getting there

4 files changed, 388 insertions, 21 deletions

diff --git a/sorsyl/test/dune b/sorsyl/test/dune
index 701e92d..e79f200 100644
--- a/sorsyl/test/dune
+++ b/sorsyl/test/dune
@@ -1,3 +1,11 @@
+; (test
+; (name test_sonorityold)
+; (libraries sorsyl))
+
 (test
  (name test_sonority)
  (libraries sorsyl))
+
+(test
+ (name test_table)
+ (libraries sorsyl))
diff --git a/sorsyl/test/test_sonority.ml b/sorsyl/test/test_sonority.ml
index 70845a6..c24f4b4 100644
--- a/sorsyl/test/test_sonority.ml
+++ b/sorsyl/test/test_sonority.ml
@@ -1,11 +1,9 @@
-(** Tests for the Sonority module *)
+(** Tests for the functional Sonority module *)
 
 open Sorsyl
 
-(** Test fixture - initialize the module once *)
-let () =
-  (* Initialize with the data directory *)
-  (* When run with dune test, the working directory is _build/default/test *)
+(** Test fixture - create the sonority calculator once *)
+let sonority_calc =
   let data_dir =
     if Sys.file_exists "./data" then "./data"
     else if Sys.file_exists "../data" then "../data"
@@ -14,13 +12,13 @@ let () =
       Printf.eprintf "Current directory: %s\n" (Sys.getcwd ());
       failwith "Cannot find data directory")
   in
-  Sonority.init data_dir
+  Sonority.create data_dir
 
 (** Test sonority value 9 - Low vowels *)
 let test_sonority_nine () =
   let segments = [ "a"; "ɑ"; "æ"; "ɒ"; "e"; "o̥" ] in
   let expected = [ 9; 9; 9; 9; 9; 9 ] in
-  let results = List.map Sonority.sonority segments in
+  let results = List.map (Sonority.sonority sonority_calc) segments in
   assert (results = expected);
   Printf.printf "test_sonority_nine: PASSED\n"
 
@@ -28,7 +26,7 @@ let test_sonority_nine () =
 let test_sonority_eight () =
   let segments = [ "i"; "y"; "ɨ"; "ʉ"; "ɯ"; "u" ] in
   let expected = [ 8; 8; 8; 8; 8; 8 ] in
-  let results = List.map Sonority.sonority segments in
+  let results = List.map (Sonority.sonority sonority_calc) segments in
   assert (results = expected);
   Printf.printf "test_sonority_eight: PASSED\n"
 
@@ -36,7 +34,7 @@ let test_sonority_eight () =
 let test_sonority_seven () =
   let segments = [ "j"; "w"; "ʋ"; "ɰ"; "ɹ"; "e̯" ] in
   let expected = [ 7; 7; 7; 7; 7; 7 ] in
-  let results = List.map Sonority.sonority segments in
+  let results = List.map (Sonority.sonority sonority_calc) segments in
   assert (results = expected);
   Printf.printf "test_sonority_seven: PASSED\n"
 
@@ -44,7 +42,7 @@ let test_sonority_seven () =
 let test_sonority_six () =
   let segments = [ "l"; "ɭ"; "r"; "ɾ" ] in
   let expected = [ 6; 6; 6; 6 ] in
-  let results = List.map Sonority.sonority segments in
+  let results = List.map (Sonority.sonority sonority_calc) segments in
   assert (results = expected);
   Printf.printf "test_sonority_six: PASSED\n"
 
@@ -52,7 +50,7 @@ let test_sonority_six () =
 let test_sonority_five () =
   let segments = [ "n"; "m"; "ŋ"; "ɴ" ] in
   let expected = [ 5; 5; 5; 5 ] in
-  let results = List.map Sonority.sonority segments in
+  let results = List.map (Sonority.sonority sonority_calc) segments in
   assert (results = expected);
   Printf.printf "test_sonority_five: PASSED\n"
 
