Managed diphtongues

8 files changed, 117 insertions, 51 deletions

diff --git a/src/lib/lexer.mll b/src/lib/lexer.mll
index f8a7f2c..07305d8 100644
--- a/src/lib/lexer.mll
+++ b/src/lib/lexer.mll
@@ -13,6 +13,8 @@ rule letter = parse
 | 'c'           { C }
 | 'd'           { D }
 | 'e'           { E }
+| '\232'        { E_ACUTE }
+| '\xC3' '\xA8' { E_AGRAVE }
 | '\233'        { E_ACUTE }
 | '\xC3' '\xA9' { E_ACUTE }
 | 'f'           { F }
diff --git a/src/lib/modifiers/nasal.ml b/src/lib/modifiers/nasal.ml
index da24863..b9874c9 100644
--- a/src/lib/modifiers/nasal.ml
+++ b/src/lib/modifiers/nasal.ml
@@ -9,15 +9,28 @@ let process
     let module T = (val m:Sounds.T with type t = el) in
     let (((v1, v2), c) , ending) = init in
     let ending = Option.bind ending (fun x -> x) in
-    match ending with
-    | None -> init
-    | Some ending ->
-      match T.is_nasal ending with
-      | false -> init
-      | true ->
-        (* Remove the ending consonant, and transform the voyel into
-           the nasal form  *)
-        ( ( (T.nasal v1, T.nasal v2)
-          , c )
-        , None )
+    let opening = Option.map (fun v -> v.Sig.opening) c in
+    let is_voyel = T.is_voyel v1 && T.is_voyel v2 in
+    match ending, is_voyel, opening with
+    | Some ending, _, _ when T.is_nasal ending ->
+      (* Remove the ending consonant, and transform the voyel into
+         the nasal form  *)
+      ( ( (T.nasal v1, T.nasal v2)
+        , c )
+      , None )
+    | None, false, Some (opening::tl) when T.is_nasal opening ->
+      (* If there is no voyel here, transform the opening consonant as an
+         ending consonant for the next syllabus *)
+      let c = Option.map
+          (fun c ->
+             { c with
+               Sig.opening = tl
+             ; Sig.ending = (Some (Some opening))
+             }) c in
+      ( ( (T.nasal v1, T.nasal v2)
+        , c )
+      , None )
+    | _ ->
+      (* Return the element unchanged *)
+      init
 
diff --git a/src/lib/parser.mly b/src/lib/parser.mly
index 2a94839..f12e2cc 100644
--- a/src/lib/parser.mly
+++ b/src/lib/parser.mly
@@ -28,7 +28,7 @@ occlusiv:
   | D               { T.d }
 
   | K               { T.k }
-  | G               { T.none }
+  | G               { T.g }
 
 fricativ:
   | S               { T.s () }
@@ -75,14 +75,18 @@ opening_consonant:
 (* Each voyel as two associated sounds, depending there is a followng sound or
    not *)
 voyels:
-  | A               { T.a  `Opened , T.a  `Opened }
+  | A               { T.a          , T.a          }
   | A I             { T.e  `Opened , T.e  `Opened }
-  | I               { T.i  `Opened , T.i  `Opened }
+  | I               { T.i          , T.i          }
   | E               { T.e  `Opened , T.schwa () }
   | E_ACUTE E?      { T.e  `Closed , T.e  `Closed }
+  | E_AGRAVE        { T.e  `Opened , T.e  `Opened }
   | E U             { T.eu `Opened , T.eu `Opened }
   | O               { T.o  `Opened , T.o  `Opened }
   | U               { T.u          , T.u          }
+  | OU              { T.u'         , T.u'         }
+  | W A             { T.diphtongue T.semi_voyel_w T.a, T.diphtongue T.semi_voyel_w T.a}
+  | W I             { T.diphtongue T.semi_voyel_w T.i, T.diphtongue T.semi_voyel_w T.a}
 
 nasal_voyels:
   | A N             { T.a' ()      , T.a' ()      }
@@ -113,12 +117,14 @@ consonant_group:
 syllable:
   | c = consonant_group?
     v = voyels
+    Sep?
     { (v, c) }
 
 
 syllables:
   | { [] }
-  | ss = syllables s = syllable { s::ss }
+  | ss = syllables 
+    s = syllable { s::ss }
     
 
 word:
diff --git a/src/lib/process.ml b/src/lib/process.ml
index a74c44b..8c463b5 100644
--- a/src/lib/process.ml
+++ b/src/lib/process.ml
@@ -32,7 +32,6 @@ module M(T:Sounds.T) = struct
     = function
       | [] -> acc
       | hd::tl ->
-
         let modifier = (Modifiers.vocalize_s) :: (Modifiers.nasal) :: m in
         let (voyel, consonants), ending_consonant =
           apply_modifiers
@@ -41,18 +40,17 @@ module M(T:Sounds.T) = struct
         let voyel = change_voyel ((voyel, consonants), ending_consonant) in
 
