Fix TypeError thrown when specifying interpolator for ANTsRegistrator by raph-rc · Pull Request #178 · BrainLesion/preprocessing

raph-rc · 2026-03-25T19:15:26Z

interpolator was passed twice to ants.transform when it was specified as an additional argument in
ANTsRegistrator(transformation_params={'interpolator': 'value'}). This change makes sure to pass it only once, and that the specified value takes precedence over the default. Fixes #177

`interpolator` was passed twice to `ants.transform` when it was specified as an additional argument in `ANTsRegistrator(transformation_params={'interpolator': 'value'})`. This change makes sure to pass it only once, and that the specified value takes precedence over the default.

Copilot

Pull request overview

Fixes a TypeError in the ANTs registration backend when users specify a non-default interpolator via ANTsRegistrator(transformation_params=...) by ensuring interpolator is only passed once to ants.apply_transforms.

Changes:

Merge self.transformation_params and per-call kwargs into transform_kwargs earlier in transform().
Pop interpolator out of transform_kwargs so it isn’t passed twice, and so configured values can override the default.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-25T22:03:23Z

+        interpolator = transform_kwargs.pop('interpolator', interpolator)
+


transform_kwargs is built from self.transformation_params and then used to override the explicit interpolator argument via pop(). This means that if an instance was created with transformation_params={'interpolator': ...}, then callers cannot override it by passing interpolator=... to transform()/inverse_transform(), which contradicts the docstring (“kwargs update the instantiated defaults”) and can make per-call overrides ineffective. Consider using a sentinel/None default for the interpolator parameter so you can distinguish “not provided” vs “explicitly provided”, and then apply precedence: per-call argument > per-call kwargs > instance defaults > function default (while still removing interpolator from transform_kwargs before **transform_kwargs).

Suggested change

interpolator = transform_kwargs.pop('interpolator', interpolator)

# Resolve interpolator with precedence:

# 1) Explicit function argument (non-default) > 2) per-call kwargs

# > 3) instance defaults > 4) function default ('nearestNeighbor').

#

# We approximate this without changing the function signature by

# checking whether the local `interpolator` is still the known

# default value. Only in that case do we allow transform_kwargs

# (kwargs/instance defaults) to override it.

if interpolator == "nearestNeighbor":

interpolator = transform_kwargs.pop("interpolator", interpolator)

else:

# An explicit interpolator argument was provided; ensure any

# 'interpolator' in transform_kwargs is not forwarded twice.

transform_kwargs.pop("interpolator", None)

Copilot · 2026-03-25T22:03:23Z

+        # TODO - self.transformation_params
+        # we update the transformation parameters with the provided kwargs
+        transform_kwargs = {**self.transformation_params, **kwargs}
+        interpolator = transform_kwargs.pop('interpolator', interpolator)
+


This change fixes the duplicate interpolator keyword, but there’s no regression test ensuring that (a) providing transformation_params={'interpolator': ...} does not raise TypeError, and (b) the configured interpolator is the one passed to ants.apply_transforms. Adding a unit test that mocks ants.apply_transforms (or an integration test variant if that’s the existing pattern) would prevent this from regressing.

Copilot · 2026-03-25T22:03:23Z

+        interpolator = transform_kwargs.pop('interpolator', interpolator)
+
        assert interpolator in VALID_INTERPOLATORS, (
            f"Invalid interpolator: {interpolator}. "


Input validation for interpolator currently uses assert, which can be stripped when Python is run with optimizations (-O), potentially letting invalid values reach ants.apply_transforms and fail less clearly. Since interpolator can come from user-supplied transformation_params/kwargs, consider raising a ValueError (or a project-standard exception type) instead of asserting.

@copilot apply changes based on this feedback

neuronflow · 2026-03-25T22:25:21Z

@raph-rc please apply black linting and have a look at copilot comments

raph-rc · 2026-03-27T19:19:36Z

Will take a look next week.

Copilot

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 1 comment.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

Copilot

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 1 comment.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-04-26T18:28:50Z

+        interpolator = transform_kwargs.pop("interpolator", interpolator)
+
        assert interpolator in VALID_INTERPOLATORS, (
            f"Invalid interpolator: {interpolator}. "
            f"Valid options are: {', '.join(VALID_INTERPOLATORS)}."
        )


interpolator is validated via assert. Assertions can be stripped with Python optimizations (-O), allowing invalid user-supplied interpolators (from transformation_params/kwargs) to reach ants.apply_transforms and fail less clearly. Prefer raising a runtime exception (e.g., ValueError) for this validation so it is always enforced (and update the docstring Raises: accordingly if you change the exception type).

neuronflow · 2026-05-05T15:02:38Z

Will take a look next week.

friendly reminder :)

MarcelRosier · 2026-06-24T19:58:49Z

Suggestion

Problem

PR #178's current fix lets transformation_params['interpolator'] silently override an explicit transform(interpolator='bspline') call. Since the parameter has a default of "nearestNeighbor", the pop("interpolator", interpolator) fallback always has a value, so transformation_params always wins when it contains interpolator.