@@ -60,15 +58,18 @@ let test_sonority_five () =
 let test_sonority_four () =
   let segments = [ "v"; "z"; "ʒ"; "ɣ" ] in
   let expected = [ 4; 4; 4; 4 ] in
-  let results = List.map Sonority.sonority segments in
+  let results = List.map (Sonority.sonority sonority_calc) segments in
+  let results_string =
+    List.fold_left (fun acc item -> Printf.sprintf "%s-%d" acc item) "" results
+  in
   assert (results = expected);
-  Printf.printf "test_sonority_four: PASSED\n"
+  Printf.printf "test_sonority_four: %s\nPASSED\n" results_string
 
 (** Test sonority value 3 - Voiceless fricatives *)
 let test_sonority_three () =
   let segments = [ "f"; "s"; "x"; "ħ"; "ʃ" ] in
   let expected = [ 3; 3; 3; 3; 3 ] in
-  let results = List.map Sonority.sonority segments in
+  let results = List.map (Sonority.sonority sonority_calc) segments in
   assert (results = expected);
   Printf.printf "test_sonority_three: PASSED\n"
 
@@ -76,7 +77,7 @@ let test_sonority_three () =
 let test_sonority_two () =
   let segments = [ "b"; "ɡ"; "d"; "ɢ" ] in
   let expected = [ 2; 2; 2; 2 ] in
-  let results = List.map Sonority.sonority segments in
+  let results = List.map (Sonority.sonority sonority_calc) segments in
   assert (results = expected);
   Printf.printf "test_sonority_two: PASSED\n"
 
@@ -84,14 +85,14 @@ let test_sonority_two () =
 let test_sonority_one () =
   let segments = [ "p"; "k"; "c"; "q" ] in
   let expected = [ 1; 1; 1; 1 ] in
-  let results = List.map Sonority.sonority segments in
+  let results = List.map (Sonority.sonority sonority_calc) segments in
   assert (results = expected);
   Printf.printf "test_sonority_one: PASSED\n"
 
 (** Test unknown segment handling *)
 let test_unknown_segment () =
   try
-    let _ = Sonority.sonority "🦆" in
+    let _ = Sonority.sonority sonority_calc "🦆" in
     assert false (* Should not reach here *)
   with
   | Failure msg when String.sub msg 0 20 = "Unknown IPA segment:" ->
@@ -110,14 +111,30 @@ let test_sonority_from_features () =
       (Feature.Voiced, Feature.Minus);
     ]
   in
-  let result = Sonority.sonority_from_features segment in
+  let result = Sonority.sonority_from_features sonority_calc segment in
   assert (result = 1);
   Printf.printf "test_sonority_from_features: PASSED\n"
 