         (* Add the last consonant and the voyel *)
-        let m, acc = match ending_consonant with
-          | None -> modifier, voyel::acc
+        let acc = match ending_consonant with
+          | None -> voyel::acc
           | Some s ->
-            let default = modifier, voyel :: acc in
+            let default = voyel :: acc in
             match s with
             | None -> default
             | Some s ->
-
-              modifier, voyel :: s::acc  in
+              voyel :: s::acc  in
 
         match consonants with
-        | None -> _rebuild ~m acc None tl
+        | None -> _rebuild ~m:[] acc None tl
         | Some {ending; opening; following} ->
 
           let acc = match following with
@@ -60,8 +58,8 @@ module M(T:Sounds.T) = struct
             | Some s -> s::acc in
 
           match opening with
-          | [] ->_rebuild ~m acc ending tl
-          | opening -> _rebuild ~m (opening @ acc) ending tl
+          | [] ->_rebuild ~m:[] acc ending tl
+          | opening -> _rebuild ~m:[] (opening @ acc) ending tl
 
   (** Rebuild the list in the normal order
 
diff --git a/src/lib/prononciation.mly b/src/lib/prononciation.mly
index 3195078..5dbf153 100644
--- a/src/lib/prononciation.mly
+++ b/src/lib/prononciation.mly
@@ -14,14 +14,15 @@
 %%
 
 initial_voyel:
-  | A   { A }
-  | A U { O }
-  | E   { E }
-  | I   { I }
-  | O   { O }
-  | U   { U }
-  | U I { UI }
-  | E_ACUTE { E_ACUTE }
+  | A           { A }
+  | A U         { O }
+  | E           { E }
+  | I           { I }
+  | O           { O }
+  | O U         { OU }
+  | U           { U }
+  | E_ACUTE     { E_ACUTE }
+  | E_AGRAVE    { E_AGRAVE }
 
 voyel: 
   | initial_voyel { $1 }
@@ -57,6 +58,8 @@ letters:
   | letters N       { N :: $1 }
   | letters N N     { N :: $1 }
 
+  | letters O I N   { N :: I :: W :: $1 }
+  | letters O I     { A :: W :: $1 }
 
   | letters P       { P :: $1 }
   | letters P H     { F :: $1 }
diff --git a/src/lib/sounds/sounds.ml b/src/lib/sounds/sounds.ml
index a2035b7..6126dd4 100644
--- a/src/lib/sounds/sounds.ml
+++ b/src/lib/sounds/sounds.ml
@@ -3,15 +3,18 @@ module type T = sig
   type t
   val muted : t -> t
 
-  val a : [`Closed | `Opened] -> t
+  val a : t
   val e : [`Closed | `Opened] -> t
   val eu : [`Closed | `Opened] -> t
 
   val o : [`Closed | `Opened] -> t
   val schwa : unit -> t
-  val i : [`Closed | `Opened] -> t
+  val i : t
   val u : t
+  val u' : t
 
+  (** Create a diphtongue from a semi-voyel and a voyel *)
+  val diphtongue: t -> t -> t
 
   val nasal: t -> t
 
@@ -22,6 +25,7 @@ module type T = sig
   val t: t
   val d: t
   val k: t
+  val g: t
   val f: t
   val s: unit -> t
   val sz: unit -> t
@@ -43,7 +47,10 @@ module type T = sig
     | None
     | SZ
     | Voyel_A
+    | Voyel_I
     | Voyel_O
+    | SemiVoyel_W
+    | Diphtonge of code * code
 
   val code : t -> code
 
@@ -58,7 +65,10 @@ module T = struct
     | None
     | SZ
     | Voyel_A
+    | Voyel_I
     | Voyel_O
+    | SemiVoyel_W
+    | Diphtonge of code * code
 
   type t =
     { code : code
@@ -79,7 +89,9 @@ module Repr = struct
   and o_nasal = "§"
 
   and i = "i"
+  and i_nasal = "5"
   and u = "y"
+  and u' = "u"
 
   and p = "p"
   and b = "b"
@@ -112,6 +124,7 @@ module S = struct
   let is_voyel t = t.kind = Voyel
   let is_nasal t = t.nasal
 
