Merge pull request #44 from inariksit/arabic

Arabic: nouns and stuff

src/arabic/CatAra.gf

@@ -47,7 +47,7 @@ concrete CatAra of Cat = CommonX - [Utt]  ** open ResAra, Prelude, ParamX in {
 
 -- Noun
 
-    CN = ResAra.Noun ** {adj : NTable};
+    CN = ResAra.Noun ** {adj : NTable ; np : Case => Str};
     NP, Pron = ResAra.NP; --{s : Case => Str ; a : Agr } ;
     Num,
     Ord,
@@ -57,10 +57,7 @@ concrete CatAra of Cat = CommonX - [Utt]  ** open ResAra, Prelude, ParamX in {
     Det = ResAra.Det ;
 --  {s : Species => Gender => Case => Str ;
 --   d : State; n : Size; isNum : Bool } ;
-    Quant = {s : ResAra.Number => Species => Gender => Case => Str;
+    Quant = ResAra.Quant ;
-             d : State;
-             isNum : Bool;
-             isPron: Bool} ;
     Art = {s : ResAra.Number => Species => Gender => Case => Str;
              d : State} ;
 

src/arabic/LexiconAra.gf

@@ -234,7 +234,7 @@ flags
   write_V2     = dirV2 (regV "يَكتُب") ;
 --  write_V2     = dirV2 (v1 "كتب" a u) ;
   yellow_A     = clrA "صفر" ;
-  young_A      = sndA "شبّ" "فاعّ" ;
+  young_A      = sndA "شبب" "فاعّ" ;
 
   do_V2        = dirV2 (regV "يَفعَل") ;
 --  do_V2        = dirV2 (v1 "فعل" a a ) ;
@@ -287,7 +287,7 @@ flags
   fingernail_N = brkN "ظفر" "فُعل" "أَفاعِل"    Masc NoHum;
   fire_N       = brkN "نور" "فاع" "فِيعان"       Fem NoHum;
   flower_N     = brkN "زهر" "فَعلة" "فُعُول"     Fem NoHum;
-  fog_N        = brkN "ضبّ" "فَعال" "فَعال"     Masc NoHum; --no plural ?
+  fog_N        = brkN "ضبب" "فَعال" "فَعال"     Masc NoHum; --no plural ?
   foot_N       = brkN "قدم" "فَعَل" "أَفعال"    Fem NoHum;
   forest_N     = sdfN "غيب" "فاعة"              Fem NoHum;
   grass_N      = brkN "عشب" "فُعلة" "أَفعال"    Fem NoHum;
@@ -305,7 +305,7 @@ flags
   liver_N      = brkN "كبد" "فَعِل" "أَفعال"    Masc NoHum ;
   louse_N      = sdfN "قمل" "فَعلة"             Fem NoHum;
   mouth_N      = brkN "فوه" "فُعل" "أَفعال"     Masc NoHum ;
-  name_N       = brkN "؟سم" "فِعل" "فَعالِي"     Masc NoHum;
+  name_N       = brkN "ءسم" "فِعل" "فَعالِي"     Masc NoHum;
   neck_N       = brkN "رقب" "فَعَلة" "فِعال"    Fem NoHum;
   night_N      = brkN "ليل" "فَعلة" "فَعالِي"    Fem NoHum;  --plural?
   nose_N       = brkN "؟نف" "فَعل" "فُعُول"      Masc NoHum;

src/arabic/MorphoAra.gf

@@ -7,9 +7,10 @@ flags optimize = all ;--noexpand;
 
     mkDet : Str -> Number -> State -> Det
       = \word,num,state ->
-      { s = \\_,_,c => word + vowel ! c ;
+      { s = \\_,_,c => word + caseTbl ! c ;
         n = numberToSize num;
         d = state;  --only Const is used now. check StructuralAra
+        is1sg = False;
         isNum = False;
         isPron = False
       };
@@ -18,7 +19,7 @@ flags optimize = all ;--noexpand;
       = \word,decl ->
       { s = \\c =>
           case decl of {
-            True => word + vowel!c;
+            True => word + caseTbl!c;
             False => word
           };
         isDecl = decl
@@ -40,11 +41,4 @@ flags optimize = all ;--noexpand;
         isNum = True
       };
 
