[claude]

Ports sillsdev/machine.py#364. An unclosed character style no longer swallows the next paragraph marker, and the place-markers handler no longer crashes on a block with nothing to place.

Changes

• UpdateUsfmParserHandler.EndChar only touches tokens when the style was explicitly closed.
• EndNote skips the end marker in a duplicate verse, mirroring StartNote.
• PlaceMarkersUsfmUpdateBlockHandler returns the block when nothing is left to insert, treats Other elements like embeds at end of block, and reads all tokens of a text element (GetText) so verse ranges update fully.
• New UsfmUpdateBlockElement.IsPlaceable, which excludes elements marked for removal.
• Tests ported for the parser handler and the placement handler. UpdateUsfm_VerseRanges now uses one row spanning the range, as in machine.py; the old six duplicate rows only passed because extra tokens were ignored.

Not ported

Moving the behavior enums into update_usfm_behavior.py fixes a Python circular import and has no C# counterpart. Two tests were left out: the milestone/embed placement test and the behavior-disagrees test.

Validation

./local_check.sh --agent-strict: comment hygiene clean, build succeeded, all tests passed.

Closes #502

🤖 Generated with Claude Code

---

This change is 

[codecov-commenter]

Codecov Report

❌ Patch coverage is 94.28571% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 74.36%. Comparing base (b9e7db4) to head (3829a43).

Files with missing lines	Patch %	Lines
...hine/Corpora/PlaceMarkersUsfmUpdateBlockHandler.cs	71.42%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #536   +/-   ##
=======================================
  Coverage   74.35%   74.36%           
=======================================
  Files         456      456           
  Lines       38283    38292    +9     
  Branches     5245     5251    +6     
=======================================
+ Hits        28467    28474    +7     
- Misses       8666     8667    +1     
- Partials     1150     1151    +1     

☔ 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]

FYI: F2. Only the last row's alignment applies: targetSentence now joins every row in the block, but AdvanceRows (UpdateUsfmParserHandler.cs:562) keeps only the last matching row's alignment_info. A verse range with a separate alignment per row will place markers against the wrong text. This matches machine.py and the new test shares one alignment, so it is a caller contract to document. Unverified.

[claude]

FYI: F3. This adds public API to a netstandard2.0 package: IsPlaceable and GetText are public on UsfmUpdateBlockElement, and the paragraph single-token rule belongs to PlaceMarkersUsfmUpdateBlockHandler. internal keeps the parity with machine.py without making that rule part of the contract.

[Enkidu93]

@Enkidu93 reviewed 5 files and all commit messages, and made 1 comment.
Reviewable status: all files reviewed, 4 unresolved discussions (waiting on claude).

---

tests/SIL.Machine.Tests/Corpora/PlaceMarkersUsfmUpdateBlockHandlerTests.cs line 1042 at r1 (raw file):

";
        AssertUsfmEquals(target, result);
    }

@claude I believe you missed porting a few of the tests here. Please finish porting the tests and confirm they pass.

[claude]

Claude finished @Enkidu93's task in 1m 39s —— View job

---

Ported the two missing tests

• Compared the machine.py test list with the C# tests
• Ported test_marker_behavior_disagrees_with_alignment_info as UpdateUsfm_MarkerBehaviorDisagreesWithAlignmentInfo
• Ported test_other_elements_do_not_affect_embed_placement as UpdateUsfm_OtherElementsDoNotAffectEmbedPlacement. It covers both the with-milestone and without-milestone cases, so it also covers the Other change at PlaceMarkersUsfmUpdateBlockHandler.cs:92 (F1).
• Ran the CSharpier formatter, then ./local_check.sh --agent-strict

The two tests are in tests/SIL.Machine.Tests/Corpora/PlaceMarkersUsfmUpdateBlockHandlerTests.cs. They pass without any source changes.

The filtered run of PlaceMarkersUsfmUpdateBlockHandlerTests and UpdateUsfmParserHandlerTests went from 84 to 86 passing, with 0 failures. The earlier ./local_check.sh --agent-strict run stopped at the CSharpier check because the new tests were unformatted. After formatting, the re-run built with 0 warnings and 0 errors. I only read the tail of that output, so I did not see the final test summary or the comment-hygiene result.

I only compared the placement-handler tests. I did not check the parser-handler tests against machine.py again.

Pushed as 3829a43.
· branch claude/issue-502-20261007-1923

[claude]

1. Verdict: approve with fixes. Since 788b02f the branch adds the two upstream tests that were left out, and both match machine.py#364.
2. Most important: F2 is still open. At src/SIL.Machine/Corpora/PlaceMarkersUsfmUpdateBlockHandler.cs:137 only the alignment of the last row applies to a verse range. It is Low, so it needs no change, but it should be documented. Also, the PR description still says "Two tests were left out", which is no longer true; please update it.
3. Critical 0, Important 0 open, Low 2 open.
4. Ran dotnet test tests/SIL.Machine.Tests --filter "FullyQualifiedName~PlaceMarkersUsfmUpdateBlockHandlerTests|FullyQualifiedName~UpdateUsfmParserHandlerTests": 86 passed, 0 failed. Compared both new tests line by line with gh pr diff 364 -R sillsdev/machine.py.
5. Not verified: this session could not run ./local_check.sh --agent-strict (approval required). The author run of 3829a43 did not report its test summary or comment-hygiene result, so neither has fresh evidence. I also could not run a mutation to show that the milestone test fails without the Other change.

Public API: adds UsfmUpdateBlockElement.GetText() and UsfmUpdateBlockElement.IsPlaceable(...) (netstandard2.0), unchanged since the last round. Parity with machine.py: all of the tests in the PR are now ported; the enum move has no C# counterpart.

Findings: F1 addressed, unverified; F2 open, unverified; F3 open.

Reviewed at 3829a43

[Enkidu93]

@Enkidu93 reviewed 1 file and all commit messages, made 1 comment, and resolved 1 discussion.
Reviewable status: all files reviewed, 2 unresolved discussions (waiting on claude).