Emulating org-mode's :noexport: behavior for Markdown.

noexport-subtrees.lua

--[[
Remove all subtrees whose headlines contain class `noexport`.

License:   MIT
Copyright: © Albert Krewinkel
]]

-- pandoc.utils.make_sections exists since pandoc 2.8
PANDOC_VERSION:must_be_at_least {2,8}

local utils = require 'pandoc.utils'

-- Returns true iff a div is a section div.
local function is_section_div (div)
  return div.t == 'Div'
    and div.classes[1] == 'section'
    and div.attributes.number
end

-- Returns the header element of a section, or nil if the argument is not a
-- section.
local function section_header (div)
  if not div.t == 'Div' then return nil end
  local header = div.content and div.content[1]
  local is_header = is_section_div(div)
    and header
    and header.t == 'Header'
  return is_header and header or nil
end

--- Remove remaining section divs
local function flatten_sections (div)
  local header = section_header(div)
  if not header then
    return nil
  else
    header.identifier = div.identifier
    div.content[1] = header
    return div.content
  end
end

function drop_noexport_sections (div)
  if div.classes:includes('noexport') then
    return {}
  end
end

--- Setup the document for further processing by wrapping all
--- sections in Div elements.
function setup_document (doc)
  local sections = utils.make_sections(false, nil, doc.blocks)
  return pandoc.Pandoc(sections, doc.meta)
end

return {
  {Pandoc = setup_document},
  {Div = drop_noexport_sections},
  {Div = flatten_sections}
}

Mark headers with {.noexport} to exclude them from the final document:

# First

First paragraph.

# This is an internal note {.noexport}

## with subsections

And words. If only a few.

# Third

## Yet more

This is more content.

Running pandoc --lua-filter=noexport-subtrees.lua test.md gives

<h1 id="first">First</h1>
<p>First paragraph.</p>
<h1 id="third">Third</h1>
<h2 id="yet-more">Yet more</h2>
<p>This is more content.</p>

Thank you for this ❤️ it works beautifully 🥳

I do believe the first line should not have a space between the -- and the [[

The script errored on my machine until I changed it to:

--[[

Thank you @whyboris! I removed the space.