@@ -1,60 +0,0 @@
-# Stub generated from the C++ (pybind11) signatures in ``src/_image_wrapper.cpp``.
-# NOT verified with matplotlib's ``stubtest`` (the extension was not importable
-# when this stub was written); treat the annotations as best-effort.
-from enum import Enum
-from typing import cast
-
-import numpy as np
-
-class _InterpolationType(Enum):
-    NEAREST = cast(int, ...)
-    BILINEAR = cast(int, ...)
-    BICUBIC = cast(int, ...)
-    SPLINE16 = cast(int, ...)
-    SPLINE36 = cast(int, ...)
-    HANNING = cast(int, ...)
-    HAMMING = cast(int, ...)
-    HERMITE = cast(int, ...)
-    KAISER = cast(int, ...)
-    QUADRIC = cast(int, ...)
-    CATROM = cast(int, ...)
-    GAUSSIAN = cast(int, ...)
-    BESSEL = cast(int, ...)
-    MITCHELL = cast(int, ...)
-    SINC = cast(int, ...)
-    LANCZOS = cast(int, ...)
-    BLACKMAN = cast(int, ...)
-
-# ``_InterpolationType`` uses ``.export_values()``, so every member is also
-# injected as a module-level constant.
-NEAREST: _InterpolationType
-BILINEAR: _InterpolationType
-BICUBIC: _InterpolationType
-SPLINE16: _InterpolationType
-SPLINE36: _InterpolationType
-HANNING: _InterpolationType
-HAMMING: _InterpolationType
-HERMITE: _InterpolationType
-KAISER: _InterpolationType
-QUADRIC: _InterpolationType
-CATROM: _InterpolationType
-GAUSSIAN: _InterpolationType
-BESSEL: _InterpolationType
-MITCHELL: _InterpolationType
-SINC: _InterpolationType
-LANCZOS: _InterpolationType
-BLACKMAN: _InterpolationType
-
-def resample(
-    input_array: np.ndarray,
-    output_array: np.ndarray,
-    transform: np.ndarray,
-    interpolation: _InterpolationType = ...,
-    resample: bool = ...,
-    alpha: float = ...,
-    norm: bool = ...,
-    radius: float = ...,
-) -> None: ...
-def calculate_rms_and_diff(
-    expected_image: np.ndarray, actual_image: np.ndarray
-) -> tuple[float, np.ndarray]: ...
@@ -1,7 +1,8 @@
 from collections.abc import Callable, Sequence
+from enum import Enum
 import os
 import pathlib
-from typing import Any, BinaryIO, Literal
+from typing import Any, BinaryIO, Literal, cast
 
 import numpy as np
 from numpy.typing import ArrayLike, NDArray
@@ -19,34 +20,58 @@ from matplotlib.transforms import Affine2D, BboxBase, Bbox, Transform
 # These names are re-exported from matplotlib._image.
 #
 
-BESSEL: int
-BICUBIC: int
-BILINEAR: int
-BLACKMAN: int
-CATROM: int
-GAUSSIAN: int
-HAMMING: int
-HANNING: int
-HERMITE: int
-KAISER: int
-LANCZOS: int
-MITCHELL: int
-NEAREST: int
-QUADRIC: int
-SINC: int
-SPLINE16: int
-SPLINE36: int
+class _InterpolationType(Enum):
+    NEAREST = cast(int, ...)
+    BILINEAR = cast(int, ...)
+    BICUBIC = cast(int, ...)
+    SPLINE16 = cast(int, ...)
+    SPLINE36 = cast(int, ...)
+    HANNING = cast(int, ...)
+    HAMMING = cast(int, ...)
+    HERMITE = cast(int, ...)
+    KAISER = cast(int, ...)
+    QUADRIC = cast(int, ...)
+    CATROM = cast(int, ...)
+    GAUSSIAN = cast(int, ...)
+    BESSEL = cast(int, ...)
+    MITCHELL = cast(int, ...)
+    SINC = cast(int, ...)
+    LANCZOS = cast(int, ...)
+    BLACKMAN = cast(int, ...)
+
+# ``_InterpolationType`` uses ``.export_values()``, so every member is also
+# injected as a module-level constant.
+NEAREST: _InterpolationType
+BILINEAR: _InterpolationType
+BICUBIC: _InterpolationType
+SPLINE16: _InterpolationType
+SPLINE36: _InterpolationType
+HANNING: _InterpolationType
+HAMMING: _InterpolationType
+HERMITE: _InterpolationType
+KAISER: _InterpolationType
+QUADRIC: _InterpolationType
+CATROM: _InterpolationType
+GAUSSIAN: _InterpolationType
+BESSEL: _InterpolationType
+MITCHELL: _InterpolationType
+SINC: _InterpolationType
+LANCZOS: _InterpolationType
+BLACKMAN: _InterpolationType
 
 def resample(
     input_array: NDArray[np.float32] | NDArray[np.float64] | NDArray[np.int8],
     output_array: NDArray[np.float32] | NDArray[np.float64] | NDArray[np.int8],
     transform: Transform,
-    interpolation: int = ...,
+    interpolation: _InterpolationType = ...,
     resample: bool = ...,
     alpha: float = ...,
     norm: bool = ...,
     radius: float = ...,
 ) -> None: ...
+def calculate_rms_and_diff(
+    expected_image: np.ndarray, actual_image: np.ndarray
+) -> tuple[float, np.ndarray]: ...
 
 #
 # END names re-exported from matplotlib._image.