|
1
|
|
|
import pytest |
|
2
|
|
|
from green_magic.data.commands_manager import CommandsManager |
|
3
|
|
|
from green_magic.data.backend import Backend, DataEngine |
|
4
|
|
|
from green_magic.data.data_manager import DataManager |
|
5
|
|
|
from green_magic.data.backend import panda_handling |
|
6
|
|
|
|
|
7
|
|
|
|
|
8
|
|
|
@pytest.fixture() |
|
9
|
|
|
def test_json_data(sample_json): |
|
10
|
|
|
return { |
|
11
|
|
|
'file_path': sample_json, |
|
12
|
|
|
'nb_lines': 100, |
|
13
|
|
|
'attributes': {'flavors', 'name', 'medical', 'description', 'image_urls', 'parents', 'negatives', 'grow_info', '_id', 'type', 'image_paths', 'effects'}, |
|
14
|
|
|
} |
|
15
|
|
|
|
|
16
|
|
|
def test_engine_registration(): |
|
17
|
|
|
from green_magic.data.backend import DataEngine |
|
18
|
|
|
from green_magic.data.backend.engine import EngineType |
|
19
|
|
|
assert type(DataEngine) == EngineType |
|
20
|
|
|
from green_magic.data.commands_manager import CommandRegistrator |
|
21
|
|
|
|
|
22
|
|
|
# type(type(DataEngine)) |
|
23
|
|
|
ATTRS = ('registry', '_commands', 'command') |
|
24
|
|
|
DataEngine.new('test_pd') |
|
25
|
|
|
assert 'test_pd' in DataEngine.subclasses |
|
26
|
|
|
assert type(DataEngine.test_pd) == type(DataEngine) |
|
27
|
|
|
assert all(hasattr(DataEngine.test_pd, x) for x in ATTRS) |
|
28
|
|
|
|
|
29
|
|
|
DataEngine.new('data_lib') |
|
30
|
|
|
assert 'data_lib' in DataEngine.subclasses |
|
31
|
|
|
assert type(DataEngine.data_lib) == type(DataEngine) |
|
32
|
|
|
assert all(hasattr(DataEngine.data_lib, x) for x in ATTRS) |
|
33
|
|
|
|
|
34
|
|
|
for x in ATTRS: |
|
35
|
|
|
print(getattr(DataEngine.test_pd, x), getattr(DataEngine.data_lib, x)) |
|
36
|
|
|
assert id(getattr(DataEngine.test_pd, x)) != id(getattr(DataEngine.data_lib, x)) != id(DataEngine.registry) |
|
37
|
|
|
|
|
38
|
|
|
assert id(DataEngine.test_pd.datapoints_factory) == id(DataEngine.data_lib.datapoints_factory) |
|
39
|
|
|
|
|
40
|
|
|
import pandas as pd |
|
41
|
|
|
@DataEngine.test_pd.command |
|
42
|
|
|
def observations(file_path): |
|
43
|
|
|
return pd.read_json(file_path, lines=True) |
|
44
|
|
|
|
|
45
|
|
|
# TODO uncomment below |
|
46
|
|
|
# assert len(DataEngine.test_pd.registry) == 1 |
|
47
|
|
|
# assert len(DataEngine.data_lib.registry) == 0 |
|
48
|
|
|
|
|
49
|
|
|
# assert all(id(getattr(DataEngine.test_pd, x) != id(getattr(DataEngine.data_lib, x)) for x in ATTRS)) |
|
50
|
|
|
|
|
51
|
|
|
|
|
52
|
|
|
def test_data_manager(test_json_data): |
|
53
|
|
|
import types |
|
54
|
|
|
from green_magic.utils.commands import Command |
|
55
|
|
|
from green_magic.data.backend.panda_handling.df_backend import PDTabularIterator, PDTabularRetriever, PDTabularReporter |
|
56
|
|
|
from green_magic.data.features.phi import PhiFunction |
|
57
|
|
|
|
|
58
|
|
|
DataEngine.new('test_pd') |
|
59
|
|
|
assert 'test_pd' in DataEngine.subclasses |
|
60
|
|
|
assert hasattr(DataEngine, 'state') |
|
61
|
|
|
assert hasattr(DataEngine, 'registry') |
|
62
|
|
|
|
|
63
|
|
|
data_api = DataManager(CommandsManager(), Backend(DataEngine.create('test_pd'))) |
|
64
|
|
|
