Cluster page: reading list, voting, performance, and build pipeline

[LukasWallrich]

Summary

Overhaul of the cluster/curated resources pages — new features, performance improvements, and a simplified build pipeline. Featured/recommended resources are disabled by default (enable_featured_resources: false) until real community recommendations are collected; all other improvements are active.

New features

• Expandable reading list panel — saved items now show APA reference (with copy button), abstract preview with show more/less, DOI link, and Open Access indicator
• PDF export for reading list (loads jsPDF from CDN on first use)
• Vote submission — thumbs-up votes are now submitted to a Google Form via hidden iframe, and existing vote counts from the response sheet are shown as the starting count (vote_base)
• pub_cards.json added to CI pipeline — enriched publication card data is now persisted and deployed alongside featured_resources.json

Performance

• Cached card data at init — search text, focus/type/specificity attributes are extracted once and stored in a JS array; filtering loops over plain objects instead of querying the DOM on every keystroke
• Two-pass filtering — pass 1 computes visibility from cached data (zero DOM reads), pass 2 batches all DOM writes, eliminating layout thrashing
• maxHeight: 'none' replaces scrollHeight + 'px' for open accordion sections during filtering, avoiding forced reflows

Build pipeline (build_featured_data.py)

• Replaced gws CLI dependency with direct published CSV URLs — no credentials needed
• Parallelized 5 CSV fetches with ThreadPoolExecutor (~3× faster)
• Added vote count fetching from Google Sheets response CSV
• Added display exclusion column support (column 16)

Featured resources toggle

• enable_featured_resources set to false in params.toml
• Both the "Recommended" accordion section and the special recommendation card layout (with testimonial carousel) are gated on this flag
• Flip to true when real recommendations are ready — no code changes needed

Test plan

• Open /clusters/ — verify no "Recommended" sections or recommendation card layouts appear
• Type in the search box — confirm filtering feels responsive with many resources
• Click filter tags (Focus, Resource Type) — cards filter correctly
• Save resources to reading list — expand items, verify APA copy, abstract toggle, DOI/OA links
• Click "Download PDF" in reading list — verify PDF generates with correct content
• Click thumbs-up vote on a card — verify count increments and persists across reload
• Run python3 scripts/build_featured_data.py — verify it completes without errors and outputs correct JSON
• Set enable_featured_resources = true in params.toml — verify recommended section and card layout reappear

🤖 Generated with Claude Code

[github-actions]

👍 All image files/references (if any) are in webp format, in line with our policy.

[github-actions]

📝 Spell Check Results

Found 4 potential spelling issue(s) when checking 16 changed file(s):

📄 static/js/clusters-page.js

Line	Issue
80	tabEl ==> table
81	tabEl ==> table
83	tabEl ==> table
85	tabEl ==> table

---

ℹ️ How to address these issues:

1. Fix the typo: If it's a genuine typo, please correct it.
2. Add to whitelist: If it's a valid word (e.g., a name, technical term), add it to .codespell-ignore.txt
3. False positive: If this is a false positive, please report it in the PR comments.

_{🤖 This check was performed by codespell}

[LukasWallrich]

✅ Staging Deployment Status

This PR has been successfully deployed to staging as part of an aggregated deployment.

Deployed at: 2026-04-15 21:21:15 UTC
Staging URL: https://staging.forrt.org

The staging site shows the combined state of all compatible open PRs.