more thai transliteration; numeral thai grammars

6 files changed, 283 insertions, 5 deletions

diff --git a/examples/numerals/mkThai.gfs b/examples/numerals/mkThai.gfs
new file mode 100644
index 000000000..8f29d4e24
--- /dev/null
+++ b/examples/numerals/mkThai.gfs
@@ -0,0 +1,5 @@
+i -old -optimize=none -abs=Numerals -cnc=decimal decimal.gf
+i -old -optimize=none -abs=Numerals -cnc=thai_decimal thaiDU.gf
+i -old -optimize=none -abs=Numerals -cnc=thai thaiU.gf
+i -old -optimize=none -abs=Numerals -cnc=thai_pronounce thaiP.gf
+sf -unlexer=unwords
diff --git a/examples/numerals/thai.gf b/examples/numerals/thai.gf
new file mode 100644
index 000000000..da6648a42
--- /dev/null
+++ b/examples/numerals/thai.gf
@@ -0,0 +1,67 @@
+include numerals.Abs.gf ;
+
+-- Thai transliteration, produces thaiU.gf by GF/Text/Thai.hs
+-- AR 28/12/2006
+
+lincat 
+  Numeral    = {s : Str} ;
+  Digit      = {s : DForm => Str} ;
+  Sub10      = {s : DForm => Str} ;
+  Sub100     = {s : NForm => Str} ;
+  Sub1000    = {s : NForm => Str} ;
+  Sub1000000 = {s : Str} ;
+
+lin 
+  num x = x ;
+
+  pot01 = mkNum "hnvg" "hnvg" "eOMd'" ;
+
+  n2 = mkNum  "s-Og" "y'i:E" "s-Og" ;
+  n3 = regNum "s-a:m" ;
+  n4 = regNum "s-i:E" ;
+  n5 = regNum "ha:E" ; 
+  n6 = regNum "ho?k" ;
+  n7 = regNum "ecMd'" ;
+  n8 = regNum "e:pd'" ; 
+  n9 = regNum "eka:" ;
+
+
+  pot0 d = d ;
+
+  pot110 = {s = sip} ;
+  pot111 = {s = table {
+    Unit => ["s'ib et"] ; 
+    Thousand => ["hnvg hmv:En hnvg p2an"]
+    }
+  } ;
+  pot1to19 d = {s = table {
+    Unit => "s'ib" ++ d.s ! After ; 
+    Thousand => ["hnvg hmv:En"] ++ d.s ! Indep ++ "p2an"
+    }
+  } ;
+  pot0as1 d = {s = \\n => d.s ! Indep ++ phan ! n} ;
+  pot1 d = {s = \\n => d.s ! ModTen ++ sip ! n} ;
+  pot1plus d e = {
+    s = \\n => d.s ! ModTen ++ sip ! n ++ e.s ! After ++ phan ! n
+  } ;
+  pot1as2 n = n ;
+  pot2 d = {s = \\n => d.s ! Indep ++ roy ! n} ;
+  pot2plus d e = {s = \\n => d.s ! Indep ++ roy ! n ++ e.s ! n} ;
+  pot2as3 n = {s = n.s ! Unit} ;
+  pot3 n = {s = n.s ! Thousand} ;
+  pot3plus n m = {s = n.s ! Thousand ++ m.s ! Unit} ;
+
+param 
+  DForm = Indep | ModTen | After ;
+  NForm = Unit | Thousand ;
+
+oper 
+  mkNum : Str -> Str -> Str -> {s : DForm => Str} = \x,y,z ->
+    {s = table {Indep => x ; ModTen => y ; After => z}} ;
+  regNum : Str -> {s : DForm => Str} = \x -> 
+    mkNum x x x ;
+
+
+  sip  = table {Unit => "s'ib"  ; Thousand => "hmv:En"} ;
+  roy  = table {Unit => "rEOy'" ; Thousand => "se:n"} ;
+  phan = table {Unit => []      ; Thousand => "p2an"} ;
diff --git a/examples/numerals/thaiDU.gf b/examples/numerals/thaiDU.gf
new file mode 100644
index 000000000..1cbf81fbc
--- /dev/null
+++ b/examples/numerals/thaiDU.gf
@@ -0,0 +1,48 @@
