Automated axe testing

[myasonik]

Summary

Closes #2231 and introduces automated axe testing!

Currently only testing the guidelines so the capability can be merged. Adding more pages over time can become a ticket an a few pages can be added at a time when people have spare cycles. (For reference, there are currently 84 errors found when running it across all pages. This does not count color contrast rules which had to be disabled due to their quantity...)

Run npm run start-test-server-and-a11y-test to run the tests locally.

TODO

• How to make this run in CI to block merge? (Added to test script)
• Make an issue for adding all the pages ([Meta][Docs] Add automated tests for all pages #2679)
• What sort of documentation does this need? (Added wiki page on it)
• On the Colors Guidelines there's a duplicate ID error in the range slider that I can't track down... (@thompsongl put up Alter id on EuiFormRow wrapping element #2588 to address this)

Checklist

- [ ] Checked in dark mode
- [ ] Checked in mobile
- [ ] Checked in IE11 and Firefox
- [ ] Props have proper autodocs
- [ ] Added documentation examples
- [ ] Added or updated jest tests
- [ ] Checked for breaking changes and labeled appropriately
- [ ] Checked for accessibility including keyboard-only and screenreader modes
- [ ] A changelog entry exists and is marked appropriately

[thompsongl]

On the Colors Guidelines there's a duplicate ID error in the range slider that I can't track down...

I'll take a look today

[thompsongl]

^ #2588 has what I think is the right way to fix the duplicate id error, but check my math there.

[chandlerprall]

Overall, these changes look good. I haven't reviewed a11y-testing.js or the readme changes in depth yet, but everything looks good functionally. A blocker for this moving forward is CI's failure when running

20:21:56 > @elastic/eui@17.0.0 test-a11y /app
20:21:56 > node ./scripts/a11y-testing
20:21:56 
20:21:57 (node:559) UnhandledPromiseRejectionWarning: Error: Failed to launch chrome!
20:21:57 /app/node_modules/puppeteer/.local-chromium/linux-706915/chrome-linux/chrome: error while loading shared libraries: libX11-xcb.so.1: cannot open shared object file: No such file or directory
20:21:57 
20:21:57 
20:21:57 TROUBLESHOOTING: https://github.com/GoogleChrome/puppeteer/blob/master/docs/troubleshooting.md
20:21:57 
20:21:57     at onClose (/app/node_modules/puppeteer/lib/Launcher.js:348:14)
20:21:57     at Interface.helper.addEventListener (/app/node_modules/puppeteer/lib/Launcher.js:337:50)
20:21:57     at Interface.emit (events.js:203:15)
20:21:57     at Interface.close (readline.js:397:8)
20:21:57     at Socket.onend (readline.js:173:10)
20:21:57     at Socket.emit (events.js:203:15)
20:21:57     at endReadableNT (_stream_readable.js:1143:12)
20:21:57     at process._tickCallback (internal/process/next_tick.js:63:19)
20:21:57 (node:559) UnhandledPromiseRejectionWarning: Unhandled promise rejection. This error originated either by throwing inside of an async function without a catch block, or by rejecting a promise which was not handled with .catch(). (rejection id: 1)
20:21:57 (node:559) [DEP0018] DeprecationWarning: Unhandled promise rejections are deprecated. In the future, promise rejections that are not handled will terminate the Node.js process with a non-zero exit code.

which demonstrates both a failure to run puppeteer and that the error didn't fail the CI job, instead it crashed "silently".

[thompsongl]

Just an update: I've been working on some Docker and script changes to get the necessary deps into EUI CI runs. Learning some things and getting closer to understanding what the limitations are. I'll continue to with infra to get this going.

[thompsongl]

jenkins test this

^ Last commit works in Docker locally, so I suspect some user/permission thing with the Jenkins setup

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/

[thompsongl]

jenkins test this

Going trigger a few more builds to see about stability.

The recent change is to run Chromium without a sandbox, which is generally only acceptable in cases like this where we fully own the content being crawled.

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/

[thompsongl]

jenkins test this

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/

[thompsongl]

jenkins test this

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/

[thompsongl]

jenkins test this

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/

[thompsongl]

jenkins test this

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/

[thompsongl]

jenkins test this

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/

[thompsongl]

jenkins test this

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/

[thompsongl]

jenkins test this

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/

[thompsongl]

I'm going to call this "stable enough" and do some clean up 🎉

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/

[thompsongl]

Ready for your review, @chandlerprall

[chandlerprall]

Changes LGreatTM, pulled & tested locally in both a passing & error states, with both the automated test server on port 9999 & against the local development server.

[thompsongl]

LGreatTM

😅

[kibanamachine]

Preview documentation changes for this PR: https://eui.elastic.co/pr_2569/