Skip to content

🐍 Geometric Computer Vision Library for Spatial AI

docs: enable ruff D101 and D102 and document the remaining methods (#6306)

Maidul Hasan Masum5fb33451afc0
diff --git a/.github/scripts/check_import_surface.py b/.github/scripts/check_import_surface.py
index 8866d4245..a90889719 100755
--- a/.github/scripts/check_import_surface.py
+++ b/.github/scripts/check_import_surface.py
@@ -623,6 +623,8 @@ def export_removals(base_ref: str, paths: list[str]) -> dict[str, set[str]]:
@dataclass
class FileReport:
+ """What one file removed from its ``__all__``, and whether that should fail the check."""
+
path: str
removed_from_all: set[str]
removed_undocumented: set[str]
diff --git a/kornia/constants.py b/kornia/constants.py
index 656655154..98828662b 100644
--- a/kornia/constants.py
+++ b/kornia/constants.py
@@ -97,6 +97,7 @@ class Resample(Enum, metaclass=_KORNIA_EnumMeta):
@classmethod
def get(cls, value: TKEnum["Resample"]) -> "Resample":
+ """Return the Resample member matching a name, a value or a member."""
return _get(cls, value)
@@ -110,6 +111,7 @@ class BorderType(Enum, metaclass=_KORNIA_EnumMeta):
@classmethod
def get(cls, value: TKEnum["BorderType"]) -> "BorderType":
+ """Return the BorderType member matching a name, a value or a member."""
return _get(cls, value)
@@ -123,6 +125,7 @@ class SamplePadding(Enum, metaclass=_KORNIA_EnumMeta):
@classmethod
def get(cls, value: TKEnum["SamplePadding"]) -> "SamplePadding":
+ """Return the SamplePadding member matching a name, a value or a member."""
return _get(cls, value)
@@ -137,6 +140,7 @@ class DType(Enum, metaclass=_KORNIA_EnumMeta):
@classmethod
def get(cls, value: Union[str, int, torch.dtype, torch.Tensor, "DType"]) -> "DType":
+ """Return the DType member matching a name, a value, a torch dtype, a tensor or a member."""
if isinstance(value, torch.dtype):
return _get_by_name(cls, str(value)[6:], value)
@@ -159,6 +163,7 @@ class DType(Enum, metaclass=_KORNIA_EnumMeta):
@classmethod
def to_torch(cls, value: TKEnum["DType"]) -> torch.dtype:
+ """Return the torch dtype matching a DType name, value or member."""
data = cls.get(value=value)
if data == DType.INT64:
@@ -181,6 +186,8 @@ class DType(Enum, metaclass=_KORNIA_EnumMeta):
# TODO: (low-priority) add INPUT3D, MASK3D, BBOX3D, LAFs etc.
class DataKey(Enum, metaclass=_KORNIA_EnumMeta):
+ """Kind of data an augmentation container transforms, one per entry of ``data_keys``."""
+
IMAGE = 0
INPUT = 0
MASK = 1
@@ -193,4 +200,5 @@ class DataKey(Enum, metaclass=_KORNIA_EnumMeta):
@classmethod
def get(cls, value: TKEnum["DataKey"]) -> "DataKey":
+ """Return the DataKey member matching a name, a value or a member."""
return _get(cls, value)
diff --git a/kornia/geometry/camera/pinhole.py b/kornia/geometry/camera/pinhole.py
index 37169696c..03fac4b24 100644
--- a/kornia/geometry/camera/pinhole.py
+++ b/kornia/geometry/camera/pinhole.py
@@ -587,9 +587,11 @@ class PinholeCamerasList(PinholeCamera):
return self
def clone(self) -> "PinholeCamerasList":
+ """Return a copy of the list, with each camera cloned."""
return PinholeCamerasList([self.get_pinhole(index).clone() for index in range(self.num_cameras)])
def scale(self, scale_factor: torch.Tensor) -> "PinholeCamerasList":
+ """Scale the cameras by a vector factor along the batch dimension, returning a new list."""
return self.clone().scale_(scale_factor)
def scale_(self, scale_factor: Union[float, torch.Tensor]) -> "PinholeCamerasList":
diff --git a/kornia/losses/hausdorff.py b/kornia/losses/hausdorff.py
index b7096e7eb..0e00360a0 100644
--- a/kornia/losses/hausdorff.py
+++ b/kornia/losses/hausdorff.py
@@ -202,7 +202,8 @@ class HausdorffERLoss(_HausdorffERLossBase):
@staticmethod
def max_pool(x: torch.Tensor) -> torch.Tensor:
- # Global max over the spatial dims; equals ``nn.AdaptiveMaxPool2d(1)`` but has an ONNX lowering.
+ """Take the global maximum over the spatial dimensions, keeping them with size one."""
+ # Equals ``nn.AdaptiveMaxPool2d(1)`` but has an ONNX lowering.
return x.amax(dim=(-2, -1), keepdim=True)
def get_kernel(self) -> torch.Tensor:
@@ -276,7 +277,8 @@ class HausdorffERLoss3D(_HausdorffERLossBase):
@staticmethod
def max_pool(x: torch.Tensor) -> torch.Tensor:
- # Global max over the volume dims; equals ``nn.AdaptiveMaxPool3d(1)`` but has an ONNX lowering.
+ """Take the global maximum over the volume dimensions, keeping them with size one."""
+ # Equals ``nn.AdaptiveMaxPool3d(1)`` but has an ONNX lowering.
return x.amax(dim=(-3, -2, -1), keepdim=True)
def get_kernel(self) -> torch.Tensor:
diff --git a/pyproject.toml b/pyproject.toml
index 76152f123..f78afd078 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -299,8 +299,6 @@ ignore = [
"UP045", # Use `X | None` for type annotations
"RUF005", # Consider `(*points_in_cam_canonical.shape[:-1], 1)` instead of concatenation. Note: breaks JIT.
"D100", # Allow Undocumented public module
- "D101", # TODO: Undocumented public class
- "D102", # TODO: Undocumented public method
"D104", # TODO: Undocumented public package
"D105", # Allow Undocumented magic method
"D107", # TODO: Undocumented public init