|
1
|
|
|
from abc import ABC, abstractmethod |
|
2
|
|
|
import os |
|
3
|
|
|
import attr |
|
4
|
|
|
|
|
5
|
|
|
from so_magic.utils import Observer |
|
6
|
|
|
|
|
7
|
|
|
|
|
8
|
|
|
@attr.s(str=True, repr=True) |
|
9
|
|
|
class Dataset: |
|
10
|
|
|
datapoints = attr.ib(init=True) |
|
11
|
|
|
name = attr.ib(init=True, default=None) |
|
12
|
|
|
|
|
13
|
|
|
_features = attr.ib(init=True, default=[]) |
|
14
|
|
|
size = attr.ib(init=False, default=attr.Factory(lambda self: len(self.datapoints) if self.datapoints else 0, takes_self=True)) |
|
15
|
|
|
|
|
16
|
|
|
@property |
|
17
|
|
|
def features(self): |
|
18
|
|
|
return self._features |
|
19
|
|
|
|
|
20
|
|
|
@features.setter |
|
21
|
|
|
def features(self, features): |
|
22
|
|
|
self._features = features |
|
23
|
|
|
|
|
24
|
|
|
|
|
25
|
|
|
class DatapointsInterface(ABC): |
|
26
|
|
|
"""The Datapoints interface gives access to the 'observations' property.""" |
|
27
|
|
|
@property |
|
28
|
|
|
@abstractmethod |
|
29
|
|
|
def observations(self): |
|
30
|
|
|
raise NotImplementedError |
|
31
|
|
|
|
|
32
|
|
|
class StructuredDataInterface(ABC): |
|
33
|
|
|
@property |
|
34
|
|
|
@abstractmethod |
|
35
|
|
|
def attributes(self): |
|
36
|
|
|
raise NotImplementedError |
|
37
|
|
|
|
|
38
|
|
|
|
|
39
|
|
|
class DatapointsFactory: |
|
40
|
|
|
constructors = {} |
|
41
|
|
|
|
|
42
|
|
|
@classmethod |
|
43
|
|
|
def register_constructor(cls, name): |
|
44
|
|
|
def wrapper(subclass): |
|
45
|
|
|
cls.constructors[name] = subclass |
|
46
|
|
|
return subclass |
|
47
|
|
|
return wrapper |
|
48
|
|
|
|
|
49
|
|
|
@classmethod |
|
50
|
|
|
def create(cls, name, *args, **kwargs) -> DatapointsInterface: |
|
51
|
|
|
if name not in cls.constructors: |
|
52
|
|
|
raise ValueError( |
|
53
|
|
|
f"Request Engine of type '{name}'; supported are [{', '.join(sorted(cls.constructors.keys()))}]") |
|
54
|
|
|
try: |
|
55
|
|
|
return cls.constructors[name](*args, **kwargs) |
|
56
|
|
|
except TypeError as e: |
|
57
|
|
|
print(f"Datapoints creation failed. Args: [{', '.join(f'{i}: {str(_)}' for i, _ in enumerate(args))}]") |
|
58
|
|
|
print(f"Kwargs: [{', '.join(f'{k}: {v}' for k, v in kwargs.items())}]") |
|
59
|
|
|
print(e) |
|
60
|
|
|
import sys |
|
61
|
|
|
sys.exit(1) |
|
62
|
|
|
|
|
63
|
|
|
@attr.s |
|
64
|
|
|
class BroadcastingDatapointsFactory(DatapointsFactory): |
|
65
|
|
|
subject = attr.ib(init=True) |
|
66
|
|
|
|
|
67
|
|
|
def create(self, datapoints_factory_type, *args, **kwargs) -> DatapointsInterface: |
|
68
|
|
|
self.subject.name = kwargs.pop('id', kwargs.pop('name', kwargs.pop('file_path', ''))) |
|
69
|
|
|
if kwargs: |
|
70
|
|
|
msg = f"Kwargs: [{', '.join(f'{k}: {v}' for k, v in kwargs.items())}]" |
|
71
|
|
|
raise RuntimeError("The 'create' method of DatapointsFactory does not support kwargs:", msg) |
|
72
|
|
|
self.subject.state = super().create(datapoints_factory_type, *args, **kwargs) |
|
73
|
|
|
if args and not hasattr(self, '.name'): |
|
74
|
|
|
self.name = getattr(args[0], 'name', '') |
|
75
|
|
|
self.subject.notify() |
|
76
|
|
|
return self.subject.state |
|
77
|
|
|
|
|
78
|
|
|
|
|
79
|
|
|
@attr.s |
|
80
|
|
|
@DatapointsFactory.register_constructor('structured-data') |
|
81
|
|
|
class StructuredData(DatapointsInterface, StructuredDataInterface): |
|
82
|
|
|
"""Structured data. There are specific attributes/variables per observation. |
|
83
|
|
|
|
|
84
|
|
|
Args: |
|
85
|
|
|
observations (object): a reference to an object that encapsulates structured data |
