The match patterns in render/base.rkt (and elsewhere) match reliably against elements created by the Commonmark parser because the set and order of attributes is consistent. But an author may generate txexprs of their own and these will amost never match.

#lang punct

`doc->html` will convert this to an `img` tag: ![Image 1](image1.jpg)

But not this: •'(image [[src "image3.jpg"]])

When writing Punct, there was a question of what guarantees I should make about the body element of document. I started by making it a very straight translation of the commonmark package’s document structure¹ into a list. Hence, Punct’s documentation specifies document-body as simply list?.

After some feedback, I’m revisiting this question. It would be more ergonomic to make this value (listof xexpr?). This way, people could make use of the familiar xml and txexpr packages to manipulate the document’s elements. Further, I could uniformly require that custom elements also be xexpr?, and the same requirements and guarantees would hold at all phases of parsing the document. I don’t think there would be any downside.