Add ClassVar type-hints.

Details :
* In `/optimizer`, `/metrics`, and `/aggregator` :`name` and `aux_name`
* In `/communications` : `protocol`
* In `/dataset` : `_type_key`
* In `/data_info` : `field`, `type`, and `doc`

Excluding all classes decorated with `@dataclasses.dataclass`,
see https://stackoverflow.com/a/52099457

declearn/aggregator/_api.py

@@ -3,13 +3,11 @@
 """Model updates aggregation API."""
 
 from abc import ABCMeta, abstractmethod
-from typing import Any, Dict
-
+from typing import Any, ClassVar, Dict
 
 from declearn.model.api import Vector
 from declearn.utils import create_types_registry, register_type
 
-
 __all__ = [
     "Aggregator",
 ]
@@ -59,7 +57,7 @@ class Aggregator(metaclass=ABCMeta):
     See `declearn.utils.register_type` for details on types registration.
     """
 
-    name: str = NotImplemented
+    name: ClassVar[str] = NotImplemented
 
     def __init_subclass__(
         cls,

declearn/aggregator/_base.py

@@ -2,12 +2,10 @@
 
 """FedAvg-like mean-aggregation class."""
 
-from typing import Any, Dict, Optional
+from typing import Any, ClassVar, Dict, Optional
 
-
-from declearn.model.api import Vector
 from declearn.aggregator._api import Aggregator
-
+from declearn.model.api import Vector
 
 __all__ = [
     "AveragingAggregator",
@@ -24,7 +22,7 @@ class AveragingAggregator(Aggregator):
     that use simple weighting schemes.
     """
 
-    name = "averaging"
+    name: ClassVar[str] = "averaging"
 
     def __init__(
         self,

declearn/aggregator/_gma.py

@@ -2,12 +2,10 @@
 
 """Gradient Masked Averaging aggregation class."""
 
-from typing import Any, Dict, Optional
+from typing import Any, ClassVar, Dict, Optional
 
-
-from declearn.model.api import Vector
 from declearn.aggregator._base import AveragingAggregator
-
+from declearn.model.api import Vector
 
 __all__ = [
     "GradientMaskedAveraging",
@@ -42,7 +40,7 @@ class GradientMaskedAveraging(AveragingAggregator):
         https://arxiv.org/abs/2201.11986
     """
 
-    name = "gradient-masked-averaging"
+    name: ClassVar[str] = "gradient-masked-averaging"
 
     def __init__(
         self,

declearn/communication/api/_client.py

@@ -5,8 +5,7 @@
 import logging
 import types
 from abc import ABCMeta, abstractmethod
-from typing import Any, Dict, Optional, Type, Union
-
+from typing import Any, ClassVar, Dict, Optional, Type, Union
 
 from declearn.communication.messaging import (
     Empty,
@@ -18,7 +17,6 @@ from declearn.communication.messaging import (
 )
 from declearn.utils import create_types_registry, get_logger, register_type
 
-
 __all__ = [
     "NetworkClient",
 ]
@@ -58,7 +56,7 @@ class NetworkClient(metaclass=ABCMeta):
     probably be rejected by the server if the client has not registered.
     """
 
-    protocol: str = NotImplemented
+    protocol: ClassVar[str] = NotImplemented
 
     def __init_subclass__(cls, register: bool = True) -> None:
         """Automate the type-registration of NetworkClient subclasses."""
@@ -107,7 +105,6 @@ class NetworkClient(metaclass=ABCMeta):
 
         The return type is communication-protocol dependent.
         """
-        return NotImplemented
 
     # similar to NetworkServer API; pylint: disable=duplicate-code
 
@@ -118,7 +115,6 @@ class NetworkClient(metaclass=ABCMeta):
         Note: this method can be called safely even if the
         client is already running (simply having no effect).
         """
-        return None
 
     @abstractmethod
     async def stop(self) -> None:

declearn/communication/api/_server.py

@@ -6,7 +6,7 @@ import asyncio
 import logging
 import types
 from abc import ABCMeta, abstractmethod
-from typing import Any, Dict, Optional, Set, Type, Union
+from typing import Any, Dict, Optional, Set, Type, Union, ClassVar
 
 
 from declearn.communication.api._service import MessagesHandler
@@ -54,7 +54,7 @@ class NetworkServer(metaclass=ABCMeta):
     of the awaitable `wait_for_clients` method.
     """
 
-    protocol: str = NotImplemented
+    protocol: ClassVar[str] = NotImplemented
 
     def __init_subclass__(cls, register: bool = True) -> None:
         """Automate the type-registration of NetworkServer subclasses."""

