(* Encoding label normalization per WHATWG Encoding Standard *)

let normalize_label label =
  if String.length label = 0 then None
  else
    let s = Astring.String.Ascii.lowercase (Astring.String.trim label) in
    if String.length s = 0 then None
    else
      (* Security: never allow utf-7 *)
      if s = "utf-7" || s = "utf7" || s = "x-utf-7" then
        Some Encoding_types.Windows_1252
      else if s = "utf-8" || s = "utf8" then
        Some Encoding_types.Utf8
      (* HTML treats latin-1 labels as windows-1252 *)
      else if s = "iso-8859-1" || s = "iso8859-1" || s = "latin1" ||
              s = "latin-1" || s = "l1" || s = "cp819" || s = "ibm819" then
        Some Encoding_types.Windows_1252
      else if s = "windows-1252" || s = "windows1252" || s = "cp1252" || s = "x-cp1252" then
        Some Encoding_types.Windows_1252
      else if s = "iso-8859-2" || s = "iso8859-2" || s = "latin2" || s = "latin-2" then
        Some Encoding_types.Iso_8859_2
      else if s = "euc-jp" || s = "eucjp" then
        Some Encoding_types.Euc_jp
      else if s = "utf-16" || s = "utf16" then
        Some Encoding_types.Utf16le  (* Default to LE for ambiguous utf-16 *)
      else if s = "utf-16le" || s = "utf16le" then
        Some Encoding_types.Utf16le
      else if s = "utf-16be" || s = "utf16be" then
        Some Encoding_types.Utf16be
      else
        None

let normalize_meta_declared label =
  match normalize_label label with
  | None -> None
  | Some enc ->
    (* Per HTML meta charset handling: ignore UTF-16/UTF-32 declarations and
       treat them as UTF-8 *)
    match enc with
    | Encoding_types.Utf16le | Encoding_types.Utf16be -> Some Encoding_types.Utf8
    | other -> Some other