scikit-learn · ogrisel · Apr 8, 2024 · Apr 3, 2024 · Apr 8, 2024 · Apr 8, 2024
diff --git a/doc/conftest.py b/doc/conftest.py
@@ -9,9 +9,8 @@
 from sklearn.datasets import get_data_home
 from sklearn.datasets._base import _pkl_filepath
 from sklearn.datasets._twenty_newsgroups import CACHE_NAME
-from sklearn.utils import IS_PYPY
 from sklearn.utils._testing import SkipTest, check_skip_network
-from sklearn.utils.fixes import np_base_version, parse_version
+from sklearn.utils.fixes import _IS_PYPY, np_base_version, parse_version
 
 
 def setup_labeled_faces():
@@ -35,7 +34,7 @@ def setup_twenty_newsgroups():
 
 
 def setup_working_with_text_data():
-    if IS_PYPY and os.environ.get("CI", None):
+    if _IS_PYPY and os.environ.get("CI", None):
         raise SkipTest("Skipping too slow test with PyPy on CI")
     check_skip_network()
     cache_path = _pkl_filepath(get_data_home(), CACHE_NAME)

diff --git a/doc/whats_new/v1.5.rst b/doc/whats_new/v1.5.rst
@@ -335,6 +335,9 @@ Changelog
 :mod:`sklearn.utils`
 ....................
 
+- |API| :data:`utils.IS_PYPY` is deprecated and will be removed in version 1.7.
+  :pr:`28768` by :user:`Jérémie du Boisberranger <jeremiedbb>`.
+
 - |API| :func:`utils.tosequence` is deprecated and will be removed in version 1.7.
   :pr:`28763` by :user:`Jérémie du Boisberranger <jeremiedbb>`.
 

diff --git a/sklearn/_loss/tests/test_loss.py b/sklearn/_loss/tests/test_loss.py
@@ -27,8 +27,9 @@
     HuberLoss,
     PinballLoss,
 )
-from sklearn.utils import _IS_WASM, assert_all_finite
+from sklearn.utils import assert_all_finite
 from sklearn.utils._testing import create_memmap_backed_data, skip_if_32bit
+from sklearn.utils.fixes import _IS_WASM
 
 ALL_LOSSES = list(_LOSSES.values())
 

diff --git a/sklearn/base.py b/sklearn/base.py
@@ -16,14 +16,14 @@
 from . import __version__
 from ._config import config_context, get_config
 from .exceptions import InconsistentVersionWarning
-from .utils import _IS_32BIT
 from .utils._estimator_html_repr import _HTMLDocumentationLinkMixin, estimator_html_repr
 from .utils._metadata_requests import _MetadataRequester, _routing_enabled
 from .utils._param_validation import validate_parameter_constraints
 from .utils._set_output import _SetOutputMixin
 from .utils._tags import (
     _DEFAULT_TAGS,
 )
+from .utils.fixes import _IS_32BIT
 from .utils.validation import (
     _check_feature_names_in,
     _check_y,

diff --git a/sklearn/conftest.py b/sklearn/conftest.py
@@ -27,9 +27,9 @@
     fetch_species_distributions,
 )
 from sklearn.tests import random_seed
-from sklearn.utils import _IS_32BIT
 from sklearn.utils._testing import get_pytest_filterwarning_lines
 from sklearn.utils.fixes import (
+    _IS_32BIT,
     np_base_version,
     parse_version,
     sp_version,

diff --git a/sklearn/datasets/_svmlight_format_io.py b/sklearn/datasets/_svmlight_format_io.py
@@ -23,10 +23,11 @@
 import scipy.sparse as sp
 
 from .. import __version__
-from ..utils import IS_PYPY, check_array
+from ..utils import check_array
 from ..utils._param_validation import HasMethods, Interval, StrOptions, validate_params
+from ..utils.fixes import _IS_PYPY
 
-if not IS_PYPY:
+if not _IS_PYPY:
     from ._svmlight_format_fast import (
         _dump_svmlight_file,
         _load_svmlight_file,

diff --git a/sklearn/ensemble/_hist_gradient_boosting/tests/test_gradient_boosting.py b/sklearn/ensemble/_hist_gradient_boosting/tests/test_gradient_boosting.py
@@ -35,9 +35,10 @@
 from sklearn.model_selection import cross_val_score, train_test_split
 from sklearn.pipeline import make_pipeline
 from sklearn.preprocessing import KBinsDiscretizer, MinMaxScaler, OneHotEncoder
-from sklearn.utils import _IS_32BIT, shuffle
+from sklearn.utils import shuffle
 from sklearn.utils._openmp_helpers import _openmp_effective_n_threads
 from sklearn.utils._testing import _convert_container
+from sklearn.utils.fixes import _IS_32BIT
 
 n_threads = _openmp_effective_n_threads()
 

diff --git a/sklearn/experimental/tests/test_enable_hist_gradient_boosting.py b/sklearn/experimental/tests/test_enable_hist_gradient_boosting.py
@@ -4,8 +4,8 @@
 
 import pytest
 
-from sklearn.utils import _IS_WASM
 from sklearn.utils._testing import assert_run_python_script_without_output
+from sklearn.utils.fixes import _IS_WASM
 
