1 files changed, 114 insertions, 136 deletions
diff --git a/sorsyl/lib/sonority.ml b/sorsyl/lib/sonority.ml
index 90bfa55..c47d4a0 100644
--- a/sorsyl/lib/sonority.ml
+++ b/sorsyl/lib/sonority.ml
@@ -1,16 +1,4 @@
-(** Sonority module for determining the sonority of phonetic segments.
-
-    This module provides functionality to determine the sonority of IPA
-    (International Phonetic Alphabet) segments on a scale of 1 to 9, where:
-    - 9: Low vowels (most sonorous)
-    - 8: High vowels
-    - 7: Glides/approximants
-    - 6: Liquids
-    - 5: Nasals
-    - 4: Voiced fricatives
-    - 3: Voiceless fricatives
-    - 2: Voiced stops
-    - 1: Voiceless stops (least sonorous) *)
+(** Functional sonority module without global state *)
 
 (** Decision tree for computing sonority values *)
 type bool_tree =
@@ -21,140 +9,130 @@ type bool_tree =
       f_branch : bool_tree;  (** Branch to follow if test is false *)
     }
 
-(** Main Sonority module functionality *)
-module Sonority = struct
-  (** Initialize the module by loading IPA data *)
-  let init data_dir =
-    let csv_file = Filename.concat data_dir "ipa_all.csv" in
-    Ipa_table.load_csv csv_file
-
-  (** Build the decision tree for sonority calculation *)
-  let build_tree () =
-    let open Feature in
-    let plusSyl = test (Syllabic, Plus) in
-    let minusHi = test (High, Minus) in
-    let minusCons = test (Consonantal, Minus) in
-    let plusSon = test (Sonorant, Plus) in
-    let minusNas = test (Nasal, Minus) in
-    let plusCont = test (Continuant, Plus) in
-    let plusVoi = test (Voiced, Plus) in
-
-    (* Build the tree bottom-up *)
-    let minusHi_branch =
-      Node
-        {
-          test = minusHi;
-          t_branch = Leaf 9;
-          (* -hi vowels = low vowels *)
-          f_branch = Leaf 8;
-          (* +hi vowels = high vowels *)
-        }
-    in
-
-    let plusVoi1_branch =
-      Node
-        {
-          test = plusVoi;
-          t_branch = Leaf 4;
-          (* +voi +cont = voiced fricatives *)
-          f_branch = Leaf 3;
-          (* -voi +cont = voiceless fricatives *)
-        }
-    in
-
-    let plusVoi2_branch =
-      Node
-        {
-          test = plusVoi;
-          t_branch = Leaf 2;
-          (* +voi -cont = voiced stops *)
-          f_branch = Leaf 1;
-          (* -voi -cont = voiceless stops *)
-        }
-    in
-
-    let plusCont_branch =
-      Node
-        {
-          test = plusCont;
-          t_branch = plusVoi1_branch;
-          (* +cont = fricatives *)
-          f_branch = plusVoi2_branch;
-          (* -cont = stops *)
-        }
-    in
-
-    let minusNas_branch =
-      Node
-        {
-          test = minusNas;
-          t_branch = Leaf 6;
-          (* -nas +son = liquids *)
-          f_branch = Leaf 5;
-          (* +nas +son = nasals *)
-        }
-    in
-
-    let plusSon_branch =
-      Node
-        {
-          test = plusSon;
-          t_branch = minusNas_branch;
-          (* +son = sonorants *)
-          f_branch = plusCont_branch;
-          (* -son = obstruents *)
-        }
-    in
-
-    let minusCons_branch =
-      Node
-        {
-          test = minusCons;
-          t_branch = Leaf 7;
-          (* -cons = glides *)
-          f_branch = plusSon_branch;
-          (* +cons = true consonants *)
-        }
-    in
+type t = { ipa_table : Ipa_table.t; decision_tree : bool_tree }
+(** Type representing a sonority calculator *)
+
+(** Build the decision tree for sonority calculation *)
+let build_tree () =
+  let open Feature in
+  let plusSyl = test (Syllabic, Plus) in
+  let minusHi = test (High, Minus) in
+  let minusCons = test (Consonantal, Minus) in
+  let plusSon = test (Sonorant, Plus) in
+  let minusNas = test (Nasal, Minus) in
+  let plusCont = test (Continuant, Plus) in
+  let plusVoi = test (Voiced, Plus) in
+
+  (* Build the tree bottom-up, matching the Python original exactly *)
+  let minusHi_branch =
+    Node
+      {
+        test = minusHi;
+        t_branch = Leaf 9;
+        (* -hi vowels = low vowels *)
+        f_branch = Leaf 8;
+        (* +hi vowels = high vowels *)
+      }
+  in
 
