|
1
|
|
|
import os |
|
2
|
|
|
import pytest |
|
3
|
|
|
|
|
4
|
|
|
|
|
5
|
|
|
my_dir = os.path.dirname(os.path.realpath(__file__)) |
|
6
|
|
|
|
|
7
|
|
|
####### Files and folders |
|
8
|
|
|
@pytest.fixture |
|
9
|
|
|
def tests_root_dir(): |
|
10
|
|
|
return my_dir |
|
11
|
|
|
|
|
12
|
|
|
@pytest.fixture |
|
13
|
|
|
def tests_data_root(tests_root_dir): |
|
14
|
|
|
return os.path.join(tests_root_dir, 'dts') |
|
15
|
|
|
|
|
16
|
|
|
# Test data |
|
17
|
|
|
@pytest.fixture |
|
18
|
|
|
def sample_json(tests_data_root): |
|
19
|
|
|
return os.path.join(tests_data_root, 'sample-data.jsonlines') |
|
20
|
|
|
|
|
21
|
|
|
@pytest.fixture |
|
22
|
|
|
def sample_collaped_json(tests_data_root): |
|
23
|
|
|
return os.path.join(tests_data_root, 'sample-data-collapsed.jsonlines') |
|
24
|
|
|
|
|
25
|
|
|
|
|
26
|
|
|
@pytest.fixture() |
|
27
|
|
|
def test_json_data(sample_json): |
|
28
|
|
|
return { |
|
29
|
|
|
'file_path': sample_json, |
|
30
|
|
|
'nb_lines': 100, |
|
31
|
|
|
'attributes': {'flavors', 'name', 'medical', 'description', 'image_urls', 'parents', 'negatives', 'grow_info', '_id', 'type', 'image_paths', 'effects'}, |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
|
|
35
|
|
|
@pytest.fixture |
|
36
|
|
|
def somagic(): |
|
37
|
|
|
from so_magic import init_so_magic |
|
38
|
|
|
_ = init_so_magic() |
|
39
|
|
|
return _ |
|
40
|
|
|
|
|
41
|
|
|
|
|
42
|
|
|
@pytest.fixture |
|
43
|
|
|
def data_manager(): |
|
44
|
|
|
def getter(): |
|
45
|
|
|
from so_magic.data import init_data_manager |
|
46
|
|
|
from so_magic.data.backend import init_engine |
|
47
|
|
|
data_manager = init_data_manager(init_engine(engine_type='pd')) |
|
48
|
|
|
return data_manager |
|
49
|
|
|
return getter |
|
50
|
|
|
|
|
51
|
|
|
|
|
52
|
|
|
@pytest.fixture |
|
53
|
|
|
def read_observations(): |
|
54
|
|
|
"""Read a json lines formatted file and create the observations object (see Datapoints class).""" |
|
55
|
|
|
def load_data(so_master, json_lines_formatted_file_path): |
|
56
|
|
|
"""Create the observations object for a Datapoints instance, given a data file. |
|
57
|
|
|
|
|
58
|
|
|
Args: |
|
59
|
|
|
so_master (so_magic.so_master.SoMaster): an instance of SoMaster |
|
60
|
|
|
json_lines_formatted_file_path (str): path to a json lines formatted file with the observations data |
|
61
|
|
|
""" |
|
62
|
|
|
cmd = so_master.command.observations_command |
|
63
|
|
|
cmd.args = [json_lines_formatted_file_path] |
|
64
|
|
|
cmd.execute() |
|
65
|
|
|
return load_data |
|
66
|
|
|
|
|
67
|
|
|
|
|
68
|
|
|
@pytest.fixture |
|
69
|
|
|
def test_datapoints(read_observations, sample_collaped_json, somagic): |
|
70
|
|
|
"""Read the designated json lines 'test file' (which contains the 'test observations') as a Datapoints instance.""" |
|
71
|
|
|
read_observations(somagic, sample_collaped_json) |
|
72
|
|
|
return somagic.datapoints |
|
73
|
|
|
|
|
74
|
|
|
|
|
75
|
|
|
@pytest.fixture |
|
76
|
|
|
def test_dataset(somagic, read_observations, sample_collaped_json): |
|
77
|
|
|
"""Dataset ready to be fed into a training/inference algorithm; feature vectors have been computed.""" |
|
78
|
|
|
read_observations(somagic, sample_collaped_json) |
|
79
|
|
|
|
|
80
|
|
|
type_values = ['hybrid', 'indica', 'sativa'] |
|
81
|
|
|
ATTRS2 = [f'type_{x}' for x in type_values] |
|
82
|
|
|
from functools import reduce |
|
83
|
|
|
UNIQUE_FLAVORS = reduce(lambda i, j: set(i).union(set(j)), |
|
84
|
|
|
[_ for _ in somagic._data_manager.datapoints.observations['flavors'] if _ is not None]) |
|
85
|
|
|
|
|
86
|
|
|
cmd = somagic._data_manager.command.select_variables_command |
|
87
|
|
|
# current limitations: |
|
88
|
|
|
# 1. client code has to know the number of distict values for the nominal variable 'type' |
|
89
|
|
|
# 2. client code has to provide the column names that will result after encoding the 'type' variable |
|
90
|
|
|
cmd.args = [[ |
|
91
|
|
|
# current limitations: |
|
92
|
|
|
# 1. client code has to know the number of distict values for the nominal variable 'type' |
|
93
|
|
|
