init

5 files changed, 375 insertions, 0 deletions

diff --git a/sorsyl/lib/dune b/sorsyl/lib/dune
new file mode 100644
index 0000000..148997f
--- /dev/null
+++ b/sorsyl/lib/dune
@@ -0,0 +1,3 @@
+(library
+ (name sorsyl)
+ (libraries csv base stdio))
diff --git a/sorsyl/lib/feature.ml b/sorsyl/lib/feature.ml
new file mode 100644
index 0000000..280977b
--- /dev/null
+++ b/sorsyl/lib/feature.ml
@@ -0,0 +1,80 @@
+(** Module for handling phonological features and segments *)
+
+(* module Feature = struct *)
+
+(** Type representing a phonological feature value *)
+type feature_value = Plus | Minus | Zero
+
+(** Type representing a phonological feature *)
+type feature =
+  | Syllabic
+  | Sonorant
+  | Consonantal
+  | Continuant
+  | DelayedRelease
+  | Lateral
+  | Nasal
+  | Strident
+  | Voiced
+  | SpreadGlottis
+  | ConstrictedGlottis
+  | Anterior
+  | Coronal
+  | Distributed
+  | Labial
+  | High
+  | Low
+  | Back
+  | Rounded
+  | Velaric
+  | Tense
+  | Long
+  | HighTone
+  | HighReg (*high registry?*)
+
+type feature_spec = feature * feature_value
+(** Type representing a feature specification as a (value, feature) pair *)
+
+type segment = feature_spec list
+
+(** Convert a string feature value to the feature_value type *)
+let value_of_string = function
+  | "+" -> Plus
+  | "-" -> Minus
+  | "0" -> Zero
+  | s -> failwith (Printf.sprintf "Invalid feature value: %s" s)
+
+let string_of_feature = function
+  | "syl" -> Syllabic
+  | "son" -> Sonorant
+  | "cons" -> Consonantal
+  | "cont" -> Continuant
+  | "delrel" -> DelayedRelease
+  | "lat" -> Lateral
+  | "nas" -> Nasal
+  | "strid" -> Strident
+  | "voi" -> Voiced
+  | "sg" -> SpreadGlottis
+  | "cg" -> ConstrictedGlottis
+  | "ant" -> Anterior
+  | "cor" -> Coronal
+  | "distr" -> Distributed
+  | "lab" -> Labial
+  | "hi" -> High
+  | "lo" -> Low
+  | "back" -> Back
+  | "round" -> Rounded
+  | "velaric" -> Velaric
+  | "tense" -> Tense
+  | "long" -> Long
+  | "hitone" -> HighTone
+  | "hireg" -> HighReg
+  | _ -> failwith "not a valid feature"
+
+(** Check if a segment has a specific feature with a given value *)
+let has_feature (value, feature_name) segment =
+  List.exists (fun (v, f) -> v = value && f = feature_name) segment
+
+(** Create a feature test function for use in the decision tree *)
+let test feature_spec segment = has_feature feature_spec segment
+(* end *)
diff --git a/sorsyl/lib/ipa_table.ml b/sorsyl/lib/ipa_table.ml
new file mode 100644
index 0000000..bee027a
--- /dev/null
+++ b/sorsyl/lib/ipa_table.ml
@@ -0,0 +1,85 @@
+(** Type representing a segment as a set of feature specifications *)
+(* an association list I guess . Use List.assoc to handle*)
+
+(** Decision tree for computing sonority values *)
+type bool_tree =
+  | Leaf of int  (** Terminal node with sonority value *)
+  | Node of {
+      test : Feature.segment -> bool;  (** Test function *)
+      t_branch : bool_tree;  (** Branch to follow if test is true *)
+      f_branch : bool_tree;  (** Branch to follow if test is false *)
+    }
+
+type ipa_entry = { ipa : string; features : Feature.segment }
+(** Type representing a row from the IPA CSV file *)
+
+(** Storage for loaded IPA data *)
+let ipa_table : (string, Feature.segment) Hashtbl.t = Hashtbl.create 1000
+
+(** Parse a single row from the CSV file *)
+let parse_row (row : string list) : ipa_entry option =
+  match row with
+  | [] -> None
+  | ipa :: features -> (
+      let feature_names =
+        [
+          "syl";
+          "son";
+          "cons";
+          "cont";
+          "delrel";
+          "lat";
+          "nas";
+          "strid";
+          "voi";
+          "sg";
+          "cg";
+          "ant";
+          "cor";
+          "distr";
+          "lab";
+          "hi";
+          "lo";
+          "back";
+          "round";
+          "velaric";
+          "tense";
+          "long";
+          "hitone";
+          "hireg";
+        ]
+      in
+      (* Skip the header row *)
+      if ipa = "ipa" then None
