How can I parse content into a custom structure?

Questions
1

Is it possible to have a function parse content into a custom structure like so?

#parse[
  [a[b[c][d]]]
] == (body: [a], children: (
  (body: [b], children: (
      (body: [c], children: ()),
      (body: [d], children: ()),
    )
  ),
))

I know how to do this for strings and have done so, but would like to write a solution which allows for any content as the nodes’ bodies. I could try going into the sequence and messing around, but I wanted to see if there’s a better way to approah this first.

As a note, square brackets are typically used for this (linguistics syntax trees), but if it’s not possible I’d still be interested in how to do this for other delimiters.

2

Since you’ve already got a way to turn a string like “a[b[c]]” into that type of structure, maybe all you need is a way to convert content to string:

#let parseAsString(it) = {
  let asString = it.fields().at("children")
    .map(c => c.fields())
    .map(v => v.at("text", default: none))
    .filter(v => v != none).join()
  
  return asString
}

A more general solution for converting content to string was provided by GitHub user lvjr here:

3

That would work if I only wanted string output. I was able to find a solution that preserves content (ie. allows [$x$[*y*][@z]]). As a result of using square brackets, all content passed into the function ends up being a sequence (as far as I can tell), where [ and ] are never grouped with anything else (they’re their own child). So I ended up taking the sequence’s children and matching on [[] and []] instead of "[" and "]", then concatenating the children which make up the body.

4

I ended up needing this for an extension to my parser, so it was super helpful. Thanks! I modified it for my use case to preserve quotation marks, so here’s that for posterity.

let stringify(it) = {
  for child in it.children {
    if child.func() == smartquote {
        if child.double { "\"" } else { "'" }
      } else {
        child.at("text", default: none)
      }
  }
}
3 Likes
5

For converting content to string, there is t4t package:

#import "@preview/t4t:0.4.2": get
#get.text[content]

Although it doesn’t always work as expected, like for math mode, but I think there are other packages that can handle math conversion.