-    vowel : Case => Str =
-      table {
-        Nom =>  "ُ";
-        Acc =>  "َ";
-        Gen =>  "ِ"
-      };
-
 }

src/arabic/NounAra.gf

@@ -6,18 +6,26 @@ lin
 
   DetCN det cn = let {
     number = sizeToNumber det.n;
+    state = possState det.d;
     determiner : Case -> Str = \c ->
       det.s ! cn.h ! (detGender cn.g det.n) ! c;
-    noun : Case -> NTable -> Str = \c,nt -> nt !
+    noun : Case -> NTable -> Str = \c,nt -> 
-      number ! (nounState det.d number) ! (nounCase c det.n det.d)
+      let cas = if_then_else Case det.is1sg Bare c -- no case vowel with 1sg poss. suff.
+       in nt ! number 
+             ! nounState det.d number 
+             ! nounCase cas det.n det.d
     } in {
       s = \\c =>
         case cnB4det det.isPron det.isNum det.n det.d of {
-          False => determiner c ++ noun c cn.s ++ noun c cn.adj ;
+          False => determiner c 
-          --FIXME use the adj -> cn -> cn rule from below instead of
+                ++ noun c cn.s -- deal with poss. suffix
-          --repeating code
+                ++ cn.adj ! number ! state ! c -- normal case+state
-          True => cn.s ! number ! det.d ! c ++ det.s ! cn.h ! cn.g ! c
+                ++ cn.np ! c ;
-            ++ cn.adj ! number ! det.d ! c
+          True => noun c cn.s -- deal with poss. suffix
+            --   ++ determiner c    -- or this? 
+               ++ det.s ! cn.h ! cn.g ! c
+               ++ cn.adj ! number ! state ! c -- normal case+state
+               ++ cn.np ! c
         };  
       a = { pgn = agrP3 cn.h cn.g number;
             isPron = False }
@@ -70,36 +78,33 @@ lin
     } ;
 -}
 
-  DetQuantOrd quant num ord = {
+  DetQuantOrd quant num ord = quant ** {
     s = \\h,g,c => quant.s ! Pl ! h ! g ! c
       ++ num.s ! g ! (toDef quant.d num.n) ! c
       --FIXME check this:
       ++ ord.s ! g ! (toDef quant.d num.n) ! c ;
     n = num.n;
-    d = quant.d;
+    isNum = orB num.isNum ord.isNum ;
-    isPron = quant.isPron;
+    -- ord may come from OrdDigits or OrdNumeral
-    isNum = case num.n of {
+    -- num may come from NumCard : Card -> Num
-        None => ord.isNum ; -- ord may come from OrdDigits or OrdNumeral
+
-        _    => True
-      }
     } ;
 
-  DetQuant quant num = {
+  DetQuant quant num = quant ** {
     s = \\h,g,c => quant.s ! Pl ! h ! g ! c
       ++ num.s ! g ! (toDef quant.d num.n) ! c ;
     n = num.n;
-    d = quant.d;
-    isPron = quant.isPron;
     isNum = -- Num may come from NumCard : Card -> Num
       case num.n of {
         None => False;
-        _    => True
+        _    => num.isNum
       }
     } ;
 
   PossPron p = {
     s = \\_,_,_,_ => p.s ! Gen;
-    d = Const;
+    d = Poss;
+    is1sg = case p.a.pgn of { Per1 _ => True ; _ => False } ;
     isPron = True;
     isNum = False } ;
 
@@ -152,13 +157,13 @@ lin
   DefArt = {
     s = \\_,_,_,_ => [];
     d = Def ;
-    isNum,isPron = False
+    isNum,isPron,is1sg = False
     } ;
 
   IndefArt = {
     s = \\_,_,_,_ => [];
     d = Indef ;
-    isNum,isPron = False
+    isNum,isPron,is1sg = False
     } ;
 
   MassNP cn = ---- AR
@@ -169,7 +174,9 @@ lin
 --             isNum = False; isPron = False} ;
 
   UseN,
-  UseN2 = \n -> n ** {adj = \\_,_,_ => []};
+  UseN2 = \n -> n ** {
+    adj = \\_,_,_ => [];
+    np  = \\_     => []};
   Use2N3 n3 = n3 ;
   Use3N3 n3 = n3 ** {c2 = n3.c3} ;
 
@@ -179,16 +186,12 @@ lin
 
   ComplN3 n3 np = ComplN2 n3 np ** {c2 = n3.c3} ;
 
-  AdjCN ap cn = {
+  AdjCN ap cn = cn ** {
-    s = \\n,d,c => cn.s ! n ! d ! c;
+    adj = \\n,d,c => ap.s ! cn.h ! cn.g ! n ! (definite ! d) ! c 
-    adj = \\n,d,c => ap.s ! cn.h ! cn.g ! n ! (definite ! d) ! c ;
-    g = cn.g;
-    h = cn.h
     };
   --    RelCN cn rs = {s = \\n,c => cn.s ! n ! c ++ rs.s ! {n = n ; p = P3}} ;
   --    AdvCN cn ad = {s = \\n,c => cn.s ! n ! c ++ ad.s} ;
   --
   --    SentCN cn sc = {s = \\n,c => cn.s ! n ! c ++ sc.s} ;
-  ApposCN cn np = cn ** {
+  ApposCN cn np = cn ** { np = np.s } ;
-    s = \\n,d,c => cn.s ! n ! d ! c ++ np.s ! c } ;
 }