 
 @pytest.mark.xfail(_IS_WASM, reason="cannot start subprocess")

diff --git a/sklearn/experimental/tests/test_enable_iterative_imputer.py b/sklearn/experimental/tests/test_enable_iterative_imputer.py
@@ -4,8 +4,8 @@
 
 import pytest
 
-from sklearn.utils import _IS_WASM
 from sklearn.utils._testing import assert_run_python_script_without_output
+from sklearn.utils.fixes import _IS_WASM
 
 
 @pytest.mark.xfail(_IS_WASM, reason="cannot start subprocess")

diff --git a/sklearn/experimental/tests/test_enable_successive_halving.py b/sklearn/experimental/tests/test_enable_successive_halving.py
@@ -4,8 +4,8 @@
 
 import pytest
 
-from sklearn.utils import _IS_WASM
 from sklearn.utils._testing import assert_run_python_script_without_output
+from sklearn.utils.fixes import _IS_WASM
 
 
 @pytest.mark.xfail(_IS_WASM, reason="cannot start subprocess")

diff --git a/sklearn/feature_extraction/tests/test_text.py b/sklearn/feature_extraction/tests/test_text.py
@@ -26,14 +26,13 @@
 from sklearn.model_selection import GridSearchCV, cross_val_score, train_test_split
 from sklearn.pipeline import Pipeline
 from sklearn.svm import LinearSVC
-from sklearn.utils import _IS_WASM, IS_PYPY
 from sklearn.utils._testing import (
     assert_allclose_dense_sparse,
     assert_almost_equal,
     fails_if_pypy,
     skip_if_32bit,
 )
-from sklearn.utils.fixes import CSC_CONTAINERS, CSR_CONTAINERS
+from sklearn.utils.fixes import _IS_PYPY, _IS_WASM, CSC_CONTAINERS, CSR_CONTAINERS
 
 JUNK_FOOD_DOCS = (
     "the pizza pizza beer copyright",
@@ -1069,7 +1068,7 @@ def test_pickling_vectorizer():
         copy = pickle.loads(s)
         assert type(copy) == orig.__class__
         assert copy.get_params() == orig.get_params()
-        if IS_PYPY and isinstance(orig, HashingVectorizer):
+        if _IS_PYPY and isinstance(orig, HashingVectorizer):
             continue
         else:
             assert_allclose_dense_sparse(
@@ -1347,7 +1346,7 @@ def test_vectorizers_invalid_ngram_range(vec):
         f"Invalid value for ngram_range={invalid_range} "
         "lower boundary larger than the upper boundary."
     )
-    if isinstance(vec, HashingVectorizer) and IS_PYPY:
+    if isinstance(vec, HashingVectorizer) and _IS_PYPY:
         pytest.xfail(reason="HashingVectorizer is not supported on PyPy")
 
     with pytest.raises(ValueError, match=message):
@@ -1462,7 +1461,7 @@ def build_preprocessor(self):
     ],
 )
 def test_callable_analyzer_error(Estimator, input_type, err_type, err_msg):
-    if issubclass(Estimator, HashingVectorizer) and IS_PYPY:
+    if issubclass(Estimator, HashingVectorizer) and _IS_PYPY:
         pytest.xfail("HashingVectorizer is not supported on PyPy")
     data = ["this is text, not file or filename"]
     with pytest.raises(err_type, match=err_msg):
@@ -1495,7 +1494,7 @@ def test_callable_analyzer_reraise_error(tmpdir, Estimator):
     def analyzer(doc):
         raise Exception("testing")
 
-    if issubclass(Estimator, HashingVectorizer) and IS_PYPY:
+    if issubclass(Estimator, HashingVectorizer) and _IS_PYPY:
         pytest.xfail("HashingVectorizer is not supported on PyPy")
 
     f = tmpdir.join("file.txt")

diff --git a/sklearn/feature_extraction/text.py b/sklearn/feature_extraction/text.py
@@ -27,8 +27,8 @@
 from ..base import BaseEstimator, OneToOneFeatureMixin, TransformerMixin, _fit_context
 from ..exceptions import NotFittedError
 from ..preprocessing import normalize
-from ..utils import _IS_32BIT
 from ..utils._param_validation import HasMethods, Interval, RealNotInt, StrOptions
+from ..utils.fixes import _IS_32BIT
 from ..utils.validation import FLOAT_DTYPES, check_array, check_is_fitted
 from ._hash import FeatureHasher
 from ._stop_words import ENGLISH_STOP_WORDS

diff --git a/sklearn/inspection/tests/test_partial_dependence.py b/sklearn/inspection/tests/test_partial_dependence.py
@@ -37,8 +37,8 @@
 )
 from sklearn.tree import DecisionTreeRegressor
 from sklearn.tree.tests.test_tree import assert_is_subtree
-from sklearn.utils import _IS_32BIT
 from sklearn.utils._testing import assert_allclose, assert_array_equal
+from sklearn.utils.fixes import _IS_32BIT
 from sklearn.utils.validation import check_random_state
 