+(** Test that we can create multiple calculators (no global state) *)
+let test_multiple_calculators () =
+  let data_dir =
+    if Sys.file_exists "./data" then "./data"
+    else if Sys.file_exists "../data" then "../data"
+    else if Sys.file_exists "../../../data" then "../../../data"
+    else failwith "Cannot find data directory"
+  in
+  let calc1 = Sonority.create data_dir in
+  let calc2 = Sonority.create data_dir in
+
+  (* Both should work independently *)
+  assert (Sonority.sonority calc1 "a" = 9);
+  assert (Sonority.sonority calc2 "a" = 9);
+  Printf.printf "test_multiple_calculators: PASSED\n"
+
 (** Run all tests *)
 let () =
   Printf.printf "Running Sonority module tests...\n";
-  Printf.printf "================================\n";
+  Printf.printf "===================================\n";
   test_sonority_nine ();
   test_sonority_eight ();
   test_sonority_seven ();
@@ -129,5 +146,6 @@ let () =
   test_sonority_one ();
   test_unknown_segment ();
   test_sonority_from_features ();
-  Printf.printf "================================\n";
-  Printf.printf "All tests passed!\n"
+  test_multiple_calculators ();
+  Printf.printf "===================================\n";
+  Printf.printf "All Sonority tests passed!\n"
diff --git a/sorsyl/test/test_sonorityold.ml b/sorsyl/test/test_sonorityold.ml
new file mode 100644
index 0000000..70845a6
--- /dev/null
+++ b/sorsyl/test/test_sonorityold.ml
@@ -0,0 +1,133 @@
+(** Tests for the Sonority module *)
+
+open Sorsyl
+
+(** Test fixture - initialize the module once *)
+let () =
+  (* Initialize with the data directory *)
+  (* When run with dune test, the working directory is _build/default/test *)
+  let data_dir =
+    if Sys.file_exists "./data" then "./data"
+    else if Sys.file_exists "../data" then "../data"
+    else if Sys.file_exists "../../../data" then "../../../data"
+    else (
+      Printf.eprintf "Current directory: %s\n" (Sys.getcwd ());
+      failwith "Cannot find data directory")
+  in
+  Sonority.init data_dir
+
+(** Test sonority value 9 - Low vowels *)
+let test_sonority_nine () =
+  let segments = [ "a"; "ɑ"; "æ"; "ɒ"; "e"; "o̥" ] in
+  let expected = [ 9; 9; 9; 9; 9; 9 ] in
+  let results = List.map Sonority.sonority segments in
+  assert (results = expected);
+  Printf.printf "test_sonority_nine: PASSED\n"
+
+(** Test sonority value 8 - High vowels *)
+let test_sonority_eight () =
+  let segments = [ "i"; "y"; "ɨ"; "ʉ"; "ɯ"; "u" ] in
+  let expected = [ 8; 8; 8; 8; 8; 8 ] in
+  let results = List.map Sonority.sonority segments in
+  assert (results = expected);
+  Printf.printf "test_sonority_eight: PASSED\n"
+
+(** Test sonority value 7 - Glides/approximants *)
+let test_sonority_seven () =
+  let segments = [ "j"; "w"; "ʋ"; "ɰ"; "ɹ"; "e̯" ] in
+  let expected = [ 7; 7; 7; 7; 7; 7 ] in
