fix(preprocessor): reject NumPy input of the wrong width at transform

docs/core_concepts/configuration.md

@@ -27,7 +27,10 @@ strategy strings is in the [representation comparison](../representations/compar
 
 `Preprocessor` also accepts a plain `numpy.ndarray`, not just a `DataFrame`. Columns are
 named `feature_0`, `feature_1`, ... in position order, then detected as numerical or
-categorical exactly as they would be for a `DataFrame`.
+categorical exactly as they would be for a `DataFrame`. Because array columns are matched by
+position, `transform` raises the usual scikit-learn error
+(`X has 4 features, but Preprocessor is expecting 3 features as input.`) when an array has a
+different number of columns than the one seen at `fit`.
 
 ```python
 import numpy as np

pretab/preprocessor.py

@@ -632,10 +632,25 @@ def transform(self, X, embeddings=None, return_array=None):
             (or ``return_array=False``); a SciPy CSR matrix (or CSR blocks) when
             ``output_format`` resolves to ``"sparse"``; or a pandas / polars DataFrame
             when configured via :meth:`set_output`.
+
+        Raises
+        ------
+        PretabDataError
+            If ``X`` is a NumPy array whose number of columns differs from
+            ``n_features_in_``. Array columns are matched by position, so such an
+            array can never be routed correctly.
         """
 
         check_is_fitted(self)
 
+        # Array columns are matched by position, so a different width can never be
+        # routed correctly; an extra column would silently shift or drop features.
+        if isinstance(X, np.ndarray) and X.ndim == 2 and X.shape[1] != self.n_features_in_:
+            raise PretabDataError(
+                f"X has {X.shape[1]} features, but {type(self).__name__} is expecting "
+                f"{self.n_features_in_} features as input."
+            )
+
         X = to_dataframe(X, copy=True)
 
         if self.missing_policy == "error":

tests/integration/test_preprocessor.py

@@ -377,3 +377,23 @@ def test_unknown_feature_preprocessing_key_raises(sample_data):
     X, y = sample_data
     with pytest.raises(InvalidParamError, match="feature_preprocessing"):
         Preprocessor(feature_preprocessing={"nnum1": "minmax"}).fit(X, y)
+
+
+# --- ndarray width check at transform (issue #72) ---------------------------------
+
+
+@pytest.mark.parametrize("width", [2, 4])
+def test_transform_rejects_an_ndarray_of_a_different_width(width):
+    """Regression guard for issue #72: an extra array column silently shifted features."""
+    rng = np.random.default_rng(0)
+    X, y = rng.normal(size=(60, 3)), rng.normal(size=60)
+    pre = Preprocessor(random_state=0).fit(X, y)
+    with pytest.raises(ValueError, match=f"X has {width} features, but Preprocessor is expecting 3"):
+        pre.transform(rng.normal(size=(5, width)))
+
+
+def test_transform_accepts_an_ndarray_of_the_fitted_width():
+    rng = np.random.default_rng(0)
+    X, y = rng.normal(size=(60, 3)), rng.normal(size=60)
+    pre = Preprocessor(random_state=0).fit(X, y)
+    assert np.asarray(pre.transform(X)).shape[0] == 60