 # toy sample

diff --git a/sklearn/linear_model/tests/test_logistic.py b/sklearn/linear_model/tests/test_logistic.py
@@ -37,9 +37,9 @@
 )
 from sklearn.preprocessing import LabelEncoder, StandardScaler, scale
 from sklearn.svm import l1_min_c
-from sklearn.utils import _IS_32BIT, compute_class_weight, shuffle
+from sklearn.utils import compute_class_weight, shuffle
 from sklearn.utils._testing import ignore_warnings, skip_if_no_parallel
-from sklearn.utils.fixes import COO_CONTAINERS, CSR_CONTAINERS
+from sklearn.utils.fixes import _IS_32BIT, COO_CONTAINERS, CSR_CONTAINERS
 
 pytestmark = pytest.mark.filterwarnings(
     "error::sklearn.exceptions.ConvergenceWarning:sklearn.*"

diff --git a/sklearn/linear_model/tests/test_ridge.py b/sklearn/linear_model/tests/test_ridge.py
@@ -39,7 +39,7 @@
     cross_val_predict,
 )
 from sklearn.preprocessing import minmax_scale
-from sklearn.utils import _IS_32BIT, check_random_state
+from sklearn.utils import check_random_state
 from sklearn.utils._testing import (
     assert_allclose,
     assert_almost_equal,
@@ -48,6 +48,7 @@
     ignore_warnings,
 )
 from sklearn.utils.fixes import (
+    _IS_32BIT,
     COO_CONTAINERS,
     CSC_CONTAINERS,
     CSR_CONTAINERS,

diff --git a/sklearn/tests/test_common.py b/sklearn/tests/test_common.py
@@ -61,7 +61,7 @@
     StandardScaler,
 )
 from sklearn.semi_supervised import LabelPropagation, LabelSpreading
-from sklearn.utils import _IS_WASM, IS_PYPY, all_estimators
+from sklearn.utils import all_estimators
 from sklearn.utils._tags import _DEFAULT_TAGS, _safe_tags
 from sklearn.utils._testing import (
     SkipTest,
@@ -87,6 +87,7 @@
     check_transformer_get_feature_names_out_pandas,
     parametrize_with_checks,
 )
+from sklearn.utils.fixes import _IS_PYPY, _IS_WASM
 
 
 def test_all_estimator_no_base_class():
@@ -221,7 +222,7 @@ def test_import_all_consistency():
         # Avoid test suite depending on setuptools
         if "sklearn._build_utils" in modname:
             continue
-        if IS_PYPY and (
+        if _IS_PYPY and (
             "_svmlight_format_io" in modname
             or "feature_extraction._hashing_fast" in modname
         ):

diff --git a/sklearn/tests/test_config.py b/sklearn/tests/test_config.py
@@ -6,7 +6,7 @@
 
 import sklearn
 from sklearn import config_context, get_config, set_config
-from sklearn.utils import _IS_WASM
+from sklearn.utils.fixes import _IS_WASM
 from sklearn.utils.parallel import Parallel, delayed
 
 

diff --git a/sklearn/tests/test_discriminant_analysis.py b/sklearn/tests/test_discriminant_analysis.py
@@ -11,14 +11,15 @@
     _cov,
 )
 from sklearn.preprocessing import StandardScaler
-from sklearn.utils import _IS_WASM, check_random_state
+from sklearn.utils import check_random_state
 from sklearn.utils._testing import (
     _convert_container,
     assert_allclose,
     assert_almost_equal,
     assert_array_almost_equal,
     assert_array_equal,
 )
+from sklearn.utils.fixes import _IS_WASM
 
 # Data is just 6 separable points in the plane
 X = np.array([[-2, -1], [-1, -1], [-1, -2], [1, 1], [1, 2], [2, 1]], dtype="f")

diff --git a/sklearn/tests/test_docstring_parameters.py b/sklearn/tests/test_docstring_parameters.py
@@ -22,7 +22,7 @@
 )
 from sklearn.linear_model import LogisticRegression
 from sklearn.preprocessing import FunctionTransformer
-from sklearn.utils import IS_PYPY, all_estimators
+from sklearn.utils import all_estimators
 from sklearn.utils._testing import (
     _get_func_name,
     check_docstring_parameters,
@@ -34,7 +34,7 @@
     _enforce_estimator_tags_X,
     _enforce_estimator_tags_y,
 )
-from sklearn.utils.fixes import parse_version, sp_version
+from sklearn.utils.fixes import _IS_PYPY, parse_version, sp_version
 