+  let results = List.map Sonority.sonority segments in
+  assert (results = expected);
+  Printf.printf "test_sonority_seven: PASSED\n"
+
+(** Test sonority value 6 - Liquids *)
+let test_sonority_six () =
+  let segments = [ "l"; "ɭ"; "r"; "ɾ" ] in
+  let expected = [ 6; 6; 6; 6 ] in
+  let results = List.map Sonority.sonority segments in
+  assert (results = expected);
+  Printf.printf "test_sonority_six: PASSED\n"
+
+(** Test sonority value 5 - Nasals *)
+let test_sonority_five () =
+  let segments = [ "n"; "m"; "ŋ"; "ɴ" ] in
+  let expected = [ 5; 5; 5; 5 ] in
+  let results = List.map Sonority.sonority segments in
+  assert (results = expected);
+  Printf.printf "test_sonority_five: PASSED\n"
+
+(** Test sonority value 4 - Voiced fricatives *)
+let test_sonority_four () =
+  let segments = [ "v"; "z"; "ʒ"; "ɣ" ] in
+  let expected = [ 4; 4; 4; 4 ] in
+  let results = List.map Sonority.sonority segments in
+  assert (results = expected);
+  Printf.printf "test_sonority_four: PASSED\n"
+
+(** Test sonority value 3 - Voiceless fricatives *)
+let test_sonority_three () =
+  let segments = [ "f"; "s"; "x"; "ħ"; "ʃ" ] in
+  let expected = [ 3; 3; 3; 3; 3 ] in
+  let results = List.map Sonority.sonority segments in
+  assert (results = expected);
+  Printf.printf "test_sonority_three: PASSED\n"
+
+(** Test sonority value 2 - Voiced stops *)
+let test_sonority_two () =
+  let segments = [ "b"; "ɡ"; "d"; "ɢ" ] in
+  let expected = [ 2; 2; 2; 2 ] in
+  let results = List.map Sonority.sonority segments in
+  assert (results = expected);
+  Printf.printf "test_sonority_two: PASSED\n"
+
+(** Test sonority value 1 - Voiceless stops *)
+let test_sonority_one () =
+  let segments = [ "p"; "k"; "c"; "q" ] in
+  let expected = [ 1; 1; 1; 1 ] in
+  let results = List.map Sonority.sonority segments in
+  assert (results = expected);
+  Printf.printf "test_sonority_one: PASSED\n"
+
+(** Test unknown segment handling *)
+let test_unknown_segment () =
+  try
+    let _ = Sonority.sonority "🦆" in
+    assert false (* Should not reach here *)
+  with
+  | Failure msg when String.sub msg 0 20 = "Unknown IPA segment:" ->
+      Printf.printf "test_unknown_segment: PASSED\n"
+  | _ -> assert false
+
+(** Test feature-based sonority calculation *)
+let test_sonority_from_features () =
+  (* Test a simple voiceless stop: -syl, +cons, -son, -cont, -voi *)
+  let segment =
+    [
+      (Feature.Syllabic, Feature.Minus);
+      (Feature.Consonantal, Feature.Plus);
+      (Feature.Sonorant, Feature.Minus);
+      (Feature.Continuant, Feature.Minus);
+      (Feature.Voiced, Feature.Minus);
+    ]
+  in
+  let result = Sonority.sonority_from_features segment in
+  assert (result = 1);
+  Printf.printf "test_sonority_from_features: PASSED\n"
+
+(** Run all tests *)
+let () =
