[d-morrison]

Splitting a file puts every line in new files, so check-term-links reflagged each first use the original already carried. It failed pds#244, the first of the one-div-per-subfile splits for gha#1176. The sibling diff-scoped checks (check-literal-math-letters, check-local-macros, check-fform-paren) already exempt moved lines; this adds the same exemption: an added line whose exact text, ignoring surrounding whitespace, the same diff also deletes is left out.

Hunk counts now decide which diff lines are body lines, so a body line that starts with ++ or -- is never read as a file header.

Tests: a real move between files is not flagged, an edited line that still carries the first use is, and the parser drops moved lines and ignores header-looking body lines. Replacing the exemption with if True turns two of them red (test_cli_moved_line_is_not_new, test_parse_added_lines_drops_lines_the_diff_also_deletes). The 40 tests in the folder pass.

The first check-term-links test here that only git mvs a file would have passed for the wrong reason: a pure rename has no hunks, so it never reaches the parser. I replaced it with a copy plus an emptied original.

The notes repos pin this check at a tag, so the fix reaches them when the tag slides.

---

Generated by Claude Code

[github-actions]

Claude finished review — View run

---

I found no defects in this PR. I read the parser change in check-term-links/check-term-links.py and the new tests, and ran the suite: 40 passed.

• Moved-line exemption: an added line is dropped only when its exact text, stripped of surrounding whitespace, was also deleted in the same diff. A genuinely edited line still has different text, so it is still flagged. The new test test_cli_edited_line_with_the_same_first_use_is_still_flagged covers that.
• Hunk-count parsing: the counts bound how many body lines are consumed. A body line starting with ++ or -- is therefore not read as a file header. This matches the model check-new-line-breaks already uses.
• +++ /dev/null (deleted files): the stale current file is harmless, because new_left is 0 for those hunks, so nothing is attributed to the previous file.
• Blank lines: they are kept as added (not text), so they cannot all be treated as moved.
• Documentation: the module docstring and a changelog.d fragment are included.

One residual: a moved line that also appears elsewhere in the base is exempted by text alone. The PR accepts that trade-off and matches the check-new-line-breaks precedent in CLAUDE.md.

I did not review the macro rules: the diff adds no math. I found no hallucinated symbols or URLs; the one external reference is gha#1176, which I did not check.

Verdict

Ready for merge

Structured Review Data (JSON)

{
  "schema_version": "1.1",
  "reviewer": "claude",
  "commit_sha": "778571d81c51ab75caea0d531f0d21a24a20f2d1",
  "verdict": "CLEAN",
  "findings": [],
  "detailed_assessment": "Traced parse_added_lines hunk-count handling, deleted-file diffs, blank lines and edited-line cases; the suite of 40 tests passes and no defect was found.",
  "holistic_assessment": "The change matches the documented moved-line diff model used by sibling checks, keeps skip behavior intact, and includes docs, a changelog fragment and regression tests."
}

Reviewed commit: 778571d

[github-actions]

💰 Cost: $0.6876 (review) — run