declearn/communication/grpc/_client.py

@@ -3,17 +3,16 @@
 """Client-side communication endpoint implementation using gRPC"""
 
 import logging
-from typing import Any, Dict, Optional, Union
+from typing import Any, ClassVar, Dict, Optional, Union
 
 import grpc  # type: ignore
 
 from declearn.communication.api import NetworkClient
-from declearn.communication.messaging import Message, parse_message_from_string
 from declearn.communication.grpc.protobufs import message_pb2
 from declearn.communication.grpc.protobufs.message_pb2_grpc import (
     MessageBoardStub,
 )
-
+from declearn.communication.messaging import Message, parse_message_from_string
 
 __all__ = [
     "GrpcClient",
@@ -26,7 +25,7 @@ CHUNK_LENGTH = 2**22 - 50  # 2**22 - sys.getsizeof("") - 1
 class GrpcClient(NetworkClient):
     """Client-side communication endpoint using gRPC."""
 
-    protocol = "grpc"
+    protocol: ClassVar[str] = "grpc"
 
     def __init__(
         self,

declearn/communication/websockets/_client.py

@@ -5,7 +5,7 @@
 import asyncio
 import logging
 import ssl
-from typing import Any, Dict, Optional, Union
+from typing import Any, Dict, Optional, Union, ClassVar
 
 import websockets as ws
 from websockets.client import WebSocketClientProtocol
@@ -25,7 +25,7 @@ CHUNK_LENGTH = 100000
 class WebsocketsClient(NetworkClient):
     """Client-side communication endpoint using WebSockets."""
 
-    protocol = "websockets"
+    protocol: ClassVar[str] = "websockets"
 
     def __init__(
         self,

declearn/data_info/_base.py

@@ -34,7 +34,7 @@ data_info fields, are implemented (although unexposed) here.
 
 import warnings
 from abc import ABCMeta, abstractmethod
-from typing import Any, Dict, List, Optional, Set, Tuple, Type
+from typing import Any, Dict, List, Optional, Set, Tuple, Type, ClassVar
 
 
 __all__ = [
@@ -75,9 +75,9 @@ class DataInfoField(metaclass=ABCMeta):
         is called, run `is_valid` on each and every input.
     """
 
-    field: str
-    types: Tuple[Type, ...]
-    doc: str
+    field: ClassVar[str] = NotImplemented
+    types: ClassVar[Tuple[Type, ...]] = NotImplemented
+    doc: ClassVar[str] = NotImplemented
 
     @classmethod
     def is_valid(
@@ -85,6 +85,7 @@ class DataInfoField(metaclass=ABCMeta):
         value: Any,
     ) -> bool:
         """Check that a given value may belong to this field."""
+        # false-pos; pylint: disable=isinstance-second-argument-not-valid-type
         return isinstance(value, cls.types)
 
     @classmethod

declearn/data_info/_fields.py

@@ -2,13 +2,12 @@
 
 """DataInfoField subclasses specifying common 'data_info' metadata fields."""
 
-from typing import Any, List, Optional, Set
+from typing import Any, ClassVar, List, Optional, Set, Tuple, Type
 
 import numpy as np
 
 from declearn.data_info._base import DataInfoField, register_data_info_field
 
-
 __all__ = [
     "ClassesField",
     "InputShapeField",
@@ -21,9 +20,9 @@ __all__ = [
 class ClassesField(DataInfoField):
     """Specifications for 'classes' data_info field."""
 
-    field = "classes"
-    types = (list, set, tuple, np.ndarray)
-    doc = "Set of classification targets, combined by union."
+    field: ClassVar[str] = "classes"
+    types: ClassVar[Tuple[Type, ...]] = (list, set, tuple, np.ndarray)
+    doc: ClassVar[str] = "Set of classification targets, combined by union."
 
     @classmethod
     def is_valid(
@@ -47,9 +46,9 @@ class ClassesField(DataInfoField):
 class InputShapeField(DataInfoField):
     """Specifications for 'input_shape' data_info field."""
 
-    field = "input_shape"
-    types = (tuple, list)
-    doc = "Input features' batched shape, checked to be equal."
+    field: ClassVar[str] = "input_shape"
+    types: ClassVar[Tuple[Type, ...]] = (tuple, list)
+    doc: ClassVar[str] = "Input features' batched shape, checked to be equal."
 
     @classmethod
     def is_valid(
@@ -96,9 +95,9 @@ class InputShapeField(DataInfoField):
 class NbFeaturesField(DataInfoField):
     """Specifications for 'n_features' data_info field."""
 
-    field = "n_features"
-    types = (int,)
-    doc = "Number of input features, checked to be equal."
+    field: ClassVar[str] = "n_features"
+    types: ClassVar[Tuple[Type, ...]] = (int,)
+    doc: ClassVar[str] = "Number of input features, checked to be equal."
 
     @classmethod
     def is_valid(
@@ -128,9 +127,9 @@ class NbFeaturesField(DataInfoField):
 class NbSamplesField(DataInfoField):
     """Specifications for 'n_samples' data_info field."""
 
