1 files changed, 55 insertions, 0 deletions
diff --git a/motus/lib/entropy.ml b/motus/lib/entropy.ml
new file mode 100644
index 0000000..5b86a9d
--- /dev/null
+++ b/motus/lib/entropy.ml
@@ -0,0 +1,55 @@
+type t = float * string
+
+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 -> Wordlist.t -> t =
+ fun base words ->
+  let max_element = Float.to_int @@ (Validity.elements ** Float.of_int base) in
+  let words_number = Float.of_int (Wordlist.list_size words) 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 |> Validity.CharSet.of_seq in
+
+          let arr =
+            Bigarray.Array1.create Bigarray.Int Bigarray.C_layout max_element
+          in
+
+          Seq.iter
+            (fun w2 ->
+              let result = Validity.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 words)