Skip to content

TINYDOC-3514 - Add scheduled link checking to CI#4276

Draft
kemister85 wants to merge 1 commit into
mainfrom
TINYDOC-3514-link-checking
Draft

TINYDOC-3514 - Add scheduled link checking to CI#4276
kemister85 wants to merge 1 commit into
mainfrom
TINYDOC-3514-link-checking

Conversation

@kemister85

Copy link
Copy Markdown
Contributor

Ticket: TINYDOC-3514

Adds automated external-link checking to the docs CI. This PR is layer 1 of 2 — the scheduled full-site scan that detects link rot across every published version. The PR-time check (validating links a PR changes) follows separately on the version branches (tinymce/8, /7, /6, /5), since that is where .adoc content changes land.

What this adds

  • .github/workflows/link-check.yml — runs every Monday 06:00 UTC (plus manual dispatch). Builds the full production site (all versions via antora-playbook.yml) and runs lychee over the built HTML. Opens a link-rot issue when dead links are found. Non-blocking by nature (scheduled, not a PR gate).
  • .lychee.toml — checker config. Key behaviours, all validated locally against a full 4-version build:
    • base_url resolves root-relative links (site chrome, /roadmap/) against the live domain instead of erroring on them locally.
    • Excludes the tiny.cloud family — those are validated by our own build/deploy, and HTTP-checking them from CI would hammer production (a naive config produced 640k+ false errors from rate-limiting).
    • Accepts redirects and 429; treats only genuine failures (4xx/5xx / no-resolve) as broken.
  • .lycheeignore — curated hosts that block automated checkers (npmjs, codepen, etc.) but resolve fine in a browser; each verified by hand.
  • package.jsoncheck-links script for local runs.
  • .gitignore — ignores the local lychee cache and report output.

Baseline

A full multi-version crawl currently surfaces ~12 genuinely dead links (the same rot found in the DOC-3513 audit: archive.tinymce.com, curl.haxx.se, commonjs.org, Facebook oEmbed, etc.). Fixing those is separate content follow-up; the weekly issue will track them.

Notes for review

  • Lands on main because scheduled workflows only fire from the default branch — and main's playbook already aggregates all versions, so one job covers 5/6/7/8.
  • Node pinned to 22 to match the current deploy workflows.

Opened as a draft for review before the version-branch PRs.

Add a weekly scheduled workflow that builds the full published site (all
versions) and runs lychee to detect dead external links, opening a
link-rot issue when any are found.

- .github/workflows/link-check.yml: weekly cron + manual dispatch
- .lychee.toml: checker config (resolves root-relative links via base_url,
  excludes the tiny.cloud family and localhost, accepts redirects and 429)
- .lycheeignore: curated hosts that block automated checkers but resolve
  in a browser
- package.json: check-links script for local use
- .gitignore: ignore the local lychee cache and report output
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant