BUG: Fix potential segfault after pd.Categorical(pd.Series(...), categories=...)

Author: batterseapower

Diff for: doc/source/whatsnew/v0.24.2.rst

@@ -27,6 +27,8 @@ Fixed Regressions
 - Fixed regression in :meth:`DataFrame.duplicated()`, where empty dataframe was not returning a boolean dtyped Series. (:issue:`25184`)
 - Fixed regression in :meth:`Series.min` and :meth:`Series.max` where ``numeric_only=True`` was ignored when the ``Series`` contained ```Categorical`` data (:issue:`25299`)
 
+- Fixed regression in ``Categorical``, where constructing it from a categorical ``Series`` and an explicit ``categories=`` that differed from that in the ``Series`` could create a object with nonsense data. In particular, the ``Categorical`` could have out-of-bounds ``codes`` which would trigger segfaults elsewhere in Pandas. (:issue:`25318`)
+
 .. _whatsnew_0242.enhancements:
 
 Enhancements

Diff for: pandas/core/arrays/categorical.py

@@ -323,14 +323,6 @@ def __init__(self, values, categories=None, ordered=None, dtype=None,
         # we may have dtype.categories be None, and we need to
         # infer categories in a factorization step futher below
 
-        if is_categorical(values):
-            # GH23814, for perf, if values._values already an instance of
-            # Categorical, set values to codes, and run fastpath
-            if (isinstance(values, (ABCSeries, ABCIndexClass)) and
-               isinstance(values._values, type(self))):
-                values = values._values.codes.copy()
-                fastpath = True
-
         if fastpath:
             self._codes = coerce_indexer_dtype(values, dtype.categories)
             self._dtype = self._dtype.update_dtype(dtype)
@@ -2625,6 +2617,9 @@ def _recode_for_categories(codes, old_categories, new_categories):
     if len(old_categories) == 0:
         # All null anyway, so just retain the nulls
         return codes.copy()
+    elif new_categories.equals(old_categories):
+        # Same categories, so no need to actually recode
+        return codes.copy()
     indexer = coerce_indexer_dtype(new_categories.get_indexer(old_categories),
                                    new_categories)
     new_codes = take_1d(indexer, codes.copy(), fill_value=-1)

Diff for: pandas/tests/arrays/categorical/test_constructors.py

@@ -149,6 +149,16 @@ def test_constructor(self):
                          categories=["a", "b", "c", "d"])
         tm.assert_categorical_equal(c1, c2)
 
+        # GH25318
+        c0 = Categorical(["a", "b", "c", "a"])
+        c1 = Categorical(["a", "b", "c", "a"], categories=["b", "c"])
+
+        c2 = Categorical(c0, categories=c1.categories)
+        tm.assert_categorical_equal(c1, c2)
+
+        c3 = Categorical(Series(c0), categories=c1.categories)
+        tm.assert_categorical_equal(c1, c3)
+
         # This should result in integer categories, not float!
         cat = Categorical([1, 2, 3, np.nan], categories=[1, 2, 3])
         assert is_integer_dtype(cat.categories)