chore: Check links with lychee

[scsmithr]

Adds an npm script for running lychee against a locally running site.

https://github.com/lycheeverse/lychee

[greyscaled]

I'd maybe add it to the check:all script as well

[tychoish]

I worry a little bit about this being flaky when run a lot if sites block robots or similar, if this ends up going in CI/blocking a deploy that could be annoying.

[scsmithr]

I'd maybe add it to the check:all script as well

Needs a running server which I think would trigger an unexpected failure.

I wrote this script:

#!/usr/bin/env bash

# Spins up the docs site (on a non-default port) and runs lychee against it to
# check for broken links.

set -e

bundle exec jekyll serve --port 4001 > /dev/null 2>&1 &
pid=$!

attempts=0
until curl --output /dev/null --silent --head --fail http://localhost:4001; do
    attempts=$((attempts+1))
    if [ $attempts -ge 10 ]; then
        echo "max attempts exceeded"
        exit 1
    fi
    echo "waiting for server"
    sleep 1
done

lychee -v http://localhost:4001

kill $pid

Which would spin up a site on 4001 and check against that. I was going to check that in and run it in ci, but got kinda lazy with actually getting lychee installed (I'm yamled out).

I worry a little bit about this being flaky when run a lot if sites block robots or similar, if this ends up going in CI/blocking a deploy that could be annoying.

The goal is really to check links against our site & docs. It's easy enough to exclude external sites as necessary. I just wanted something that we could use to quickly check that our links are working, and give us confidence when moving/updating docs.

I plan on setting this up on glaredb.com too since we have links to docs in the blog posts, and I'd rather be proactive about updating links than not know we broke them in the first place.

[tychoish]

should have it run more in a cron on the main site just so that broken docs links won't linger too long in the main site, but this seems reasonable,