Check untyped defs with mypy

Add missing type annotations, fix a bug found by mypy,
enable --check-untyped-defs in tox.ini

eli5/_feature_weights.py

@@ -24,7 +24,7 @@ def _get_top_features(feature_names, coef, top):
       no more than ``num_neg`` negative features.
     """
     if isinstance(top, (list, tuple)):
-        num_pos, num_neg = top
+        num_pos, num_neg = list(top)  # "list" is just for mypy
         pos = _get_top_positive_features(feature_names, coef, num_pos)
         neg = _get_top_negative_features(feature_names, coef, num_neg)
     else:

eli5/formatters/text.py

@@ -1,6 +1,7 @@
 # -*- coding: utf-8 -*-
 from __future__ import absolute_import
 import six
+from typing import List
 
 from . import fields
 from .features import FormattedFeatureName
@@ -14,7 +15,7 @@
 
 
 def format_as_text(expl, show=fields.ALL):
-    lines = []
+    lines = []  # type: List[str]
 
     if expl.error:  # always shown
         lines.extend(_error_lines(expl))

eli5/lime/samplers.py

@@ -2,7 +2,7 @@
 from __future__ import absolute_import
 import abc
 import six
-from typing import Tuple
+from typing import List, Tuple
 
 import numpy as np
 
@@ -145,8 +145,8 @@ class UnivariateKernelDensitySampler(_BaseKernelDensitySampler):
     of the features instead of generating totally new examples.
     """
     def fit(self, X, y=None):
-        self.kdes_ = []
-        self.grids_ = []
+        self.kdes_ = []  # type: List[KernelDensity]
+        self.grids_ = []  # type: List[GridSearchCV]
         num_features = X.shape[-1]
         for i in range(num_features):
             grid, kde = self._fit_kde(self.kde, X[:, i].reshape(-1, 1))

eli5/sklearn/text.py

@@ -1,4 +1,5 @@
 import re
+from typing import Set, Tuple
 
 from six.moves import xrange
 from sklearn.feature_extraction.text import VectorizerMixin
@@ -56,7 +57,7 @@ def _get_features(feature):
 def _get_other(feature_weights, feature_weights_dict, found_features):
     # search for items that were not accounted at all.
     other_items = []
-    accounted_keys = set()
+    accounted_keys = set()  # type: Set[Tuple[str, int]]
     for feature, (_, key) in feature_weights_dict.items():
         if key not in found_features and key not in accounted_keys:
             group, idx = key

eli5/sklearn/unhashing.py

@@ -6,7 +6,7 @@
 
 from collections import defaultdict, Counter
 from itertools import chain
-from typing import List, Iterable, Any
+from typing import List, Iterable, Any, Dict
 
 import numpy as np
 from sklearn.base import BaseEstimator, TransformerMixin
@@ -183,7 +183,7 @@ def _get_collisions(indices):
     Return a dict ``{column_id: [possible term ids]}``
     with collision information.
     """
-    collisions = defaultdict(list)
+    collisions = defaultdict(list)  # type: Dict[int, List[int]]
     for term_id, hash_id in enumerate(indices):
         collisions[hash_id].append(term_id)
     return dict(collisions)

eli5/sklearn/utils.py

@@ -96,7 +96,7 @@ def get_feature_names(clf, vec=None, bias_name='<BIAS>', feature_names=None):
         if feature_names.n_features != num_features:
             raise ValueError("feature_names has a wrong n_features: "
                              "expected=%d, got=%d" % (num_features,
-                                                      len(feature_names)))
+                                                      feature_names.n_features))
         # Make a shallow copy setting proper bias_name
         return FeatureNames(
             feature_names.feature_names,

tox.ini

@@ -33,4 +33,4 @@ deps=
     {[testenv]deps}
     mypy-lang
 commands=
-    mypy --silent-imports eli5
+    mypy --silent-imports --check-untyped-defs eli5