+include numerals.Abs.gf ;
+
+-- Thai digits. AR 28/12/2006
+
+flags lexer=chars ; unlexer=concat ; flags coding=utf8 ;
+
+param Zeros = noz  | zz  ;
+
+lincat Numeral =    { s : Str } ;
+lincat Digit =      { s : Str } ;
+lincat Sub10 =      { s : Str } ;
+lincat Sub100 = {s : Zeros => Str} ;
+lincat Sub1000 = {s : Zeros => Str} ;
+lincat Sub1000000 = {s : Zeros => Str} ;
+
+oper ss : Str -> {s : Str} = \s -> {s = s} ;
+oper mkz : Str -> {s : Zeros => Str} = \s -> {s = table {_ => s}} ;
+
+lin num n = {s = n.s ! noz} ;
+lin n2 = ss "๒" ;
+lin n3 = ss "๓" ;
+lin n4 = ss "๔" ;
+lin n5 = ss "๕" ;
+lin n6 = ss "๖" ;
+lin n7 = ss "๗" ;
+lin n8 = ss "๘" ;
+lin n9 = ss "๙" ;
+
+lin pot01 = ss "๑" ;
+lin pot0 d = d ;
+
+lin pot110 = mkz ("๑" ++ "๐") ;
+lin pot111 = mkz ("๑" ++ "๑") ;
+lin pot1to19 d = mkz ("๑" ++ d.s) ;
+
+lin pot0as1 n = {s = table {noz => n.s ; zz => "๐" ++ n.s}} ;
+
+lin pot1 d = mkz (d.s ++ "๐") ;
+lin pot1plus d e = mkz (d.s ++ e.s) ;
+
+lin pot1as2 n = {s = table {noz => n.s ! noz ; zz => "๐" ++ n.s ! zz}} ;
+lin pot2 d = mkz (d.s ++ "๐" ++ "๐") ;
+lin pot2plus d e = mkz (d.s ++ e.s ! zz) ;
+
+lin pot2as3 n = {s = table {noz => n.s ! noz ; zz => "๐" ++ n.s ! zz}} ;
+
+lin pot3 n = mkz (n.s ! noz ++ "๐" ++ "๐" ++ "๐") ;
+lin pot3plus n m = {s = table {z => n.s ! z ++ m.s ! zz}} ;
diff --git a/examples/numerals/thaiP.gf b/examples/numerals/thaiP.gf
new file mode 100644
index 000000000..c35b2dfbf
--- /dev/null
+++ b/examples/numerals/thaiP.gf
@@ -0,0 +1,69 @@
+include numerals.Abs.gf ;
+
+-- Thai pronunciation (mostly following Smyth's Essential Grammar) 
+-- AR 28/12/2006
+
+flags coding=utf8 ;
+
+lincat 
+  Numeral    = {s : Str} ;
+  Digit      = {s : DForm => Str} ;
+  Sub10      = {s : DForm => Str} ;
+  Sub100     = {s : NForm => Str} ;
+  Sub1000    = {s : NForm => Str} ;
+  Sub1000000 = {s : Str} ;
+
+lin 
+  num x = x ;
+
+  pot01 = mkNum "nỳng" "nỳng" "èt" ;
+
+  n2 = mkNum  "söong" "yîi" "söong" ;
+  n3 = regNum "säam" ;
+  n4 = regNum "sìi" ;
+  n5 = regNum "hâa" ; 
+  n6 = regNum "hòk" ;
+  n7 = regNum "cèt" ;
+  n8 = regNum "pèet" ; 
+  n9 = regNum "kâaw" ;
+
+
+  pot0 d = d ;
+
+  pot110 = {s = sip} ;
+  pot111 = {s = table {
+    Unit => ["sìp èt"] ; 
+    Thousand => ["nỳng mỳyn nỳng phan"]
+    }
+  } ;
+  pot1to19 d = {s = table {
+    Unit => "sìp" ++ d.s ! After ; 
+    Thousand => ["nỳng mỳyn"] ++ d.s ! Indep ++ "phan"
+    }
+  } ;
+  pot0as1 d = {s = \\n => d.s ! Indep ++ phan ! n} ;
+  pot1 d = {s = \\n => d.s ! ModTen ++ sip ! n} ;
+  pot1plus d e = {
+    s = \\n => d.s ! ModTen ++ sip ! n ++ e.s ! After ++ phan ! n
+  } ;
+  pot1as2 n = n ;
+  pot2 d = {s = \\n => d.s ! Indep ++ roy ! n} ;
+  pot2plus d e = {s = \\n => d.s ! Indep ++ roy ! n ++ e.s ! n} ;
