3561 Fix new mypy errors for numpy 1.22.0

monai/apps/datasets.py

@@ -122,7 +122,7 @@ def __init__(
             copy_cache=copy_cache,
         )
 
-    def randomize(self, data: List[int]) -> None:
+    def randomize(self, data: np.ndarray) -> None:
         self.R.shuffle(data)
 
     def get_num_classes(self) -> int:
@@ -331,7 +331,7 @@ def get_indices(self) -> np.ndarray:
         """
         return self.indices
 
-    def randomize(self, data: List[int]) -> None:
+    def randomize(self, data: np.ndarray) -> None:
         self.R.shuffle(data)
 
     def get_properties(self, keys: Optional[Union[Sequence[str], str]] = None):

monai/apps/pathology/transforms/spatial/array.py

@@ -60,7 +60,7 @@ def __call__(self, image: NdarrayOrTensor) -> NdarrayOrTensor:
             if isinstance(image, torch.Tensor):
                 return torch.stack([image])
             elif isinstance(image, np.ndarray):
-                return np.stack([image])
+                return np.stack([image])  # type: ignore
             else:
                 raise ValueError(f"Input type [{type(image)}] is not supported.")
 

monai/data/utils.py

@@ -539,7 +539,7 @@ def rectify_header_sform_qform(img_nii):
     return img_nii
 
 
-def zoom_affine(affine: np.ndarray, scale: Sequence[float], diagonal: bool = True):
+def zoom_affine(affine: np.ndarray, scale: Union[np.ndarray, Sequence[float]], diagonal: bool = True):
     """
     To make column norm of `affine` the same as `scale`.  If diagonal is False,
     returns an affine that combines orthogonal rotation and the new scale.

monai/transforms/intensity/array.py

@@ -443,6 +443,7 @@ def __call__(self, img: NdarrayOrTensor) -> NdarrayOrTensor:
             ValueError: When ``self.minv=None`` or ``self.maxv=None`` and ``self.factor=None``. Incompatible values.
 
         """
+        ret: NdarrayOrTensor
         if self.minv is not None or self.maxv is not None:
             if self.channel_wise:
                 out = [rescale_array(d, self.minv, self.maxv, dtype=self.dtype) for d in img]
@@ -591,7 +592,7 @@ def __call__(self, img: NdarrayOrTensor, randomize: bool = True) -> NdarrayOrTen
             axis=0,
         )
         img_np, *_ = convert_data_type(img, np.ndarray)
-        out = img_np * np.exp(_bias_fields)
+        out: NdarrayOrTensor = img_np * np.exp(_bias_fields)
         out, *_ = convert_to_dst_type(src=out, dst=img, dtype=self.dtype or img.dtype)
         return out
 

monai/transforms/spatial/array.py

@@ -13,7 +13,7 @@
 https://github.com/Project-MONAI/MONAI/wiki/MONAI_Design
 """
 import warnings
-from typing import Any, List, Optional, Sequence, Tuple, Union
+from typing import Any, Callable, List, Optional, Sequence, Tuple, Union
 
 import numpy as np
 import torch
@@ -700,7 +700,7 @@ def __call__(self, img: NdarrayOrTensor) -> NdarrayOrTensor:
         Args:
             img: channel first array, must have shape: (num_channels, H[, W, ..., ]),
         """
-        rot90 = torch.rot90 if isinstance(img, torch.Tensor) else np.rot90
+        rot90: Callable = torch.rot90 if isinstance(img, torch.Tensor) else np.rot90  # type: ignore
         out: NdarrayOrTensor = rot90(img, self.k, map_spatial_axes(img.ndim, self.spatial_axes))
         out, *_ = convert_data_type(out, dtype=img.dtype)
         return out
@@ -1373,14 +1373,15 @@ def __call__(
             raise ValueError("Unknown grid.")
         _device = img.device if isinstance(img, torch.Tensor) else self.device
         img_t: torch.Tensor
+        grid_t: torch.Tensor
         img_t, *_ = convert_data_type(img, torch.Tensor, device=_device, dtype=torch.float32)  # type: ignore
-        grid, *_ = convert_to_dst_type(grid, img_t)
+        grid_t, *_ = convert_to_dst_type(grid, img_t)  # type: ignore
 