|
86
|
|
|
""" |
|
87
|
|
|
_observations = attr.ib(init=True) |
|
88
|
|
|
_attributes = attr.ib(init=True, converter=lambda input_value: [x for x in input_value]) |
|
89
|
|
|
|
|
90
|
|
|
@property |
|
91
|
|
|
def attributes(self): |
|
92
|
|
|
return self._attributes |
|
93
|
|
|
|
|
94
|
|
|
@property |
|
95
|
|
|
def observations(self): |
|
96
|
|
|
return self._observations |
|
97
|
|
|
|
|
98
|
|
|
@observations.setter |
|
99
|
|
|
def observations(self, observations): |
|
100
|
|
|
self._observations = observations |
|
101
|
|
|
|
|
102
|
|
|
@attr.s |
|
103
|
|
|
@DatapointsFactory.register_constructor('tabular-data') |
|
104
|
|
|
class TabularData(StructuredData): |
|
105
|
|
|
"""Table-like datapoints that are loaded in memory""" |
|
106
|
|
|
retriever = attr.ib(init=True) |
|
107
|
|
|
iterator = attr.ib(init=True) |
|
108
|
|
|
mutator = attr.ib(init=True) |
|
109
|
|
|
|
|
110
|
|
|
@property |
|
111
|
|
|
def attributes(self): |
|
112
|
|
|
return self.iterator.columnnames(self) |
|
113
|
|
|
|
|
114
|
|
|
def column(self, identifier): |
|
115
|
|
|
return self.retriever.column(identifier, self) |
|
116
|
|
|
|
|
117
|
|
|
def row(self, identifier): |
|
118
|
|
|
return self.retriever.row(identifier, self) |
|
119
|
|
|
|
|
120
|
|
|
def get_numerical_attributes(self): |
|
121
|
|
|
return self.retriever.get_numerical_attributes(self) |
|
122
|
|
|
|
|
123
|
|
|
def get_categorical_attributes(self): |
|
124
|
|
|
return iter(set(self.attributes) - set([_ for _ in self.retriever.get_numerical_attributes(self)])) |
|
125
|
|
|
|
|
126
|
|
|
@property |
|
127
|
|
|
def nb_columns(self): |
|
128
|
|
|
return self.retriever.nb_columns(self) |
|
129
|
|
|
|
|
130
|
|
|
@property |
|
131
|
|
|
def nb_rows(self): |
|
132
|
|
|
return self.retriever.nb_rows(self) |
|
133
|
|
|
|
|
134
|
|
|
def __len__(self): |
|
135
|
|
|
return self.retriever.nb_rows(self) |
|
136
|
|
|
|
|
137
|
|
|
def __iter__(self): |
|
138
|
|
|
return self.iterator.iterrows(self) |
|
139
|
|
|
|
|
140
|
|
|
def iterrows(self): |
|
141
|
|
|
return self.iterator.iterrows(self) |
|
142
|
|
|
|
|
143
|
|
|
def itercolumns(self): |
|
144
|
|
|
return self.iterator.itercolumns(self) |
|
145
|
|
|
|
|
146
|
|
|
|
|
147
|
|
|
@attr.s |
|
148
|
|
|
class DatapointsManager(Observer): |
|
149
|
|
|
datapoints_objects = attr.ib(init=True, default={}) |
|
150
|
|
|
_last_key = attr.ib(init=False, default='') |
|
151
|
|
|
|
|
152
|
|
|
def update(self, subject): |
|
153
|
|
|
datapoints_object = subject.state |
|
154
|
|
|
key = getattr(subject, 'name', '') |
|
155
|
|
|
if key == '': |
|
156
|
|
|
raise RuntimeError(f"Subject {Subject} with state {str(subject.state)} resulted in an empty string as key (to use in dict/hash).") |
|
157
|
|
|
if key in self.datapoints_objects: |
|
158
|
|
|
raise RuntimeError(f"Attempted to register a new Datapoints object at the existing key '{key}'.") |
|
159
|
|
|
self.datapoints_objects[key] = datapoints_object |
|
160
|
|
|
self._last_key = key |
|
161
|
|
|
|
|
162
|
|
|
@property |
|
163
|
|
|
def state(self): |
|
164
|
|
|
return self._last_key |
|
165
|
|
|
|
|
166
|
|
|
@property |
|
167
|
|
|
def datapoints(self): |
|
168
|
|
|
try: |
|
169
|
|
|
return self.datapoints_objects[self._last_key] |
|
170
|
|
|
except KeyError as e: |
|
171
|
|
|
print(f"{e}. Requested datapoints with id '{self._last_key}', but was not found in registered [{', '.join(_ for _ in self.datapoints_objects.keys())}]") |
|
172
|
|
|
|
boromir674 /
so-magic
