MAINT conversion old->new/new->old tags (bis) (#30327)

Co-authored-by: Adrin Jalali <[email protected]> Co-authored-by: Thomas J. Fan <[email protected]> Co-authored-by: Loïc Estève <[email protected]>
scikit-learn · Nov 23, 2024 · 46a7c9a · 46a7c9a
1 parent 27a903b
commit 46a7c9a
Show file tree

Hide file tree

Showing 3 changed files with 840 additions and 12 deletions.
diff --git a/sklearn/base.py b/sklearn/base.py
@@ -389,6 +389,33 @@ def __setstate__(self, state):
         except AttributeError:
             self.__dict__.update(state)
 
+    # TODO(1.7): Remove this method
+    def _more_tags(self):
+        """This code should never be reached since our `get_tags` will fallback on
+        `__sklearn_tags__` implemented below. We keep it for backward compatibility.
+        It is tested in `test_base_estimator_more_tags` in
+        `sklearn/utils/testing/test_tags.py`."""
+        from sklearn.utils._tags import _to_old_tags, default_tags
+
+        warnings.warn(
+            "The `_more_tags` method is deprecated in 1.6 and will be removed in "
+            "1.7. Please implement the `__sklearn_tags__` method.",
+            category=FutureWarning,
+        )
+        return _to_old_tags(default_tags(self))
+
+    # TODO(1.7): Remove this method
+    def _get_tags(self):
+        from sklearn.utils._tags import _to_old_tags, get_tags
+
+        warnings.warn(
+            "The `_get_tags` method is deprecated in 1.6 and will be removed in "
+            "1.7. Please implement the `__sklearn_tags__` method.",
+            category=FutureWarning,
+        )
+
+        return _to_old_tags(get_tags(self))
+
     def __sklearn_tags__(self):
         return Tags(
             estimator_type=None,

diff --git a/sklearn/utils/_tags.py b/sklearn/utils/_tags.py
@@ -1,7 +1,9 @@
 from __future__ import annotations
 
 import warnings
+from collections import OrderedDict
 from dataclasses import dataclass, field
+from itertools import chain
 
 from .fixes import _dataclass_args
 
@@ -290,6 +292,71 @@ def default_tags(estimator) -> Tags:
     )
 
 
+# TODO(1.7): Remove this function
+def _find_tags_provider(estimator, warn=True):
+    """Find the tags provider for an estimator.
+
+    Parameters
+    ----------
+    estimator : estimator object
+        The estimator to find the tags provider for.
+
+    warn : bool, default=True
+        Whether to warn if the tags provider is not found.
+
+    Returns
+    -------
+    tag_provider : str
+        The tags provider for the estimator. Can be one of:
+        - "_get_tags": to use the old tags infrastructure
+        - "__sklearn_tags__": to use the new tags infrastructure
+    """
+    mro_model = type(estimator).mro()
+    tags_mro = OrderedDict()
+    for klass in mro_model:
+        tags_provider = []
+        if "_more_tags" in vars(klass):
+            tags_provider.append("_more_tags")
+        if "_get_tags" in vars(klass):
+            tags_provider.append("_get_tags")
+        if "__sklearn_tags__" in vars(klass):
+            tags_provider.append("__sklearn_tags__")
+        tags_mro[klass.__name__] = tags_provider
+
+    all_providers = set(chain.from_iterable(tags_mro.values()))
+    if "__sklearn_tags__" not in all_providers:
+        # default on the old tags infrastructure
+        return "_get_tags"
+
+    tag_provider = "__sklearn_tags__"
+    for klass in tags_mro:
+        has_get_or_more_tags = any(
+            provider in tags_mro[klass] for provider in ("_get_tags", "_more_tags")
+        )
+        has_sklearn_tags = "__sklearn_tags__" in tags_mro[klass]
+
+        if tags_mro[klass] and tag_provider == "__sklearn_tags__":  # is it empty
+            if has_get_or_more_tags and not has_sklearn_tags:
+                # Case where a class does not implement __sklearn_tags__ and we fallback
+                # to _get_tags. We should therefore warn for implementing
+                # __sklearn_tags__.
+                tag_provider = "_get_tags"
+                break
+
+    if warn and tag_provider == "_get_tags":
+        warnings.warn(
+            f"The {estimator.__class__.__name__} or classes from which it inherits "
+            "use `_get_tags` and `_more_tags`. Please define the "
+            "`__sklearn_tags__` method, or inherit from `sklearn.base.BaseEstimator` "
+            "and/or other appropriate mixins such as `sklearn.base.TransformerMixin`, "
+            "`sklearn.base.ClassifierMixin`, `sklearn.base.RegressorMixin`, and "
+            "`sklearn.base.OutlierMixin`. From scikit-learn 1.7, not defining "
+            "`__sklearn_tags__` will raise an error.",
+            category=FutureWarning,
+        )
+    return tag_provider
+
+
 def get_tags(estimator) -> Tags:
     """Get estimator tags.
 
