Fix misplaced anchors for remark and proof-like divs

[d-morrison]

Closes #159

Summary

In Quarto, theorem environments render their headings with .theorem-title while proof-like environments (proof, remark, solution) render their headings with .proof-title. Previously, div-anchors.js only queried .theorem-title when moving AnchorJS anchor links inline, causing remark and other proof-like divs to leave the anchor link at the bottom of the div rather than beside the heading.

Changes

1. Title selector: Update moveTheoremDivAnchorsInline to select .theorem-title, .proof-title.
2. Whitespace handling: Quarto appends ". " inside .proof-title and immediately follows with the body text without leading whitespace. We trim trailing whitespace from the title's last text node before appending the non-breaking space separator (\u00A0) and anchor link, and ensure a trailing space separates the anchor from subsequent body content.
3. Class and ID aliases:
   • Added "rem" and "sol" aliases to theorem_div_classes and theoremLikeClasses.
   • Added crossref ID prefix matching (rem-, sol-, thm-, etc.) to is_theorem_div in both div-anchors.lua and div-anchors.js.

Verification

• Rendered chapters/theorem-examples.qmd to HTML.
• Verified DOM with headless Chrome and BeautifulSoup: anchor links for rem-convergence and sol-integration are placed directly inline after the heading inside .proof-title, with no misplaced link at the bottom of the div.
• Verified theorems (thm-mean-value, def-derivative, etc.) continue to render correctly with inline anchors.
• Ran devtools::check(), lintr::lint_dir(), and package spellcheck — all passed cleanly.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[github-actions]

Claude finished review — View run

---

Review: PR #160 — Fix misplaced anchors for remark and proof-like divs

Scope: Only _extensions/div-anchors/div-anchors.js and _extensions/div-anchors/div-anchors.lua are touched — tightly scoped to issue #159, consistent with the "keep PRs scoped" guidance in CLAUDE.md. chapters/theorem-examples.qmd (used for manual verification per the PR description) is unchanged, and no _site/_freeze artifacts are included.

Verification performed

I don't have quarto/pandoc/R in this sandbox, so I couldn't render the site directly. Instead I pulled the actual upstream quarto-cli source for theorem.lua and proof.lua (src/resources/filters/customnodes/{theorem,proof}.lua on quarto-dev/quarto-cli@main) to confirm the real DOM shape Quarto emits, then reconstructed that exact shape in a scratch jsdom harness (under /tmp, not committed) and ran the new trimming logic verbatim:

• For .proof-title: Quarto appends the span as the first inline child of the body paragraph, ending with a bare ". " text node, immediately followed (no leading space) by the paragraph's original text as the title span's nextSibling. Running the PR's trim/insert logic against this exact structure correctly produces Remark 1 (...).<nbsp><anchor> The limit in … — the anchor lands inline and a space correctly separates it from the following prose. This confirms the fix in div-anchors.js:126-139 works as intended.
• For .theorem-title: Quarto's renderer inserts the separator as a sibling pandoc.Space() node rather than trailing whitespace inside the span, so the new trim branch correctly never fires for theorem titles (no trailing whitespace found), preserving prior passing behavior — i.e., this change is backward-compatible with the existing theorem-title path.
• theoremLikeClasses/theorem_div_classes stay in sync between the .js and .lua copies (both gained "rem"/"sol"), and theoremLikePrefixes/theorem_id_prefixes list the same 10 prefixes in the same order in both files — good parity between the JS and Lua implementations, which matters since they must agree on what counts as theorem-like.
• Lua id_is_theorem_like correctly guards nil identifiers and uses 1-indexed sub prefix comparison correctly.

Observations (non-blocking)

• ID-prefix heuristic's necessity is not fully demonstrated. Tracing quarto-cli's theorem.lua, the built-in renderer unconditionally inserts class="theorem" plus the full env name (lemma, corollary, definition, …) on any div it recognizes as a Theorem node — and theoremLikeClasses/theorem_div_classes already contain those full env names, so class-based matching alone already covers thm-/lem-/etc. in the un-wrapped rendering path. Because this repo wraps every theorem/proof type in a callout via sun123zxy/callouty-theorem (_quarto-website.yml:67-130), the outer div the user sees may instead carry callout classes, which is plausibly why the author added the ID-prefix fallback. I can't confirm from the diff alone whether the prefix check is load-bearing or just defensive redundancy, since I can't render the site to check the final callout DOM — the PR description's claim of having verified this with headless Chrome/BeautifulSoup is the best evidence available and I have no basis to contradict it. Worth a quick sanity check by the author (temporarily removing the prefix branch and re-rendering) only if there's any doubt — not a merge blocker.
• _extensions/div-anchors/_extension.yml keeps version: 1.0.0 unchanged. The repo has no established convention requiring a version bump for extension edits (confirmed via git log on that file and no NEWS/CHANGELOG), so this isn't a violation of anything documented — just flagging for awareness in case the maintainer wants one.
• The PR description says prefix matching was "added to is_theorem_div... in both div-anchors.lua and div-anchors.js" — the JS function is actually named isTheoremLikeDiv (camelCase, matching JS convention), not is_theorem_div. Purely a wording nit in the description, not a code issue.
• New ID-prefix matching is scoped to 10 specific 3-4 letter dash-suffixed prefixes (thm-, lem-, …, rem-, sol-), mirroring Quarto's own crossref refType convention, so collision risk with unrelated div IDs is the same risk Quarto itself already accepts — not a new concern introduced by this PR.

CLAUDE.md / lab-manual compliance

No R, .qmd, bibliography, or build-artifact changes — the R-specific (lintr, tidyverse, function-length) and Quarto-prose sections of the lab manual don't apply here. No new dependencies introduced. Comment added at div-anchors.js:126-128 explains non-obvious Quarto whitespace behavior (the "why"), consistent with good comment practice rather than restating the "what."

I found no correctness bugs, no CLAUDE.md violations, and no hallucinated APIs/symbols — callouty-theorem and custom-callout are real vendored extensions present in _extensions/, and the Quarto rendering behavior described in the PR matches the actual upstream quarto-cli source.

Verdict

Ready for merge — no blocking findings; one non-blocking observation about unverified necessity of the ID-prefix fallback (not a correctness issue, just an open question for the author's own confidence).

Structured Review Data (JSON)

{
  "schema_version": "1.1",
  "reviewer": "claude",
  "commit_sha": "71bb81772d081b9d2bf725d44b3d6ce9a681201d",
  "verdict": "CLEAN",
  "findings": [],
  "detailed_assessment": "Reconstructed quarto-cli's real proof-title DOM shape in a jsdom harness and confirmed the trim-and-reinsert logic in div-anchors.js correctly places the anchor inline without breaking the pre-existing theorem-title path.",
  "holistic_assessment": "Scope stays limited to the two div-anchors extension files for issue 159, parity between the JS and Lua class/prefix lists is maintained, and no regression risk was found for downstream template consumers."
}

Reviewed commit: 71bb817

[github-actions]

💰 Cost: $1.2615 (review) — run

[d-morrison]

Done with my local session — unclaiming.

Posted by Antigravity (AI agent) --- not written by a human.

[github-actions]

PR Preview Action v1.8.1
Preview removed because the pull request was closed.
2026-10-06 17:34 PDT