# make the datapoint_manager listen to newly created Datapoints objects events |
|
65
|
|
|
assert hasattr(data_api, 'commands_manager') |
|
66
|
|
|
assert hasattr(data_api.commands_manager, 'command') |
|
67
|
|
|
assert hasattr(data_api.backend, 'datapoints_manager') |
|
68
|
|
|
|
|
69
|
|
|
data_api.backend.engine.__class__.datapoints_factory.subject.attach(data_api.backend.datapoints_manager) |
|
70
|
|
|
DataEngine.test_pd.command_factory.attach(data_api.commands_manager.command.accumulator) |
|
71
|
|
|
PhiFunction.subject.attach(data_api.phis) |
|
72
|
|
|
|
|
73
|
|
|
assert data_api.backend.engine.__class__.datapoints_factory not in PhiFunction.subject._observers |
|
74
|
|
|
|
|
75
|
|
|
# test runtime command registration |
|
76
|
|
|
import pandas as pd |
|
77
|
|
|
@DataEngine.test_pd.dec() |
|
78
|
|
|
def observations(file_path): |
|
79
|
|
|
return pd.read_json(file_path, lines=True) |
|
80
|
|
|
|
|
81
|
|
|
@DataEngine.test_pd.dec() |
|
82
|
|
|
def add_attribute(_datapoints, values, new_attribute): |
|
83
|
|
|
print("CORRECT") |
|
84
|
|
|
_datapoints.observations[new_attribute] = values |
|
85
|
|
|
|
|
86
|
|
|
assert type(DataEngine.test_pd) == type(DataEngine) |
|
87
|
|
|
assert type(DataEngine.test_pd.registry) == dict |
|
88
|
|
|
assert type(DataEngine.test_pd._commands) == dict |
|
89
|
|
|
assert 'observations' in DataEngine.test_pd.registry |
|
90
|
|
|
assert 'observations' in DataEngine.test_pd._commands |
|
91
|
|
|
assert 'observations' not in DataEngine.registry |
|
92
|
|
|
|
|
93
|
|
|
assert type(DataEngine.test_pd._commands['observations']) == Command |
|
94
|
|
|
|
|
95
|
|
|
cmd = DataEngine.test_pd._commands['observations'] |
|
96
|
|
|
# cmd = data_api.command.observations |
|
97
|
|
|
cmd.args = [test_json_data['file_path']] |
|
98
|
|
|
|
|
99
|
|
|
assert type(cmd._receiver) == types.FunctionType |
|
100
|
|
|
assert cmd._method == '__call__' |
|
101
|
|
|
assert cmd.args == [test_json_data['file_path']] |
|
102
|
|
|
|
|
103
|
|
|
from green_magic.utils.commands import Invoker, CommandHistory |
|
104
|
|
|
|
|
105
|
|
|
DataEngine.test_pd.retriever = PDTabularRetriever |
|
106
|
|
|
DataEngine.test_pd.iterator = PDTabularIterator |
|
107
|
|
|
DataEngine.test_pd.reporter = PDTabularReporter |
|
108
|
|
|
|
|
109
|
|
|
inv = Invoker(CommandHistory()) |
|
110
|
|
|
inv.execute_command(cmd) |
|
111
|
|
|
|
|
112
|
|
|
datapoints = data_api.backend.datapoints_manager.datapoints |
|
113
|
|
|
assert len(datapoints) == test_json_data['nb_lines'] |
|
114
|
|
|
print(datapoints.attributes) |
|
115
|
|
|
|
|
116
|
|
|
assert set(datapoints.attributes) == test_json_data['attributes'] |
|
117
|
|
|
|
|
118
|
|
|
assert 'add_attribute' in DataEngine.test_pd.registry |
|
119
|
|
|
|
|
120
|
|
|
cmd1 = DataEngine.test_pd._commands['add_attribute'] |
|
121
|
|
|
# cmd1 = data_api.command.observations |
|
122
|
|
|
cmd1.args = [datapoints, [_ for _ in range(1, len(datapoints) + 1)], 'test_attr'] |
|
123
|
|
|
|
|
124
|
|
|
cmd1.execute() |
|
125
|
|
|
|
|
126
|
|
|
assert set(datapoints.attributes) == set(_ for _ in list(test_json_data['attributes']) + ['test_attr']) |
|
127
|
|
|
|