xplainable-preprocessing 0.3.0__tar.gz → 0.3.1__tar.gz

{xplainable_preprocessing-0.3.0 → xplainable_preprocessing-0.3.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: xplainable-preprocessing
-Version: 0.3.0
+Version: 0.3.1
 Summary: Shared preprocessing pipeline package for xplainable
 Requires-Python: >=3.9
 Requires-Dist: cloudpickle>=3.0

{xplainable_preprocessing-0.3.0 → xplainable_preprocessing-0.3.1}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "xplainable-preprocessing"
-version = "0.3.0"
+version = "0.3.1"
 description = "Shared preprocessing pipeline package for xplainable"
 requires-python = ">=3.9"
 dependencies = [

{xplainable_preprocessing-0.3.0 → xplainable_preprocessing-0.3.1}/src/xplainable_preprocessing/transformers/expression.py

@@ -5,6 +5,17 @@ from __future__ import annotations
 import pandas as pd
 from sklearn.base import BaseEstimator, TransformerMixin
 
+try:
+    # Semi-private pandas API (stable since 1.0, present through 2.x). It is
+    # the canonical way to reproduce how pandas mangles backtick-quoted names
+    # (`Monthly Charges` -> BACKTICK_QUOTED_STRING_Monthly_Charges) before
+    # resolving them; the mangling table is version-specific, so we do not
+    # re-implement it. If a future pandas removes it, the package still
+    # imports and only backtick-quoted column references lose support.
+    from pandas.core.computation.parsing import clean_column_name
+except ImportError:  # pragma: no cover
+    clean_column_name = None
+
 
 class ExpressionTransformer(BaseEstimator, TransformerMixin):
     """Create new columns using pandas.eval() expressions.
@@ -30,6 +41,11 @@ class ExpressionTransformer(BaseEstimator, TransformerMixin):
         Xt = X.copy()
         # Support both bare column names ("age * salary") and df reference ("df['age'] * df['salary']")
         local_dict = {col: Xt[col] for col in Xt.columns}
+        # Backtick-quoted names ("`Monthly Charges` * Tenure") are rewritten by
+        # pandas' parser into mangled identifiers regardless of engine, so the
+        # same Series must also be reachable under the mangled alias.
+        if clean_column_name is not None:
+            local_dict.update({clean_column_name(col): Xt[col] for col in Xt.columns})
         local_dict["df"] = Xt
         Xt[self.output_column] = pd.eval(self.expression, local_dict=local_dict, engine="python")
         return Xt

{xplainable_preprocessing-0.3.0 → xplainable_preprocessing-0.3.1}/tests/test_transformers/test_expression.py

@@ -28,3 +28,26 @@ class TestExpressionTransformer:
         df = pd.DataFrame({"a": [100, 200], "b": [50, 60]})
         result = t.fit_transform(df)
         assert list(result["result"]) == [5.0, 12.0]
+
+    def test_backtick_quoted_column_with_spaces(self):
+        # pandas' expression parser mangles `Monthly Charges` into the
+        # identifier BACKTICK_QUOTED_STRING_Monthly_Charges before lookup, so
+        # local_dict must also be reachable under that mangled name.
+        t = ExpressionTransformer(
+            expression="`Monthly Charges` * Tenure",
+            output_column="Total Charges",
+        )
+        df = pd.DataFrame({"Monthly Charges": [10.0, 20.0, 30.0], "Tenure": [1, 2, 3]})
+        result = t.fit_transform(df)
+        assert list(result["Total Charges"]) == [10.0, 40.0, 90.0]
+        assert "Monthly Charges" in result.columns
+        assert "Tenure" in result.columns
+
+    def test_df_indexing_style_with_spaced_column_still_works(self):
+        t = ExpressionTransformer(
+            expression="df['Monthly Charges'] * 2",
+            output_column="Doubled",
+        )
+        df = pd.DataFrame({"Monthly Charges": [10.0, 20.0, 30.0]})
+        result = t.fit_transform(df)
+        assert list(result["Doubled"]) == [20.0, 40.0, 60.0]