+  Printf.printf "Running Sonority module tests...\n";
+  Printf.printf "================================\n";
+  test_sonority_nine ();
+  test_sonority_eight ();
+  test_sonority_seven ();
+  test_sonority_six ();
+  test_sonority_five ();
+  test_sonority_four ();
+  test_sonority_three ();
+  test_sonority_two ();
+  test_sonority_one ();
+  test_unknown_segment ();
+  test_sonority_from_features ();
+  Printf.printf "================================\n";
+  Printf.printf "All tests passed!\n"
diff --git a/sorsyl/test/test_table.ml b/sorsyl/test/test_table.ml
new file mode 100644
index 0000000..89cb4a2
--- /dev/null
+++ b/sorsyl/test/test_table.ml
@@ -0,0 +1,208 @@
+open Sorsyl
+
+(* let es = *)
+(* String.split_on_char ' ' *)
+(* "la ˌinteɾnˌaθjonˌaliθaθjˈon del kˌoɾaθˈon i el ðˌikθjonˈaɾjo" *)
+
+(* let de = *)
+(* [ *)
+(* "kɔmˈpjuːtɐ"; *)
+(* "teleˈfoːn"; *)
+(* "fɑˈmiːliːjə"; *)
+(* "ˈʔɑːpɔtekə"; *)
+(* "ˈʃoːkolɑdə"; *)
+(* "ˈtoːmɑtən"; *)
+(* "ˈbananə"; *)
+(* "ˈpoːlitsae"; *)
+(* "ˈmuːzɔøm"; *)
+(* "ˈbʏçɐ̯ae"; *)
+(* "mediˈt͡siːn"; *)
+(* "ˈpɾoːfɛszoɾ"; *)
+(* "ʔeleˈfɑnt"; *)
+(* "dokumɛnt"; *)
+(* "ˈʔɪntɛɾnət"; *)
+(* "ˈʔʊnʔivɛɾziˈtɛːt"; *)
+(* "ˈkaɾtɔffɛln"; *)
+(* "ˈmatemɑtik"; *)
+(* "gəˈbʊɾtstɑk"; *)
+(* "ˈvœʁtɐˌbuːx"; *)
+(* "ˈbɪbli̯oːtək"; *)
+(* "demɔkɾɑˈtiːjə"; *)
+(* "fotɔgɾɑˈfiːjə"; *)
+(* "tɛçnoloˈgiːjə"; *)
+(* "bioloˈgiːjə"; *)
+(* "psʏçoloˈgiːjə"; *)
+(* "filozɔfiːjə"; *)
+(* "ˈʃoːkolɑdə"; *)
+(* "ˈmaɾmelɑdə"; *)
+(* "ˈzɛkɾetɛɾɪn"; *)
+(* "ʔɛntˈʃʊldɪgʊŋ"; *)
+(* "ˈkɾaŋkɛnvɛɾzɪçɐ̯ʊŋ"; *)
+(* "gəˈbʊɾtstagspaɾty"; *)
+(* "kɔmmunikɑˈtsĭoːn"; *)
+(* "ʔɔɾgɑnizɑˈtsĭoːn"; *)
+(* "bʏɾgɛɾˈməɪstɐ̯"; *)
+(* "zeɛnsvʏɾdɪkˈkəiːt"; *)
+(* "ˈmiːneɾalvaszɐ̯"; *)
+(* "ˈtsuːzammenaɾˈbəiːt"; *)
+(* "mœglɪçˈkəiːtən"; *)
+(* "gəlegɛnˈəiːtən"; *)
+(* "naxˈmɪtˌtɑːk"; *)
+(* "ʔɪnfɔɾmɑˈtsĭoːn"; *)
+(* "televiˈzĭoːn"; *)
+(* "gəʃvɪndɪkkaetsbɛgɾentsʊŋ"; *)
+(* "ˈkɾaŋkɛnaozaofɛntalt"; *)
+(* "ʔaɾbaetslozɪkˈkəiːt"; *)
+(* "fɛːɐ̯ˈʔantvɔɾtlɪçˈkəiːt"; *)
+(* "zeɛnsvʏɾdɪkˈkəiːtən"; *)
+(* "ˈzɛlpstvɛɾstɛntlɪç"; *)
+(* "ˈvaenaxtsgɛʃɛŋkə"; *)
+(* "gəˈbʊɾtstaksgɛʃɛŋk"; *)
+(* "tuɾɪstenɪnfɔɾmɑˈtsĭoːn"; *)
+(* "ˈʔʊnˌʔiːvɛɾzitɛtspɾofɛszoɾ"; *)
+(* "ˈleːbɛnsmɪtɛlgɛʃɛft"; *)
+(* "ˈfɑɾɾatvɛɾlae"; *)
+(* "ˈbʊndɛstakzapgeɔɾdnətɐ̯"; *)
+(* "ˈʃtɾaeçɔltsʃɛçtɛlçən"; *)
+(* "ˈfɾɔøntʃaftsbetsiːʊŋən"; *)
+(* "ˈɾɛçtsʃʊtsvɛɾzɪçɐ̯ʊŋ"; *)
+(* "nɑɾʊŋsmɪtelʊnvɛɾtɾɛglɪçˈkəiːt"; *)
+(* ] *)
+
+(* let en1 = *)
+(* [ *)
+(* "ˈæpəɫ"; *)
+(* "ˈðɪs"; *)
+(* "ˈɪs"; *)
+(* "ˈeɪ"; *)
+(* "ɫɪŋˈɡwɪstɪks"; *)
+(* "kəˈtæstɹəfi"; *)
+(* "wɪˈθaʊt"; *)
+(* "ˈpɹɛsədənt"; *)
+(* "ˈtu"; *)
+(* "ədˈmɪt"; *)
+(* "ˈænd"; *)
+(* "ˈɪts"; *)
+(* "ˌɪnstəˈɡeɪʃən"; *)
+(* "ˈʃʊd"; *)
+(* "ˈbi"; *)
+(* "ˈpənɪʃt"; *)
+(* ] *)
+
+let en2 =
+  [
+    "d͡ʒɹ̩mən";
+    "ˈpɹɛzənt";
+    "ˈɑɹtɪkəɫ";
+    "pɹəˈvaɪdz";
+    "ˌɹiəˈnæɫəsəs";