src/arabic/ParadigmsAra.gf

@@ -428,9 +428,9 @@ resource ParadigmsAra = open
   mkPron : (_,_,_ : Str) -> PerGenNum -> NP = \ana,nI,I,pgn ->
     { s =
         table {
-          Nom => ana;
           Acc => nI;
-          Gen => I
+          Gen => I;
+          _   => ana
         };
       a = {pgn = pgn; isPron = True };
       lock_NP = <>
@@ -451,7 +451,7 @@ resource ParadigmsAra = open
     };
 
   mkQuant7 : (_,_,_,_,_,_,_ : Str) -> State -> Quant =
-    \hava,havihi,havAn,havayn,hAtAn,hAtayn,hA'ulA,det ->
+    \hava,havihi,havAn,havayn,hAtAn,hAtayn,hA'ulA,det -> lin Quant (baseQuant **
     { s = \\n,s,g,c =>
         case <s,g,c,n> of {
           <_,Masc,_,Sg>  => hava;
@@ -463,14 +463,11 @@ resource ParadigmsAra = open
           <Hum,_,_,Pl>   => hA'ulA;
           _              => havihi
         };
-      d = Def;
+      d = det
-      isPron = False;
+    });
-      isNum = False;
-      lock_Quant = <>
-    };
 
   mkQuant3 : (_,_,_ : Str) -> State -> Quant =
-    \dalika,tilka,ula'ika,det ->
+    \dalika,tilka,ula'ika,det -> lin Quant (baseQuant **
     { s = \\n,s,g,c =>
         case <s,g,c,n> of {
           <_,Masc,_,Sg>  => dalika;
@@ -478,11 +475,8 @@ resource ParadigmsAra = open
           <Hum,_,_,_>   => ula'ika;
           _              => tilka
         };
-      d = det;
+      d = det
-      isPron = False;
+    });
-      isNum = False;
-      lock_Quant = <>
-    };
 
   degrA : (posit,compar,plur : Str) -> A
     = \posit,compar,plur -> lin A {s = clr posit compar plur} ;

src/arabic/ResAra.gf

@@ -17,10 +17,14 @@ resource ResAra = PatternsAra ** open  Prelude, Predef, OrthoAra, ParamX  in {
     Vowel   = u | a | i ;
     Number  = Sg | Dl | Pl;
     Gender  = Masc | Fem ;
-    Case    = Nom | Acc | Gen ;
+    Case    = Nom | Acc | Gen 
+            | Bare ; -- 1st person poss. suff. overrides case
     Person  = P1 | P2 | P3 ;
     Species = NoHum | Hum ;
-    State   = Def | Indef | Const ;
+    State   = Def | Indef | Const
+            | Poss ; -- ة turns into ت
+                     -- sound masculine plural drops ن
+                     -- case vowel retained
     Mood    = Ind | Cnj | Jus ;
     Voice   = Act | Pas ;
     Tense   = Pres | Past | Fut ;
@@ -783,6 +787,14 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
 
 --Nominal Morphology
 
+  caseTbl : Case => Str =
+    table {
+      Bare => [] ;
+      Nom  => "ُ";
+      Acc  => "َ";
+      Gen  => "ِ"
+    };
+
 --takes the adjective lemma and gives the Posit table
   positAdj : Str -> Gender => NTable  =
     \kabIr ->
@@ -848,20 +860,22 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
       \\s,c => defArt s (case word of {
          lemma + "ِي"  => fixShd lemma (dec2sg ! s ! c) ;
          _ + ("ا"|"ى") => fixShd word  (dec3sg ! s ! c) ;
+         lemma + "ة"   => case s of { 
+                            Poss => lemma + "ت" + dec1sg ! s ! c ;
+                            _    => word        + dec1sg ! s ! c
+                          } ;
          _             => fixShd word  (dec1sg ! s ! c) 
       }) ;
 
 
     -- takes a singular word and tests the ending to
     -- determine the declension and gives the corresponding dual inf table
-    dual : Str -> State => Case => Str =
+    dual : Str -> State => Case => Str = \caSaA ->
-      \caSaA ->
+      \\s,c => defArt s (case caSaA of {
-      case caSaA of {
+        lemma + ("ا"|"ى") => lemma + "ي" + dl ! s ! c ;
-        lemma + ("ا"|"ى") => \\s,c => defArt s lemma + "ي" + dl ! s ! c ;
+        lemma + "ة"       => lemma + "ت" + dl ! s ! c ;
-        lemma + "ة" =>
+        _                 => fixShd caSaA (dl ! s ! c)
-          \\s,c => defArt s (lemma + "ت") + dl ! s ! c ;
+      });
-        _  => \\s,c => defArt s caSaA + dl ! s ! c
-      };
 