-    field = "n_samples"
-    types = (int,)
-    doc = "Number of data samples, combined by summation."
+    field: ClassVar[str] = "n_samples"
+    types: ClassVar[Tuple[Type, ...]] = (int,)
+    doc: ClassVar[str] = "Number of data samples, combined by summation."
 
     @classmethod
     def is_valid(

declearn/dataset/_base.py

@@ -4,12 +4,11 @@
 
 from abc import ABCMeta, abstractmethod
 from dataclasses import dataclass
-from typing import Any, Iterator, Optional, Set
+from typing import Any, ClassVar, Iterator, Optional, Set
 
 from declearn.typing import Batch
 from declearn.utils import access_registered, create_types_registry, json_load
 
-
 __all__ = [
     "DataSpecs",
     "Dataset",
@@ -41,7 +40,7 @@ class Dataset(metaclass=ABCMeta):
     straightforward to specify as part of FL algorithms.
     """
 
-    _type_key: str = NotImplemented
+    _type_key: ClassVar[str] = NotImplemented
 
     @abstractmethod
     def save_to_json(

declearn/dataset/_inmemory.py

@@ -4,7 +4,7 @@
 
 import functools
 import os
-from typing import Any, Dict, Iterator, List, Optional, Set, Union
+from typing import Any, ClassVar, Dict, Iterator, List, Optional, Set, Union
 
 import numpy as np
 import pandas as pd  # type: ignore
@@ -17,7 +17,6 @@ from declearn.dataset._sparse import sparse_from_file, sparse_to_file
 from declearn.typing import Batch
 from declearn.utils import json_dump, json_load, register_type
 
-
 __all__ = [
     "InMemoryDataset",
 ]
@@ -56,7 +55,7 @@ class InMemoryDataset(Dataset):
 
     # attributes serve clarity; pylint: disable=too-many-instance-attributes
 
-    _type_key = "InMemoryDataset"
+    _type_key: ClassVar[str] = "InMemoryDataset"
 
     def __init__(
         self,

declearn/main/utils/_training.py

@@ -3,7 +3,7 @@
 """Wrapper to run local training and evaluation rounds in a FL process."""
 
 import logging
-from typing import Any, Dict, List, Optional, Union
+from typing import Any, ClassVar, Dict, List, Optional, Union
 
 import numpy as np
 
@@ -96,7 +96,7 @@ class TrainingManager:
         class LossMetric(MeanMetric, register=False):
             """Ad hoc Metric wrapping a model's loss function."""
 
-            name = "loss"
+            name: ClassVar[str] = "loss"
 
             def metric_func(
                 self, y_true: np.ndarray, y_pred: np.ndarray

declearn/metrics/_api.py

@@ -104,7 +104,7 @@ class Metric(metaclass=ABCMeta):
     See `declearn.utils.register_type` for details on types registration.
     """
 
-    name: ClassVar[str]
+    name: ClassVar[str] = NotImplemented
 
     def __init__(
         self,

declearn/metrics/_classif.py

@@ -2,7 +2,7 @@
 
 """Iterative and federative classification evaluation metrics."""
 
-from typing import Any, Collection, Dict, Optional, Union
+from typing import Any, ClassVar, Collection, Dict, Optional, Union
 
 import numpy as np
 import sklearn  # type: ignore
@@ -41,7 +41,7 @@ class BinaryAccuracyPrecisionRecall(Metric):
         Confusion matrix of predictions. Values: [[TN, FP], [FN, TP]]
     """
 
-    name = "binary-classif"
+    name: ClassVar[str] = "binary-classif"
 
     def __init__(
         self,
@@ -131,7 +131,7 @@ class MulticlassAccuracyPrecisionRecall(Metric):
         were predicted to belong to label j.
     """
 
-    name = "multi-classif"
+    name: ClassVar[str] = "multi-classif"
 
     def __init__(
         self,

declearn/metrics/_mean.py

@@ -3,7 +3,7 @@
 """Iterative and federative generic evaluation metrics."""
 
 from abc import ABCMeta, abstractmethod
-from typing import Dict, Optional, Union
+from typing import ClassVar, Dict, Optional, Union
 
 import numpy as np
 
@@ -110,7 +110,7 @@ class MeanAbsoluteError(MeanMetric):
         summed over channels for (>=2)-dimensional inputs).
     """
 
-    name = "mae"
+    name: ClassVar[str] = "mae"
 
     def metric_func(
         self,
@@ -138,7 +138,7 @@ class MeanSquaredError(MeanMetric):
         summed over channels for (>=2)-dimensional inputs).
     """
 
-    name = "mse"
+    name: ClassVar[str] = "mse"
 
     def metric_func(
         self,

declearn/metrics/_roc_auc.py

@@ -2,7 +2,7 @@
 
 """Iterative and federative ROC AUC evaluation metrics."""
 
