|
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
|
|
|
|
|
48
|
|
|
data_manager = init_data_manager(init_engine(engine_type='pd')) |
|
49
|
|
|
|
|
50
|
|
|
datapoints_fact = data_manager.engine.backend.datapoints_factory |
|
51
|
|
|
cmd_fact = data_manager.engine.backend.command_factory |
|
52
|
|
|
|
|
53
|
|
|
# test 1 |
|
54
|
|
|
from so_magic.data.datapoints.datapoints import DatapointsFactory |
|
55
|
|
|
from so_magic.data.backend.engine_command_factory import MagicCommandFactory |
|
56
|
|
|
|
|
57
|
|
|
assert isinstance(datapoints_fact, DatapointsFactory) |
|
58
|
|
|
assert isinstance(cmd_fact, MagicCommandFactory) |
|
59
|
|
|
|
|
60
|
|
|
subjects = [datapoints_fact.subject, cmd_fact, data_manager.phi_class.subject] |
|
61
|
|
|
assert len(set([id(x._observers) for x in subjects])) == len(subjects) |
|
62
|
|
|
|
|
63
|
|
|
assert datapoints_fact.subject._observers[0] == data_manager.engine.datapoints_manager |
|
64
|
|
|
assert cmd_fact._observers[0] == data_manager.commands_manager.command.accumulator |
|
65
|
|
|
assert data_manager.phi_class.subject._observers[0] == data_manager.built_phis |
|
66
|
|
|
|
|
67
|
|
|
print(f"DTP FCT OBS: [{', '.join(str(_) for _ in datapoints_fact.subject._observers)}]") |
|
68
|
|
|
print(f"CMD FCT OBS: [{', '.join(str(_) for _ in cmd_fact._observers)}]") |
|
69
|
|
|
print(f"PHIFUNC class OBS: [{', '.join(str(_) for _ in data_manager.phi_class.subject._observers)}]") |
|
70
|
|
|
assert all([len(x._observers) == 1 for x in subjects]) |
|
71
|
|
|
return data_manager |
|
72
|
|
|
|
|
73
|
|
|
return getter |
|
74
|
|
|
|
|
75
|
|
|
|
|
76
|
|
|
@pytest.fixture |
|
77
|
|
|
def test_data_manager(data_manager): |
|
78
|
|
|
return data_manager() |
|
79
|
|
|
|
|
80
|
|
|
|
|
81
|
|
|
@pytest.fixture |
|
82
|
|
|
def load_test_data(test_data_manager, sample_json): |
|
83
|
|
|
def load_data(json_lines_formatted_file_path): |
|
84
|
|
|
cmd = test_data_manager.command.observations_command |
|
85
|
|
|
cmd.args = [json_lines_formatted_file_path] |
|
86
|
|
|
cmd.execute() |
|
87
|
|
|
return lambda: load_data(sample_json) |
|
88
|
|
|
|
|
89
|
|
|
|
|
90
|
|
|
@pytest.fixture |
|
91
|
|
|
def read_observations(): |
|
92
|
|
|
"""Read a json lines formatted file and create the observations object (see Datapoints class).""" |
|
93
|
|
|
def load_data(so_master, json_lines_formatted_file_path): |
|
94
|
|
|
"""Create the observations object for a Datapoints instance, given a data file. |
|
95
|
|
|
|
|
96
|
|
|
Args: |
|
97
|
|
|
so_master (so_magic.so_master.SoMaster): an instance of SoMaster |
|
98
|
|
|
json_lines_formatted_file_path (str): path to a json lines formatted file with the observations data |
|
99
|
|
|
""" |
|
100
|
|
|
cmd = so_master.command.observations_command |
|
101
|
|
|
cmd.args = [json_lines_formatted_file_path] |
|
102
|
|
|
cmd.execute() |
|
103
|
|
|
return load_data |
|
104
|
|
|
|
|
105
|
|
|
|
|
106
|
|
|
@pytest.fixture |
|
107
|
|
|
def test_datapoints(read_observations, sample_collaped_json, somagic): |
|
108
|
|
|
"""Read the designated json lines 'test file' (which contains the 'test observations') as a Datapoints instance.""" |
|
109
|
|
|
read_observations(somagic, sample_collaped_json) |
|
110
|
|
|
return somagic.datapoints |
|
111
|
|
|
|
|
112
|
|
|
|
|
113
|
|
|
@pytest.fixture |
|
114
|
|
|
def test_dataset(somagic, read_observations, sample_collaped_json): |
|
115
|
|
|
"""Dataset ready to be fed into a training/inference algorithm; feature vectors have been computed.""" |
|
116
|
|
|
read_observations(somagic, sample_collaped_json) |
|
117
|
|
|
|
|
118
|
|
|
ATTRS2 = ['type_hybrid', 'type_indica', 'type_sativa'] |
|
119
|
|
|
from functools import reduce |
|
120
|
|
|
UNIQUE_FLAVORS = reduce(lambda i, j: set(i).union(set(j)), |
|
121
|
|
|
[_ for _ in somagic._data_manager.datapoints.observations['flavors'] if _ is not None]) |
|
122
|
|
|
|
|
123
|
|
|
cmd = somagic._data_manager.command.select_variables_command |
|
124
|
|
|
# current limitations: |
|
125
|
|
|
# 1. client code has to know the number of distict values for the nominal variable 'type' |
|
126
|
|
|
# 2. client code has to provide the column names that will result after encoding the 'type' variable |
|
127
|
|
|
cmd.args = [[ |
|
128
|
|
|
# current limitations: |
|
129
|
|
|
# 1. client code has to know the number of distict values for the nominal variable 'type' |
|
130
|
|
|
# 2. client code has to provide the column names that will result after encoding the 'type' variable |
|
131
|
|
|
{'variable': 'type', 'columns': ATTRS2}, |
|
132
|
|
|
# current limitations: |
|
133
|
|
|
# 1. client code has to know the number of distict values for the nominal variable 'flavors' |
|
134
|
|
|
# 2. client code has to provide the column names that will result after encoding the 'flavors' variable |
|
135
|
|
|
{'variable': 'flavors', 'columns': list(UNIQUE_FLAVORS)}]] |
|
136
|
|
|
cmd.execute() |
|
137
|
|
|
|
|
138
|
|
|
cmd = somagic._data_manager.command.one_hot_encoding_command |
|
139
|
|
|
cmd.args = [somagic._data_manager.datapoints, 'type'] |
|
140
|
|
|
cmd.execute() |
|
141
|
|
|
|
|
142
|
|
|
cmd = somagic._data_manager.command.one_hot_encoding_list_command |
|
143
|
|
|
cmd.args = [somagic._data_manager.datapoints, 'flavors'] |
|
144
|
|
|
cmd.execute() |
|
145
|
|
|
|
|
146
|
|
|
import numpy as np |
|
147
|
|
|
|
|
148
|
|
|
setattr(somagic.dataset, 'feature_vectors', |
|
149
|
|
|
np.array(somagic._data_manager.datapoints.observations[ATTRS2 + list(UNIQUE_FLAVORS)])) |
|
150
|
|
|
|
|
151
|
|
|
return somagic.dataset |
|
152
|
|
|
|
|
153
|
|
|
|
|
154
|
|
|
@pytest.fixture |
|
155
|
|
|
def built_in_backends(): |
|
156
|
|
|
from so_magic.data.backend.panda_handling.df_backend import magic_backends |
|
157
|
|
|
engine_backends = magic_backends() |
|
158
|
|
|
return engine_backends |
|
159
|
|
|
|
boromir674 /
so-magic