+    "kˈɔːɹɑːnəl";
+    "kɝˈoʊnəɫ";
+    "ˈɑptɪks";
+    "ˈɛksəɫəns";
+    "əbˈstɹuənts";
+    "ˈdʒɝmən";
+    "ˈɛŋɡɫɪʃ";
+    "ˈkɑmənɫi";
+    "əˈsumd";
+    "ˈdʒɪps";
+    "ˈpɫæstɝ";
+    "ˈɛŋɡɫɪʃ";
+    (* "ˈɫæps"; *)
+    "ˈɑɹɡjud";
+    "bɪˈɫoʊ";
+    "ˌɛkstɹəsɪˈɫæbɪk";
+    "ˈkɑnsənənts";
+    "ˌdɛɹəˈveɪʃənəɫ";
+    "ˈsteɪdʒ";
+    "ˌɛkstɹəsɪɫəˈbɪsɪti";
+    "ˈaɪðɝ";
+    "ˈɫæŋɡwɪdʒɪz";
+    "ɪɡˈzɪsts";
+    "ˈɛvədəns";
+    "ˈkɑmənɫi";
+    "pɹɪˈzɛntɪd";
+    "səˈpɔɹt";
+    "ˌɛkstɹəsɪˈɫæbɪk";
+    "ˈkɑnsənənts";
+    "kəmˈpætəbəɫ";
+    "ˈfʊɫi";
+    "ˈsɝfəs";
+    "ˌɑptəˈmæɫəti";
+    "ˌθiɝˈɛtɪk";
+    "ˈtɹitmənt";
+    "kənˈstɹeɪnts";
+    "ɹɪˈfɝɪŋ";
+    "ˈfʊɫi";
+    "səˈɫæbəˌfaɪd";
+    "ˈaʊtˌpʊt";
+    "ˌɹɛpɹəzɛnˈteɪʃənz";
+    "pɹəˈpoʊzd";
+  ]
+
+(* let zh = [ "/t͡ɕi⁵¹ ti⁵¹ pʰi⁵¹/" ] *)
+
+(* let ws = *)
+(* [ *)
+(* ( "/nuˌmɑ.noʊ.ʌl.tɹə.maɪ.kɹoʊˈskɑ.pɪkˌsɪ.lɪ.koʊ.vɑl.keɪ.noʊ.koʊ.niˈoʊ.sɪs/", *)
+(* "/əˈbæn.dn̩.əd.li/", *)
+(* " /əˈbæn.dn̩.əd.li/", *)
+(* "/əˈbæn.dn̩.mn̩t/", *)
+(* "/-ˌbiːə-/", *)
+(* "/əˈbluː.ʃn̩/xx" ); *)
+(* ] *)
+
+let get_data =
+  Printf.printf "Getting data\n";
+  let data_dir =
+    if Sys.file_exists "./data" then "./data"
+    else if Sys.file_exists "../data" then "../data"
+    else if Sys.file_exists "../../../data" then "../../../data"
+    else (
+      Printf.eprintf "Current directory: %s\n" (Sys.getcwd ());
+      failwith "Cannot find data directory")
+  in
+  Ipa_table.load_csv data_dir
+
+let test_fts table =
+  let result = Ipa_table.fts table "s" in
+  match result with
+  | None -> ()
+  | Some seg -> Printf.printf "fts\n %s\n" (Feature.string_of_segment seg)
+(* let expected = None in *)
+(* let results = Ipa_table.fts *)
+(* assert (result = expected); *)
+(* Printf.printf "test_fts: PASSED\n" *)
+
+let test_segs data =
+  let words = en2 in
+  (* let expected = [] in *)
+  let _results =
+    Base.List.map words ~f:(fun word ->
+        let res = Ipa_table.ipa_segs data word in
+        Printf.printf "%s\n" word;
+        let xl =
+          Base.List.fold res ~init:"" ~f:(fun acc char ->
+              Printf.sprintf "%s-%s" acc char)
+        in
+        Printf.printf "%s\n" xl;
+        res)
+  in
+  (* Base.List.iter results ~f:(fun x -> *)
+  (* let xl = *)
+  (* Base.List.fold x ~init:"" ~f:(fun acc char -> *)
+  (* Printf.sprintf "%s-%s" acc char) *)
+  (* in *)
+  (* Printf.printf "%s\n" xl) *)
+  (* assert (result = expected) *)
+  Printf.printf "test_fts: PASSED\n"
+
+(** Run all tests *)
+let () =
+  let data = get_data in
+  Printf.printf "Running IPA Table module tests...\n";
+  Printf.printf "===================================\n";
+  test_fts data.table;
+  test_segs data;
+  Printf.printf "===================================\n";
+  Printf.printf "All IPA Table tests passed!\n"