Custom filter for processing Github/Obsidian style callouts stopped working - any tips?

[danieltomasz]

Description

In order to process a obsidian/new github style callouts

> [!NOTE]
> Highlights information that users should take into account, even when skimming.

I added a custom lua filter to my _quarto.yml
with

filters:
- obsidian-callouts.lua

local stringify = (require "pandoc.utils").stringify

function BlockQuote (el)
    start = el.content[1]
    if (start.t == "Para" and start.content[1].t == "Str" and
        start.content[1].text:match("^%[!%w+%][-+]?$")) then
        _, _, ctype = start.content[1].text:find("%[!(%w+)%]")
        el.content:remove(1)
        start.content:remove(1)
        local class = "callout-" .. ctype:lower()
        div = pandoc.Div(el.content, {class = class})
        div.attributes["data-callout"] = ctype:lower()
        div.attributes["title"] = stringify(start.content):gsub("^ ", "")
        return div
    else
        return el
    end
end

-- source https://forum.obsidian.md/t/rendering-callouts-similarly-in-pandoc/40020/6

the filter worked a few months ago with 1.3 and previous pandoc, but now the filter produce empty output (no text from callout)
Anyone has any idea how to fix it?

[cscheid]

Can you provide a full example?

[danieltomasz]

I have now a problem with fully reproducing it. The behaviour I am expecting was working in May (colorful callouts), I even wrote a post about it
https://danielborek.me/2023/obsidian-quarto-callouts/

To partially replicate here are file

---
title: "Untitled"
format: html
filters:
  - callouts.lua
---

# Test filter

> [!note] My note
>
> Note content

And again the callout included

local stringify = (require "pandoc.utils").stringify

function BlockQuote (el)
    start = el.content[1]
    if (start.t == "Para" and start.content[1].t == "Str" and
        start.content[1].text:match("^%[!%w+%][-+]?$")) then
        _, _, ctype = start.content[1].text:find("%[!(%w+)%]")
        el.content:remove(1)
        start.content:remove(1)
        div = pandoc.Div(el.content, {class = "callout"})
        div.attributes["data-callout"] = ctype:lower()
        div.attributes["title"] = stringify(start.content):gsub("^ ", "")
        return div
    else
        return el
    end
end

The closest what I want to achieve I am getting with Quarto 1.2.475 and v1.1.189 (but previously I was able to get colour and title as well whem those are one of 5 defined in quarto types)

with the corresponding latex (I copying only the body, what is important 1.3 and 1.4 dont' produce tcolorbox)

\begin{document}
\maketitle
\ifdefined\Shaded\renewenvironment{Shaded}{\begin{tcolorbox}[interior hidden, borderline west={3pt}{0pt}{shadecolor}, frame hidden, enhanced, boxrule=0pt, sharp corners, breakable]}{\end{tcolorbox}}\fi

\hypertarget{test-filter}{%
\section{Test filter}\label{test-filter}}

\begin{tcolorbox}[enhanced jigsaw, opacityback=0, colback=white, rightrule=.15mm, arc=.35mm, colframe=quarto-callout-color-frame, toprule=.15mm, breakable, left=2mm, bottomrule=.15mm, leftrule=.75mm]

Note content

\end{tcolorbox}



\end{document}

The latest 1.3 and 1.4 releases render just text without a callout around it

\begin{document}
\maketitle
\hypertarget{test-filter}{%
\section{Test filter}\label{test-filter}}

Note content



\end{document}

Both rendering from qmd and ipynb gave me the same results,

[cscheid]

Ok, I understand now what you're asking. Callout is now a node type in the Quarto AST, and you just need to use quarto.Callout explicitly:

https://quarto.org/docs/prerelease/1.3/custom-ast-nodes/callout.html

function BlockQuote (el)
    start = el.content[1]
    if (start.t == "Para" and start.content[1].t == "Str" and
        start.content[1].text:match("^%[!%w+%][-+]?$")) then
        _, _, ctype = start.content[1].text:find("%[!(%w+)%]")
        el.content:remove(1)
        return quarto.Callout({
            type = ctype:lower(),
            title = stringify(start.content):gsub("^ ", "")
            content = el.content
        })
    else
        return el
    end

I haven't tested this, so it might need tweaking. But the general idea should work.

[danieltomasz]

Thanks, this is really informative.

I almost got it working :)
I have the filter as below

local stringify = (require "pandoc.utils").stringify

function BlockQuote(el)

    local start = el.content[1]

    if (start.t == "Para" and start.content[1].t == "Str" and
        start.content[1].text:match("^%[!%w+%][-+]?$")) then
        local firstline = stringify(start.content[1])
        local  _, _, ctype = firstline:find("%[!(%w+)%]")
        local titlevar =stringify(start.content):match("^%[!%w+%](.-)$")
        el.content:remove(1)
        -- Create Quarto callout
        return quarto.Callout({
          type = ctype:lower(),
          title = titlevar,
          content = el.content
        })
    else
        return el
    end
end

with it, how to make the content from 2 lines

> Line 1
> Line 2

To appear in the final document as separate lines

Line 1
Line 2

not

Line 1 Line 2

?

[danieltomasz]

I am don't know lua, how I can declare quarto dependency at the beginning to make it "proper' ?:P

[cscheid]

I assume this will work:

> Line 1
>
> Line 2

https://pandoc.org/lua-filters.html#type-blockquote shows that a BlockQuote holds a list of Block elements, so this should work (I again haven't tested).

If you're going down this rabbit hole, Pandoc's Lua filters page is indispensable. You should also get in the habit of reading the AST produced by Pandoc:

-- This is Lua code

-- this works, but might be a little hard to read
print(blockquote)
-- this produces indented output, but it only works for entire Pandoc documents,
-- so you'll need to be a bit more careful with it.
print(pandoc.write(pandoc.Pandoc({blockquote}), "native"))

[cscheid]

I am don't know lua, how I can declare quarto dependency at the beginning to make it "proper' ?:P

Sorry, I truly don't understand your question. What do you think is not proper about your filter declaration?

[RoyiAvital]

@danieltomasz , Is this available as an official extension?
It would be great!

[danieltomasz]

@RoyiAvital I might try to work on it, in meantime you might just use downloaded filter https://gist.github.com/danieltomasz/31d298aca2969adaf60d8841b68005e2

[f0nzie]

@RoyiAvital I might try to work on it, in meantime you might just use downloaded filter https://gist.github.com/danieltomasz/31d298aca2969adaf60d8841b68005e2

The script has two extras end at the bottom. After removing them, it works perfect!