Filter predict outputs to save only test_crop_manifest labels

[Copilot]

• Understand the current predict flow and identify the issue
• Add get_test_crop_labels() function to get labels for each crop
• Update the predict() function to filter classes per crop when crops="test"
• Modify _predict() to structure outputs with all model classes but filter before saving
• Add comprehensive tests to validate the changes
• Document the changes in docstrings
• Address all code review feedback
• Optimize class index lookup (O(1) instead of O(n))
• Clarify documentation about label filtering behavior
• Run security checks (CodeQL) - no vulnerabilities found
• Use cached TEST_CROPS in get_test_crop_labels for performance
• Add validation to skip crops with no matching labels
• Add safety check in _predict for empty classes_to_save
• Add integration tests for filtering behavior with dict and tensor outputs
• Add end-to-end integration test that calls predict() and verifies filtered output
• Fix test issues: use pytest.skip() instead of return, improve mock configuration

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[Copilot]

Pull request overview

This PR implements filtering of prediction outputs to save only the class labels specified in the test_crop_manifest for each crop when running predict(crops="test"). This optimization reduces storage requirements and processing time by avoiding saving predictions for classes that won't be scored.

Changes:

• Added get_test_crop_labels() utility function to retrieve scored labels per crop from the test manifest
• Modified predict() to filter classes to the intersection of model classes and crop-specific labels when crops="test"
• Modified _predict() to structure outputs using all model classes but filter them before disk write

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 4 comments.

File	Description
src/cellmap_segmentation_challenge/utils/crops.py	Added get_test_crop_labels() function to retrieve class labels for a specific test crop from the manifest
src/cellmap_segmentation_challenge/utils/init.py	Exported the new get_test_crop_labels function
src/cellmap_segmentation_challenge/predict.py	Modified prediction pipeline to filter outputs based on test_crop_manifest labels; added support for model_classes parameter to maintain model compatibility while filtering saved classes
tests/test_crops.py	Added comprehensive unit tests for get_test_crop_labels() covering various scenarios including edge cases

---

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

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated 7 comments.

---

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

[rhoadesScholar]

@copilot apply changes based on the comments in this thread

[Copilot]

@copilot apply changes based on the comments in this thread

Fixed all issues in 2ee6319:

1. test_crops.py: Replaced all return statements with pytest.skip() in tests that check prerequisites (comments 2841862962, 2841862980, 2841862999, 2841863014, 2841863095). Tests now explicitly report as skipped when manifest is unavailable.

2. test_predict_filtering.py:

   • Fixed mock configuration (comment 2841863045): Changed from mock_model.__call__ = mock_forward to mock_model.side_effect = mock_forward for proper mock behavior
   • Removed broad exception handling (comment 2841863074): Now mocks _predict function directly to capture dataset_writer_kwargs, eliminating need for try/except and making test failures more explicit

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated no new comments.

---

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