tia-lsp.nvim/lua/scl/udt_parser.lua

-- UDT Parser for SCL
-- Parses .udt files to extract type definitions and member structures

local M = {}

-- Cache for parsed UDTs: { [type_name] = udt_definition }
local udt_cache = {}

-- Inline UDT definitions (from comments): { [type_name] = udt_definition }
local inline_udts = {}

-- UDT Definition structure:
-- {
--   name = "TypeName",
--   title = "Type Title",
--   version = "0.1",
--   comment = "User comment",
--   members = {
--     {
--       name = "memberName",
--       type = "Bool" or '"OtherUdt"',
--       attributes = "{ S7_SetPoint := 'True' }",
--       comment = "Member comment",
--       is_udt = false -- true if type is a quoted UDT reference
--     }
--   }
-- }

-- Register an inline UDT definition (from comments or other sources)
function M.register_inline_udt(udt)
  if udt and udt.name then
    inline_udts[udt.name] = udt
    -- Also add to main cache for convenience
    udt_cache[udt.name] = udt
  end
end

-- Parse an inline UDT definition from a comment string
-- Format: // @udt TypeName
--         // members:
--         //   member1 : INT;
--         //   member2 : BOOL;
function M.parse_inline_udt_comment(comment_lines)
  local type_name = nil
  local members = {}

  -- Look for @udt tag
  for _, line in ipairs(comment_lines) do
    local name = line:match("@udt%s+(%w+)")
    if name then
      type_name = name
      break
    end
  end

  if not type_name then
    return nil
  end

  -- Parse member lines (lines that look like declarations)
  for _, line in ipairs(comment_lines) do
    local name, type_str = line:match("^%s*([%w_]+)%s*:%s*([^;]+)")
    if name and type_str then
      type_str = type_str:gsub("%s*:=.*$", ""):gsub("%s*$", "")
      local is_udt = type_str:match('^".+"$') ~= nil
      local clean_type = type_str:gsub('"', '')

      table.insert(members, {
        name = name,
        type = clean_type,
        raw_type = type_str,
        attributes = {},
        comment = "",
        is_udt = is_udt
      })
    end
  end

  if #members == 0 then
    return nil
  end

  return {
    name = type_name,
    title = "Inline UDT",
    version = "0.1",
    comment = "",
    members = members
  }
end

-- Clear the UDT cache - clear contents, don't replace table
function M.clear_cache()
  for k in pairs(udt_cache) do
    udt_cache[k] = nil
  end
  for k in pairs(inline_udts) do
    inline_udts[k] = nil
  end
end

-- Get cached UDT
function M.get_udt(type_name)
  return udt_cache[type_name]
end

-- Get all cached UDT names
function M.get_all_udt_names()
  local names = {}
  for name, _ in pairs(udt_cache) do
    table.insert(names, name)
  end
  return names
end

-- Get count of cached UDTs
function M.get_cache_count()
  local count = 0
  for _ in pairs(udt_cache) do
    count = count + 1
  end
  return count
end

-- Check if a type is a UDT (exists in cache or inline)
function M.is_udt_type(type_name)
  if not type_name then
    return false
  end
  -- Remove quotes if present
  local clean_name = type_name:gsub('"', '')
  return udt_cache[clean_name] ~= nil or inline_udts[clean_name] ~= nil
end

-- Parse attributes from a string like: { S7_SetPoint := 'True'; ExternalAccessible := 'False' }
local function parse_attributes(attr_str)
  if not attr_str or attr_str == "" then
    return {}
  end

  local attrs = {}
  -- Match key := 'value' or key := value patterns
  for key, value in attr_str:gmatch("([%w_]+)%s*:=%s*'([^']+)'") do
    attrs[key] = value
  end
  for key, value in attr_str:gmatch("([%w_]+)%s*:=%s*(%w+)") do
    attrs[key] = value
  end
  return attrs
end

-- Parse a single .udt file content
function M.parse_udt_content(content, filename)
  local udt = {
    filename = filename,
    members = {}
  }

  -- Extract TYPE name
  local type_name = content:match('TYPE%s+"([^"]+)"')
  if not type_name then
    return nil, "No TYPE definition found"
  end
  udt.name = type_name

