N') ==
expected_ans)
@nose.tools.raises(ValueError)
def test_interpolate_intervals_badtime():
"""Check that interpolate_intervals throws an exception if
input is unordered.
"""
labels = list('abc')
intervals = np.array([(n, n + 1.0) for n in range(len(labels))])
time_points = [-1.0, 0.1, 0.9, 0.8, 2.3, 4.0]
mir_eval.util.interpolate_intervals(intervals, labels, time_points)
def test_intervals_to_samples():
"""Check that an interval set is sampled properly, with boundaries
conditions and out-of-range values.
"""
labels = list('abc')
intervals = np.array([(n, n + 1.0) for n in range(len(labels))])
expected_times = [0.0, 0.5, 1.0, 1.5, 2.0, 2.5]
expected_labels = ['a', 'a', 'b', 'b', 'c', 'c']
result = util.intervals_to_samples(
intervals, labels, offset=0, sample_size=0.5, fill_value='N')
assert result[0] == expected_times
assert result[1] == expected_labels
expected_times = [0.25, 0.75, 1.25, 1.75, 2.25, 2.75]
expected_labels = ['a', 'a', 'b', 'b', 'c', 'c']
result = util.intervals_to_samples(
intervals, labels, offset=0.25, sample_size=0.5, fill_value='N')
assert result[0] == expected_times
assert result[1] == expected_labels
def test_intersect_files():
"""Check that two non-identical yield correct results.
"""
flist1 = ['/a/b/abc.lab', '/c/d/123.lab', '/e/f/xyz.lab']
flist2 = ['/g/h/xyz.npy', '/i/j/123.txt', '/k/l/456.lab']
sublist1, sublist2 = util.intersect_files(flist1, flist2)
assert sublist1 == ['/e/f/xyz.lab', '/c/d/123.lab']
assert sublist2 == ['/g/h/xyz.npy', '/i/j/123.txt']
sublist1, sublist2 = util.intersect_files(flist1[:1], flist2[:1])
assert sublist1 == []
assert sublist2 == []
def test_merge_labeled_intervals():
"""Check that two labeled interval sequences merge correctly.
"""
x_intvs = np.array([
[0.0, 0.44],
[0.44, 2.537],
[2.537, 4.511],
[4.511, 6.409]])
x_labels = ['A', 'B', 'C', 'D']
y_intvs = np.array([
[0.0, 0.464],
[0.464, 2.415],
[2.415, 4.737],
[4.737, 6.409]])
y_labels = [0, 1, 2, 3]
expected_intvs = [
[0.0, 0.44],
[0.44, 0.464],
[0.464, 2.415],
[2.415, 2.537],
[2.537, 4.511],
[4.511, 4.737],
[4.737, 6.409]]
expected_x_labels = ['A', 'B', 'B', 'B', 'C', 'D', 'D']
expected_y_labels = [0, 0, 1, 2, 2, 2, 3]
new_intvs, new_x_labels, new_y_labels = util.merge_labeled_intervals(
x_intvs, x_labels, y_intvs, y_labels)
assert new_x_labels == expected_x_labels
assert new_y_labels == expected_y_labels
assert new_intvs.tolist() == expected_intvs
# Check that invalid inputs raise a ValueError
y_intvs[-1, -1] = 10.0
nose.tools.assert_raises(ValueError, util.merge_labeled_intervals, x_intvs,
x_labels, y_intvs, y_labels)
def test_boundaries_to_intervals():
# Basic tests
boundaries = np.arange(10)
correct_intervals = np.array([np.arange(10 - 1), np.arange(1, 10)]).T
intervals = mir_eval.util.boundaries_to_intervals(boundaries)
assert np.all(intervals == correct_intervals)
def test_adjust_events():
# Test appending at the end
events = np.arange(1, 11)
labels = [str(n) for n in range(10)]
new_e, new_l = mir_eval.util.adjust_events(events, labels, 0.0, 11.)
assert new_e[0] == 0.
assert new_l[0] == '__T_MIN'
assert new_e[-1] == 11.
assert new_l[-1] == '__T_MAX'
assert np.all(new_e[1:-1] == events)
assert new_l[1:-1] == labels
# Test trimming
new_e, new_l = mir_eval.util.adjust_events(events, labels, 0.0, 9.)
assert new_e[0] == 0.
assert new_l[0] == '__T_MIN'
assert new_e[-1] == 9.
assert np.all(new_e[1:] == events[:-1])
assert new_l[1:] == labels[:-1]
def test_bipartite_match():
# This test constructs a graph as follows:
# v9 -- (u0)
# v8 -- (u0, u1)
# v7 -- (u0, u1, u2)
# ...
# v0 -- (u0, u1, ..., u9)
#
# This structure and ordering of this graph should force Hopcroft-Karp to
# hit each algorithm/layering phase
#
G = collections.defaultdict(list)
u_set = ['u{:d}'.format(_) for _ in range(10)]
v_set = ['v{:d}'.format(_) for _ in range(len(u_set)+1)]
for i, u in enumerate(u_set):
for v in v_set[:-i-1]:
G[v].append(u)
matching = util._bipartite_match(G)
# Make sure that each u vertex is matched
nose.tools.eq_(len(matching), len(u_set))
# Make sure that there are no duplicate keys
lhs = set([k for k in matching])
rhs = set([matching[k] for k in matching])
nose.tools.eq_(len(matching), len(lhs))
nose.tools.eq_(len(matching), len(rhs))
# Finally, make sure that all detected edges are present in G
for k in matching:
v = matching[k]
assert v in G[k] or k in G[v]
def test_outer_distance_mod_n():
ref = [1., 2., 3.]
