|
1
|
|
|
import pickle |
|
2
|
|
|
import logging |
|
3
|
|
|
from .strain_dataset import create_dataset_from_pickle |
|
4
|
|
|
from .clustering import get_model_quality_reporter |
|
5
|
|
|
from .data.dataset import DatapointsManager |
|
6
|
|
|
from green_magic.utils import Invoker, CommandHistory |
|
7
|
|
|
from .data.backend.engine import DataEngine |
|
8
|
|
|
|
|
9
|
|
|
_log = logging.getLogger(__name__) |
|
10
|
|
|
|
|
11
|
|
|
|
|
12
|
|
|
class StrainMaster: |
|
13
|
|
|
__instance = None |
|
14
|
|
|
|
|
15
|
|
|
def __new__(cls, *args, **kwargs): |
|
16
|
|
|
if not cls.__instance: |
|
17
|
|
|
from green_magic.data.commands_manager import CommandsManager |
|
18
|
|
|
from green_magic.data.backend import Backend |
|
19
|
|
|
from green_magic.data.data_manager import DataManager |
|
20
|
|
|
from green_magic.data.backend import panda_handling |
|
21
|
|
|
|
|
22
|
|
|
print("!1111111", DataEngine.subclasses) |
|
23
|
|
|
|
|
24
|
|
|
cls.__instance = super().__new__(cls) |
|
25
|
|
|
DataEngine.new('pd') |
|
26
|
|
|
cls.__instance.data_api = DataManager(CommandsManager(), Backend(DataEngine.create('pd'))) |
|
27
|
|
|
# make the datapoint_manager listen to newly created Datapoints objects events |
|
28
|
|
|
cls.__instance.data_api.backend.engine.datapoints_factory.subject.attach(cls.__instance.data_api.backend.datapoints_manager) |
|
29
|
|
|
return cls.__instance |
|
30
|
|
|
|
|
31
|
|
|
def __call__(self, *args, **kwargs): |
|
32
|
|
|
""" |
|
33
|
|
|
Call to update any of 'datasets_dir' and/or 'maps_dir' |
|
34
|
|
|
""" |
|
35
|
|
|
self._datasets_dir = kwargs.get('datasets_dir', self._datasets_dir) |
|
36
|
|
|
self._maps_dir = kwargs.get('maps_dir', self._maps_dir) |
|
37
|
|
|
self.map_manager.maps_dir = self._maps_dir |
|
38
|
|
|
return self |
|
39
|
|
|
|
|
40
|
|
|
def __init__(self, datasets_dir=None, maps_dir=None): pass |
|
41
|
|
|
|
|
42
|
|
|
@property |
|
43
|
|
|
def commands(self): |
|
44
|
|
|
"""Get a Command object from the pool of Command prototypes""" |
|
45
|
|
|
return self.data_api.command |
|
46
|
|
|
|
|
47
|
|
|
@property |
|
48
|
|
|
def datasets_dir(self): |
|
49
|
|
|
return self._datasets_dir |
|
50
|
|
|
|
|
51
|
|
|
@datasets_dir.setter |
|
52
|
|
|
def datasets_dir(self, dataset_directory_path): |
|
53
|
|
|
self._datasets_dir = dataset_directory_path |
|
54
|
|
|
# self.map_manager.maps_dir = dataset_directory_path |
|
55
|
|
|
|
|
56
|
|
|
def strain_names(self, coordinates): |
|
57
|
|
|
g = ((self.dt.datapoint_index2_id[_], self.som.bmus[_]) for _ in range(len(self.dt))) |
|
|
|
|
|
|
58
|
|
|
return [n for n, c in g if c[0] == coordinates['x'] and c[1] == coordinates['y']] |
|
59
|
|
|
|
|
60
|
|
|
@property |
|
61
|
|
|
def dt(self): |
|
62
|
|
|
""" |
|
63
|
|
|
Returns the currently selected/active dataset as a reference to a StrainDataset object.\n |
|
64
|
|
|
:return: the reference to the dataset |
|
65
|
|
|
:rtype: green_magic.strain_dataset.StrainDataset |
|
66
|
|
|
""" |
|
67
|
|
|
return self.data_api.backend.datapoints_manager.datapoints |
|
68
|
|
|
|
|
69
|
|
|
# if self.selected_dt_id not in self._id2dataset: |
|
70
|
|
|
# raise InvalidDatasetSelectionError("Requested dataset with id '{}', but StrainMaster knows only of [{}].".format(self.selected_dt_id, ', '.join(self._id2dataset.keys()))) |
|
71
|
|
|
# return self._id2dataset[self.selected_dt_id] |
|
72
|
|
|
|
|
73
|
|
|
@property |
|
74
|
|
|
def som(self): |
|
75
|
|
|
""" |
|
76
|
|
|
Returns the currently selected/active som instance, as a reference to a som object.\n |
|
77
|
|
|
:return: the reference to the self-organizing map |
|
78
