Fix #150: remove duplicate osipi_check_required_bvalues definition - #171

Describe the changes you have made in this PR

OsipiBase defined osipi_check_required_bvalues twice. The second definition was a bare pass stub that shadowed the first, so the method returned None instead of a boolean.

This PR removes the shadowing stub and adds a regression test so that a duplicate method definition cannot silently reappear.

@@ -535,11 +535,6 @@ class OsipiBase:
             #return True
         return True
 
-    
-    def osipi_check_required_bvalues(self):
-        """Minimum number of b-values required"""
-        pass
-
     def osipi_author(self):
         """Author identification"""
         return ''

The surviving definition returns True and is consistent with its three siblings osipi_check_required_thresholds, osipi_check_required_bounds and osipi_check_required_initial_guess.

Verification

The pre-fix class was loaded directly from git and compared against the fixed one, so the result does not depend on reading the diff correctly:

The public API is unchanged, and every remaining method is byte-identical (__code__.co_code) to before, so this change cannot affect any other behaviour.

Regression test

tests/IVIMmodels/unit_tests/test_ivim_fit.py::test_no_duplicate_method_definitions_in_osipibase

It parses OsipiBase with ast and asserts that no method name is defined more than once — guarding against any future duplicate, not only this one — then asserts that all four osipi_check_required_* methods return a boolean rather than None.

The test was confirmed to fail against the pre-fix code and pass after, so it is a real guard rather than one that passes vacuously. np.bool_ is accepted alongside bool so the test does not obstruct a future real implementation of these checks.

Test suite

Run from the repository root (Windows, Python 3.11.3, pytest 9.0.2):

python -m pytest -k "not test_volume and not test_parallel" -q

672 passed, 89 skipped, 432 deselected, 11 xfailed, 4 xpassed in 2852.13s (0:47:32)

No failures. The skips are the MATLAB-dependent algorithms (no matlab.engine locally) and the deep-learning skips; the xfails/xpasses are the pre-existing TCML_TechnionIIT_lsqBOBYQA entries from algorithms.json.

test_volume and test_parallel (54 tests) were not run locally, as they require the XCAT phantom downloaded from Zenodo which is not present in my environment. CI covers them, and per the byte-identity check above this change cannot affect them.

Link this PR to an issue [optional]

Fixes #150

Checklist

• Self-review of changed code
• Added automated tests where applicable
• Update Docs & Guides — not applicable, no user-facing behaviour change