est = [1.1, 6., 1.9, 5., 10.]
expected = np.array([
[0.1, 5., 0.9, 4., 3.],
[0.9, 4., 0.1, 3., 4.],
[1.9, 3., 1.1, 2., 5.]])
actual = mir_eval.util._outer_distance_mod_n(ref, est)
assert np.allclose(actual, expected)
ref = [13., 14., 15.]
est = [1.1, 6., 1.9, 5., 10.]
expected = np.array([
[0.1, 5., 0.9, 4., 3.],
[0.9, 4., 0.1, 3., 4.],
[1.9, 3., 1.1, 2., 5.]])
actual = mir_eval.util._outer_distance_mod_n(ref, est)
assert np.allclose(actual, expected)
def test_outer_distance():
ref = [1., 2., 3.]
est = [1.1, 6., 1.9, 5., 10.]
expected = np.array([
[0.1, 5., 0.9, 4., 9.],
[0.9, 4., 0.1, 3., 8.],
[1.9, 3., 1.1, 2., 7.]])
actual = mir_eval.util._outer_distance(ref, est)
assert np.allclose(actual, expected)
def test_match_events():
ref = [1., 2., 3.]
est = [1.1, 6., 1.9, 5., 10.]
expected = [(0, 0), (1, 2)]
actual = mir_eval.util.match_events(ref, est, 0.5)
assert actual == expected
ref = [1., 2., 3., 11.9]
est = [1.1, 6., 1.9, 5., 10., 0.]
expected = [(0, 0), (1, 2), (3, 5)]
actual = mir_eval.util.match_events(
ref, est, 0.5, distance=mir_eval.util._outer_distance_mod_n)
assert actual == expected
def test_validate_intervals():
# Test for ValueError when interval shape is invalid
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_intervals,
np.array([[1.], [2.5], [5.]]))
# Test for ValueError when times are negative
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_intervals,
np.array([[1., -2.], [2.5, 3.], [5., 6.]]))
# Test for ValueError when duration is zero
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_intervals,
np.array([[1., 2.], [2.5, 2.5], [5., 6.]]))
# Test for ValueError when duration is negative
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_intervals,
np.array([[1., 2.], [2.5, 1.5], [5., 6.]]))
def test_validate_events():
# Test for ValueError when max_time is violated
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_events, np.array([100., 100000.]))
# Test for ValueError when events aren't 1-d arrays
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_events,
np.array([[1., 2.], [3., 4.]]))
# Test for ValueError when event times are not increasing
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_events,
np.array([1., 2., 5., 3.]))
def test_validate_frequencies():
# Test for ValueError when max_freq is violated
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_frequencies,
np.array([100., 100000.]), 5000., 20.)
# Test for ValueError when min_freq is violated
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_frequencies,
np.array([2., 200.]), 5000., 20.)
# Test for ValueError when events aren't 1-d arrays
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_frequencies,
np.array([[100., 200.], [300., 400.]]), 5000., 20.)
# Test for ValueError when allow_negatives is false and negative values
# are passed
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_frequencies,
np.array([[-100., 200.], [300., 400.]]), 5000., 20.,
allow_negatives=False)
# Test for ValueError when max_freq is violated and allow_negatives=True
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_frequencies,
np.array([100., -100000.]), 5000., 20., allow_negatives=True)
# Test for ValueError when min_freq is violated and allow_negatives=True
nose.tools.assert_raises(
ValueError, mir_eval.util.validate_frequencies,
np.array([-2., 200.]), 5000., 20., allow_negatives=True)
def test_has_kwargs():
def __test(target, f):
assert target == mir_eval.util.has_kwargs(f)
def f1(_):
return None
def f2(_=5):
return None
def f3(*_):
return None
def f4(_, **kw):
return None
def f5(_=5, **kw):
return None
yield __test, False, f1
yield __test, False, f2
yield __test, False, f3
yield __test, True, f4
yield __test, True, f5
def test_sort_labeled_intervals():
def __test_labeled(x, labels, x_true, lab_true):
xs, ls = mir_eval.util.sort_labeled_intervals(x, labels)
assert np.allclose(xs, x_true)
nose.tools.eq_(ls, lab_true)
def __test(x, x_true):
xs = mir_eval.util.sort_labeled_intervals(x)
assert np.allclose(xs, x_true)
x1 = np.asarray([[10, 20], [0, 10]])
x1_true = np.asarray([[0, 10], [10, 20]])
labels = ['a', 'b']
labels_true = ['b', 'a']
yield __test_labeled, x1, labels, x1_true, labels_true
yield __test, x1, x1_true
yield __test_labeled, x1_true, labels_true, x1_true, labels_true
yield __test, x1_true, x1_true