|
|
|
:rtype: somoclu.Somoclu |
|
79
|
|
|
""" |
|
80
|
|
|
return self.map_manager.som |
|
81
|
|
|
|
|
82
|
|
|
@property |
|
83
|
|
|
def model_quality(self): |
|
84
|
|
|
return get_model_quality_reporter(self, self.selected_dt_id) |
|
85
|
|
|
|
|
86
|
|
|
def set_feature_vectors(self, list_of_variables=None): |
|
87
|
|
|
_ = self.get_feature_vectors(self.dt, list_of_variables=list_of_variables) |
|
88
|
|
|
|
|
89
|
|
|
def get_feature_vectors(self, strain_dataset, list_of_variables=None): |
|
90
|
|
|
"""Call this function to get the encoded feature as a list of vectors |
|
91
|
|
|
This method must be called |
|
92
|
|
|
:param strain_dataset: |
|
93
|
|
|
:param list_of_variables: |
|
94
|
|
|
:return: |
|
95
|
|
|
""" |
|
96
|
|
|
if not list_of_variables: |
|
97
|
|
|
return strain_dataset.load_feature_vectors() |
|
98
|
|
|
else: |
|
99
|
|
|
strain_dataset.use_variables(list_of_variables) |
|
100
|
|
|
return strain_dataset.load_feature_vectors() |
|
101
|
|
|
|
|
102
|
|
|
# def create_strain_dataset(self, jl_file, dataset_id, ffilter=''): |
|
103
|
|
|
# data_set = StrainDataset(dataset_id) |
|
104
|
|
|
# with open(jl_file, 'r') as json_lines_file: |
|
105
|
|
|
# for line in json_lines_file: |
|
106
|
|
|
# strain_dict = json.loads(line) |
|
107
|
|
|
# if ffilter.split(':')[0] in strain_dict: |
|
108
|
|
|
# if strain_dict[ffilter.split(':')[0]] == ffilter.split(':')[1]: # if datapoint meets criteria, add it |
|
109
|
|
|
# data_set.add(strain_dict) |
|
110
|
|
|
# if 'description' in strain_dict: |
|
111
|
|
|
# self.lexicon.munch(strain_dict['description']) |
|
112
|
|
|
# else: |
|
113
|
|
|
# data_set.add(strain_dict) |
|
114
|
|
|
# if 'description' in strain_dict: |
|
115
|
|
|
# self.lexicon.munch(strain_dict['description']) |
|
116
|
|
|
# data_set.load_feature_indexes() |
|
117
|
|
|
# self._id2dataset[dataset_id] = data_set |
|
118
|
|
|
# self.selected_dt_id = dataset_id |
|
119
|
|
|
# _log.info("Created StrainDataset object with id '{}'".format(data_set.name)) |
|
120
|
|
|
# assert data_set.name == dataset_id |
|
121
|
|
|
# return data_set |
|
122
|
|
|
|
|
123
|
|
|
def load_dataset(self, a_file): |
|
124
|
|
|
strain_dataset = create_dataset_from_pickle(self._datasets_dir + '/' + a_file) |
|
125
|
|
|
self._id2dataset[strain_dataset.name] = strain_dataset |
|
126
|
|
|
self.selected_dt_id = strain_dataset.name |
|
127
|
|
|
_log.info("Loaded dataset with id '{}'".format(strain_dataset.name)) |
|
128
|
|
|
return strain_dataset |
|
129
|
|
|
|
|
130
|
|
|
def save_active_dataset(self): |
|
131
|
|
|
self.save_dataset(self.selected_dt_id) |
|
132
|
|
|
|
|
133
|
|
|
def save_dataset(self, strain_dataset_id): |
|
134
|
|
|
dataset = self._id2dataset[strain_dataset_id] |
|
135
|
|
|
if dataset.has_missing_values: |
|
136
|
|
|
name = '-not-clean' |
|
137
|
|
|
else: |
|
138
|
|
|
name = '-clean' |
|
139
|
|
|
name = self._datasets_dir + '/' + dataset.name + name + '.pk' |
|
140
|
|
|
try: |
|
141
|
|
|
with open(name, 'wb') as pickled_dataset: |
|
142
|
|
|
pickle.dump(dataset, pickled_dataset, protocol=pickle.HIGHEST_PROTOCOL) |
|
143
|
|
|
_log.info("Saved dataset with id '{}' as {}".format(strain_dataset_id, name)) |
|
144
|
|
|
except RuntimeError as e: |
|
145
|
|
|
_log.debug(e) |
|
146
|
|
|
_log.info("Failed to save dataset wtih id {}".format(strain_dataset_id)) |
|
147
|
|
|
|
|
148
|
|
|
def __getitem__(self, wd_id): |
|
149
|
|
|
self.selected_dt_id = wd_id |
|
150
|
|
|
return self |
|
151
|
|
|
|
|
152
|
|
|
|
|
153
|
|
|
class InvalidDatasetSelectionError(Exception): pass |
|
154
|
|
|
|
boromir674 /
so-magic
Loading history...