so_magic.data.features.features.BaseAttributeReporter.values() - Code Metrics - Inspection of "Release v0.6.0" - boromir674/so-magic - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#31)

by Konstantinos

created 2021-05-30 17:32 UTC

BaseAttributeReporter.values() A

↳ Parent: so_magic.data.features.features

Complexity

Conditions

Size

Total Lines	2
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	1
eloc	2
nop	4
dl	0
loc	2
rs	10
c	0
b	0
f	0

from abc import ABC, abstractmethod
import attr
from so_magic.data.variables.types import VariableTypeFactory


class AttributeReporterInterface(ABC):
    """A class implementing this interface has the ability to report information on an attribute/variable
    of some structured data (observations)
    """
    @abstractmethod
    def values(self, datapoints, attribute, **kwargs):
        """Get the values ([N x 1] vector) of all datapoints (N x D) corresponding to the input variable/attribute.

        Args:
            datapoints (Datapoints): [description]
            attribute (str): [description]

        Return:
            (numpy.ndarray): the values in a [N x 1] vector
        """
        raise NotImplementedError

    @abstractmethod
    def variable_type(self, datapoints, attribute, **kwargs):
        """Call to get the variable type of the datapoints, given the attribute.

        Args:
            datapoints (Datapoints): [description]
            attribute (str): [description]

        Return:
            (str): [description]
        """
        raise NotImplementedError

    @abstractmethod
    def value_set(self, datapoints, attribute, **kwargs):
        raise NotImplementedError


class BaseAttributeReporter(AttributeReporterInterface):

    def values(self, datapoints, attribute, **kwargs):
        return datapoints[attribute]

    def variable_type(self, datapoints, attribute, **kwargs):
        return VariableTypeFactory.infer(datapoints, attribute, **kwargs)

    def value_set(self, datapoints, attribute, **kwargs):
        return set(datapoints.column(attribute))


#### HELPERS
def _list_validator(_self, _attribute, value):
    if not isinstance(value, list):
        raise ValueError(f'Expected a list; instead a {type(value).__name__} was given.')


def _string_validator(_self, _attribute, value):
    if not isinstance(value, str):
        raise ValueError(f'Expected a string; instead a {type(value).__name__} was given.')


@attr.s
class AttributeReporter:
    label = attr.ib(init=True)
    reporter = attr.ib(init=True, default=BaseAttributeReporter())

    def values(self, datapoints):
        """A default implementation of the values method"""
        return self.reporter.values(datapoints, self.label)

    def variable_type(self, datapoints):
        """A default implementation of the values method"""
        return self.reporter.variable_type(datapoints, self.label)

    def value_set(self, datapoints):
        return self.reporter.value_set(datapoints, self.label)

    def __str__(self):
        return self.label

@attr.s
class FeatureState:
    key = attr.ib(init=True)
    reporter = attr.ib(init=True)

    def __str__(self):
        return self.key


def is_callable(_self, _attribute, value):
    if not callable(value):
        raise ValueError(f"Expected a callable object; instead {type(value)} was given.")
    if value.func_code.co_argcount < 1:
        raise ValueError("Expected a callable that takes at least 1 argument; "
                            "instead a callable that takes no arguments was given.")

@attr.s
class FeatureFunction:
    """Example: Assume we have a datapoint v = [v_1, v_2, .., v_n, and 2 feature functions f_1, f_2\n
    Then we can produce an encoded vector (eg to feed for training a ML model) like: encoded_vector = [f_1(v), f_2(v)]
    """
    function = attr.ib(init=True, validator=is_callable)

    label = attr.ib(init=True, default=None)
    @label.validator
    def is_label(self, _attribute, value):
        if value is None:
            self.label = self.function.func_name

    def values(self, dataset):
        return self.function(dataset)

    @property
    def state(self):
        return FeatureState(self.label, self.function)


@attr.s
class StateMachine:
    states = attr.ib(init=True)
    init_state = attr.ib(init=True)
    _current = attr.ib(init=False, default=attr.Factory(lambda self: self.init_state, takes_self=True))

    @property
    def current(self):
        return self._current

    def update(self, *args, **kwargs):
        if len(args) > 1:
            self.states[args[0]] = args[1]
            self._current = args[0]
        elif len(args) > 0:
            if args[0] in self.states:
                self._current = args[0]
            else:
                raise RuntimeError(f"Requested to set the current state to '{args[0]}', "
                                   f"it is not in existing [{', '.join(sorted(self.states))}]")

    @property
    def state(self):
        """Construct an object representing the current state"""
        return FeatureState(self._current, self.states[self._current])


@attr.s
class TrackingFeature:
    feature = attr.ib(init=True)
    state_machine = attr.ib(init=True)
    variable_type = attr.ib(init=True, default=None)

    @classmethod
    def from_callable(cls, a_callable, label=None, variable_type=None):
        """Construct a feature that has one extract/report capability.
        Input id is correlated to the features position on the vector (see FeatureFunction above)"""
        return TrackingFeature(FeatureFunction(a_callable, label), StateMachine({'raw': a_callable}, 'raw'),
                               variable_type)

    def values(self, dataset):
        return self.state_machine.state.reporter(dataset)

    def label(self):
        return self.feature.label

    @property
    def state(self):
        """Returns the current state"""
        return self.state_machine.state

    def update(self, *args, **kwargs):
        self.state_machine.update(*args, **kwargs)


@attr.s
class FeatureIndex:
    keys = attr.ib(init=True, validator=_list_validator)


class PhiFeatureFunction:
    def __call__(self, *args, **kwargs):
        raise NotImplementedError


