paragraph-headers.lua

-- SPDX-License-Identifier: 0BSD
-- Copyright (C) 2020 Roland Hieber
-- Convert headers of level >= 3 to LaTeX \paragraph{}s (usable with koma-script classes)
--

if FORMAT ~= "latex" and
   FORMAT ~= "beamer" and
   FORMAT ~= "json" and
   FORMAT ~= "native"
then
  error("unsupported format: " .. FORMAT)
end

local List = require 'pandoc.List'

-- The identifier in the native representation can still contain Unicode, and
-- pandoc's LaTeX writer takes additional care to escape unicode characters
-- when creating the label in the header and when referencing it in links (see
-- the function `toLabel` in pandoc's Text/Pandoc/Writers/LaTeX.hs).
-- Sadly this function is not exposed to Lua, so we have to make our own
-- compatible version.
--
-- Input: (string) identifier from native AST (`attr.identifier`)
-- Output: (string) label, compatible with the LaTeX writer
function to_label(str)
  local function do_error(msg, n, c)
    error(string.format("%s at input byte %d (0x%x)", msg, n, c))
  end

  local function uc_escape(val)
    if (val >= 0x30 and val <= 0x39) or -- isDigit x && isAscii x
       (val >= 0x41 and val <= 0x5A) or 
       (val >= 0x61 and val <= 0x7a) or -- isLetter x && isAscii x
       val == 0x5F or val == 0x2D or 
       val == 0x2B or val == 0x3D or
       val == 0x3A or val == 0x3B or
       val == 0x2E                      -- x `elemText` "_-+=:;."
    then
      return string.char(val)
    else
      return string.format("ux%x", val)
    end
  end

  -- no Unicode support in Lua... :( decode codepoint from UTF-8 manually
  local res = ""
  local seq = 0
  local val = nil
  local c = nil
  local n = 1
  while n <= #str do
    c = string.byte(str, n)
    if seq == 0 then
      -- leader byte: 0b0xxx xxxx, 0b110x xxxx, 0b1110 xxxx, 0b1111 0xxx
      if c >= 0x80 and c < 0xC0 then
        do_error("invalid UTF-8 sequence: expecting leader", n-1, c)
        return nil
      end
      seq = (c < 0x80 and 1) or (c < 0xE0 and 2) or
            (c < 0xF0 and 3) or (c < 0xF8 and 4)
      val = c & (2^(8-seq) - 1)
    elseif seq > 0 and c >= 0x80 and c < 0xC0 then
      -- continuation byte: 0b10xx xxxx
      val = (val << 6) | (c & 0x3F)
    else
      do_error("invalid UTF-8 continuation byte", n-1, c)
      return nil
    end
    seq = seq - 1

    if seq == 0 then
      res = res .. uc_escape(val)
    end

    n = n+1
  end

  if seq > 0 then
    do_error("invalid UTF-8 sequence: leader without continuation bytes", n-1, c)
    return nil
  end

  return res
end

-- Temporary pandoc.List of the elements making up the previous header
saved_header = nil

--
-- Handle Headers
--
function Header(elem)
  if elem.level < 3 then
    return elem
  else

    -- Fallback: if more than one Header occurs directly after another, put the
    -- elements of the previous one in an extra Para
    local result = {}
    if saved_header then
      result = pandoc.Para(saved_header)
    end

    -- maybe we need to merge Header with the next Para, save it
    local label = to_label(elem.identifier)
    saved_header = List:new {
        pandoc.RawInline("tex", "\\hypertarget{" .. label .. "}{%\n")
      , pandoc.RawInline("tex", "\\paragraph{")
      , pandoc.Span(elem.content)
      , pandoc.RawInline("tex", "}\\label{" .. label .. "}}\n")
    }

    return result
  end
end

-- If Para directly after Header, merge Para with header to get an in-line
-- paragraph header in LaTeX, i.e.:
--
--     \paragraph{Title} Lorem ipsum dolor sit amet...
--
-- instead of:
--
--     \paragraph{Title}
--
--     Lorem ipsum dolor sit amet...
--
function Para(elem)
  if saved_header then
    saved_header:extend(elem.content)
    local p = pandoc.Para(saved_header)
    saved_header = nil
    return p
  else
    return elem
  end
end

-- Fallback: In-line LaTeX paragraphs look bad when followed by e.g. a list, so
-- just output the saved header as its own block, followed by this Para.
function Block(elem)
  if saved_header then
    local p = pandoc.Para(saved_header)
    saved_header = nil
    return { p, elem }
  else
    return elem
  end
end

-- Fallback for one lonely header at the end with nothing after it
function Pandoc(doc)
  if saved_header then
    table.insert(doc.blocks, pandoc.Para(saved_header))
  end

  return doc
end

return {
  { Header = Header },
  { Para = Para },
  { Block = Block },
  { Pandoc = Pandoc },
}