Merge pull request #77 from inariksit/arabic

Arabic

src/arabic/AdverbAra.gf

@@ -14,7 +14,8 @@ concrete AdverbAra of Adverb = CatAra ** open ResAra, Prelude in {
 
 --    AdAdv = cc2 ;
 --
---    SubjS = cc2 ;
+   -- : Subj -> S -> Adv ;              -- when she sleeps
+   SubjS subj s = {s = subj.s ++ s.s ! Subord} ;
 --    AdvSC s = s ; --- this rule give stack overflow in ordinary parsing
 --
 --    AdnCAdv cadv = {s = cadv.s ++ "تهَن"} ;

src/arabic/LexiconAra.gf

@@ -73,7 +73,7 @@ flags
   door_N       = brkN "بوب" "فاع" "أَفعَال" Masc NoHum ;
   drink_V2     = dirV2 (regV "شَرِب") ;
 --  drink_V2     = dirV2 (v1 "شرب" i a) ;
-  easy_A2V     = mkA2 (sndA "سهل" "فَعل") "لِ" ;
+  easy_A2V     = mkA2 (sndA "سهل" "فَعل") datPrep ;
   eat_V2       = dirV2 (mkV "ءكل" FormI) ;
   empty_A      = sndA "فرغ" "فاعِل" ;
   enemy_N      = brkN "عدو" "فَعُلّ" "أَفعَاء" Masc Hum ;
@@ -203,7 +203,7 @@ flags
   switch8off_V2 = dirV2 (v4 "طفء") ;
   switch8on_V2 = dirV2 (v4 "شعل") ;
   table_N      = sdfN "طول" "فاعِلة" Fem NoHum ;
-  talk_V3      = mkV3 (v5 "حدث") "لِ" "عَن" ;
+  talk_V3      = mkV3 (v5 "حدث") datPrep (mkPrep "عَن") ;
   teacher_N    = sdmN "علم" "مُفَعِّل" Masc Hum ; --mucal~imö
   teach_V2     = dirV2 (v2 "علم") ;
   television_N = mkN (sndf "تِلِفِزيُون") Masc NoHum ;

src/arabic/MorphoAra.gf

@@ -34,7 +34,7 @@ flags optimize = all ;--noexpand;
           case g of {
             Masc => waHid;
             Fem => waHida
-          } in defArt state waHid + word + dec1sg ! state ! c;
+          } in defArt state c waHid + word + dec1sg ! state ! c;
         n = num;
         d = state;
         isPron = False;

src/arabic/ResAra.gf

@@ -18,7 +18,8 @@ resource ResAra = PatternsAra ** open  Prelude, Predef, OrthoAra, ParamX  in {
     Number  = Sg | Dl | Pl;
     Gender  = Masc | Fem ;
     Case    = Nom | Acc | Gen
-            | Bare ; -- 1st person poss. suff. overrides case
+            | Bare -- 1st person poss. suff. overrides case
+            | Dat ; -- Hack to make the preposition لِ contract
     Species = NoHum | Hum ;
     State   = Def | Indef | Const
             | Poss ; -- ة turns into ت
@@ -126,6 +127,7 @@ resource ResAra = PatternsAra ** open  Prelude, Predef, OrthoAra, ParamX  in {
     } ;
 
     noPrep : Preposition = mkPreposition [] Nom ;
+    datPrep : Preposition = mkPreposition ("لِ" ++ BIND) Dat ;
 
     Adj  : Type = {s : AForm => Str} ;
     Adj2 : Type = Adj ** {c2 : Preposition} ;
@@ -901,7 +903,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
       Bare => [] ;
       Nom  => "ُ";
       Acc  => "َ";
-      Gen  => "ِ"
+      _Gen  => "ِ" -- dat is the same as gen, except in definite before لِ
     };
 
 --takes the adjective lemma and gives the Posit table
@@ -931,7 +933,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
 
   -- indeclinable nominal word (mamnuu3 mina S-Sarf)
   indeclN : Str -> State => Case => Str =
-    \aHmar -> \\s,c => defArt s aHmar + indecl!c;
+    \aHmar -> \\s,c => defArt s c aHmar + indecl!c;
 
     -- takes 2 words, singular and broken plural, and gives the
     -- complete noun inflection table
@@ -966,7 +968,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
     -- takes a singular or broken plural word and tests the ending to
     -- determine the declension and gives the corresponding inf table
     sing : Str -> State => Case => Str = \word ->
-      \\s,c => defArt s (case word of {
+      \\s,c => defArt s c (case word of {
         lemma + "ِيّ" => fixShd word  (decNisba ! s ! c) ;
         lemma + "ِي"  => fixShd lemma (dec2sg ! s ! c) ;
         _ + ("ا"|"ى") => fixShd word  (dec3sg ! s ! c) ;
@@ -981,7 +983,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
     -- takes a singular word and tests the ending to
     -- determine the declension and gives the corresponding dual inf table
     dual : Str -> State => Case => Str = \caSaA ->
-      \\s,c => defArt s (case caSaA of {
+      \\s,c => defArt s c (case caSaA of {
         lemma + ("ا"|"ى") => lemma + "ي" + dl ! s ! c ;
         lemma + "ة"       => lemma + "ت" + dl ! s ! c ;
         _                 => fixShd caSaA (dl ! s ! c)
@@ -991,13 +993,13 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
     --plural feminine table
     plurF : Str -> State => Case => Str =
       \kalima ->
-      \\s,c => defArt s (mkAt kalima) + f_pl ! s ! c ;
+      \\s,c => defArt s c (mkAt kalima) + f_pl ! s ! c ;
 
