nvim-lua-cmark-ast.lua

--- This used to be in <https://github.com/dmitmel/dotfiles/blob/master/nvim/lua/dotfiles/lsp/markup.lua>.
local M = require('dotfiles.autoload')('dotfiles.lsp.markdown')

local utils = require('dotfiles.utils')
local ffi = require('ffi')

---@type table<string, any>
local lib = ffi.load('cmark-gfm')
---@type table<string, any>
local lib_ext = ffi.load('cmark-gfm-extensions')

ffi.cdef(utils.read_file(vim.fn.fnamemodify(utils.script_path(), ':h') .. '/cmark_ffi_defs.h'))

function M.cmark_text_to_ast(text)
  local options = lib.CMARK_OPT_DEFAULT

  local parser = ffi.gc(lib.cmark_parser_new(options), lib.cmark_parser_free)
  for _, ext_name in ipairs({
    'table',
    'autolink',
    'strikethrough',
    -- 'tagfilter',
    'tasklist',
  }) do
    local ext = lib_ext.cmark_find_syntax_extension(ext_name)
    if ext == nil then
      error('extension not available: ' .. ext_name)
    end
    lib_ext.cmark_parser_attach_syntax_extension(parser, ext)
  end

  lib.cmark_parser_feed(parser, text, #text)
  local doc = ffi.gc(lib.cmark_parser_finish(parser), lib.cmark_node_free)

  return M.cmark_node_to_ast(doc)
end

-- Based on <https://github.com/github/cmark-gfm/blob/0.29.0.gfm.2/src/xml.c#L34-L155>.
function M.cmark_node_to_ast(root_node)
  local ast_root_node = nil
  local ast_nodes_stack = {}

  local iter = ffi.gc(lib.cmark_iter_new(root_node), lib.cmark_iter_free)
  while true do
    local event = lib.cmark_iter_next(iter)
    if event == lib.CMARK_EVENT_DONE then
      break
    end
    local node = lib.cmark_iter_get_node(iter)
    local node_type = lib.cmark_node_get_type(node)

    if event == lib.CMARK_EVENT_ENTER then
      local ast_node = {
        __type = ffi.string(lib.cmark_node_get_type_string(node)),
        _source_pos = {
          lib.cmark_node_get_start_line(node),
          lib.cmark_node_get_start_column(node),
          lib.cmark_node_get_end_line(node),
          lib.cmark_node_get_end_column(node),
        },
      }

      if
        node_type == lib.CMARK_NODE_TEXT
        or node_type == lib.CMARK_NODE_CODE
        or node_type == lib.CMARK_NODE_HTML_BLOCK
        or node_type == lib.CMARK_NODE_HTML_INLINE
        or node_type == lib.CMARK_NODE_FOOTNOTE_REFERENCE
      then
        ast_node._literal = ffi.string(lib.cmark_node_get_literal(node))

        --
      elseif node_type == lib.CMARK_NODE_HEADING then
        ast_node._heading_level = lib.cmark_node_get_heading_level(node)

        --
      elseif node_type == lib.CMARK_NODE_LIST then
        local list_type = lib.cmark_node_get_list_type(node)
        if list_type == lib.CMARK_BULLET_LIST then
          ast_node._list_type = 'bullet'
        elseif list_type == lib.CMARK_ORDERED_LIST then
          ast_node._list_type = 'ordered'
          ast_node._list_start = lib.cmark_node_get_list_start(node)
          local list_delim = lib.cmark_node_get_list_delim(node)
          if list_delim == lib.CMARK_PAREN_DELIM then
            ast_node._list_delim = 'paren'
          elseif list_delim == lib.CMARK_PERIOD_DELIM then
            ast_node._list_delim = 'period'
          end
          ast_node._list_start = lib.cmark_node_get_list_start(node)
        end
        ast_node._list_tight = lib.cmark_node_get_list_tight(node) ~= 0

        --
      elseif node_type == lib.CMARK_NODE_CODE_BLOCK then
        ast_node._fence_info = ffi.string(lib.cmark_node_get_fence_info(node))
        ast_node._literal = ffi.string(lib.cmark_node_get_literal(node))
        local out_len, out_offset, out_char =
          ffi.new('int[1]'), ffi.new('int[1]'), ffi.new('char[1]')
        lib.cmark_node_get_fenced(node, out_len, out_offset, out_char)
        ast_node._fence_length, ast_node._fence_offset = out_len[0], out_offset[0]
        ast_node._fence_char = string.char(out_char[0])

        --
      elseif
        node_type == lib.CMARK_NODE_CUSTOM_BLOCK or node_type == lib.CMARK_NODE_CUSTOM_INLINE
      then
        ast_node._on_enter = ffi.string(lib.cmark_node_get_on_enter(node))
        ast_node._on_exit = ffi.string(lib.cmark_node_get_on_exit(node))

        --
      elseif node_type == lib.CMARK_NODE_LINK or node_type == lib.CMARK_NODE_IMAGE then
        ast_node._title = ffi.string(lib.cmark_node_get_url(node))
        ast_node._url = ffi.string(lib.cmark_node_get_url(node))

        --
      end

      if ast_root_node == nil then
        ast_root_node = ast_node
      else
        local parent_ast_node = ast_nodes_stack[#ast_nodes_stack]
        table.insert(parent_ast_node.children, ast_node)
      end

      if lib.cmark_node_first_child(node) ~= nil then
        ast_node.children = {}
        table.insert(ast_nodes_stack, ast_node)
      end
    elseif event == lib.CMARK_EVENT_EXIT then
      if lib.cmark_node_first_child(node) ~= nil then
        table.remove(ast_nodes_stack)
      end
    end
  end

  assert(#ast_nodes_stack == 0)
  return ast_root_node
end

return M