finnish updates

examples/uusisuomi/Makefile

@@ -21,6 +21,10 @@ nouns:
 	cat all-diff-* >all-differences
 	cat all-differences
 
+CSC: 
+	export LEX=NCSC ; make -e tests
+	cat all-diff-NCSC
+
 gf-files:
 	runghc MkLex.hs 0 $(CAT) $(LEX) > $(LEX)Abs.gf
 	runghc MkLex.hs 1 $(CAT) $(LEX) > $(LEX)1.gf

examples/uusisuomi/MkLex.hs

@@ -38,7 +38,23 @@ initiate tgt cat i = mapM_ putStrLn [
      ]
 
 nums = map prt [1 ..] where
-  prt i = (if i < 10 then "0" else "") ++ show i ++ ". "
+----  prt i = (if i < 10 then "0" else "") ++ show i ++ ". "
+  prt i = let n = show i in replicate (4-length n) '0' ++ n ++ ". "
+
+-- W is the flag for mixed-class word lists
+mkLex "W" 0 line = case words line of
+  num:cat:sana:_ -> do
+    let nimi = "n" ++ init num ++ "_" ++ sana 
+    putStrLn $ "fun " ++ nimi ++ "_" ++ cat ++ " : " ++ cat ++ " ;"
+  _ -> return ()
+
+mkLex "W" 1 line = case words line of
+  num:cat:sanat@(sana:_) -> do
+    let nimi = "n" ++ init num ++ "_" ++ sana 
+    putStrLn $ "lin " ++ nimi ++ 
+               "_" ++ cat ++ " = mk" ++ cat ++ " " ++ 
+               unwords (map prQuoted sanat) ++" ;"
+  _ -> return ()
 
 mkLex cat 0 line = case words line of
   num:sana:_ -> do
@@ -68,6 +84,7 @@ mkLex "N" 2 line = case words line of
   _ -> return ()
 
 mkLex "N" 3 line = case words line of
+----  num:sana:sanan:sanoja:_ -> do
   num:sana:sanan:_:_:_:_:sanoja:_ -> do
     let nimi = "n" ++ init num ++ "_" ++ sana 
     putStrLn $ "lin " ++ nimi ++ 
@@ -82,7 +99,6 @@ mkLex "N" 4 line = case words line of
                  "\" \"" ++ sanoja ++ "\" \"" ++ sanaa ++ "\" ;"
   _ -> return ()
 
-
 -- to initiate from a noun list that has compounds
 
 mkLex "N" 11 line = case words line of
@@ -94,6 +110,8 @@ mkLex "N" 11 line = case words line of
     putStrLn $ "lin " ++ nimi ++ "_N = mkN \"" ++ sana ++ "\" ;"
   _ -> return ()
 