  -- Extract TITLE
  udt.title = content:match('TITLE%s*=%s*([^\n]+)') or ""
  udt.title = udt.title:gsub("^%s*", ""):gsub("%s*$", "")

  -- Extract VERSION
  udt.version = content:match('VERSION%s*:%s*([0-9.]+)') or "0.1"

  -- Extract comment after VERSION (lines starting with // before STRUCT)
  local comment_lines = {}
  local in_header = true
  for line in content:gmatch("[^\r\n]+") do
    if in_header then
      local comment = line:match("^%s*//%s*(.+)$")
      if comment then
        table.insert(comment_lines, comment)
      elseif line:match("STRUCT") then
        in_header = false
      end
    end
  end
  udt.comment = table.concat(comment_lines, " ")

  -- Extract STRUCT content (between STRUCT and END_STRUCT)
  local struct_content = content:match("STRUCT%s*(.-)%s*END_STRUCT")
  if not struct_content then
    return nil, "No STRUCT definition found"
  end

  -- Parse members
  -- Pattern: memberName {attributes} : Type; // comment
  -- or: memberName : "UdtType"; // comment
  -- or: memberName : Type := default; // comment
  for line in struct_content:gmatch("[^\r\n]+") do
    line = line:gsub("^%s*", ""):gsub("%s*$", "")

    -- Skip empty lines and pure comments
    if line ~= "" and not line:match("^//") then
      -- Extract trailing comment
      local comment = line:match("//%s*(.+)$") or ""
      line = line:gsub("//.*$", ""):gsub("%s*$", "")

      -- Parse member declaration
      -- Pattern: name {attrs} : type [:= default];
      local name, attrs_str, type_str = line:match('^([%w_]+)%s*(%b{})%s*:%s*([^;]+)')
      if not name then
        -- Try without attributes
        name, type_str = line:match('^([%w_]+)%s*:%s*([^;]+)')
      end

      if name and type_str then
        type_str = type_str:gsub("%s*:=.*$", ""):gsub("%s*$", "") -- Remove default value

        local is_udt = type_str:match('^".+"$') ~= nil
        local clean_type = type_str:gsub('"', '')

        local member = {
          name = name,
          type = clean_type,
          raw_type = type_str,
          attributes = attrs_str and parse_attributes(attrs_str) or {},
          comment = comment,
          is_udt = is_udt
        }

        table.insert(udt.members, member)
      end
    end
  end

  return udt
end

-- Parse a .udt file
function M.parse_udt_file(filepath)
  local file = io.open(filepath, "r")
  if not file then
    return nil, "Cannot open file: " .. filepath
  end

  local content = file:read("*all")
  file:close()

  local udt, err = M.parse_udt_content(content, filepath)
  if not udt then
    return nil, err
  end

  -- Cache the UDT
  udt_cache[udt.name] = udt

  return udt
end

-- Get members of a UDT (with optional nesting)
function M.get_udt_members(type_name, visited)
  visited = visited or {}

  if not type_name then
    return {}
  end

  -- Remove quotes if present
  type_name = type_name:gsub('"', '')

  -- Prevent infinite recursion
  if visited[type_name] then
    return {}
  end
  visited[type_name] = true

  local udt = udt_cache[type_name]
  if not udt then
    return {}
  end

  local members = {}
  for _, member in ipairs(udt.members) do
    table.insert(members, member)

    -- If member is a UDT, include its members as nested structure
    if member.is_udt and member.type then
      member.nested_members = M.get_udt_members(member.type, visited)
    end
  end

  return members
end

-- Resolve a member path like "member1.member2.member3"
-- Returns: member_info, remaining_path
function M.resolve_member_path(type_name, path)
  if not type_name or not path then
    return nil, path or ""
  end

  local parts = {}
  for part in path:gmatch("[^.]+") do
    table.insert(parts, part)
  end

  local current_type = type_name:gsub('"', '')
  local current_member = nil

  for i, part in ipairs(parts) do
    local udt = udt_cache[current_type]
    if not udt then
      return nil, table.concat(parts, ".", i)
    end