     -- takes a singular word and gives the corresponding sound
     --plural masculine table. FIXME: consider declension 3
     plurM : Str -> State => Case => Str =
       \mucallim ->
-      \\s,c => defArt s mucallim + m_pl ! s ! c ;
+      \\s,c => defArt s c mucallim + m_pl ! s ! c ;
 
     -- to add the Al prefix for Definite words
     Al : State => Str =
@@ -1006,13 +1008,14 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
         _   => ""
       };
 
-    defArt : State -> Str -> Str = \st,stem -> -- IL -- to be checked
+    defArt : State -> Case -> Str -> Str = \st,c,stem -> -- IL -- to be checked
       let al = "ال" in
-      case st of {
+      case <st,c> of {
-        Def =>
+        <Def,Dat> => "ل" + stem ; -- only happens before the preposition لِ
+        <Def> =>
           case stem of {
-            s@#sun + x          => fixShd (al + s) ("ّ" + x) ;
+            s@#sun + x  => fixShd (al + s) ("ّ" + x) ;
-            x                   => al + x } ;
+            x           => al + x } ;
         _   => stem
       };
 
@@ -1024,7 +1027,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
             Bare => [];
             Nom => "ٌ";
             Acc => "ً";
-            Gen => "ٍ"
+            _Gen => "ٍ"
           };
         _ => caseTbl --think of ?axU, ?axA, (the five nouns)
 
@@ -1033,8 +1036,8 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
     --indeclinables (mamnuu3 mina S-Sarf)
     indecl :  Case => Str =
       table {
-        Gen => "َ" ;
+        (Gen|Dat) => "َ" ;
-        x   => caseTbl ! x
+        x         => caseTbl ! x
       };
 
 
@@ -1284,9 +1287,10 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
     mkPron : (_,_,_ : Str) -> PerGenNum -> NP = \ana,nI,I,pgn ->
      { s =
         table {
+          Nom => ana;
           Acc => BIND ++ nI; -- object suffix
-          Gen => BIND ++ I;  -- possessive suffix
+          Gen => BIND ++ I;   -- possessive suffix
-          _   => ana
+          Dat => I -- will only be used with preposition لِ, which already has a BIND
         };
       a = {pgn = pgn; isPron = True };
       empty = []
@@ -1612,16 +1616,16 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
             NCard => table {
               Masc => \\s,c => (sing wAhid) ! s ! c ;
               --all fem are first declension:
-              Fem => \\s,c => defArt s wAhida + dec1sg ! s ! c
+              Fem => \\s,c => defArt s c wAhida + dec1sg ! s ! c
               };
             NOrd => table {
-              Masc => \\s,c => defArt s awwal + dec1sg ! s ! c;
+              Masc => \\s,c => defArt s c awwal + dec1sg ! s ! c;
               Fem => \\s,c => (sing Ula) ! s ! c
               }
             };
           ten => table {
-            NCard => \\_,s,c => defArt s wAhid + m_pl ! Indef ! c;
+            NCard => \\_,s,c => defArt s c wAhid + m_pl ! Indef ! c;
-            NOrd => \\_,s,c => defArt s awwal + m_pl ! Indef ! c
+            NOrd => \\_,s,c => defArt s c awwal + m_pl ! Indef ! c
             }
           }
       };

src/arabic/StructuralAra.gf

@@ -94,7 +94,7 @@ concrete StructuralAra of Structural = CatAra **
   to_Prep = mkPrep "إِلى" ;
   under_Prep = mkPrep "تَحْتَ" ;
 --  very_AdA = ss "ثري" ;
---  want_VV = P.mkVV (P.regV "وَنت") ;
+  want_VV = mkVV (mkV "رود" FormIV) ;
   we_Pron = ResAra.we_Pron ;
   whatPl_IP = mkIP "ما" "ماذا" Pl ;
   whatSg_IP = mkIP "ما" "ماذا" Sg ;
@@ -108,8 +108,8 @@ concrete StructuralAra of Structural = CatAra **
              <Nom,_>     => "أيُّ" ;
              <Acc,Indef> => "أيّاً" ;
              <Acc,_>     => "أيَّ" ;
-             <Gen,Indef> => "أيٍّ" ;
+             <_Gen,Indef> => "أيٍّ" ;
-             <Gen,_>     => "أيِّ"
+             <_Gen,_>     => "أيِّ"
              }
     } ;
   whoSg_IP = mkIP "مَنْ" "مَنْ" Sg ;