PERF/TYP: typing cast in __getitem__ gives quite some overhead (#44624)

* PERF/TYP: typing cast in __getitem__ gives quite some overhead

* stringify

* categorical as well

Author: jorisvandenbossche

pandas/core/arrays/categorical.py

@@ -509,7 +509,7 @@ def astype(self, dtype: AstypeArg, copy: bool = True) -> ArrayLike:
             result = self.copy() if copy else self
 
         elif is_categorical_dtype(dtype):
-            dtype = cast(Union[str, CategoricalDtype], dtype)
+            dtype = cast("Union[str, CategoricalDtype]", dtype)
 
             # GH 10696/18593/18630
             dtype = self.dtype.update_dtype(dtype)

pandas/core/arrays/datetimelike.py

@@ -331,10 +331,13 @@ def __getitem__(
         This getitem defers to the underlying array, which by-definition can
         only handle list-likes, slices, and integer scalars
         """
-        # Use cast as we know we will get back a DatetimeLikeArray or DTScalar
+        # Use cast as we know we will get back a DatetimeLikeArray or DTScalar,
+        # but skip evaluating the Union at runtime for performance
+        # (see https://github.com/pandas-dev/pandas/pull/44624)
         result = cast(
-            Union[DatetimeLikeArrayT, DTScalarOrNaT], super().__getitem__(key)
+            "Union[DatetimeLikeArrayT, DTScalarOrNaT]", super().__getitem__(key)
         )
+        result = super().__getitem__(key)
         if lib.is_scalar(result):
             return result
         else: