I see. I think you're spot on and I have a conceptual misunderstanding.

As to my use case, it's pretty simple. I have a markdown editor on the client. I just want to load markdown files and preserve the structure, so

foo 

bar

stays as is, rather than becoming

foo
bar

But I suppose, judging by what you're telling me, this is a problem with the transformer rather than the markdown AST?

Oops I missed Christian's comment above which actually helps a lot, thanks!

As to my use case, it's pretty simple. I have a markdown editor on the client. I just want to load markdown files and preserve the structure, so ...

If you want the raw text, then why are you parsing it with a markdown parser? If you want the plain text, get the file and don’t pass it to a markdown parser, but display that raw file instead?

No, I do parse and render the AST rather than the raw text. But Christian is right I think it's up to the transformer/client. For example, since I render in HTML so I would either have to add bottom margin to show the split or something like that (maybe add an extra p tag).

I think the confusion came from the fact that I had one conceptual understanding of what a paragraph is based on HTML and that is slightly different with Markdown.

In HTML you can do

<p>foo</p>
<p></p>
<p>bar</p>
<p></p>

and this is what I wanted my pipeline to produce. But it turns out those empty lines signify the split of the paragraph in Markdown so in the end you get only two. So I will need to think about how to manage that difference.

Nice. Sounds like you’ve got a good understanding of what’s going on and how to solve it.

One thing though: In HTML you also shouldn’t add empty paragraphs. HTML has meaning. So this might be more of a CSS thing, which adds presentation!