From 3235260a3dca98e96ab50458a5daf3baf3f238d0 Mon Sep 17 00:00:00 2001 From: Sébastien Dailly Date: Sun, 27 Feb 2022 10:31:27 +0100 Subject: Do not evaluate the entropy when only 2 words are still presents --- motus/lib/entropy.ml | 75 ++++++++++++++++++++++++++++++---------------------- 1 file changed, 43 insertions(+), 32 deletions(-) (limited to 'motus/lib') diff --git a/motus/lib/entropy.ml b/motus/lib/entropy.ml index 043d0c8..aec5d86 100644 --- a/motus/lib/entropy.ml +++ b/motus/lib/entropy.ml @@ -43,41 +43,52 @@ let get_entropy max_element words_number arr = 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 + (* If we have only two elements, just pick one of them *) + if Wordlist.list_size words <= 2 + then (1.0, Option.get @@ Wordlist.pick words) + else + 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 + 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; + 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 + 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); + 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 + (* 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) + 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) -- cgit v1.2.3