|
1
|
|
|
import pytest |
|
2
|
|
|
|
|
3
|
|
|
|
|
4
|
|
|
@pytest.fixture |
|
5
|
|
|
def assert_selected_variables_are(somagic): |
|
6
|
|
|
def _assert_selected_variables_are(variables: set): |
|
7
|
|
|
assert set([x['variable'] for x in somagic._data_manager.feature_manager.feature_configuration.variables]) == variables |
|
8
|
|
|
return _assert_selected_variables_are |
|
9
|
|
|
|
|
10
|
|
|
|
|
11
|
|
|
@pytest.fixture |
|
12
|
|
|
def assert_column_values(test_dataset): |
|
13
|
|
|
import pandas as pd |
|
14
|
|
|
|
|
15
|
|
|
def _assert_column_values_are(attribute, expected_values): |
|
16
|
|
|
assert set([_ for _ in test_dataset[0].datapoints.observations[attribute]]) == set(expected_values) |
|
17
|
|
|
assert set(pd.unique(test_dataset[0].datapoints.observations[attribute])) == set(expected_values) |
|
18
|
|
|
return _assert_column_values_are |
|
19
|
|
|
|
|
20
|
|
|
|
|
21
|
|
|
@pytest.fixture |
|
22
|
|
|
def assert_correct_nominal_variable_encoding(test_dataset): |
|
23
|
|
|
"""Test a column with each row having a string representing one of the possible values of an Attrbiute. |
|
24
|
|
|
|
|
25
|
|
|
Useful when an Attribute corresponds to a discreet Variable of type Nominal (ordering does not matter) and its |
|
26
|
|
|
observation (row) can have only one of the possible values. |
|
27
|
|
|
""" |
|
28
|
|
|
from collections import Counter |
|
29
|
|
|
|
|
30
|
|
|
def _assert_nominal_variable_encoded_as_expected(expected_feature_columns): |
|
31
|
|
|
assert all(Counter([datarow[_] for _ in expected_feature_columns]) == |
|
32
|
|
|
Counter({0: len(expected_feature_columns) - 1, 1: 1}) |
|
33
|
|
|
for index, datarow in test_dataset[0].datapoints.observations[expected_feature_columns].iterrows()) |
|
34
|
|
|
return _assert_nominal_variable_encoded_as_expected |
|
35
|
|
|
|
|
36
|
|
|
|
|
37
|
|
|
def test_sanity_checks_on_dataset(test_dataset, assert_selected_variables_are, assert_column_values, |
|
38
|
|
|
assert_correct_nominal_variable_encoding): |
|
39
|
|
|
ATTRS2 = [f'type_{x}' for x in test_dataset[1]] |
|
40
|
|
|
|
|
41
|
|
|
datapoints = test_dataset[0].datapoints |
|
42
|
|
|
assert_selected_variables_are({'type', 'flavors'}) |
|
43
|
|
|
|
|
44
|
|
|
assert all(type(x) == str for x in datapoints.observations['type']) |
|
45
|
|
|
|
|
46
|
|
|
assert_column_values('type', expected_values=test_dataset[1]) |
|
47
|
|
|
|
|
48
|
|
|
assert_correct_nominal_variable_encoding(ATTRS2) |
|
49
|
|
|
|
|
50
|
|
|
# the below is expected because test_dataset invokes the 'one_hot_encoding_list_command' command which unfortunately |
|
51
|
|
|
# at the moment has a side effect on the attribute it operates on. |
|
52
|
|
|
# side effect: _data_manager.datapoints.observations[_attribute].fillna(value=np.nan, inplace=True) |
|
53
|
|
|
assert set([type(x) for x in datapoints.observations['flavors']]) == {list, float} |
|
54
|
|
|
|
|
55
|
|
|
assert len(test_dataset[2]) > 5 |
|
56
|
|
|
|
|
57
|
|
|
assert all(x in datapoints.observations.columns for x in test_dataset[2]) |
|
58
|
|
|
assert all(0 <= sum([datarow[_] for _ in test_dataset[2]]) <= test_dataset[3] |
|
59
|
|
|
for index, datarow in datapoints.observations[list(test_dataset[2])].iterrows()) |
|
60
|
|
|
|
|
61
|
|
|
assert hasattr(test_dataset[0], 'feature_vectors') |
|
62
|
|
|
|
|
63
|
|
|
|
|
64
|
|
|
@pytest.mark.parametrize('train_args', [ |
|
65
|
|
|
([6, 8, 'toroid', 'hexagonal']), |
|
66
|
|
|
# ([12, 12, 'toroid', 'rectangular']) |
|
67
|
|
|
]) |
|
68
|
|
|
def test_somagic_scenario(train_args, somagic, test_dataset, sample_collaped_json): |
|
69
|
|
|