-from typing import Any, Dict, Optional, Tuple, Union
+from typing import Any, ClassVar, Dict, Optional, Tuple, Union
 
 import numpy as np
 import sklearn  # type: ignore
@@ -10,7 +10,6 @@ import sklearn.metrics  # type: ignore
 
 from declearn.metrics._api import Metric
 
-
 __all__ = [
     "BinaryRocAUC",
 ]
@@ -42,7 +41,7 @@ class BinaryRocAUC(Metric):
     unless its
     """
 
-    name = "binary-roc"
+    name: ClassVar[str] = "binary-roc"
 
     def __init__(
         self,

declearn/optimizer/modules/_adaptive.py

@@ -2,13 +2,12 @@
 
 """Adaptive algorithms for optimizers, implemented as plug-in modules."""
 
-from typing import Any, Dict, Optional, Union
+from typing import Any, ClassVar, Dict, Optional, Union
 
 from declearn.model.api import Vector
 from declearn.optimizer.modules._api import OptiModule
 from declearn.optimizer.modules._momentum import EWMAModule, YogiMomentumModule
 
-
 __all__ = [
     "AdaGradModule",
     "AdamModule",
@@ -38,7 +37,7 @@ class AdaGradModule(OptiModule):
         https://jmlr.org/papers/v12/duchi11a.html
     """
 
-    name = "adagrad"
+    name: ClassVar[str] = "adagrad"
 
     def __init__(
         self,
@@ -102,7 +101,7 @@ class RMSPropModule(OptiModule):
         Average of its Recent Magnitude.
     """
 
-    name = "rmsprop"
+    name: ClassVar[str] = "rmsprop"
 
     def __init__(
         self,
@@ -184,7 +183,7 @@ class AdamModule(OptiModule):
         https://arxiv.org/abs/1904.09237
     """
 
-    name = "adam"
+    name: ClassVar[str] = "adam"
 
     def __init__(
         self,
@@ -304,7 +303,7 @@ class YogiModule(AdamModule):
         https://arxiv.org/abs/1904.09237
     """
 
-    name = "yogi"
+    name: ClassVar[str] = "yogi"
 
     def __init__(
         self,

declearn/optimizer/modules/_api.py

@@ -3,7 +3,7 @@
 """Base API for plug-in gradients-alteration algorithms."""
 
 from abc import ABCMeta, abstractmethod
-from typing import Any, Dict, Optional
+from typing import Any, ClassVar, Dict, Optional
 
 from declearn.model.api import Vector
 from declearn.utils import (
@@ -76,9 +76,8 @@ class OptiModule(metaclass=ABCMeta):
     See `declearn.utils.register_type` for details on types registration.
     """
 
-    name: str = NotImplemented
-
-    aux_name: Optional[str] = None
+    name: ClassVar[str] = NotImplemented
+    aux_name: ClassVar[Optional[str]] = None
 
     def __init_subclass__(
         cls,

declearn/optimizer/modules/_clipping.py

@@ -2,13 +2,11 @@
 
 """Batch-averaged gradients clipping plug-in modules."""
 
-from typing import Any, Dict
-
+from typing import Any, ClassVar, Dict
 
 from declearn.model.api import Vector
 from declearn.optimizer.modules._api import OptiModule
 
-
 __all__ = ["L2Clipping"]
 
 
@@ -32,7 +30,7 @@ class L2Clipping(OptiModule):
     to bound the sensitivity associated to that action.
     """
 
-    name = "l2-clipping"
+    name: ClassVar[str] = "l2-clipping"
 
     def __init__(
         self,

declearn/optimizer/modules/_momentum.py

@@ -2,7 +2,7 @@
 
 """Base API and common examples of plug-in gradients-alteration algorithms."""
 
-from typing import Any, Dict, Union
+from typing import Any, ClassVar, Dict, Union
 
 from declearn.model.api import Vector
 from declearn.optimizer.modules._api import OptiModule
@@ -48,7 +48,7 @@ class MomentumModule(OptiModule):
         https://proceedings.mlr.press/v28/sutskever13.pdf
     """
 
-    name = "momentum"
+    name: ClassVar[str] = "momentum"
 
     def __init__(
         self,
@@ -114,7 +114,7 @@ class EWMAModule(OptiModule):
     decaying moving-average of past gradients.
     """
 
-    name = "ewma"
+    name: ClassVar[str] = "ewma"
 
     def __init__(
         self,
@@ -184,7 +184,7 @@ class YogiMomentumModule(EWMAModule):
         Adaptive Methods for Nonconvex Optimization.
     """
 
-    name = "yogi-momentum"
+    name: ClassVar[str] = "yogi-momentum"
 
     def run(
         self,