@@ -316,19 +383,201 @@ def get_tags(estimator) -> Tags:
         The estimator tags.
     """
 
-    if hasattr(estimator, "__sklearn_tags__"):
+    tag_provider = _find_tags_provider(estimator)
+
+    if tag_provider == "__sklearn_tags__":
         tags = estimator.__sklearn_tags__()
     else:
-        warnings.warn(
-            f"Estimator {estimator} has no __sklearn_tags__ attribute, which is "
-            "defined in `sklearn.base.BaseEstimator`. This will raise an error in "
-            "scikit-learn 1.8. Please define the __sklearn_tags__ method, or inherit "
-            "from `sklearn.base.BaseEstimator` and other appropriate mixins such as "
-            "`sklearn.base.TransformerMixin`, `sklearn.base.ClassifierMixin`, "
-            "`sklearn.base.RegressorMixin`, and `sklearn.base.ClusterMixin`, and "
-            "`sklearn.base.OutlierMixin`.",
-            category=FutureWarning,
+        # TODO(1.7): Remove this branch of the code
+        # Let's go through the MRO and patch each class implementing _more_tags
+        sklearn_tags_provider = {}
+        more_tags_provider = {}
+        class_order = []
+        for klass in reversed(type(estimator).mro()):
+            if "__sklearn_tags__" in vars(klass):
+                sklearn_tags_provider[klass] = klass.__sklearn_tags__(estimator)  # type: ignore[attr-defined]
+                class_order.append(klass)
+            elif "_more_tags" in vars(klass):
+                more_tags_provider[klass] = klass._more_tags(estimator)  # type: ignore[attr-defined]
+                class_order.append(klass)
+
+        # Find differences between consecutive in the case of __sklearn_tags__
+        # inheritance
+        sklearn_tags_diff = {}
+        items = list(sklearn_tags_provider.items())
+        for current_item, next_item in zip(items[:-1], items[1:]):
+            current_name, current_tags = current_item
+            next_name, next_tags = next_item
+            current_tags = _to_old_tags(current_tags)
+            next_tags = _to_old_tags(next_tags)
+
+            # Compare tags and store differences
+            diff = {}
+            for key in current_tags:
+                if current_tags[key] != next_tags[key]:
+                    diff[key] = next_tags[key]
+
+            sklearn_tags_diff[next_name] = diff
+
+        tags = {}
+        for klass in class_order:
+            if klass in sklearn_tags_diff:
+                tags.update(sklearn_tags_diff[klass])
+            elif klass in more_tags_provider:
+                tags.update(more_tags_provider[klass])
+
+        tags = _to_new_tags(
+            {**_to_old_tags(default_tags(estimator)), **tags}, estimator
         )
-        tags = default_tags(estimator)
 
