asw/packs/pandoc/asw.lua

-- asw.lua — Pandoc Lua filter for Agentic Semantic Web (ASW)
-- Converts GFM markdown constructs to ASW data-attribute HTML
--
-- Usage:
--   pandoc input.md --from gfm --lua-filter asw.lua --template asw.html5 -o output.html
--
-- Transforms:
--   - [ ] item      → <li data-task="todo">
--   - [x] item      → <li data-task="done">
--   > [!NOTE]       → <div data-callout="note">
--   > [!WARNING]    → <div data-callout="warning">
--   > [!TIP]        → <div data-callout="tip">
--   > [!ERROR]      → <div data-callout="error">
--   [[target]]      → <a data-wikilink href="#target">target</a>
--   [[target|label]] → <a data-wikilink href="#target">label</a>

-- ── Callouts ─────────────────────────────────────────────────────────────────
-- GitHub callout syntax: > [!TYPE]\n> content → <div data-callout="type">

local CALLOUT_MAP = {
  NOTE      = "note",
  INFO      = "note",
  IMPORTANT = "note",
  WARNING   = "warning",
  CAUTION   = "warning",
  TIP       = "tip",
  ERROR     = "error",
  DANGER    = "error",
}

local CALLOUT_LABEL = {
  note    = "Note",
  warning = "Warning",
  tip     = "Tip",
  error   = "Error",
}

function BlockQuote(el)
  local first = el.content[1]
  if not first or (first.t ~= "Para" and first.t ~= "Plain") then return el end

  -- The first inline should be a Str like "[!NOTE]"
  local first_inline = first.content and first.content[1]
  if not first_inline or first_inline.t ~= "Str" then return el end

  local marker = first_inline.text:match("^%[!(%u+)%]$")
  if not marker then return el end

  local callout_type = CALLOUT_MAP[marker]
  if not callout_type then return el end

  local label = CALLOUT_LABEL[callout_type] or marker:sub(1,1) .. marker:sub(2):lower()

  local blocks = {
    pandoc.RawBlock("html", '<div data-callout="' .. callout_type .. '">'),
    pandoc.RawBlock("html", '<p data-callout-title>' .. label .. '</p>'),
  }

  -- First para: strip the [!MARKER] token and optional following SoftBreak/Space
  local inlines = first.content
  local start_i = 2
  if inlines[start_i] and
     (inlines[start_i].t == "SoftBreak" or inlines[start_i].t == "Space") then
    start_i = start_i + 1
  end

  local rest_inlines = {}
  for i = start_i, #inlines do
    table.insert(rest_inlines, inlines[i])
  end
  if #rest_inlines > 0 then
    table.insert(blocks, pandoc.Para(rest_inlines))
  end

  -- Remaining blocks in the blockquote
  for i = 2, #el.content do
    table.insert(blocks, el.content[i])
  end

  table.insert(blocks, pandoc.RawBlock("html", '</div>'))
  return blocks
end

-- ── Task Lists ───────────────────────────────────────────────────────────────
-- GFM: - [ ] → <li data-task="todo">, - [x] → <li data-task="done">
-- Pandoc (GFM mode) emits checkboxes as RawInline html: <input type="checkbox" ...>

-- Pandoc 3.x encodes GFM task list checkboxes as Unicode Str:
-- ☒ (U+2612) = checked/done, ☐ (U+2610) = unchecked/todo
local CHECKBOX_DONE = "\xe2\x98\x92" -- ☒ UTF-8
local CHECKBOX_TODO = "\xe2\x98\x90" -- ☐ UTF-8

local function checkbox_state(inline)
  if inline and inline.t == "Str" then
    if inline.text == CHECKBOX_DONE then return "done" end
    if inline.text == CHECKBOX_TODO then return "todo" end
  end
  return nil
end

function BulletList(el)
  -- First pass: check if any item is a task
  local has_tasks = false
  for _, item in ipairs(el.content) do
    local first = item[1]
    if first and (first.t == "Plain" or first.t == "Para") and first.content[1] then
      if checkbox_state(first.content[1]) then
        has_tasks = true
        break
      end
    end
  end
  if not has_tasks then return el end