+prQuoted s = concat ["\"",s,"\""]
+
 -- from sora+tie to tie
 
 uncompound = reverse . takeWhile (/= '+') . reverse

examples/uusisuomi/Nominal.gf

@@ -44,6 +44,9 @@ resource Nominal = ResFin ** open MorphoFin,Declensions,CatFin,Prelude in {
     --  \s,t,u -> nForms2N (nForms3 s t u) ;
     mkN : (talo,talon,taloja,taloa : Str) -> N = mk4N ;
     --  \s,t,u,v -> nForms2N (nForms4 s t u v) ;
+    mkN : 
+      (talo,talon,taloa,talona,taloon,talojen,taloja,taloina,taloissa,taloihin
+        : Str) -> N = mk10N ;
     mkN : (sora : Str) -> (tie : N) -> N = mkStrN ;
     mkN : (oma,tunto : N) -> N = mkNN ;
   } ;
@@ -56,6 +59,11 @@ resource Nominal = ResFin ** open MorphoFin,Declensions,CatFin,Prelude in {
   mk3N : (talo,talon,taloja : Str) -> N = \s,t,u -> nForms2N (nForms3 s t u) ;
   mk4N : (talo,talon,taloa,taloja : Str) -> N = \s,t,u,v -> 
       nForms2N (nForms4 s t u v) ;
+  mk10N : 
+      (talo,talon,taloa,talona,taloon,talojen,taloja,taloina,taloissa,taloihin
+        : Str) -> N = \a,b,c,d,e,f,g,h,i,j -> 
+        nForms2N (nForms10 a b c d e f g h i j) ;
+
   mkStrN : Str -> N -> N = \sora,tie -> {
     s = \\c => sora + tie.s ! c ; lock_N = <>
     } ;
@@ -68,6 +76,7 @@ resource Nominal = ResFin ** open MorphoFin,Declensions,CatFin,Prelude in {
       ukk = init ukko ;
       uko = weakGrade ukko ;
       ukon = uko + "n" ;
+      o = case last ukko of {"ä" => "ö" ; "a" => "o"} ; -- only used then 
       renka = Declensions.strongGrade (init ukko) ;
       rake = Declensions.strongGrade ukko ;
     in
@@ -83,14 +92,29 @@ resource Nominal = ResFin ** open MorphoFin,Declensions,CatFin,Prelude in {
       _ + "e" => dRae ukko (rake + "en") ;
       _ + ("ut" | "yt") => dRae ukko (ukk + "en") ;
       _ + ("as" | "äs") => dRae ukko (renka + last renka + "n") ;
-      _ + ("uus" | "yys") => dLujuus ukko ;
+      _ + ("uus" | "yys" | "eus" | "eys") => dLujuus ukko ;
       _ + "s" => dJalas ukko ; 
-      _ + ("a" | "e" | "i") + C_ + _ + "aja" =>  -- opettaja correct autom. 
+
+{- heuristics for 3-syllable nouns ending a/ä
+      _ + ("a" | "e" | "i" | "o" | "u" | "y" | "ä" | "ö") + C_ + 
+          _ + "i" + C_ + a@("a" | "ä") =>  
+          dSilakka ukko (ukko + "n") (ukk + o + "it" + a) ;
+      _ + ("a" | "e" | "i" | "o" | "u" | "y" | "ä" | "ö") + C_ + _ + 
+          ("a" | "e" | "o" | "u" | "y" | "ä" | "ö") + 
+          ("l" | "r" | "n") + a@("a" | "ä") =>  
+          dSilakka ukko (ukko + "n") (ukk + o + "it" + a) ;
+      _ + ("a" | "e" | "i" | "o" | "u" | "y" | "ä" | "ö") + C_ + _ + 
+          ("a" | "e" | "i" | "o" | "u" | "y" | "ä" | "ö") + 
+          ("n" | "k" | "s") + "k" + a@("a" | "ä") =>  
+          dSilakka ukko (uko + "n") (init uko + o + "it" + a) ;
+      _ + ("a" | "e" | "i" | "o" | "u" | "y" | "ä" | "ö") + C_ + _ + 
+          ("a" | "e" | "i" | "o" | "u" | "y" | "ä" | "ö") + 
+          ("n" | "t" | "s") + "t" + a@("a" | "ä") =>  
+          dSilakka ukko (uko + "n") (ukk + o + "j" + a) ;
+      _ + ("a" | "e" | "i" | "o" | "u") + C_ + _ + 
+          ("a" | "e" | "o" | "u") + C_ + "a" =>  
           dSilakka ukko (ukko + "n") (ukk + "ia") ;
-      _ + ("a" | "e" | "i" | "o" | "u") + C_ + _ + "ija" =>  
-          dSilakka ukko (ukko + "n") (ukk + "oita") ;
-      _ + ("e" | "i" | "y" | "ä" | "ö") + C_ + _ + "ijä" =>  
-          dSilakka ukko (ukko + "n") (ukk + "öitä") ;
+-}
       _ + "i" +o@("o"|"ö") => dSilakka ukko (ukko+"n") (ukko+"it"+getHarmony o);
       _ + "i" + "a" => dSilakka ukko (ukko + "n") (ukk + "oita") ;
       _ + "i" + "ä" => dSilakka ukko (ukko + "n") (ukk + "öitä") ;

examples/uusisuomi/Verbal.gf

@@ -10,6 +10,9 @@ resource Verbal = ResFin **
   mkV = overload {
     mkV : (huutaa : Str) -> V = mk1V ;
     mkV : (huutaa,huusi : Str) -> V = mk2V ;
+    mkV : (
+      huutaa,huudan,huutaa,huutavat,huutakaa,huudetaan,
+      huusin,huusi,huusisi,huutanut,huudettu,huutanee : Str) -> V = mk12V ;
   } ;
 
   showV : V -> Utt = \v -> ss (
@@ -29,6 +32,12 @@ resource Verbal = ResFin **
   mk1V : Str -> V = \s -> vforms2V (vForms1 s) ;
   mk2V : (_,_ : Str) -> V = \s,t -> vforms2V (vForms2 s t) ;
 
+  mk12V : (
+      huutaa,huudan,huutaa,huutavat,huutakaa,huudetaan,
+      huusin,huusi,huusisi,huutanut,huudettu,huutanee : Str) -> V = 
+     \a,b,c,d,e,f,g,h,i,j,k,l -> 
+        vforms2V (vForms12 a b c d e f g h i j k l) ;
+
   vForms1 : Str -> VForms = \ottaa ->
     let
       a = last ottaa ;

lib/resource/arabic/LangAra.gf

@@ -5,7 +5,7 @@ concrete LangAra of Lang =
   LexiconAra
   ** {
 
-  flags startcat = Phr ; unlexer = text ; lexer = text ;
+  flags startcat = Phr ; unlexer = text ; lexer = text ; coding = utf8 ;
 
 }