# 2. client code has to provide the column names that will result after encoding the 'type' variable |
|
94
|
|
|
{'variable': 'type', 'columns': ATTRS2}, |
|
95
|
|
|
# current limitations: |
|
96
|
|
|
# 1. client code has to know the number of distict values for the nominal variable 'flavors' |
|
97
|
|
|
# 2. client code has to provide the column names that will result after encoding the 'flavors' variable |
|
98
|
|
|
{'variable': 'flavors', 'columns': list(UNIQUE_FLAVORS)}]] |
|
99
|
|
|
cmd.execute() |
|
100
|
|
|
|
|
101
|
|
|
cmd = somagic._data_manager.command.one_hot_encoding_command |
|
102
|
|
|
cmd.args = [somagic._data_manager.datapoints, 'type'] |
|
103
|
|
|
cmd.execute() |
|
104
|
|
|
|
|
105
|
|
|
assert set([type(x) for x in somagic._data_manager.datapoints.observations['flavors']]) == {list, type(None)} |
|
106
|
|
|
|
|
107
|
|
|
nb_columns_before = len(somagic._data_manager.datapoints.observations.columns) |
|
108
|
|
|
|
|
109
|
|
|
cmd = somagic._data_manager.command.one_hot_encoding_list_command |
|
110
|
|
|
cmd.args = [somagic._data_manager.datapoints, 'flavors'] |
|
111
|
|
|
cmd.execute() |
|
112
|
|
|
|
|
113
|
|
|
assert nb_columns_before + len(UNIQUE_FLAVORS) == len(somagic._data_manager.datapoints.observations.columns) |
|
114
|
|
|
|
|
115
|
|
|
import numpy as np |
|
116
|
|
|
setattr(somagic.dataset, 'feature_vectors', |
|
117
|
|
|
np.array(somagic._data_manager.datapoints.observations[ATTRS2 + list(UNIQUE_FLAVORS)])) |
|
118
|
|
|
|
|
119
|
|
|
MAX_FLAVORS_PER_DAATPOINT = max( |
|
120
|
|
|
[len(x) for x in [_ for _ in somagic._data_manager.datapoints.observations['flavors'] if type(_) is list]]) |
|
121
|
|
|
return somagic.dataset, type_values, UNIQUE_FLAVORS, MAX_FLAVORS_PER_DAATPOINT, nb_columns_before |
|
122
|
|
|
|
|
123
|
|
|
|
|
124
|
|
|
@pytest.fixture |
|
125
|
|
|
def built_in_backends(): |
|
126
|
|
|
from so_magic.data.backend.panda_handling.df_backend import magic_backends |
|
127
|
|
|
engine_backends = magic_backends() |
|
128
|
|
|
return engine_backends |
|
129
|
|
|
|
|
130
|
|
|
|
|
131
|
|
|
@pytest.fixture |
|
132
|
|
|
def tabular_operators(built_in_backends): |
|
133
|
|
|
operators = { |
|
134
|
|
|
'retriever': { |
|
135
|
|
|
'class': built_in_backends.backend_interfaces['retriever']['class_registry'].subclasses['pd'], |
|
136
|
|
|
'interface': { |
|
137
|
|
|
'column': '(identifier, data)', |
|
138
|
|
|
'row': '(identifier, data)', |
|
139
|
|
|
'nb_columns': '(data)', |
|
140
|
|
|
'nb_rows': '(data)', |
|
141
|
|
|
'get_numerical_attributes': '(data)', |
|
142
|
|
|
} |
|
143
|
|
|
}, |
|
144
|
|
|
'iterator': { |
|
145
|
|
|
'class': built_in_backends.backend_interfaces['iterator']['class_registry'].subclasses['pd'], |
|
146
|
|
|
'interface': { |
|
147
|
|
|
'columnnames': '(data)', |
|
148
|
|
|
'itercolumns': '(data)', |
|
149
|
|
|
'iterrows': '(data)', |
|
150
|
|
|
}, |
|
151
|
|
|
}, |
|
152
|
|
|
'mutator': { |
|
153
|
|
|
'class': built_in_backends.backend_interfaces['mutator']['class_registry'].subclasses['pd'], |
|
154
|
|
|
'interface': { |
|
155
|
|
|
'add_column': '(datapoints, values, new_attribute, **kwargs)', |
|
156
|
|
|
}, |
|
157
|
|
|
}, |
|
158
|
|
|
} |
|
159
|
|
|
return { |
|
160
|
|
|
'operators': operators, |
|
161
|
|
|
'reverse_dict': {operator_dict['class']: key for key, operator_dict in operators.items()}, |
|
162
|
|
|
'get_nb_args': lambda operator_interface_name, method_name: len(operators[operator_interface_name]['interface'][method_name].replace(', **kwargs', '').split(',')), |
|
163
|
|
|
# operator_name_2_required_methods |
|
164
|
|
|
'required_methods': iter(((operator_interface_name, v['interface'].keys()) |
|
|
|
|
|
|
165
|
|
|
for operator_interface_name, v in operators.items())) |
|
166
|
|
|
} |
|
167
|
|
|
|
|
168
|
|
|
|
|
169
|
|
|
@pytest.fixture |
|
170
|
|
|
def assert_different_objects(): |
|
171
|
|
|
def _assert_different_objects(objects): |
|
172
|
|
|
assert len(set([id(x) for x in objects])) == len(objects) |
|
173
|
|
|
return _assert_different_objects |
|
174
|
|
|
|
boromir674 /
so-magic
Loading history...