+  pot2as3 n = {s = n.s ! Unit} ;
+  pot3 n = {s = n.s ! Thousand} ;
+  pot3plus n m = {s = n.s ! Thousand ++ m.s ! Unit} ;
+
+param 
+  DForm = Indep | ModTen | After ;
+  NForm = Unit | Thousand ;
+
+oper 
+  mkNum : Str -> Str -> Str -> {s : DForm => Str} = \x,y,z ->
+    {s = table {Indep => x ; ModTen => y ; After => z}} ;
+  regNum : Str -> {s : DForm => Str} = \x -> 
+    mkNum x x x ;
+
+
+  sip  = table {Unit => "sìp"  ; Thousand => "mỳyn"} ;
+  roy  = table {Unit => "róoy" ; Thousand => "sëen"} ;
+  phan = table {Unit => []      ; Thousand => "phan"} ;
diff --git a/examples/numerals/thaiU.gf b/examples/numerals/thaiU.gf
new file mode 100644
index 000000000..61777416d
--- /dev/null
+++ b/examples/numerals/thaiU.gf
@@ -0,0 +1,68 @@
+include numerals.Abs.gf ;
+
+-- Thai UTF8, produced from thai.gf by GF/Text/Thai.hs.  AR 28/12/2006
+
+flags coding=utf8 ;
+
+lincat 
+  Numeral    = {s : Str} ;
+  Digit      = {s : DForm => Str} ;
+  Sub10      = {s : DForm => Str} ;
+  Sub100     = {s : NForm => Str} ;
+  Sub1000    = {s : NForm => Str} ;
+  Sub1000000 = {s : Str} ;
+
+lin 
+  num x = x ;
+
+  pot01 = mkNum "หนึง" "หนึง" "เอ็ด" ;
+
+  n2 = mkNum  "สอง" "ยี่" "สอง" ;
+  n3 = regNum "สาม" ;
+  n4 = regNum "สี่" ;
+  n5 = regNum "หา่" ; 
+  n6 = regNum "หก" ;
+  n7 = regNum "เจ็ด" ;
+  n8 = regNum "แปด" ; 
+  n9 = regNum "เกา" ;
+
+
+  pot0 d = d ;
+
+  pot110 = {s = sip} ;
+  pot111 = {s = table {
+    Unit => ["ศิบเฏ"] ; 
+    Thousand => ["หนึงหมื่นหนึงพะน"]
+    }
+  } ;
+  pot1to19 d = {s = table {
+    Unit => "ศิบ" ++ d.s ! After ; 
+    Thousand => ["หนึงหมื่น"] ++ d.s ! Indep ++ "พะน"
+    }
+  } ;
+  pot0as1 d = {s = \\n => d.s ! Indep ++ phan ! n} ;
+  pot1 d = {s = \\n => d.s ! ModTen ++ sip ! n} ;
+  pot1plus d e = {
+    s = \\n => d.s ! ModTen ++ sip ! n ++ e.s ! After ++ phan ! n
+  } ;
+  pot1as2 n = n ;
+  pot2 d = {s = \\n => d.s ! Indep ++ roy ! n} ;
+  pot2plus d e = {s = \\n => d.s ! Indep ++ roy ! n ++ e.s ! n} ;
+  pot2as3 n = {s = n.s ! Unit} ;
+  pot3 n = {s = n.s ! Thousand} ;
+  pot3plus n m = {s = n.s ! Thousand ++ m.s ! Unit} ;
+
+param 
+  DForm = Indep | ModTen | After ;
+  NForm = Unit | Thousand ;
+
+oper 
+  mkNum : Str -> Str -> Str -> {s : DForm => Str} = \x,y,z ->
+    {s = table {Indep => x ; ModTen => y ; After => z}} ;
+  regNum : Str -> {s : DForm => Str} = \x -> 
+    mkNum x x x ;
+
+
+  sip  = table {Unit => "ศิบ"  ; Thousand => "หมื่น"} ;
+  roy  = table {Unit => "ร่อย" ; Thousand => "ซแน"} ;
+  phan = table {Unit => []      ; Thousand => "พะน"} ;
diff --git a/src/GF/Text/Thai.hs b/src/GF/Text/Thai.hs
index 8e344cc06..7fede0676 100644
--- a/src/GF/Text/Thai.hs
+++ b/src/GF/Text/Thai.hs
@@ -22,7 +22,7 @@ import Data.List
 
