thai phrasebook example started

examples/travel/Travel.gf

@@ -0,0 +1,54 @@
+abstract Travel = {
+
+  flags startcat=Phrase ;
+
+  cat
+    Phrase ;
+
+    Greeting ;
+    Order ; 
+    Question ;
+    Sentence ;
+
+    Object ;
+    Kind ;
+    Quality ;
+    Number ;
+
+    Speaker ;
+    Hearer ;
+    Gender ;
+    Quantity ;
+
+  fun
+    PGreeting : Greeting -> Speaker -> Hearer -> Phrase ;
+    POrder    : Order    -> Speaker -> Hearer -> Phrase ;
+    PQuestion : Question -> Speaker -> Hearer -> Phrase ;
+    PSentence : Sentence -> Speaker -> Hearer -> Phrase ;
+
+    MkSpeaker : Gender -> Quantity -> Speaker ;
+    MkHearer  : Gender -> Quantity -> Hearer ;
+
+    Male, Female : Gender ;
+    Single, Many : Quantity ;
+
+    Hello : Greeting ;
+    Thanks : Greeting ;
+
+    IWant : Object -> Order ;
+    
+    DoYouHave : Kind -> Question ;
+    IsIt : Object -> Quality -> Sentence ;
+
+    ItIs : Object -> Quality -> Sentence ;
+    
+    Indef : Kind -> Object ;
+    This  : Kind -> Object ;
+    NumberObjects : Number -> Kind -> Object ;
+
+    One, Two, Five, Ten : Number ;
+
+    Mango : Kind ;
+    Green : Quality ;
+
+}

examples/travel/TravelTha.gf

@@ -0,0 +1,73 @@
+--# -path=.:prelude
+
+concrete TravelTha of Travel = open Prelude in {
+
+  flags coding=utf8 ;
+
+-- this file is processed by
+--  GF.Text.Thai.thaiFile "../examples/travel/TravelTha.gf" 
+      (Just "../examples/travel/TravelThai.gf")
+--  GF.Text.Thai.thaiPronFile "../examples/travel/TravelTha.gf" 
+      (Just "../examples/travel/TravelThaiPron.gf")
+-- to produce target Thai script and pronunciation files.
+
+  lincat
+    Phrase,
+
+    Greeting,
+    Order,
+    Question,
+    Sentence,
+    Object = SS ;
+
+    Kind = {s,c : Str} ; -- c is classifier
+
+    Quality,
+    Number,
+
+    Speaker,
+    Hearer,
+    Gender,
+    Quantity = SS ;
+
+  lin
+    PGreeting g s h = ss (g.s ++ s.s ++ h.s) ;
+    POrder    g s h = ss (g.s ++ s.s ++ h.s) ;
+    PQuestion g s h = ss (g.s ++ s.s ++ h.s) ;
+    PSentence g s h = ss (g.s ++ s.s ++ h.s) ;
+
+    MkSpeaker = cc2 ;
+    MkHearer _ _ = ss [] ;
+
+    Male = ss "k2rab" ;
+    Female = ss "k2T1a" ;
+
+    Single, Many = ss [] ;
+
+    Hello = ss ["swa.s di:"] ;
+    Thanks = ss ["k1Ob k2un'"] ;
+
+    IWant = prefixSS "eOa:" ;
+    
+    DoYouHave = postfixSS ["mi: a&hm"] ;
+    IsIt x q = ss (x.s ++ q.s ++ "a&hm") ;
+
+    ItIs = cc2 ;
+    
+    Indef k = ss (k.s ++ k.c ++ "nvT1g") ;
+    This  k = ss (k.s ++ k.c ++ "ni:T2") ;
+
+    NumberObjects n k = ss (k.s ++ n.s ++ k.c) ;
+
+    One  = ss "nvT1g" ;
+    Two  = ss "sOg" ;
+    Five = ss "ha:" ;
+    Ten  = ss "sib" ;
+
+    Mango = cls ["ma mT1wg"] "lu:k" ;
+    Green = ss ["si: ek1i:yw"] ;
+
+  oper
+    cls : Str -> Str -> {s,c : Str} = \s,c -> {s = s ; c = c} ;
+
+}

