Fix: downlit_html_*() keeps all pre classes

[etiennebacher]

Close #149, close #156 (and v2 of #155).

downlit_html() now keeps all pre classes, which allows for customization of code blocks in pkgdown using chunk options class.source and attr.source.

Current behavior (.my-class and .another-class are dropped):

library(downlit)

html <- xml2::read_xml("
<div>
  <div class='sourceCode'>
    <pre class='sourceCode r my-class'>
      <code class='sourceCode r'>
        <span>Hello</span>
      </code>
    </pre>
  </div>
  <div class='sourceCode'>
    <pre class='sourceCode r another-class'>
      <code class='sourceCode r'>
        <span>Hello</span>
      </code>
    </pre>
  </div>
</div>")
downlit_html_node(html)
html
#> {xml_document}
#> <div>
#> [1] <div class="sourceCode">\n  <pre class="downlit sourceCode r">\n<code cla ...
#> [2] <div class="sourceCode">\n  <pre class="downlit sourceCode r">\n<code cla ...

New behavior:

library(downlit)

html <- xml2::read_xml("
<div>
  <div class='sourceCode'>
    <pre class='sourceCode r my-class'>
      <code class='sourceCode r'>
        <span>Hello</span>
      </code>
    </pre>
  </div>
  <div class='sourceCode'>
    <pre class='sourceCode r another-class'>
      <code class='sourceCode r'>
        <span>Hello</span>
      </code>
    </pre>
  </div>
</div>")
downlit_html_node(html)
html
#> {xml_document}
#> <div>
#> [1] <div class="sourceCode">\n  <pre class="downlit sourceCode r my-class">\n ...
#> [2] <div class="sourceCode">\n  <pre class="downlit sourceCode r another-clas ...

[etiennebacher]

Test failures are unrelated to this PR.

[hadley]

I think you could simplify this code by adding pre_class as an explicit argument to tweak_children() — then you don't need to go through gymnastics to figure out whether or not it's set.

[etiennebacher]

I'm not sure I understand. If I add an explicit argument pre_class in tweak_children(), then I will have to check whether this argument is NULL (or FALSE) before applying strsplit(), so I don't think this would result in a simpler code. But maybe I misunderstood you?

[sgvignali]

Do you have any updates?

[etiennebacher]

@sgvignali I don't think there's anything more I can do on my side, so I'm just waiting for someone from r-lib to review this

[froggleston]

Apologies for necroing this old PR and related issue, but I have been trying to fix a failing WCAG check and found this. Current accessibility standards state that scrollable elements need a tabindex to allow keyboard focus and scrolling. Given that downlit removes any user-added pre classes and attributes (e.g. tabindex="0"), the WCAG check will fail:

https://dequeuniversity.com/rules/axe-devtools/4.6/scrollable-region-focusable

Output from the AC Lite 1.0.7 Accessibility Checker extension for Chrome