+  let plusVoi1_branch =
     Node
       {
-        test = plusSyl;
-        t_branch = minusHi_branch;
-        (* +syl = vowels *)
-        f_branch = minusCons_branch;
-        (* -syl = non-vowels *)
+        test = plusVoi;
+        t_branch = Leaf 4;
+        (* +voi +cont = voiced fricatives *)
+        f_branch = Leaf 3;
+        (* -voi +cont = voiceless fricatives *)
       }
+  in
 
-  (** Evaluate the decision tree for a segment *)
-  let rec eval_tree tree segment =
-    match tree with
-    | Leaf value -> value
-    | Node { test; t_branch; f_branch } ->
-        if test segment then eval_tree t_branch segment
-        else eval_tree f_branch segment
+  let plusVoi2_branch =
+    Node
+      {
+        test = plusVoi;
+        t_branch = Leaf 2;
+        (* +voi -cont = voiced stops *)
+        f_branch = Leaf 1;
+        (* -voi -cont = voiceless stops *)
+      }
+  in
 
-  (** The main decision tree instance *)
-  let sonority_tree = lazy (build_tree ())
+  let plusCont_branch =
+    Node
+      {
+        test = plusCont;
+        t_branch = plusVoi1_branch;
+        (* +cont = fricatives *)
+        f_branch = plusVoi2_branch;
+        (* -cont = stops *)
+      }
+  in
 
-  (** Get sonority value from feature specifications *)
-  let sonority_from_features segment =
-    eval_tree (Lazy.force sonority_tree) segment
+  let minusNas_branch =
+    Node
+      {
+        test = minusNas;
+        t_branch = Leaf 6;
+        (* -nas +son = liquids *)
+        f_branch = Leaf 5;
+        (* +nas +son = nasals *)
+      }
+  in
+
+  let plusSon_branch =
+    Node
+      {
+        test = plusSon;
+        t_branch = minusNas_branch;
+        (* +son = sonorants *)
+        f_branch = plusCont_branch;
+        (* -son = obstruents *)
+      }
+  in
 
-  (** Get sonority value from an IPA character *)
-  let sonority ipa =
-    match Ipa_table.lookup_segment ipa with
-    | Some features -> sonority_from_features features
-    | None -> failwith (Printf.sprintf "Unknown IPA segment: %s" ipa)
-end
+  let minusCons_branch =
+    Node
+      {
+        test = minusCons;
+        t_branch = Leaf 7;
+        (* -cons = glides *)
+        f_branch = plusSon_branch;
+        (* +cons = true consonants *)
+      }
+  in
+
+  Node
+    {
+      test = plusSyl;
+      t_branch = minusHi_branch;
+      (* +syl = vowels *)
+      f_branch = minusCons_branch;
+      (* -syl = non-vowels *)
+    }
 
-(** Public interface *)
+(** Create a sonority calculator from data directory *)
+let create (data_dir : string) : t =
+  let ipa_table = Ipa_table.load_csv data_dir in
+  let decision_tree = build_tree () in
+  { ipa_table; decision_tree }
 
-(** Initialize the sonority module with the data directory *)
-let init = Sonority.init
+(** Traverse the decision tree to get sonority value *)
+let rec traverse_tree (tree : bool_tree) (segment : Feature.segment) : int =
+  match tree with
+  | Leaf value -> value
+  | Node { test; t_branch; f_branch } ->
+      if test segment then traverse_tree t_branch segment
+      else traverse_tree f_branch segment
 
 (** Get the sonority value (1-9) for an IPA character *)
-let sonority = Sonority.sonority
+let sonority (calc : t) (ipa : string) : int =
+  match Ipa_table.lookup_segment calc.ipa_table.table ipa with
+  | Some features -> traverse_tree calc.decision_tree features
+  | None -> failwith (Printf.sprintf "Unknown IPA segment: %s" ipa)
 
 (** Get the sonority value from a feature specification *)
-let sonority_from_features = Sonority.sonority_from_features
+let sonority_from_features (calc : t) (segment : Feature.segment) : int =
+  traverse_tree calc.decision_tree segment
+
+(** Get the underlying IPA table *)
+let get_ipa_table (calc : t) : Ipa_table.t = calc.ipa_table