 # walk_packages() ignores DeprecationWarnings, now we need to ignore
 # FutureWarnings
@@ -75,7 +75,7 @@
 # Python 3.7
 @pytest.mark.filterwarnings("ignore::FutureWarning")
 @pytest.mark.filterwarnings("ignore::DeprecationWarning")
-@pytest.mark.skipif(IS_PYPY, reason="test segfaults on PyPy")
+@pytest.mark.skipif(_IS_PYPY, reason="test segfaults on PyPy")
 def test_docstring_parameters():
     # Test module docstring formatting
 

diff --git a/sklearn/tree/tests/test_tree.py b/sklearn/tree/tests/test_tree.py
@@ -44,7 +44,7 @@
     _check_value_ndarray,
 )
 from sklearn.tree._tree import Tree as CythonTree
-from sklearn.utils import _IS_32BIT, compute_sample_weight
+from sklearn.utils import compute_sample_weight
 from sklearn.utils._testing import (
     assert_almost_equal,
     assert_array_almost_equal,
@@ -54,7 +54,12 @@
     skip_if_32bit,
 )
 from sklearn.utils.estimator_checks import check_sample_weights_invariance
-from sklearn.utils.fixes import COO_CONTAINERS, CSC_CONTAINERS, CSR_CONTAINERS
+from sklearn.utils.fixes import (
+    _IS_32BIT,
+    COO_CONTAINERS,
+    CSC_CONTAINERS,
+    CSR_CONTAINERS,
+)
 from sklearn.utils.validation import check_random_state
 
 CLF_CRITERIONS = ("gini", "log_loss")

diff --git a/sklearn/utils/__init__.py b/sklearn/utils/__init__.py
@@ -2,8 +2,7 @@
 The :mod:`sklearn.utils` module includes various utilities.
 """
 
-import platform
-import struct
+import warnings
 from collections.abc import Sequence
 
 import numpy as np
@@ -81,9 +80,18 @@
     "gen_even_slices",
 ]
 
-IS_PYPY = platform.python_implementation() == "PyPy"
-_IS_32BIT = 8 * struct.calcsize("P") == 32
-_IS_WASM = platform.machine() in ["wasm32", "wasm64"]
+
+# TODO(1.7): remove
+def __getattr__(name):
+    if name == "IS_PYPY":
+        warnings.warn(
+            "IS_PYPY is deprecated and will be removed in 1.7.",
+            FutureWarning,
+        )
+        from .fixes import _IS_PYPY
+
+        return _IS_PYPY
+    raise AttributeError(f"module {__name__} has no attribute {name}")
 
 
 # TODO(1.7): remove tosequence

diff --git a/sklearn/utils/_testing.py b/sklearn/utils/_testing.py
@@ -44,9 +44,10 @@
 )
 
 import sklearn
-from sklearn.utils import _IS_32BIT, IS_PYPY
 from sklearn.utils._array_api import _check_array_api_dispatch
 from sklearn.utils.fixes import (
+    _IS_32BIT,
+    _IS_PYPY,
     VisibleDeprecationWarning,
     _in_unstable_openblas_configuration,
     parse_version,
@@ -369,7 +370,7 @@ def set_random_state(estimator, random_state=0):
     import pytest
 
     skip_if_32bit = pytest.mark.skipif(_IS_32BIT, reason="skipped on 32bit platforms")
-    fails_if_pypy = pytest.mark.xfail(IS_PYPY, reason="not compatible with PyPy")
+    fails_if_pypy = pytest.mark.xfail(_IS_PYPY, reason="not compatible with PyPy")
     fails_if_unstable_openblas = pytest.mark.xfail(
         _in_unstable_openblas_configuration(),
         reason="OpenBLAS is unstable for this configuration",

diff --git a/sklearn/utils/discovery.py b/sklearn/utils/discovery.py
@@ -78,8 +78,8 @@ def all_estimators(type_filter=None):
         RegressorMixin,
         TransformerMixin,
     )
-    from . import IS_PYPY
     from ._testing import ignore_warnings
+    from .fixes import _IS_PYPY
 
     def is_abstract(c):
         if not (hasattr(c, "__abstractmethods__")):
@@ -108,7 +108,7 @@ def is_abstract(c):
 
             # TODO: Remove when FeatureHasher is implemented in PYPY
             # Skips FeatureHasher for PYPY
-            if IS_PYPY and "feature_extraction" in module_name:
+            if _IS_PYPY and "feature_extraction" in module_name:
                 classes = [
                     (name, est_cls)
                     for name, est_cls in classes