 
 mkThai :: String -> String
-mkThai = unwords . map mkThaiWord . words
+mkThai = concat . map mkThaiWord . words
 
 type ThaiChar = Char
 
@@ -35,6 +35,15 @@ mkThaiChar c = maybe 0 id $ Map.lookup c thaiMap
 thaiMap :: Map.Map String Int
 thaiMap = Map.fromList $ zip allThaiTrans allThaiCodes
 
+-- convert all string literals in a text
+
+thaiStrings :: String -> String
+thaiStrings s = case s of
+  '"':cs -> let (t,_:r) = span (/='"') cs in
+            '"':mkThai t ++ "\"" ++ thaiStrings r
+  c:cs -> c:thaiStrings cs
+  _ -> s
+
 
 -- each character is either [letter] or [letter+nonletter]
 
@@ -42,6 +51,7 @@ unchar :: String -> [String]
 unchar s = case s of
   c:d:cs 
    | isAlpha d -> [c]   : unchar (d:cs)
+   | d == '?'  ->         unchar cs -- use "o?" to represent implicit 'o'
    | otherwise -> [c,d] : unchar cs
   [_]          -> [s]
   _            -> []
@@ -52,7 +62,9 @@ allThaiTrans = words $
   "t1 t2 t3 n  d' t' t4 t5 t6 n  b  p  p1 f  p2 f' " ++
   "p3 m  y' r  -  l  -  w  s' r' s- h  l' O  h' -  " ++
   "a  a. a: a+ i  i: v  v: u  u: -  -  -  -  -  -  " ++
-  "e  e: o: a% a& "
+  "e  e: o: a% a& L  R  M  E  T  -  -  -  -  -  -  " ++
+  "N0 N1 N2 N3 N4 N5 N6 N7 N8 N9 -  -  -  -  -  -  "
+
 
 allThaiCodes :: [Int]
 allThaiCodes = [0x0e00 .. 0x0e7f]
@@ -73,6 +85,12 @@ testThai s = do
   putStrLn $ encodeUTF8 $ mkThai s
   putStrLn $ unwords $ map mkThaiPron $ words s
 
+thaiFile :: FilePath -> Maybe FilePath -> IO ()
+thaiFile f mo = do
+  s <- readFile f
+  let put = maybe putStr writeFile mo
+  put $ encodeUTF8 $ thaiStrings s
+
 mkThaiPron = concat . render . unchar where
   render s = case s of
     [c]  -> maybe c return (Map.lookup c thaiFinalMap): []
@@ -101,12 +119,13 @@ showThai s = case s of
 
 pronThai s = case s of
   [c,p]
+    | isUpper c && isDigit p -> show p
     | isDigit p   -> c:"h"
     | p==':'      -> c:[c]
     | elem p "%&" -> c:"y"
     | p=='+'      -> c:"m"
     | otherwise   -> [c]
-  "O"             -> ""
+  [c] | isUpper c -> "" --- O
   _ -> s
 
 hex = map hx . reverse . digs where
@@ -116,5 +135,7 @@ hex = map hx . reverse . digs where
 
 heights :: String
 finals  :: String
-heights = " MHHLLLLMHLLLLMMHLLLMMHLLLMMHHLLLLLL-L-LHHHHLML                   "
-finals  = " kkkkkkgt-tt-ntttttntttttnpp--pppmyn-n-wttt-n--                   "
+heights = 
+  " MHHLLLLMHLLLLMMHLLLMMHLLLMMHHLLLLLL-L-LHHHHLML" ++ replicate 99 ' '
+finals  = 
+  " kkkkkkgt-tt-ntttttntttttnpp--pppmyn-n-wttt-n--" ++ replicate 99 ' '