1			from abc import ABC, abstractmethod
2			import attr
3			from so_magic.data.variables.types import VariableTypeFactory
4
5
6			class AttributeReporterInterface(ABC):
7			"""A class implementing this interface has the ability to report information on an attribute/variable
8			of some structured data (observations)
9			"""
10			@abstractmethod
11			def values(self, datapoints, attribute, **kwargs):
12			"""Get the values ([N x 1] vector) of all datapoints (N x D) corresponding to the input variable/attribute.
13
14			Args:
15			datapoints (Datapoints): [description]
16			attribute (str): [description]
17
18			Return:
19			(numpy.ndarray): the values in a [N x 1] vector
20			"""
21			raise NotImplementedError
22
23			@abstractmethod
24			def variable_type(self, datapoints, attribute, **kwargs):
25			"""Call to get the variable type of the datapoints, given the attribute.
26
27			Args:
28			datapoints (Datapoints): [description]
29			attribute (str): [description]
30
31			Return:
32			(str): [description]
33			"""
34			raise NotImplementedError
35
36			@abstractmethod
37			def value_set(self, datapoints, attribute, **kwargs):
38			raise NotImplementedError
39
40
41			class BaseAttributeReporter(AttributeReporterInterface):
42
43			def values(self, datapoints, attribute, **kwargs):
44			return datapoints[attribute]
45
46			def variable_type(self, datapoints, attribute, **kwargs):
47			return VariableTypeFactory.infer(datapoints, attribute, **kwargs)
48
49			def value_set(self, datapoints, attribute, **kwargs):
50			return set(datapoints.column(attribute))
51
52
53			#### HELPERS
54			def _list_validator(_self, _attribute, value):
55			if not isinstance(value, list):
56			raise ValueError(f'Expected a list; instead a {type(value).__name__} was given.')
57
58
59			def _string_validator(_self, _attribute, value):
60			if not isinstance(value, str):
61			raise ValueError(f'Expected a string; instead a {type(value).__name__} was given.')
62
63
64			@attr.s
65			class AttributeReporter:
66			label = attr.ib(init=True)
67			reporter = attr.ib(init=True, default=BaseAttributeReporter())
68
69			def values(self, datapoints):
70			"""A default implementation of the values method"""
71			return self.reporter.values(datapoints, self.label)
72
73			def variable_type(self, datapoints):
74			"""A default implementation of the values method"""
75			return self.reporter.variable_type(datapoints, self.label)
76
77			def value_set(self, datapoints):
78			return self.reporter.value_set(datapoints, self.label)
79
80			def __str__(self):
81			return self.label
82
83			@attr.s
84			class FeatureState:
85			key = attr.ib(init=True)
86			reporter = attr.ib(init=True)
87
88			def __str__(self):
89			return self.key
90
91
92			def is_callable(_self, _attribute, value):
93			if not callable(value):
94			raise ValueError(f"Expected a callable object; instead {type(value)} was given.")
95			if value.func_code.co_argcount < 1:
96			raise ValueError("Expected a callable that takes at least 1 argument; "
97			"instead a callable that takes no arguments was given.")
98
99			@attr.s
100			class FeatureFunction:
101			"""Example: Assume we have a datapoint v = [v_1, v_2, .., v_n, and 2 feature functions f_1, f_2\n
102			Then we can produce an encoded vector (eg to feed for training a ML model) like: encoded_vector = [f_1(v), f_2(v)]
103			"""
104			function = attr.ib(init=True, validator=is_callable)
105
106			label = attr.ib(init=True, default=None)
107			@label.validator
108			def is_label(self, _attribute, value):
109			if value is None:
110			self.label = self.function.func_name
111
112			def values(self, dataset):
113			return self.function(dataset)
114
115			@property
116			def state(self):
117			return FeatureState(self.label, self.function)
118
119
120			@attr.s
121			class StateMachine:
122			states = attr.ib(init=True)
123			init_state = attr.ib(init=True)
124			_current = attr.ib(init=False, default=attr.Factory(lambda self: self.init_state, takes_self=True))
125
126			@property
127			def current(self):
128			return self._current
129
130			def update(self, args, *kwargs):
131			if len(args) > 1:
132			self.states[args[0]] = args[1]
133			self._current = args[0]
134			elif len(args) > 0:
135			if args[0] in self.states:
136			self._current = args[0]
137			else:
138			raise RuntimeError(f"Requested to set the current state to '{args[0]}', "
139			f"it is not in existing [{', '.join(sorted(self.states))}]")
140
141			@property
142			def state(self):
143			"""Construct an object representing the current state"""
144			return FeatureState(self._current, self.states[self._current])
145
146
147			@attr.s
148			class TrackingFeature:
149			feature = attr.ib(init=True)
150			state_machine = attr.ib(init=True)
151			variable_type = attr.ib(init=True, default=None)
152
153			@classmethod
154			def from_callable(cls, a_callable, label=None, variable_type=None):
155			"""Construct a feature that has one extract/report capability.
156			Input id is correlated to the features position on the vector (see FeatureFunction above)"""
157			return TrackingFeature(FeatureFunction(a_callable, label), StateMachine({'raw': a_callable}, 'raw'),
158			variable_type)
159
160			def values(self, dataset):
161			return self.state_machine.state.reporter(dataset)
162
163			def label(self):
164			return self.feature.label
165
166			@property
167			def state(self):
168			"""Returns the current state"""
169			return self.state_machine.state
170
171			def update(self, args, *kwargs):
172			self.state_machine.update(args, *kwargs)
173
174
175			@attr.s
176			class FeatureIndex:
177			keys = attr.ib(init=True, validator=_list_validator)
178
179
180			class PhiFeatureFunction:
181			def __call__(self, args, *kwargs):
182			raise NotImplementedError
183

boromir674 / so-magic

Pull Request — master (#31)

BaseAttributeReporter.values() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like