openbraininstitute / neurodamus / 23431330571

91%

push

github

fix stale section counts and small rework of `modification_manager.py` (#499)

## Context
Fix: #496 #497 #498
Given what I changed it made sense to solve all these issues in 1 pr. 

## Scope

Introduce `_SECTION_TYPES` as the single source of truth for section
name ↔ section list mappings, replacing the duplicated definitions
across `metype.py` and `modification_manager.py`.

### metype.py

- Replaced `_section_layout` (list of tuples with lambdas) with
`_SECTION_TYPES` (list of `(section_name, section_list)` tuples)
- Refactored `get_section_counts` to iterate `_SECTION_TYPES` directly,
with all section lists treated as optional via `getattr` and the axon
special-cased to use `nSecAxonalOrig`
- Refactored `get_section_id` and `get_sec` to use `_SECTION_TYPES`
instead of `_section_layout`
- Added documentation explaining the stale hoc accessor problem and the
axon numbering gap

### modification_manager.py

- Removed `BaseSectionModification` class; inlined section type
validation directly into `SectionListModification` and
`SectionModification`
- Replaced local `SECTION_TYPES` definitions with lookups derived from
`BaseCell._SECTION_TYPES`
- Extended allowed section types from 4 to all 7 (added `ais`, `node`,
`myelin`)
- Renamed all modification classes for clarity:
  - `TTX` → `TTXModification`
  - `ConfigureAllSections` → `ConfigureAllSectionsModification`
  - `SectionList` → `SectionListModification`
  - `Section` → `SectionModification`
  - `CompartmentSet` → `CompartmentSetModification`

### coreneuron_configuration.py

- `register_mapping` is using `BaseCell.SECTION_TYPES` now. More DRY

### tests/unit/test_metype.py

- Renamed `test_get_cell_section_counts` →
`test_section_counts_vs_stale_hoc_accessor` with docstring explaining
the stale accessor scenario
- Replaced magic index `[3]` with a lookup via `_SECTION_TYPES` so the
test won't break if layout order changes
- Fixed mock attributes to match what the section list accessors
actually read ... (continued)

33 of 36 new or added lines in 3 files covered. (91.67%)

7737 of 8494 relevant lines covered (91.09%)

2.3 hits per line