Render URLs and DOIs in dataset names and descriptions as links

[skearnes]

Dataset names and descriptions routinely cite a publication, but the reference was plain text everywhere it appeared. This makes it clickable in the browse table (Name and Description columns) and in the dataset details modal.

LinkedText handles the reference styles that actually occur in ord-data:

• http(s)://… URLs and bare www. hosts
• bare DOIs — 10.1021/jacs.8b01523, with or without a DOI: / doi: prefix — linked through doi.org
• the DOI inside a doi.org URL stays part of the URL rather than matching twice

Trailing punctuation is left out of the link, so …available at https://doi.org/10.1021/jacs.6c05959. doesn't swallow the period and (DOI: 10.15227/orgsyn.095.0080) drops the paren, while a link that opened its own bracket (…/Foo_(bar)) keeps it. Links open in a new tab with rel="noopener noreferrer".

Tests

This adds the first JavaScript tests in the repo: vitest renders the component with react-dom/server, so no DOM environment (jsdom, testing-library) is needed. npm test runs them, and the check_javascript job now runs it alongside prettier.

The cases come from real ord-data text: URLs, bare hosts, bare DOIs with and without a prefix, a DOI inside a doi.org URL, trailing sentence punctuation, and brackets the link did or didn't open.

Verification

Checked against every dataset description available in ord-data (via the parquet metadata): 30 distinct links, all correct, with the rendered text byte-identical to the input in every case. Exercised end to end against the local stack (ORD_INTERFACE_TESTING=TRUE backend + Vite) in headless Chrome — the bundled test datasets cover both a bare URL and a bare DOI.

Two descriptions in ord-data are missing a space after the reference, so the link absorbs the following word:

• ord_dataset-ac78456835404910b3a4c840248b6ac9 — https://doi.org/10.1021/acs.accounts.0c00760Data from Figure S12…
• ord_dataset-d319c2a22ecf4ce59db1a18ae71d529c — doi: 10.1039/c5sc04751jThis palladium catalyzed…

Any heuristic that splits those also breaks legitimate mixed-case DOIs such as 10.48550/arXiv.2506.07619, so the fix belongs upstream in the data.

Also bumps the footer copyright year to 2026.

🤖 Generated with Claude Code

Greptile Summary

Adds automatic URL and DOI linking to dataset names and descriptions.

• Introduces the LinkedText component with punctuation-aware link parsing and styling.
• Applies linked text rendering to the browse table and dataset details modal.
• Adds Vitest coverage and runs the tests in the JavaScript CI job.
• Updates the footer copyright year to 2026.

Confidence Score: 5/5

The PR appears safe to merge because no blocking failures eligible for this follow-up review remain.

No blocking failure remains.

Important Files Changed

Filename	Overview
app/src/components/LinkedText.tsx	Adds URL and DOI detection, punctuation trimming, and secure external-link rendering.
app/src/components/LinkedText.test.tsx	Covers supported reference formats, punctuation handling, text preservation, and external-link attributes.
app/src/views/browse/MainBrowse.tsx	Uses LinkedText for dataset names and descriptions in browse cells and the details modal.
app/package.json	Adds the Vitest test command and development dependency.
.github/workflows/checks.yml	Runs the new JavaScript test suite after installing locked dependencies.

_{Reviews (3): Last reviewed commit: "Record optional platform packages in pac..." | Re-trigger Greptile}