1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
|
From: Aditya Mehra <adityamehra@mac.mynetworksettings.com>
Date: Fri, 17 Oct 2025 23:44:58 -0400
Subject: Fix ImportError with sklearn 1.7.0+ by replacing tosequence
- Remove deprecated sklearn.utils.tosequence import (removed in sklearn 1.7)
- Convert steps to list at start of __init__ to handle all input types
- Replace tosequence(steps) with direct list assignment
- Fixes compatibility with scikit-learn >= 1.7.0
- Maintains backward compatibility with older sklearn versions
- Add regression tests for list, tuple, and generator inputs
- Verify steps attribute is always a list type
The tosequence utility was deprecated in sklearn 1.5 and removed in 1.7.
Using list() directly provides the same functionality without depending
on sklearn internals.
Fixes #267
Origin: other, https://github.com/scikit-learn-contrib/sklearn-pandas/pull/269
Bug: https://github.com/scikit-learn-contrib/sklearn-pandas/issues/267
Bug-Debian: https://bugs.debian.org/1117989
Last-Update: 2025-10-19
---
sklearn_pandas/pipeline.py | 7 +++--
tests/test_pipeline.py | 70 ++++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 75 insertions(+), 2 deletions(-)
diff --git a/sklearn_pandas/pipeline.py b/sklearn_pandas/pipeline.py
index cde2fea..34c6043 100644
--- a/sklearn_pandas/pipeline.py
+++ b/sklearn_pandas/pipeline.py
@@ -1,5 +1,4 @@
from sklearn.pipeline import _name_estimators, Pipeline
-from sklearn.utils import tosequence
def _call_fit(fit_method, X, y=None, **kwargs):
@@ -35,13 +34,17 @@ class TransformerPipeline(Pipeline):
"""
def __init__(self, steps):
+ # Convert to list first to handle generators/iterators
+ # (replacement for removed sklearn.utils.tosequence)
+ steps = list(steps)
+
names, estimators = zip(*steps)
if len(dict(steps)) != len(steps):
raise ValueError(
"Provided step names are not unique: %s" % (names,))
# shallow copy of steps
- self.steps = tosequence(steps)
+ self.steps = steps
estimator = estimators[-1]
for e in estimators:
diff --git a/tests/test_pipeline.py b/tests/test_pipeline.py
index ee57b57..dd300c6 100644
--- a/tests/test_pipeline.py
+++ b/tests/test_pipeline.py
@@ -98,3 +98,73 @@ def test_raises_type_error(mock_fit):
"""
with pytest.raises(TypeError):
_call_fit(Trans().fit, 'X', 'y', kwarg='kwarg')
+
+
+def test_transformer_pipeline_accepts_list():
+ """
+ Test that TransformerPipeline accepts list of steps.
+ Regression test for issue #267 (tosequence removal in sklearn 1.7.0)
+ """
+ steps = [
+ ('trans1', Trans()),
+ ('trans2', Trans())
+ ]
+
+ pipeline = TransformerPipeline(steps)
+ assert len(pipeline.steps) == 2
+ assert pipeline.steps[0][0] == 'trans1'
+ assert pipeline.steps[1][0] == 'trans2'
+ assert isinstance(pipeline.steps, list)
+
+
+def test_transformer_pipeline_accepts_tuple():
+ """
+ Test that TransformerPipeline accepts tuple of steps.
+ Regression test for issue #267 (tosequence removal in sklearn 1.7.0)
+ """
+ steps = (
+ ('trans1', Trans()),
+ ('trans2', Trans())
+ )
+
+ pipeline = TransformerPipeline(steps)
+ assert len(pipeline.steps) == 2
+ assert pipeline.steps[0][0] == 'trans1'
+ assert pipeline.steps[1][0] == 'trans2'
+ assert isinstance(pipeline.steps, list)
+
+
+def test_transformer_pipeline_accepts_generator():
+ """
+ Test that TransformerPipeline accepts generator of steps.
+ Regression test for issue #267 (tosequence removal in sklearn 1.7.0)
+ """
+
+ def step_generator():
+ yield ('trans1', Trans())
+ yield ('trans2', Trans())
+
+ pipeline = TransformerPipeline(step_generator())
+ assert len(pipeline.steps) == 2
+ assert pipeline.steps[0][0] == 'trans1'
+ assert pipeline.steps[1][0] == 'trans2'
+ assert isinstance(pipeline.steps, list)
+
+
+def test_transformer_pipeline_steps_is_list():
+ """
+ Test that steps attribute is always a list after initialization.
+ This ensures list() conversion works correctly (replacing tosequence).
+ Regression test for issue #267 (tosequence removal in sklearn 1.7.0)
+ """
+ # Test with list input
+ pipeline1 = TransformerPipeline([('trans', Trans())])
+ assert isinstance(pipeline1.steps, list)
+
+ # Test with tuple input
+ pipeline2 = TransformerPipeline((('trans', Trans()),))
+ assert isinstance(pipeline2.steps, list)
+
+ # Test with iterator input
+ pipeline3 = TransformerPipeline(iter([('trans', Trans())]))
+ assert isinstance(pipeline3.steps, list)
\ No newline at end of file
|