     -- takes a singular word and gives the corresponding sound
     --plural feminine table
@@ -898,52 +912,39 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
       table {
         Indef =>
           table {
+            Bare => [];
             Nom => "ٌ";
             Acc => "ً";
             Gen => "ٍ"
           };
-        _ =>
+        _ => caseTbl --think of ?axU, ?axA, (the five nouns)
-          table { --think of ?axU, ?axA, (the five nouns)
+
-            Nom => "ُ";
-            Acc => "َ";
-            Gen => "ِ"
-          }
       };
 
     --indeclinables (mamnuu3 mina S-Sarf)
     indecl :  Case => Str =
       table {
-        Nom => "ُ";
+        Gen => "َ" ;
-        _ => "َ"
+        x   => caseTbl ! x
       };
 
 
-    --declection 2 (ends in yaa')
+    --declension 2 (ends in yaa')
-    dec2sg : State => Case => Str =
+    dec2sg : State => Case => Str = \\s,c =>
-      table {
+      case <s,c> of {
-        Indef =>
+        <_,   Bare> => [] ;
-          table {
+        <Indef,Acc> => "ِياً" ;
-            Acc => "ِياً";
+        <Indef>     => "ٍ" ;
-            _ => "ٍ"
+        <_,    Acc> => "ِيَ" ;
-          };
-        _ =>
-          table {
-            Acc => "ِيَ";
         _           => "ِي"
-          }
       };
 
-    --declention 3 (ending in alif)
+    --declension 3 (ending in alif)
-    dec3sg : State => Case => Str =
+    dec3sg : State => Case => Str = \\s,c =>
-      table {
+      case <s,c> of {
-        Indef =>
+        <Indef,Bare> => [] ;
-          table {
+        <Indef>      => "ً" ;
-            _ => "ً"
+        _            => []
-          };
-        _ =>
-          table {
-            _ => ""
-          }
       };
 
 
@@ -958,11 +959,12 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
         _ =>
           table {
             Nom  => "َانِ";
+            Bare => "َيْن"; 
             _    => "َيْنِ"
           }
       };
 
-    --sound mascualine plural suffixes
+    --sound masculine plural suffixes
     m_pl : State => Case => Str =
       table {
         Const =>
@@ -972,6 +974,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
           };
         _ =>
           table {
+            Bare => "ِين"; 
             Nom => "ُونَ";
             _   => "ِينَ"
           }
@@ -982,11 +985,13 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
       table {
         Indef =>
           table {
+            Bare => [];
             Nom => "ٌ";
             _   => "ٍ"
           };
         _ =>
           table {
+            Bare => [];
             Nom => "ُ";
             _   => "ِ"
           }
@@ -1030,11 +1035,14 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
       case <s,n> of {
         <Const,Pl> => Def;   --kullu l-kutubi, bacDu l-kutubi
         <Const,Sg> => Indef; --kullu kitaabin
-        <Indef,_> => Indef;  --kitaabun
+        <Indef> => Indef;    --kitaabun
+        <Poss>  => Poss;
         _       => Def       --Lkitaabu
       };
 
-
+    possState : State -> State = \s ->
+      case s of { Poss => Def ;
+                  x    => x } ;
     --FIXME needs testing
     nounCase : Case -> Size -> State -> Case =
       \c,size,s ->
@@ -1112,14 +1120,24 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
 
   oper
 
-    Det : Type = {
+    BaseQuant : Type = {
-      s : Species => Gender => Case => Str ;
       d : State;
-      n : Size;
+      is1sg : Bool; -- To force no case marker for 1st person poss. suff.
       isNum : Bool;
       -- for genitive pronouns (suffixes). if true, then "cn ++ det"
       --should be used instead of "det ++ cn" when constructing the NP
-      isPron : Bool
+      isPron: Bool} ;
+
+    baseQuant = { d = Indef ;
+                  is1sg,isNum,isPron = False } ;
+
+    Quant : Type = BaseQuant ** {
+      s : ResAra.Number => Species => Gender => Case => Str
+      } ;
+
+    Det : Type = BaseQuant ** {
+      s : Species => Gender => Case => Str ;
+      n : Size
       } ;
 
     Predet : Type = {

src/arabic/StructuralAra.gf

@@ -98,6 +98,7 @@ concrete StructuralAra of Structural = CatAra **
   where_IAdv = ss "أَينَ" ;
   which_IQuant = {
     s = \\s,c => case <c,s> of {
+             <Bare,_>    => "أيّ" ;
              <Nom,Indef> => "أيٌّ" ;
              <Nom,_>     => "أيُّ" ;
              <Acc,Indef> => "أيّاً" ;