Add try_get_chapters to scripture parsing

[claude]

Ports machine PR 497 (sillsdev/machine#497): adds try_get_chapters, a non-raising counterpart to get_chapters. It returns (True, chapters) on success and (False, None) when get_chapters raises ValueError, the Python form of the C# out parameter. Exported from machine.scripture. The C# namespace move and SIL.Scripture qualification fixes are C#-only and skipped. Added test_try_get_chapters. ./local_check.sh --agent-strict passed (864 passed, 3 skipped). Closes issue 380 - Closes #380. 🤖 Generated with Claude Code

---

This change is 

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.07%. Comparing base (cc99576) to head (59401fb).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #407   +/-   ##
=======================================
  Coverage   92.07%   92.07%           
=======================================
  Files         394      394           
  Lines       24896    24911   +15     
=======================================
+ Hits        22922    22937   +15     
  Misses       1974     1974           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[claude]

Minor: F1. try_get_chapters raises on an empty segment instead of returning (False, None). For "MAT;;MRK", ";" or " ; ", parse_selection("") evaluates selection[-1] (line 62), which raises IndexError, and that error is not caught here. C# PR 497 added an Empty book range. guard to ParseSection and test cases that expect TryGetChapters to return false for these three inputs. Port the guard as a ValueError in parse_selection.

[claude]

Minor: F2. The failure cases leave out the empty-segment inputs "MAT;;MRK", " ; " and ";". C# PR 497 added those three to the shared GetCases, and they are the inputs that break F1. Add them here, and to test_get_chapters with raises(ValueError).

[claude]

Minor: F3. The (bool, Optional[Dict]) return shape goes into the published wheel through __all__, and it differs from the existing try-API VerseRef.try_from_string (verse_ref.py:98), which returns Optional[...]. The bool tells pyright nothing about the dict, so ok, ch = ...; if ok: ch[40] still gets flagged as possibly None. Returning Optional[Dict[int, List[int]]] would match the existing convention, and {} stays distinct from failure. Once this is released, changing the shape is a breaking change.

[claude]

Verdict: approve with fixes.

The fix that matters most is F1 at machine/scripture/parse.py:189: try_get_chapters(";") raises IndexError instead of returning (False, None), because the port left out the empty-section guard that C# PR 497 added.

Counts: Critical 0, Important 3, Low 0.

What I ran: poetry run pytest tests/scripture/test_parse.py passed (3 tests). I read gh pr diff 497 -R sillsdev/machine and confirmed the C# changes: the section.Length == 0 guard in ParseSection, and the "MAT;;MRK", " ; " and ";" cases expecting failure.

What I could not verify: executing the F1 inputs was blocked in this environment, so the IndexError is read from parse.py:59-62 ("".strip()[-1]) rather than observed. I did not run ./local_check.sh.

Public API: try_get_chapters is a new export in machine.scripture.__all__, part of the published-wheel surface (F3). Parity with sillsdev/machine: the missing empty-section guard (F1) is a divergence, and the PR body does not list it as deliberate.

Dropped as pre-existing or out of scope: get_chapters returns unsorted merged chapter lists (parse.py:175, untouched by this diff), and versification=None (the Python signature does not accept None).

• F1: new, Important
• F2: new, Important
• F3: new, Important

Reviewed at 59401fb