Keep custom attribute prefixes when converting HTML to Markdown

[qingnianz]

Is there any alternative method to retain custom attribute names prefixed with 'data-' when converting HTML to Markdown, aside from utilizing filters to strip off the 'data-' prefix?
like:

now:
::: {outlineid="731098383993"}
:::
expect:
::: {data-outlineid="731098383993"}
:::

[rnwst]

I'm not sure I understand the question. Pandoc prefixes attributes that are not HTML5 attributes with data- when converting from Markdown to HTML. When converting from HTML back to Markdown, those data- prefixes are removed again. Therefore, during a roundtrip conversion from Markdown to HTML and back to Markdown, attribute names are preserved.

If I understand correctly, you want to preserve the data- prefix when converting from HTML to Markdown? AFAIK there is no writer option to accomplish this. However, I don't understand how stripping off the data- prefix using a filter would help with this? The data- prefix is removed by the HTML reader, so by the time it gets to a filter the prefix is already gone. To prevent this, you could either write your own reader that doesn't remove the data- prefix (complicated), or you could write a filter (much easier) which checks if an attribute is a known HTML5 attribute, and if not, data- is prefixed to the attribute name (however, this won't work for attributes which had data- prefixed but are actually known HTML5 attributes -- that being said, such a case should be rare). Let me know if you need help with this.