open StdLabels

type t = float * string

module CharSet = Set.Make (Char)

let compare_words : string -> ref:string * CharSet.t -> Validity.t array option
    =
 fun w1 ~ref ->
  let wordRef = fst ref in
  let l1 = String.length w1 in
  if l1 <> String.length wordRef
  then None
  else
    let result =
      Array.init l1 ~f:(fun i ->
          let c1 = String.get w1 i
          and c2 = String.get wordRef i in

          let state =
            if Char.equal c1 c2
            then Validity.Wellplaced
            else if CharSet.mem c1 (snd ref)
            then Validity.Misplaced
            else Validity.Missing
          in
          state )
    in
    Some result


let get_entropy max_element words_number arr =
  let entropy = ref 0. in
  for idx = 0 to max_element - 1 do
    let content = Float.of_int (Bigarray.Array1.get arr idx) in
    if content > 0.
    then
      let ratio = content /. words_number in
      entropy := !entropy -. (ratio *. Float.log2 ratio)
  done;
  entropy


let analyse : int -> catalog:Wordlist.t -> Wordlist.t -> t =
 fun base ~catalog words ->
  let max_element = Float.to_int @@ (Validity.elements ** Float.of_int base) in
  let words_number = Float.of_int (Wordlist.list_size catalog) in

  let arr = Bigarray.Array1.create Bigarray.Int Bigarray.C_layout max_element in
  match Wordlist.pick words with
  | None -> (0., "")
  | Some v ->
      (* Build the array  *)
      Seq.fold_left
        (fun (score, word) word_ref ->
          (* Reinitialize the array (we use the same in the successive
             iterations *)
          let set_ref = String.to_seq word_ref |> CharSet.of_seq in

          Bigarray.Array1.fill arr 0;

          Seq.iter
            (fun w2 ->
              let result = compare_words ~ref:(word_ref, set_ref) w2 in
              match result with
              | None -> ()
              | Some r ->
                  let idx = Validity.index_of_result r in

                  let content = Bigarray.Array1.get arr idx in
                  Bigarray.Array1.set arr idx (succ content) )
            (Wordlist.words words);

          (* Now evaluate the entropy in the array *)
          let entropy = get_entropy max_element words_number arr in

          if !entropy > score
          then (
            Printf.printf "Entropy for selecting %s : %.2f\n" word_ref !entropy;
            (!entropy, word_ref) )
          else (score, word) )
        (-0., v)
        (Wordlist.words catalog)