+      else
+        let rec build_features names values acc =
+          match (names, values) with
+          | [], [] -> Some (List.rev acc)
+          | name :: ns, value :: vs ->
+              let fval = Feature.value_of_string value in
+              let fname = Feature.string_of_feature name in
+              build_features ns vs ((fname, fval) :: acc)
+          | _ -> None (* Mismatched lengths *)
+        in
+        match build_features feature_names features [] with
+        | Some feature_list -> Some { ipa; features = feature_list }
+        | None -> None)
+
+(** Load IPA data from CSV file *)
+let load_csv filename =
+  let ic = open_in filename in
+  let csv = Csv.of_channel ic in
+  try
+    Csv.iter
+      ~f:(fun row ->
+        match parse_row row with
+        | Some entry -> Hashtbl.add ipa_table entry.ipa entry.features
+        | None -> ())
+      csv;
+    close_in ic
+  with e ->
+    close_in ic;
+    raise e
+
+(** Look up features for an IPA segment *)
+let lookup_segment ipa = Hashtbl.find_opt ipa_table ipa
diff --git a/sorsyl/lib/sonority.ml b/sorsyl/lib/sonority.ml
new file mode 100644
index 0000000..90bfa55
--- /dev/null
+++ b/sorsyl/lib/sonority.ml
@@ -0,0 +1,160 @@
+(** 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) *)
+
+(** Decision tree for computing sonority values *)
+type bool_tree =
+  | Leaf of int  (** Terminal node with sonority value *)
+  | Node of {
+      test : Feature.segment -> bool;  (** Test function *)
+      t_branch : bool_tree;  (** Branch to follow if test is true *)
+      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
+
+    Node
+      {
+        test = plusSyl;
+        t_branch = minusHi_branch;
+        (* +syl = vowels *)
+        f_branch = minusCons_branch;
+        (* -syl = non-vowels *)
+      }
+
+  (** 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
+
+  (** The main decision tree instance *)
+  let sonority_tree = lazy (build_tree ())
+
+  (** Get sonority value from feature specifications *)
+  let sonority_from_features segment =
+    eval_tree (Lazy.force sonority_tree) segment
+
+  (** 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
+
+(** Public interface *)
+
+(** Initialize the sonority module with the data directory *)
+let init = Sonority.init
+
+(** Get the sonority value (1-9) for an IPA character *)
+let sonority = Sonority.sonority
+
+(** Get the sonority value from a feature specification *)
+let sonority_from_features = Sonority.sonority_from_features
diff --git a/sorsyl/lib/sonority.mli b/sorsyl/lib/sonority.mli
new file mode 100644
index 0000000..3e9166e
--- /dev/null
+++ b/sorsyl/lib/sonority.mli
@@ -0,0 +1,47 @@
+(** 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)
+
+    Example usage:
+    {[
+      (* Initialize the module with the data directory *)
+      Sonority.init "./data";;
+
+      (* Get sonority values for IPA segments *)
+      Sonority.sonority "a";;
+
+      (* Returns 9 - low vowel *)
+      Sonority.sonority "p";;
+
+      (* Returns 1 - voiceless stop *)
+      Sonority.sonority "l" (* Returns 6 - liquid *)
+    ]} *)
+
+val init : string -> unit
+(** Initialize the sonority module with the data directory. This must be called
+    before using other functions.
+    @param data_dir The directory containing the ipa_all.csv file
+    @raise Sys_error if the CSV file cannot be found or read *)
+
+val sonority : string -> int
+(** Get the sonority value (1-9) for an IPA character.
+    @param ipa The IPA character/segment to analyze
+    @return The sonority value between 1 and 9
+    @raise Failure if the IPA segment is not recognized *)
+
+val sonority_from_features : Feature.segment -> int
+(** Get the sonority value from a feature specification. This is useful when you
+    already have the phonological features of a segment and don't need to look
+    it up by IPA symbol.
+    @param segment The list of feature specifications
+    @return The sonority value between 1 and 9 *)