attrs = ('width', 'height', 'type', 'grid_type') |
|
70
|
|
|
som = somagic.map.train(*train_args[:2], maptype=train_args[2], gridtype=train_args[3]) |
|
71
|
|
|
assert som.dataset_name == sample_collaped_json |
|
72
|
|
|
assert all(parameter == getattr(som, attribute) for attribute, parameter in zip(attrs, train_args)) |
|
73
|
|
|
|
|
74
|
|
|
|
|
75
|
|
|
@pytest.fixture |
|
76
|
|
|
def so_magic_instances(): |
|
77
|
|
|
from so_magic import init_so_magic |
|
78
|
|
|
return [init_so_magic(), init_so_magic()] |
|
79
|
|
|
|
|
80
|
|
|
|
|
81
|
|
|
@pytest.mark.parametrize('nb_objects, nb_observers', [ |
|
82
|
|
|
(2, [(1, 1, 1), |
|
83
|
|
|
(1, 1, 1)]), |
|
84
|
|
|
]) |
|
85
|
|
|
def test_somagic_objects(nb_objects, so_magic_instances, nb_observers): |
|
86
|
|
|
assert id(so_magic_instances[0]) != id(so_magic_instances[1]) |
|
87
|
|
|
assert id(so_magic_instances[0]._data_manager) != id(so_magic_instances[1]._data_manager) |
|
88
|
|
|
assert id(so_magic_instances[0]._data_manager.engine) != id(so_magic_instances[1]._data_manager.engine) |
|
89
|
|
|
assert id(so_magic_instances[0]._data_manager.engine.backend) != id(so_magic_instances[1]._data_manager.engine.backend) |
|
90
|
|
|
assert id(so_magic_instances[0]._data_manager.engine.datapoints_manager) != id(so_magic_instances[1]._data_manager.engine.datapoints_manager) |
|
91
|
|
|
|
|
92
|
|
|
assert id(so_magic_instances[0]._data_manager.engine.backend.datapoints_factory) != id(so_magic_instances[1]._data_manager.engine.backend.datapoints_factory) |
|
93
|
|
|
assert id(so_magic_instances[0]._data_manager.engine.backend.datapoints_factory.subject) != id(so_magic_instances[1]._data_manager.engine.backend.datapoints_factory.subject) |
|
94
|
|
|
assert id(so_magic_instances[0]._data_manager.engine.backend.datapoints_factory.subject._observers) != id(so_magic_instances[1]._data_manager.engine.backend.datapoints_factory.subject._observers) |
|
95
|
|
|
|
|
96
|
|
|
assert id(so_magic_instances[0]._data_manager.engine.backend.command_factory) != id(so_magic_instances[1]._data_manager.engine.backend.command_factory) |
|
97
|
|
|
|
|
98
|
|
|
assert so_magic_instances[0]._data_manager.phi_class != so_magic_instances[1]._data_manager.phi_class |
|
99
|
|
|
assert id(so_magic_instances[0]._data_manager.phi_class) != id(so_magic_instances[1]._data_manager.phi_class) |
|
100
|
|
|
assert id(so_magic_instances[0]._data_manager.phi_class.subject) != id(so_magic_instances[1]._data_manager.phi_class.subject) |
|
101
|
|
|
assert id(so_magic_instances[0]._data_manager.phi_class.subject._observers) != id(so_magic_instances[1]._data_manager.phi_class.subject._observers) |
|
102
|
|
|
|
|
103
|
|
|
|
|
104
|
|
|
def test_subscriptions(so_magic_instances): |
|
105
|
|
|
s = so_magic_instances[0] |
|
106
|
|
|
datapoints_fact = s._data_manager.engine.backend.datapoints_factory |
|
107
|
|
|
cmd_fact = s._data_manager.engine.backend.command_factory |
|
108
|
|
|
phi_class = s._data_manager.phi_class |
|
109
|
|
|
nb_observers = (1, 1, 1) |
|
110
|
|
|
subjects = [datapoints_fact.subject, |
|
111
|
|
|
cmd_fact.subject, |
|
112
|
|
|
phi_class.subject |
|
113
|
|
|
] |
|
114
|
|
|
|
|
115
|
|
|
assert datapoints_fact.subject._observers[0] == s._data_manager.engine.datapoints_manager |
|
116
|
|
|
assert cmd_fact.subject._observers[0] == s._data_manager.commands_manager.command.accumulator |
|
117
|
|
|
assert phi_class.subject._observers[0] == s._data_manager.built_phis |
|
118
|
|
|
assert all([len(subject._observers) == column for subject, column in zip(subjects, nb_observers)]) |
|
119
|
|
|
|
boromir674 /
so-magic