         if USE_COMPILED:
             for i, dim in enumerate(img_t.shape[1:]):
-                grid[i] += (dim - 1.0) / 2.0
-            grid = grid[:-1] / grid[-1:]
-            grid = grid.permute(list(range(grid.ndimension()))[1:] + [0])
+                grid_t[i] += (dim - 1.0) / 2.0
+            grid_t = grid_t[:-1] / grid_t[-1:]
+            grid_t = grid_t.permute(list(range(grid_t.ndimension()))[1:] + [0])
             _padding_mode = look_up_option(
                 self.padding_mode if padding_mode is None else padding_mode, GridSamplePadMode
             ).value
@@ -1393,21 +1394,21 @@ def __call__(
             _interp_mode = look_up_option(self.mode if mode is None else mode, GridSampleMode).value
             out = grid_pull(
                 img_t.unsqueeze(0),
-                grid.unsqueeze(0),
+                grid_t.unsqueeze(0),
                 bound=bound,
                 extrapolate=True,
                 interpolation=1 if _interp_mode == "bilinear" else _interp_mode,
             )[0]
         else:
             for i, dim in enumerate(img_t.shape[1:]):
-                grid[i] = 2.0 * grid[i] / (dim - 1.0)
-            grid = grid[:-1] / grid[-1:]
+                grid_t[i] = 2.0 * grid_t[i] / (dim - 1.0)
+            grid_t = grid_t[:-1] / grid_t[-1:]
             index_ordering: List[int] = list(range(img_t.ndimension() - 2, -1, -1))
-            grid = grid[index_ordering]
-            grid = grid.permute(list(range(grid.ndimension()))[1:] + [0])
+            grid_t = grid_t[index_ordering]
+            grid_t = grid_t.permute(list(range(grid_t.ndimension()))[1:] + [0])
             out = torch.nn.functional.grid_sample(
                 img_t.unsqueeze(0),
-                grid.unsqueeze(0),
+                grid_t.unsqueeze(0),
                 mode=self.mode.value if mode is None else GridSampleMode(mode).value,
                 padding_mode=self.padding_mode.value if padding_mode is None else GridSamplePadMode(padding_mode).value,
                 align_corners=True,

monai/transforms/utility/array.py

@@ -789,7 +789,7 @@ def __call__(
         if img.shape[0] > 1:
             data = img[[*select_labels]]
         else:
-            where = np.where if isinstance(img, np.ndarray) else torch.where
+            where: Callable = np.where if isinstance(img, np.ndarray) else torch.where  # type: ignore
             if isinstance(img, np.ndarray) or is_module_ver_at_least(torch, (1, 8, 0)):
                 data = where(in1d(img, select_labels), True, False).reshape(img.shape)
             # pre pytorch 1.8.0, need to use 1/0 instead of True/False

monai/transforms/utils_pytorch_numpy_unification.py

@@ -101,7 +101,7 @@ def percentile(x: NdarrayOrTensor, q, dim: Optional[int] = None) -> Union[Ndarra
         Resulting value (scalar)
     """
     if np.isscalar(q):
-        if not 0 <= q <= 100:
+        if not 0 <= q <= 100:  # type: ignore
             raise ValueError
     elif any(q < 0) or any(q > 100):
         raise ValueError
@@ -134,7 +134,7 @@ def where(condition: NdarrayOrTensor, x=None, y=None) -> NdarrayOrTensor:
         if x is not None:
             result = np.where(condition, x, y)
         else:
-            result = np.where(condition)
+            result = np.where(condition)  # type: ignore
     else:
         if x is not None:
             x = torch.as_tensor(x, device=condition.device)