examples/travel/TravelThai.gf

@@ -0,0 +1,71 @@
+--# -path=.:prelude
+
+concrete TravelThai of Travel = open Prelude in {
+
+  flags coding=utf8 ;
+
+-- this file is processed by
+--  GF.Text.Thai.thaiFile "À€À€เÀ€à¸°à¸¡à¸›à¸¥à¹€À€à¸•à¸£à¸°à¸¶à¹€À€À€ระึเลÀ€à¸«à¸±à¸‡à¸<C3A0>" (Just TGT)
+--  GF.Text.Thai.thaiPronFile "À€À€เÀ€à¸°à¸¡à¸›à¸¥à¹€À€à¸•à¸£à¸°à¸¶à¹€À€À€ระึเลÀ€à¸«à¸±à¸‡à¸<C3A0>" (Just TGT)
+-- to produce target Thai script and pronunciation files.
+
+  lincat
+    Phrase,
+
+    Greeting,
+    Order,
+    Question,
+    Sentence,
+    Object = SS ;
+
+    Kind = {s,c : Str} ; -- c is classifier
+
+    Quality,
+    Number,
+
+    Speaker,
+    Hearer,
+    Gender,
+    Quantity = SS ;
+
+  lin
+    PGreeting g s h = ss (g.s ++ s.s ++ h.s) ;
+    POrder    g s h = ss (g.s ++ s.s ++ h.s) ;
+    PQuestion g s h = ss (g.s ++ s.s ++ h.s) ;
+    PSentence g s h = ss (g.s ++ s.s ++ h.s) ;
+
+    MkSpeaker = cc2 ;
+    MkHearer _ _ = ss [] ;
+
+    Male = ss "คระบ" ;
+    Female = ss "ค่ะ" ;
+
+    Single, Many = ss [] ;
+
+    Hello = ss ["สวัสดี"] ;
+    Thanks = ss ["ขอบคุณ"] ;
+
+    IWant = prefixSS "เอา" ;
+    
+    DoYouHave = postfixSS ["มีไหม"] ;
+    IsIt x q = ss (x.s ++ q.s ++ "ไหม") ;
+
+    ItIs = cc2 ;
+    
+    Indef k = ss (k.s ++ k.c ++ "นึ่ง") ;
+    This  k = ss (k.s ++ k.c ++ "นี้") ;
+
+    NumberObjects n k = ss (k.s ++ n.s ++ k.c) ;
+
+    One  = ss "นึ่ง" ;
+    Two  = ss "สอง" ;
+    Five = ss "หา" ;
+    Ten  = ss "สิบ" ;
+
+    Mango = cls ["มะม่วง"] "ลูà¸<C3A0>" ;
+    Green = ss ["สีเขียว"] ;
+
+  oper
+    cls : Str -> Str -> {s,c : Str} = \s,c -> {s = s ; c = c} ;
+
+}

examples/travel/TravelThaiPron.gf

@@ -0,0 +1,71 @@
+--# -path=.:prelude
+
+concrete TravelThaiPron of Travel = open Prelude in {
+
+  flags coding=utf8 ;
+
+-- this file is processed by
+--  GF.Text.Thai.thaiFile "m^eaeaveaveapltrrlhg-" (Just TGT)
+--  GF.Text.Thai.thaiPronFile "m^eaeaveaveapltrrlhg-" (Just TGT)
+-- to produce target Thai script and pronunciation files.
+
+  lincat
+    Phrase,
+
+    Greeting,
+    Order,
+    Question,
+    Sentence,
+    Object = SS ;
+
+    Kind = {s,c : Str} ; -- c is classifier
+
+    Quality,
+    Number,
+
+    Speaker,
+    Hearer,
+    Gender,
+    Quantity = SS ;
+
+  lin
+    PGreeting g s h = ss (g.s ++ s.s ++ h.s) ;
+    POrder    g s h = ss (g.s ++ s.s ++ h.s) ;
+    PQuestion g s h = ss (g.s ++ s.s ++ h.s) ;
+    PSentence g s h = ss (g.s ++ s.s ++ h.s) ;
+
+    MkSpeaker = cc2 ;
+    MkHearer _ _ = ss [] ;
+
+    Male = ss "khr'ap" ;
+    Female = ss "kh^a" ;
+
+    Single, Many = ss [] ;
+
+    Hello = ss ["s`awt d-ii"] ;
+    Thanks = ss ["kh`Op kh-un"] ;
+
+    IWant = prefixSS "O-eaa" ;
+    
+    DoYouHave = postfixSS ["m-ii m~ay"] ;
+    IsIt x q = ss (x.s ++ q.s ++ "m~ay") ;
+
+    ItIs = cc2 ;
+    
+    Indef k = ss (k.s ++ k.c ++ "n^vg") ;
+    This  k = ss (k.s ++ k.c ++ "n~ii") ;
+
+    NumberObjects n k = ss (k.s ++ n.s ++ k.c) ;
+
+    One  = ss "n^vg" ;
+    Two  = ss "s~Og" ;
+    Five = ss "h~aa" ;
+    Ten  = ss "s`ip" ;
+
+    Mango = cls ["m'a m^owg"] "l^uuk" ;
+    Green = ss ["s~ii kh`eiiyw"] ;
+
+  oper
+    cls : Str -> Str -> {s,c : Str} = \s,c -> {s = s ; c = c} ;
+
+}

src/GF/Text/Thai.hs

@@ -121,6 +121,7 @@ pronSyllable s =
   vowel = case (initv s, midv s, finalv s, shorten s, tone s) of
     ([0x0e40],[0x0e30,0x0e2d],_,_,_) -> "ö" -- eOa
     ([0x0e40],[0x0e30,0x0e32],_,_,_) -> "o" -- ea:a 
+    ([],[],[],_,_) -> "o"
     (i,m,f,_,_) -> concatMap pronThaiChar (reverse $ f ++ m ++ i) ----
 
   initCons = concatMap pronThaiChar $ case (reverse $ initc s) of
@@ -179,12 +180,14 @@ getSyllable = foldl get (Syll [] [] [] [] [] [] False False) where
       | isVowel c -> if null (initc syll) 
                        then syll {initv = c : initv syll}
                        else syll {midv  = c : midv syll}
-      | isCons c  -> if null (midv syll) 
+      | isCons c  -> if  null (initc syll) || 
+                        (null (midv syll) && isCluster (initc syll) c) 
                        then syll {initc  = c : initc syll}
                        else syll {finalc = c : finalc syll}
       | isTone c  -> syll {tone = [c]}
     _ -> syll ---- check this
 
+  isCluster s c = length s == 1 && (c == 0x0e23 || s == [0x0e2b])
 
 -- to test