Icons

[reakaleek]

Closes #409

Changes

• Copied SVG files from https://github.com/elastic/eui/tree/main/packages/eui/src/components/icon/svgs
• Created role parser with the syntax {icon}`icon-name`

Notes

• Using the syntax :icon-name: did not fully work as expected. Especially, in cases where the icon would be the first element in a list or other v3 markdown items.
  I suspect it's conflicting with the :::{} blocks and :option: syntax.

Future improvements

Somehow automatically copy the files from https://github.com/elastic/eui/tree/main/packages/eui/src/components/icon/svgs during the build process.

Preview

https://docs-v3-preview.elastic.dev/elastic/docs-builder/pull/1413/syntax/icons

[Mpdreamz]

It'd be great if we can get :icon: to work to match Github flavored markdown.

I can take a stab at that in a follow up if needed?

Otherwise we need a clear new syntax i:<icon>: is not quite it!

Other than that love this!

[reakaleek]

It'd be great if we can get :icon: to work to match Github flavored markdown.

I can take a stab at that in a follow up if needed?

Otherwise we need a clear new syntax i:<icon>: is not quite it!

Other than that love this!

I got it working. Let me know what you think.

e45efec

The main change is in the src/Elastic.Markdown/Myst/Directives/DirectiveParagraphParser.cs

But I'm actually worried, this will have other unintended sideeffects. E.g. if a directive option has the same name as an icon.

IMO, the safest option is another syntax.

Looking at the Myst documentation the most logical syntax would be:

{icon}`icon-name`

[cotti]

(Not an issue in the PR, but in GH) What a weird false positive... If you open this in a hex editor, U+2009 (0xE2 0x80 0x89) is nowhere to be seen.

[reakaleek]

Ah, you mean the error in the failing test. I was not able to find a difference and I don't think this error will cause any troubles in production.

We might feel the pain when we add tests in the future 😅

But I think the simplified test is testing the necessary parts. Which is fine.

Thank you for checking.

[theletterf]

If you're referring to the docs-build hints, it's not a false positive, the character really is invisible. It should be a space or not a space. Easy to catch with a regex such as \u2009 if I remember well. I fixed most of those in another PR, so a rebase should do the trick.

[reakaleek]

No. I had a test that was failing only in windows.

The error said that the output doesn't match the expected. But it was exactly the same.

Hence, @cotti suspected that there must be a weird space or something.

[reakaleek]

Ah sorry, I actually misunderstood this haha :D

[cotti]

I think the Suppress is not needed anymore here

[reakaleek]

Oh. Thanks!

The copy-pasta syndrome (:

[reakaleek]

91fc82b

[cotti]

🚂

PTO