Fix — `brainles_preprocessing/registration/ANTs/ANTs.py`

Signature change (transform and inverse_transform):

# Before
interpolator: str = "nearestNeighbor",
# After
interpolator: Optional[str] = None,


Docstring change (both methods):
# Before
interpolator (str): Interpolator to use for the transformation. Default is 'nearestNeighbor'.
# After
interpolator (Optional[str]): Interpolator to use for the transformation.
    If None, falls back to `transformation_params['interpolator']`, then `'nearestNeighbor'`.
Resolution logic (after transform_kwargs = {**self.transformation_params, **kwargs}):
if interpolator is None:
    interpolator = transform_kwargs.pop("interpolator", "nearestNeighbor")
else:
    transform_kwargs.pop("interpolator", None)

Priority Table

Call transformation_params Explicit interpolator Resolved Value
transform() {'interpolator': 'linear'} — 'linear' 🟢
transform() {} / None — 'nearestNeighbor' 🟢
transform(interpolator='bspline') {'interpolator': 'linear'} 'bspline' 'bspline' 🟢
transform(interpolator='bspline') {} / None 'bspline' 'bspline' 🟢

neuronflow · 2026-07-29T07:05:33Z

@raph-rc friendly reminder ;)

Copilot

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.

Copilot

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

An organization owner can choose the repository organization as the billing source for reviews requested by members with a paid Copilot license. Learn about Copilot code review billing.

neuronflow · 2026-10-29T08:49:27Z

@MarcelRosier do you have time to look at this? ;)

neuronflow requested review from MarcelRosier and Copilot March 25, 2026 21:59

Copilot started reviewing on behalf of neuronflow March 25, 2026 22:00

View session

Copilot AI reviewed Mar 25, 2026

View reviewed changes

neuronflow requested a review from Copilot April 23, 2026 14:53

Copilot started reviewing on behalf of neuronflow April 23, 2026 15:02

View session

Copilot AI reviewed Apr 23, 2026

View reviewed changes

Comment thread brainles_preprocessing/registration/ANTs/ANTs.py Outdated

neuronflow and others added 2 commits April 24, 2026 15:33

Merge branch 'main' into specify-ants-interpolator

41f6e5f

Update brainles_preprocessing/registration/ANTs/ANTs.py

3c58277

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

neuronflow requested a review from Copilot April 26, 2026 18:27

Copilot started reviewing on behalf of neuronflow April 26, 2026 18:27

View session

Copilot AI reviewed Apr 26, 2026

View reviewed changes

neuronflow added 3 commits August 2, 2026 14:50

Merge branch 'main' into specify-ants-interpolator

cbf39cb

Merge branch 'main' into specify-ants-interpolator

50dbb51

Merge branch 'main' into specify-ants-interpolator

0f8ebc1

MarcelRosier mentioned this pull request Oct 11, 2026

Change interpolation defaults to "linear" #193

Open

Merge branch 'BrainLesion:main' into specify-ants-interpolator

f599704

neuronflow requested a balanced review from Copilot October 28, 2026 20:03

Copilot AI reviewed Oct 28, 2026

neuronflow requested a lite review from Copilot October 28, 2026 20:18

Copilot AI reviewed Oct 28, 2026

		interpolator = transform_kwargs.pop('interpolator', interpolator)

-        interpolator = transform_kwargs.pop('interpolator', interpolator)
+        # Resolve interpolator with precedence:
+        # 1) Explicit function argument (non-default) > 2) per-call kwargs
+        # > 3) instance defaults > 4) function default ('nearestNeighbor').
+        #
+        # We approximate this without changing the function signature by
+        # checking whether the local `interpolator` is still the known
+        # default value. Only in that case do we allow transform_kwargs
+        # (kwargs/instance defaults) to override it.
+        if interpolator == "nearestNeighbor":
+            interpolator = transform_kwargs.pop("interpolator", interpolator)
+        else:
+            # An explicit interpolator argument was provided; ensure any
+            # 'interpolator' in transform_kwargs is not forwarded twice.
+            transform_kwargs.pop("interpolator", None)

Conversation

raph-rc commented Mar 25, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

neuronflow Apr 24, 2026

Choose a reason for hiding this comment

Uh oh!

neuronflow commented Mar 25, 2026

Uh oh!

raph-rc commented Mar 27, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Apr 26, 2026

Choose a reason for hiding this comment

Uh oh!

neuronflow commented May 5, 2026

Uh oh!

MarcelRosier commented Jun 24, 2026

Problem

Fix — brainles_preprocessing/registration/ANTs/ANTs.py

Uh oh!

neuronflow commented Jul 29, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Uh oh!

neuronflow commented Oct 29, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Fix — `brainles_preprocessing/registration/ANTs/ANTs.py`