AsciiDoc writer: Support for roles with Block elements

[yanntrividic]

Hello,

Thank you for all the work on the AsciiDoc writer, and now on the reader as well. That's fantastic to have both of them now!

A feature that would be super great to have in the writer, in my opinion, would be element attributes. Currently, the support for this is limited, as most Block elements can only carry identifiers, and inlines can only have identifiers and roles.

For example, when I input this in the Pandoc-flavored Markdown reader:

# Header {#identifier .class .class key=value key=value}

::::: {#special .sidebar}
Here is a paragraph.

And another.
:::::

This is a [class]{.class #test key=value}.

I get this AsciiDoc output:

[[identifier]]
== Header

[[special]]
Here is a paragraph.

And another.

This is a [#test .class]#class#.

(We can also appreciate a bug here, as the [[special]] identifier only applies to the first paragraph.) Edit: As noted in the next two messages, I was not using the latest version, hence this bug.

AsciiDoc has different types of element attributes, and even though all of them are useful, it feels to me that the most interesting one currently missing is the role attribute for Block elements, which is parsed as a class by Asciidoctor.

I think this might be a bit too hard of a problem for me, so I am not sure if I will be able to help, but I would definitely be thankful if somebody had the motivation to have a look at this issue!

[jgm]

Here is what I get with your input:

[[identifier]]
== Header

[[special]]
[SIDEBAR]
****
Here is a paragraph.

And another.
****

This is a [#test .class]#class#.

What version of pandoc are you using?

[yanntrividic]

My bad, I thought I used the latest version for this snippet but I forgot the leading ./ in front of my command. I got the same output as you! I will edit my first message accordingly.

[jgm]

So, what is the expected asciidoc for the header with attributes in your example?

[yanntrividic]

Okay, I'll just refine my example because it was definitely messy and with a few errors. I'll focus on the id and role attributes, as the other named attributes are not for every type of elements (even though in my original example I suggested it was the case).

If I had, let's say, this Markdown input:

# Header {#identifier .class1 .class2}

::::: {#special .sidebar .class3}
Here is a paragraph.

And another.
:::::

::::: {.class4}
This is another test.
:::::

This is a [class]{.class #test1}.

Which currently produces this AsciiDoc output:

[[identifier]]
== Header

[[special]]
[SIDEBAR]
****
Here is a paragraph.

And another.
****

This is another test.

This is a [#test1 .class]#class#.

Currently, the header and sidebar elements are using the double bracket syntax, but there are three types of syntax available for ID and role assignments, and the "shorthand" syntax is common to both block and inline elements.

Also, before continuing, the space between #test1 and .class should not exist. I think it is a bug. Asciidoctor processes it as <span id="test1 " class="class">class</span> with an extra space at the end of the id value. Maybe I should open another issue targeting this in particular?

So, I would expect it to give me this AsciiDoc output instead:

[#identifier.class1.class2]
== Header

[SIDEBAR#special.class3]
****
Here is a paragraph.

And another.
****

[.class4]
This is another test.

This is a [#test1.class]#class#.

I know Para elements in Pandoc cannot have a class, though in that case, it can be considered as a  Div with a class that contains a Para element. It would anyway be then converted like this to HTML by Asciidoctor (I indented the output for readability):

<div class="sect1 class1 class2">
	<h2 id="identifier">Header</h2>
	<div class="sectionbody">
		<div id="special" class="sidebarblock class3">
			<div class="content">
				<div class="paragraph">
					<p>Here is a paragraph.</p>
				</div>
				<div class="paragraph">
					<p>And another.</p>
				</div>
			</div>
		</div>
		<div class="paragraph class4">
			<p>This is another test.</p>
		</div>
		<div class="paragraph">
			<p>This is a <span id="test1" class="class">class</span>.</p>
		</div>
	</div>
</div>ht

I voluntarily put aside the other delimited blocks for now, as they follow the same logics as the sidebar blocks.