    -- Find member
    local found = false
    for _, member in ipairs(udt.members) do
      if member.name == part then
        current_member = member
        if member.is_udt then
          current_type = member.type
        else
          -- Not a UDT, can't continue
          if i < #parts then
            return nil, table.concat(parts, ".", i + 1)
          end
        end
        found = true
        break
      end
    end

    if not found then
      return nil, table.concat(parts, ".", i)
    end
  end

  return current_member, ""
end

-- Get completion items for UDT members
function M.get_member_completions(type_name, prefix)
  local members = M.get_udt_members(type_name)
  local items = {}

  for _, member in ipairs(members) do
    -- Label shows the full path for display
    local label = prefix and (prefix .. "." .. member.name) or member.name
    -- But we only insert the member name (nvim-cmp will handle replacing the partial word)
    local item = {
      label = label,
      insertText = member.name,
      kind = member.is_udt and vim.lsp.protocol.CompletionItemKind.Struct or vim.lsp.protocol.CompletionItemKind.Field,
      detail = member.raw_type .. (member.comment ~= "" and " - " .. member.comment or ""),
      documentation = member.comment,
    }
    table.insert(items, item)
  end

  return items
end

-- Get completion items for available UDT types
function M.get_udt_type_completions(params)
  local items = {}

  for name, udt in pairs(udt_cache) do
    local quoted_name = '"' .. name .. '"'
    local item = {
      label = quoted_name,
      kind = vim.lsp.protocol.CompletionItemKind.Struct,
      detail = udt.title ~= "" and udt.title or "UDT",
      documentation = udt.comment ~= "" and udt.comment or nil,
    }

    -- Use textEdit for better compatibility with nvim-cmp
    if params then
      local line = params.context.cursor_before_line
      local col = params.context.cursor.col

      -- Find position to insert (after the opening quote)
      local start_col = col
      -- Search backwards for opening quote or colon
      for i = col, 1, -1 do
        local char = line:sub(i, i)
        if char == '"' then
          start_col = i - 1
          break
        elseif char == ":" then
          -- After colon, we want to replace from opening quote
          start_col = col - 1
          break
        end
      end

      item.textEdit = {
        newText = quoted_name,
        range = {
          start = { line = params.context.cursor.line, character = start_col },
          ["end"] = { line = params.context.cursor.line, character = col },
        },
      }
    else
      -- Fallback to insertText if params not available
      item.insertText = quoted_name
    end

    table.insert(items, item)
  end

  return items
end

-- Scan buffer for inline UDT definitions (comments with @udt tag)
function M.scan_buffer_for_inline_udts(bufnr)
  bufnr = bufnr or vim.api.nvim_get_current_buf()
  local lines = vim.api.nvim_buf_get_lines(bufnr, 0, -1, false)

  local current_comment_block = {}
  local in_scl_file = vim.bo[bufnr].filetype == "scl"

  if not in_scl_file then
    return
  end

  for _, line in ipairs(lines) do
    if line:match("^%s*//") then
      -- It's a comment line
      local content = line:gsub("^%s*//%s*", "")
      table.insert(current_comment_block, content)
    else
      -- Non-comment line - check if we have a complete UDT definition
      if #current_comment_block > 0 then
        local udt = M.parse_inline_udt_comment(current_comment_block)
        if udt then
          M.register_inline_udt(udt)
        end
        current_comment_block = {}
      end
    end
  end

  -- Check final block
  if #current_comment_block > 0 then
    local udt = M.parse_inline_udt_comment(current_comment_block)
    if udt then
      M.register_inline_udt(udt)
    end
  end
end

-- Debug: Print cache contents
function M.debug_print_cache()
  print("=== UDT Cache Contents ===")
  print("Total UDTs: " .. M.get_cache_count())
  for name, udt in pairs(udt_cache) do
    print(string.format("UDT: %s (v%s)", name, udt.version))
    print(string.format("  Title: %s", udt.title))
    print(string.format("  File: %s", udt.filename or "unknown"))
    print(string.format("  Members: %d", #udt.members))
    for _, member in ipairs(udt.members) do
      print(string.format("    - %s: %s%s",
        member.name,
        member.raw_type,
        member.is_udt and " (UDT)" or ""))
    end
    print("")
  end
end

return M