[dzenanz]

This is a backport of #6396 to 5.4 release branch.

PR Checklist

• No API changes were made (or the changes have been approved)
• No major design changes were made (or the changes have been approved)
• Added test (or behavior not changed)
• Added Python wrapping to new files (if any) as described in ITK Software Guide Section 9.5

[greptile-apps]

Greptile Summary

This change adds dimension-equals-spline-order BSpline Python wrappers and instantiation tests. Two wrapping failures remain: the new 2D test is registered even when 2D wrapping is disabled, and the new 4D order-4 transform exposes APIs requiring 625-element FixedArrays that are not generated. The introduced variable names also do not meet the repository naming requirement.

Confidence Score: 3/5

Not safe to merge until the two Python wrapping failures are fixed and the repository naming requirement is satisfied.

Two independently reproduced failures leave valid reduced wrapping configurations with a failing test and leave the new 4D order-4 Python API incomplete.

Files Needing Attention: Modules/Core/Transform/wrapping/test/CMakeLists.txt, Modules/Core/Transform/wrapping/itkBSplineTransform.wrap, and Wrapping/WrapITKTypes.cmake

T-Rex Logs

What T-Rex did

• T-Rex produced proofs for multiple posted P1 findings and linked them to the corresponding review comments.
• T-Rex produced a proof for another P1 finding, but no artifacts were provided.
• T-Rex performed a general contract validation for bspline-2d2d-condition, capturing the before-state results and the after-state result using the authored script.
• T-Rex produced a P1 finding proof for a different case, with its supporting artifact available for review.
• T-Rex performed a separate bspline4d_fixedarray_signature_validation, recording the before/after captures and the executable-path details.

_{Ran code and verified through T-Rex}

Comments Outside Diff (2)

1. General comment

   Guard the 2D-2D Python expression test by configured wrapping dimensions

   • Bug
     • With ITK_WRAP_IMAGE_DIMS=3, candidate 0 still registers PythonInstantiateBSplineTransform2D2D, but its required BSplineTransform<double,2,2> specialization is absent. The focused candidate validation exited 1 with this mismatch.
   • Cause
     • The added 2D-2D test has no list(FIND ITK_WRAP_IMAGE_DIMS 2 ...) guard, unlike the newly added 4D-4D test. The wrapper file emits the 2D-2D specialization only within foreach(d ${ITK_WRAP_IMAGE_DIMS}) when d equals 2.
   • Fix
     • Wrap PythonInstantiateBSplineTransform2D2D in a check that ITK_WRAP_IMAGE_DIMS contains 2, matching the conditional registration used for the 4D-4D test.

   _{Ran code and verified through T-Rex}

2. General comment

   4D order-4 BSplineTransform wrapper has ungenerated FixedArray dependencies

   • Bug
     • The added BSplineTransform<double,4,4> wrapper inherits public TransformPoint and Jacobian-related signatures with WeightsType and ParameterIndexArrayType. For spline order 4 in dimension 4, NumberOfWeights is (4 + 1)^4 = 625, so those types are FixedArray<double,625> and FixedArray<unsigned long,625>. The changed FixedArray configuration retains the order-3 4^d computation and produces 16, 64, and 256 for dimensions 2, 3, and 4; neither 625 type is generated. Generated Python signatures therefore become unavailable or incomplete.
   • Cause
     • Wrapping/WrapITKTypes.cmake calculates BSpline FixedArray sizes with a hard-coded multiplication by 4, which matches spline order 3 but not the newly exposed order-4 template.
   • Fix
     • Generate the FixedArray specializations required by every exposed BSpline order, specifically add double and unsigned-long size 625 when wrapping BSplineTransform<double,4,4> (preferably derive the size from VSplineOrder + 1 rather than hard-coding 4).

   _{Ran code and verified through T-Rex}

_{Reviews (1): Last reviewed commit: "ENH: Wrap BSplineTransform for 2D-2D and..." | Re-trigger Greptile}

[dzenanz]

/azp run ITK.macOS.Python