+
   let none =
     { repr = ""
     ; muted = false
@@ -122,20 +135,17 @@ module S = struct
   let voyel =
     { none with kind = Voyel }
 
-  let code t = t.code
 
-  let nasal t =
-    match t.code with
-    | Voyel_A -> { t with repr = Repr.a_nasal ; nasal = true }
-    | Voyel_O -> { t with repr = Repr.o_nasal ; nasal = true }
-    | _ -> t
+  let diphtongue v1 v2 =
+    { voyel with
+      repr = (v1.repr) ^ (v2.repr)
+    ; code = Diphtonge (v1.code, v2.code)
+    }
 
-  let muted f =
-    { none with
-      repr = "(" ^ f.repr ^ ")"
-    ; muted = true }
 
-  let a _ =
+  let code t = t.code
+
+  let a =
     { voyel with repr = Repr.a ; code = Voyel_A }
 
   let e = function
@@ -153,12 +163,15 @@ module S = struct
   let o _ =
     { voyel with repr = Repr.o ; code = Voyel_O }
 
-  let i _ =
-    { voyel with repr = Repr.i }
+  let i =
+    { voyel with repr = Repr.i ; code = Voyel_I }
 
   let u =
     { voyel with repr = Repr.u }
 
+  let u' =
+    { voyel with repr = Repr.u' }
+
   let p =
     { none with repr = Repr.p }
 
@@ -174,6 +187,9 @@ module S = struct
   let k =
     { none with repr = Repr.k }
 
+  let g =
+    { none with repr = Repr.g }
+
   let f =
     { none with repr = Repr.f }
 
@@ -202,7 +218,26 @@ module S = struct
     { none with repr = Repr.r }
 
   let semi_voyel_w =
-    { none with repr = Repr.w }
+    { none with
+      repr = Repr.w
+    ; code = SemiVoyel_W}
+
+  let nasal t =
+    match t.code with
+    | Voyel_A -> { t with repr = Repr.a_nasal ; nasal = true }
+    | Voyel_O -> { t with repr = Repr.o_nasal ; nasal = true }
+    | Voyel_I -> { t with repr = Repr.i_nasal ; nasal = true }
+    | Diphtonge (SemiVoyel_W, Voyel_A) ->
+      diphtongue
+        semi_voyel_w
+        { t with repr = Repr.i_nasal ; nasal = true }
+    | _ -> t
+
+  let muted f =
+    { none with
+      repr = "(" ^ f.repr ^ ")"
+    ; muted = true }
+
 end
 
 include S
diff --git a/src/lib/tokens.mly b/src/lib/tokens.mly
index 5466e90..0bceeac 100644
--- a/src/lib/tokens.mly
+++ b/src/lib/tokens.mly
@@ -6,6 +6,7 @@
 %token D
 %token E
 %token E_ACUTE
+%token E_AGRAVE
 %token F
 %token G
 %token H
@@ -16,6 +17,7 @@
 %token M
 %token N
 %token O
+%token OU
 %token Q
 %token P
 %token R
@@ -23,7 +25,6 @@
 %token SZ
 %token T
 %token U
-%token UI
 %token V
 %token W (* semi voyel w *)
 %token X
@@ -34,7 +35,7 @@
 
 %nonassoc Low
 
-%right A E E_ACUTE I O U 
+%right A E E_ACUTE I O U OU
 
 %right C H J Q V W X Y Z
 %right P B T D K G
diff --git a/src/test/test.ml b/src/test/test.ml
index a8ac890..852349d 100644
--- a/src/test/test.ml
+++ b/src/test/test.ml
@@ -58,12 +58,20 @@ let tests =
   ; "chanci",       "S@si"
   ; "chat",         "Sa(t)"
   ; "chipant",      "Sip@(t)"
+  ; "co|incidant",  "ko5sid@(t)"
+  ; "croire",       "kRwaR°"
+  ; "ébrouas",      "ebRua(s)"
+  ; "loin",         "Lw5"
+  ; "groin",        "gRw5"
   ; "pacha",        "paSa"
   ; "péché",        "peSe"
   ; "persai",       "pERsE"
+  ; "plan",         "pL@"
   ; "plat",         "pLa(t)"
   ; "platte",       "pLat°"
-  ; "web",          "wEb"
+  ; "toiture",      "twatyR°"
+  ; "trois",        "tRwa(s)"
+  ; "wèb",          "wE(b)"
   ]
 
 let () =