Improve typehint (#92)

* Add pycln

* Use typevar.

* Embedding traits should not inherit from Baserecommender.

Author: tohtsky

.pre-commit-config.yaml

@@ -26,3 +26,8 @@ repos:
     rev: 20.8b1
     hooks:
       - id: black
+  - repo: https://github.com/hadialqattan/pycln
+    rev: v1.2.4 # Possible releases: https://github.com/hadialqattan/pycln/releases
+    hooks:
+      - id: pycln
+        args: [--config=pyproject.toml]

docs/source/conf.py

@@ -2,8 +2,6 @@
 # add these directories to sys.path here. If the directory is relative to the
 # documentation root, use os.path.abspath to make it absolute, like shown here.
 #
-import os
-import sys
 from typing import List
 
 # Configuration file for the Sphinx documentation builder.

examples/movielens/movielens_1m.py

@@ -1,6 +1,4 @@
 import json
-import logging
-import os
 from typing import List, Tuple, Type
 
 from scipy import sparse as sps

irspack/dataset/downloader.py

@@ -1,4 +1,3 @@
-import os
 import urllib.request
 from abc import ABCMeta, abstractmethod
 from io import BytesIO

irspack/dataset/movielens/base.py

@@ -1,4 +1,4 @@
-from abc import ABCMeta, abstractmethod
+from abc import abstractmethod
 from io import BytesIO
 
 import pandas as pd

irspack/definitions.py

@@ -1,4 +1,4 @@
-from typing import Any, List, Optional, Union
+from typing import Union
 
 import numpy as np
 from scipy import sparse as sps

irspack/optimizers/_optimizers.py

@@ -1,5 +1,5 @@
 import logging
-from typing import Any, Dict, List, Optional, Type
+from typing import Any, Dict, List, Optional
 
 from irspack.definitions import InteractionMatrix
 

irspack/recommenders/__init__.py

@@ -5,7 +5,7 @@
 )
 from irspack.recommenders.base_earlystop import BaseRecommenderWithEarlyStopping
 from irspack.recommenders.dense_slim import DenseSLIMConfig, DenseSLIMRecommender
-from irspack.recommenders.edlae import EDLAERecommender
+from irspack.recommenders.edlae import EDLAEConfig, EDLAERecommender
 from irspack.recommenders.ials import IALSConfig, IALSRecommender
 from irspack.recommenders.knn import (
     AsymmetricCosineKNNConfig,
@@ -41,6 +41,8 @@
     "RP3betaRecommender",
     "DenseSLIMConfig",
     "DenseSLIMRecommender",
+    "EDLAERecommender",
+    "EDLAEConfig",
     "SLIMConfig",
     "SLIMRecommender",
     "IALSConfig",

irspack/recommenders/base.py

@@ -1,5 +1,14 @@
 from abc import ABCMeta, abstractmethod
-from typing import TYPE_CHECKING, Any, Dict, Optional, Type, Union, no_type_check
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Dict,
+    Optional,
+    Type,
+    TypeVar,
+    Union,
+    no_type_check,
+)
 
 import numpy as np
 from optuna.trial import Trial
@@ -16,6 +25,8 @@
     UserIndexArray,
 )
 
+R = TypeVar("R", bound="BaseRecommender")
+
 
 def _sparse_to_array(U: Any) -> np.ndarray:
     if sps.issparse(U):
@@ -78,15 +89,17 @@ def __init__(self, X_train_all: InteractionMatrix, **kwargs: Any) -> None:
 
     @classmethod
     def from_config(
-        cls, X_train_all: InteractionMatrix, config: RecommenderConfig
-    ) -> "BaseRecommender":
+        cls: Type[R],
+        X_train_all: InteractionMatrix,
+        config: RecommenderConfig,
+    ) -> R:
         if not isinstance(config, cls.config_class):
             raise ValueError(
                 f"Different config has been given. config must be {cls.config_class}"
             )
         return cls(X_train_all, **config.dict())
 
-    def learn(self) -> "BaseRecommender":
+    def learn(self: R) -> R:
         """Learns and returns itself.
 
         Returns:
@@ -245,7 +258,7 @@ def get_score_block(self, begin: int, end: int) -> DenseScoreArray:
         return _sparse_to_array(self.U[begin:end].dot(self._X_csc))
 
 
-class BaseRecommenderWithUserEmbedding(BaseRecommender):
+class BaseRecommenderWithUserEmbedding:
     """Defines a recommender with user embedding (e.g., matrix factorization.).
     These class can be a base CF estimator for CB2CF (with user profile -> user embedding NN).
     """
@@ -276,7 +289,7 @@ def get_score_from_user_embedding(
         raise NotImplementedError("get_score_from_item_embedding must be implemtented.")
 
 
-class BaseRecommenderWithItemEmbedding(BaseRecommender):
+class BaseRecommenderWithItemEmbedding:
     """Defines a recommender with item embedding (e.g., matrix factorization.).
     These class can be a base CF estimator for CB2CF (with item profile -> item embedding NN).
     """

irspack/recommenders/edlae.py

@@ -1,7 +1,7 @@
 import gc
 
 import numpy as np
-from scipy import linalg, sparse
+from scipy import linalg
 
 from ..definitions import InteractionMatrix
 from .base import BaseSimilarityRecommender, RecommenderConfig