Pandoc script to fix links in articles downloaded with MarkDownload

fix_links.lua

-- Track paragraphs to remove
local paragraphs_to_remove = {}

-- First pass: identify broken link patterns
function identify_broken_links(doc)
  local blocks = doc.blocks
  
  for i = 1, #blocks do
    if blocks[i].t == "Para" then
      local text = pandoc.utils.stringify(blocks[i])
      
      -- Mark standalone [ paragraphs for removal
      if text:match("^%s*%[%s*$") then
        paragraphs_to_remove[i] = true
      end
      
      -- Mark standalone ](...) paragraphs for removal
      if text:match("^%s*%]%(") then
        paragraphs_to_remove[i] = true
      end
    end
  end
  
  return doc
end

-- Second pass: filter paragraphs and process content
function Para(el)
  local text = pandoc.utils.stringify(el)

  -- Remove paragraph with only `[`
  if text:match("^%s*%[%s*$") then
    return {}
  end

  -- Remove paragraph starting with `](` — broken link tail
  if text:match("^%s*%]%(") then
    return {}
  end

  local new_inlines = {}

  for _, inline in ipairs(el.content) do
    if inline.t == "Str" then
      -- Convert ![[...]] to markdown image (wikilink format)
      local filename = inline.text:match("^!%[%[(.-)%]%]$")
      if filename then
        table.insert(new_inlines, pandoc.Image("", filename))
      else
        -- Handle partial wikilink conversions in strings
        local text = inline.text
        -- Convert ![[filename]] patterns within text
        text = text:gsub("!%[%[(.-)%]%]", function(filename)
          return "![" .. filename .. "](" .. filename .. ")"
        end)
        table.insert(new_inlines, pandoc.Str(text))
      end
    else
      table.insert(new_inlines, inline)
    end
  end

  return pandoc.Para(new_inlines)
end

-- Filter for the entire document to handle multi-paragraph patterns
function Pandoc(doc)
  local new_blocks = {}
  local blocks = doc.blocks
  
  for i = 1, #blocks do
    if blocks[i].t == "Para" then
      local text = pandoc.utils.stringify(blocks[i])
      
      -- Skip standalone [ paragraphs
      if text:match("^%s*%[%s*$") then
        -- Skip this paragraph
      -- Skip standalone ](...) paragraphs  
      elseif text:match("^%s*%]%(") then
        -- Skip this paragraph
      else
        table.insert(new_blocks, blocks[i])
      end
    else
      table.insert(new_blocks, blocks[i])
    end
  end
  
  return pandoc.Pandoc(new_blocks, doc.meta)
end