  -- Rebuild as explicit HTML with data-task attributes
  local parts = { pandoc.RawBlock("html", "<ul>") }

  for _, item in ipairs(el.content) do
    local first = item[1]
    local state = nil
    if first and (first.t == "Plain" or first.t == "Para") and first.content[1] then
      state = checkbox_state(first.content[1])
    end

    if state then
      table.insert(parts, pandoc.RawBlock("html", '<li data-task="' .. state .. '">'))
      -- Strip checkbox inline (index 1) and optional space (index 2)
      local trimmed = {}
      for i, inline in ipairs(first.content) do
        if i == 1 then
          -- skip checkbox
        elseif i == 2 and inline.t == "Space" then
          -- skip leading space after checkbox
        else
          table.insert(trimmed, inline)
        end
      end
      if #trimmed > 0 then
        table.insert(parts, pandoc.Plain(trimmed))
      end
      -- Remaining blocks in this item
      for i = 2, #item do
        table.insert(parts, item[i])
      end
      table.insert(parts, pandoc.RawBlock("html", "</li>"))
    else
      -- Regular list item — emit verbatim
      table.insert(parts, pandoc.RawBlock("html", "<li>"))
      for _, block in ipairs(item) do
        table.insert(parts, block)
      end
      table.insert(parts, pandoc.RawBlock("html", "</li>"))
    end
  end

  table.insert(parts, pandoc.RawBlock("html", "</ul>"))
  return parts
end

-- ── Wikilinks ─────────────────────────────────────────────────────────────────
-- [[target|label]] or [[target]] → <a data-wikilink href="#slug">label</a>
--
-- Pandoc 3.x tokenizes at whitespace, so [[target|label with spaces]] arrives as
-- multiple Str/Space tokens. process_wikilinks coalesces adjacent Str+Space runs
-- into a combined string before scanning for [[...]] patterns, then reconstructs
-- the inline list. Non-Str/Space inlines (Bold, Emph, Code, etc.) pass through.

local function slugify(text)
  return text:lower():gsub("[%s_]+", "-"):gsub("[^%w%-]", "")
end

-- Scan combined for [[...]] patterns and emit inlines.
-- Returns a list of inlines with wikilinks replaced, plus a changed flag.
local function scan_for_wikilinks(combined)
  local parts = {}
  local pos = 1
  local changed = false

  while pos <= #combined do
    local s, e = combined:find("%[%[", pos)
    if not s then
      if pos <= #combined then
        table.insert(parts, pandoc.Str(combined:sub(pos)))
      end
      break
    end
    if s > pos then
      table.insert(parts, pandoc.Str(combined:sub(pos, s - 1)))
    end
    local close_s, close_e = combined:find("%]%]", e + 1)
    if not close_s then
      table.insert(parts, pandoc.Str("[["))
      pos = e + 1
    else
      local inner = combined:sub(e + 1, close_s - 1)
      local target, label = inner:match("^([^|]+)|(.+)$")
      if not target then
        target = inner
        label = inner
      end
      target = target:match("^%s*(.-)%s*$")
      label  = label:match("^%s*(.-)%s*$")
      table.insert(parts, pandoc.RawInline("html",
        '<a data-wikilink href="#' .. slugify(target) .. '">' .. label .. '</a>'))
      pos = close_e + 1
      changed = true
    end
  end

  return parts, changed
end

local function process_wikilinks(inlines)
  -- Pass 1: coalesce runs of Str+Space into combined strings, scan for wikilinks.
  -- Non-text inlines (Emph, Strong, Code, RawInline, etc.) flush any pending run first.
  local result = {}
  local changed = false
  local i = 1

  while i <= #inlines do
    local el = inlines[i]

    if el.t == "Str" or el.t == "Space" then
      -- Collect a contiguous Str+Space run
      local run = {}
      local combined = ""
      while i <= #inlines and (inlines[i].t == "Str" or inlines[i].t == "Space") do
        if inlines[i].t == "Str" then
          combined = combined .. inlines[i].text
        else
          combined = combined .. " "
        end
        table.insert(run, inlines[i])
        i = i + 1
      end

      if combined:find("%[%[") then
        local parts, any_changed = scan_for_wikilinks(combined)
        if any_changed then
          for _, part in ipairs(parts) do table.insert(result, part) end
          changed = true
        else
          -- No wikilinks resolved — emit original tokens unchanged
          for _, tok in ipairs(run) do table.insert(result, tok) end
        end
      else
        -- No [[ at all — emit original tokens unchanged
        for _, tok in ipairs(run) do table.insert(result, tok) end
      end
    else
      table.insert(result, el)
      i = i + 1
    end
  end

  return changed and result or nil
end

function Para(el)
  local r = process_wikilinks(el.content)
  if r then return pandoc.Para(r) end
end

function Plain(el)
  local r = process_wikilinks(el.content)
  if r then return pandoc.Plain(r) end
end

function Header(el)
  local r = process_wikilinks(el.content)
  if r then return pandoc.Header(el.level, r, el.attr) end
end