test: add nightly markdown checking

[bnb]

Adds nightly markdown link checking. Helps prevent broken links that are left alone forever.

[bnb]

currently failing because we're getting throttled by GitHub since we have so many GitHub links in this repository 😅

[bnb]

This is currently completely blocked unless I can either ignore a path (/meetings) or batch requests. I've made issues (https://github.com/tcort/markdown-link-check/issues/102 and https://github.com/tcort/markdown-link-check/issues/101, respectively) for these in the correct repository. Until they're addressed, I don't think we can merge this. I'm totally open to alternative Actions that allow us to do this, this was just the best one I've found.

[mmarchini]

I think we can do both using the config-file option, it takes a JSON file to configure the underlying markdown-link-check npm package: https://github.com/tcort/markdown-link-check#config-file-format.

{
	"ignorePatterns": [
		{
			"pattern": "^http://example.net"
		}
	],
	"replacementPatterns": [
		{
			"pattern": "^.attachments",
			"replacement": "file://some/conventional/folder/.attachments"
		}
	],
	"httpHeaders": [
		{
			"urls": [
				"https://example.com"
			],
			"headers": {
				"Authorization": "Basic Zm9vOmJhcg==",
				"Foo": "Bar"
			}
		}
	]
}

The file probably would need to be generated on each run. Something like (not tested, this could be wildly wrong):

  - run: |
    cat <<EOF > config.json
    {
      "ignorePatterns": [
      {
        "pattern": "^http://github.com/${{ github.repository }}/tree/.*/meetings/.*"
      }
      ],
      "httpHeaders": [
        {
          "urls": [
            "https://github.com"
          ],
          "headers": {
            "Authorization": "Bearer ${{ secrets.GITHUB_TOKEN }}",
          }
        }
      ]
    }
    EOF

[bnb]

@mmarchini could you expand on why it'd need to be generated on each run? The HTTP Header part may be enough itself, I wasn't able to find any docs saying that was enabled/allowed on GitHub's side 🤔

[mmarchini]

Because we receive a new token on each run, and since the file is a JSON we can't set the authentication headers programatically on the config file. If markdown-link-check accepted JS files as well for config (like webpack and eslint do), we wouldn't need to generate on each run.

[bnb]

ah, that makes sense 🤦🏻

[bnb]

Added that cat command - I need to read up more on that usage of cat since it doesn't seem to have worked precisely how it should have (the build should have failed, but it succeeded 🤔 ).

I could have also put it in the wrong spot.

[bnb]

theoretically the PR linked below should have fixed the issue, I'll see if I can update the PR soon.

https://github.com/tcort/markdown-link-check/pull/106

[Trott]

I've unarchived this repo so I can close all PRs and issues before re-archiving.