Merging Code Blocks using Lua Filter?

[JoeyEremondi]

I'm trying to convert an Epub to org-mode, and the issue is that code blocks are just marked in the epub as having a certain class <c28> or <c41>.

I've been able to recognize these with a lua filter, but the problem is it turns each line of code into an inline code declaration, and I'd like to merge them into a single begin_src/end_src block. Likewise, if I do markdown instead of org, they end up as inline code instead of code blocks.

Is there a way to write a Haskell or Lua filter that merges these blocks?

I can code, but I'm extremely new to lua, and relatively new to pandoc and document stuff in general. I'm not new to Haskell, so if that's the easiest way to solve this then that's fine by me.

[badumont]

The best way may be to improve the filter you already have. Could you please share it with a sample input?

[JoeyEremondi]

Here's what I've got:

function Div(el)
  if el.classes:includes("c41") or el.classes:includes("c28") then
    return { pandoc.CodeBlock(pandoc.utils.stringify(el), "racket") }
  end
end

function Span(el)
  if el.classes:includes("c41") or el.classes:includes("c28") then
    return pandoc.Code(pandoc.utils.stringify(el), "racket")
  end
end


function Code(el)
  local lang = el.attr.class or ""
  local headers = table.concat(el.attr.attributes, " ")

  local header_part = headers ~= "" and "[" .. headers .. "]" or ""
  local lang_part = lang ~= "" and lang or "text"

  return pandoc.RawInline("org", string.format("src_%s%s{%s}", lang_part, header_part, el.text))
end

As an example, the input:

<p><span class="c4"></span></p>
<p><span class="c41">(define-type Exp</span></p>
<p><span class="c41">  [num (n : Number)]</span></p>
<p><span class="c41">  [plus (left : Exp) (right : Exp)])</span></p>
<p><span class="c4"></span></p>

produces the following org:

=(define-type Exp=

= [num (n : Number)]=

= [plus (left : Exp) (right : Exp)])=

where I want it to produce:

#+begin_src racket
(define-type Exp
  [num (n : Number)]
  [plus (left : Exp) (right : Exp)])
#+end_src

[badumont]

You need to accumulate the lines in a variable and to format and output them when certain conditions are reached. Fortunately, the formatting of your sample makes this quite easy (e.g. no need to iterate over Inlines). Here is a proof of concept: ```lua traverse = "topdown" -- Otherwise all Spans will be treated before all Paras. local accum_code = "" local function to_code(el) -- The HTML code uses non breakable spaces to force indentation. return string.gsub(pandoc.utils.stringify(el), " ", " ") end local function to_code_block(s, lang) return "#+begin_src " .. lang .. "\n" .. s .. "\n#+end_src" end local function append_line(orig, appended) if orig ~= "" then orig = orig .. '\n' end return orig .. appended end function Para(el) -- We cannot return a RawBlock from a Span: only Blocks can return -- a List of Blocks. So we return from the surrounding Para. local first = el.content[1] if first.classes:includes("c41") or first.classes:includes("c28") then accum_code = append_line(accum_code, to_code(first)) return {} elseif first.classes:includes("c4") or first.classes:includes("c2") then if accum_code ~= "" then local to_print = accum_code accum_code = "" return { pandoc.RawBlock("org", to_code_block(to_print, "racket")) } else return {} end end end ```