     return tags
+
+
+# TODO(1.7): Remove this function
+def _safe_tags(estimator, key=None):
+    warnings.warn(
+        "The `_safe_tags` function is deprecated in 1.6 and will be removed in "
+        "1.7. Use the public `get_tags` function instead and make sure to implement "
+        "the `__sklearn_tags__` method.",
+        category=FutureWarning,
+    )
+    tags = _to_old_tags(get_tags(estimator))
+
+    if key is not None:
+        if key not in tags:
+            raise ValueError(
+                f"The key {key} is not defined for the class "
+                f"{estimator.__class__.__name__}."
+            )
+        return tags[key]
+    return tags
+
+
+# TODO(1.7): Remove this function
+def _to_new_tags(old_tags, estimator=None):
+    """Utility function convert old tags (dictionary) to new tags (dataclass)."""
+    input_tags = InputTags(
+        one_d_array="1darray" in old_tags["X_types"],
+        two_d_array="2darray" in old_tags["X_types"],
+        three_d_array="3darray" in old_tags["X_types"],
+        sparse="sparse" in old_tags["X_types"],
+        categorical="categorical" in old_tags["X_types"],
+        string="string" in old_tags["X_types"],
+        dict="dict" in old_tags["X_types"],
+        positive_only=old_tags["requires_positive_X"],
+        allow_nan=old_tags["allow_nan"],
+        pairwise=old_tags["pairwise"],
+    )
+    target_tags = TargetTags(
+        required=old_tags["requires_y"],
+        one_d_labels="1dlabels" in old_tags["X_types"],
+        two_d_labels="2dlabels" in old_tags["X_types"],
+        positive_only=old_tags["requires_positive_y"],
+        multi_output=old_tags["multioutput"] or old_tags["multioutput_only"],
+        single_output=not old_tags["multioutput_only"],
+    )
+    if estimator is not None and (
+        hasattr(estimator, "transform") or hasattr(estimator, "fit_transform")
+    ):
+        transformer_tags = TransformerTags(
+            preserves_dtype=old_tags["preserves_dtype"],
+        )
+    else:
+        transformer_tags = None
+    estimator_type = getattr(estimator, "_estimator_type", None)
+    if estimator_type == "classifier":
+        classifier_tags = ClassifierTags(
+            poor_score=old_tags["poor_score"],
+            multi_class=not old_tags["binary_only"],
+            multi_label=old_tags["multilabel"],
+        )
+    else:
+        classifier_tags = None
+    if estimator_type == "regressor":
+        regressor_tags = RegressorTags(
+            poor_score=old_tags["poor_score"],
+            multi_label=old_tags["multilabel"],
+        )
+    else:
+        regressor_tags = None
+    return Tags(
+        estimator_type=estimator_type,
+        target_tags=target_tags,
+        transformer_tags=transformer_tags,
+        classifier_tags=classifier_tags,
+        regressor_tags=regressor_tags,
+        input_tags=input_tags,
+        array_api_support=old_tags["array_api_support"],
+        no_validation=old_tags["no_validation"],
+        non_deterministic=old_tags["non_deterministic"],
+        requires_fit=old_tags["requires_fit"],
+        _skip_test=old_tags["_skip_test"],
+    )
+
+
+# TODO(1.7): Remove this function
+def _to_old_tags(new_tags):
+    """Utility function convert old tags (dictionary) to new tags (dataclass)."""
+    if new_tags.classifier_tags:
+        binary_only = not new_tags.classifier_tags.multi_class
+        multilabel_clf = new_tags.classifier_tags.multi_label
+        poor_score_clf = new_tags.classifier_tags.poor_score
+    else:
+        binary_only = False
+        multilabel_clf = False
+        poor_score_clf = False
+
+    if new_tags.regressor_tags:
+        multilabel_reg = new_tags.regressor_tags.multi_label
+        poor_score_reg = new_tags.regressor_tags.poor_score
+    else:
+        multilabel_reg = False
+        poor_score_reg = False
+
+    if new_tags.transformer_tags:
+        preserves_dtype = new_tags.transformer_tags.preserves_dtype
+    else:
+        preserves_dtype = ["float64"]
+
+    tags = {
+        "allow_nan": new_tags.input_tags.allow_nan,
+        "array_api_support": new_tags.array_api_support,
+        "binary_only": binary_only,
+        "multilabel": multilabel_clf or multilabel_reg,
+        "multioutput": new_tags.target_tags.multi_output,
+        "multioutput_only": (
+            not new_tags.target_tags.single_output and new_tags.target_tags.multi_output
+        ),
+        "no_validation": new_tags.no_validation,
+        "non_deterministic": new_tags.non_deterministic,
+        "pairwise": new_tags.input_tags.pairwise,
+        "preserves_dtype": preserves_dtype,
+        "poor_score": poor_score_clf or poor_score_reg,
+        "requires_fit": new_tags.requires_fit,
+        "requires_positive_X": new_tags.input_tags.positive_only,
+        "requires_y": new_tags.target_tags.required,
+        "requires_positive_y": new_tags.target_tags.positive_only,
+        "_skip_test": new_tags._skip_test,
+        "stateless": new_tags.requires_fit,
+    }
+    X_types = []
+    if new_tags.input_tags.one_d_array:
+        X_types.append("1darray")
+    if new_tags.input_tags.two_d_array:
+        X_types.append("2darray")
+    if new_tags.input_tags.three_d_array:
+        X_types.append("3darray")
+    if new_tags.input_tags.sparse:
+        X_types.append("sparse")
+    if new_tags.input_tags.categorical:
+        X_types.append("categorical")
+    if new_tags.input_tags.string:
+        X_types.append("string")
+    if new_tags.input_tags.dict:
+        X_types.append("dict")
+    if new_tags.target_tags.one_d_labels:
+        X_types.append("1dlabels")
+    if new_tags.target_tags.two_d_labels:
+        X_types.append("2dlabels")
+    tags["X_types"] = X_types
+    return tags