Merge pull request #78 from inariksit/arabic

Arabic

src/arabic/AdverbAra.gf

@@ -2,22 +2,24 @@ concrete AdverbAra of Adverb = CatAra ** open ResAra, Prelude in {
   flags coding=utf8;
 
   lin
+
     PositAdvAdj a = {s = a.s ! APosit Masc Sg Indef Acc} ;
---    ComparAdvAdj cadv a np = {
+    -- ComparAdvAdj cadv a np = {
---      s = cadv.s ++ a.s ! AAdv ++ "مِنْ" ++ np.s ! Gen
+    --   s = cadv.s ++ a.s ! AAdv ++ "مِنْ" ++ np.s ! Gen
---      } ;
+    --   } ;
---    ComparAdvAdjS cadv a s = {
+    -- ComparAdvAdjS cadv a s = {
---      s = cadv.s ++ a.s ! AAdv ++ "تهَن" ++ s.s
+    --   s = cadv.s ++ a.s ! AAdv ++ "مِنْ" ++ s.s
---      } ;
+    --   } ;
 
     PrepNP prep np = {s = prep.s ++ np.s ! prep.c} ;
 
---    AdAdv = cc2 ;
+    AdAdv ad av = cc2 av ad ;
---
+
-   -- : Subj -> S -> Adv ;              -- when she sleeps
+    -- : Subj -> S -> Adv ;              -- when she sleeps
-   SubjS subj s = {s = subj.s ++ s.s ! Subord} ;
+    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"} ;
+    AdnCAdv cadv = {s = cadv.s ++ "مِنْ"} ;
---
+
 }

src/arabic/IdiomAra.gf

@@ -49,14 +49,12 @@ concrete IdiomAra of Idiom = CatAra ** open
   -- : VP -> VP ;        -- is at home himself; is himself at home
   SelfAdvVP,
   SelfAdVVP = \vp -> vp ** {
-    s = \\pgn,vf => let pron : ResAra.NP = pgn2pron pgn in
+    s = \\pgn,vf => vp.s ! pgn ! vf ++ reflPron Nom pgn
-        vp.s ! pgn ! vf ++ refl ! Nom ++ pron.s ! Gen
     } ;
 
   -- : NP -> NP ;        -- the president himself (is at home)
   SelfNP np = np ** {
-    s = let pron : ResAra.NP = np2pron np ;
+    s = \\c => np.s ! c ++ reflPron c (np.a.pgn)
-        in \\c => np.s ! c ++ refl ! c ++ pron.s ! Gen
     } ;
 
 }

src/arabic/ParadigmsAra.gf

@@ -185,51 +185,33 @@ resource ParadigmsAra = open
 -- Overloaded operations
 
   mkV : overload {
-    mkV : (imperfect : Str) -> V ;
+    mkV : (imperfect : Str) -> V ; -- The verb in Per3 Sg Masc imperfect tense gives the most information
     mkV : (root : Str) -> (perf,impf : Vowel) -> V ; -- verb form I ; vowel = a|i|u
     mkV : (root : Str) -> VerbForm -> V ;  -- FormI .. FormX (no VII, IX) ; default vowels a u for I
     mkV : V -> (particle : Str) -> V -- V with a non-inflecting particle/phrasal verb
     } ;
 
--- The verb in the imperfect tense gives the most information
+  -- regV : Str -> V ;
 
-  regV : Str -> V ;
+  reflV : V -> V ; -- نَفْس in the proper case and with possessive suffix, e.g.  نَفْسَكِ
 
---Verb Form I : fa`ala, fa`ila, fa`ula
+  v1 : Str -> Vowel -> Vowel -> V ; --Verb Form I : fa`ala, fa`ila, fa`ula
 
-  v1 : Str -> Vowel -> Vowel -> V ;
+  v2 : Str -> V ; --Verb Form II : fa``ala
 
---Verb Form II : fa``ala
+  v3 : Str -> V ; --Verb Form III : faa`ala
 
-  v2 : Str -> V ;
+  v4 : Str -> V ; --Verb Form IV : 'af`ala
 
---Verb Form III : faa`ala
+  v5 : Str -> V ; --Verb Form V : tafa``ala
 
-  v3 : Str -> V ;
+  v6 : Str -> V ; --Verb Form VI : tafaa`ala
 
---Verb Form IV : 'af`ala
+  v7 : Str -> V ; --Verb Form VII : infa`ala
 
-  v4 : Str -> V ;
+  v8 : Str -> V ; --Verb Form VIII ifta`ala
 
---Verb Form V : tafa``ala
+  v10 : Str -> V ; -- Verb Form X 'istaf`ala
-
-  v5 : Str -> V ;
-
---Verb Form VI : tafaa`ala
-
-  v6 : Str -> V ;
-
---Verb Form VII : tafaa`ala
-
-  v7 : Str -> V ;
-
---Verb Form VIII infa`ala
-
-  v8 : Str -> V ;
-
--- Verb Form X 'istaf`ala
-
-  v10 : Str -> V ;
 
 --3 Two-place verbs
 
@@ -359,10 +341,11 @@ resource ParadigmsAra = open
    mkN : N -> (attr : Str) -> N                                       -- Compound nouns
      = \n,attr -> n ** {s2 = \\n,s,c => attr} ;
    mkN : N -> N -> N                                                  -- Compound nouns
-     = \n1,n2 -> n1 ** {s2 =
+     = \n1,n2 -> n1 ** {
-      \\n,s,c => n1.s2 ! n ! s ! c
+          s  = \\n,_,c => n1.s ! n ! Const ! c ;
-              ++ n2.s  ! n ! s ! c
+          s2 = \\n,s,c => n1.s2 ! n ! s ! c
-              ++ n2.s2 ! n ! s ! c} ;
+                       ++ n2.s  ! n ! s ! c
+                       ++ n2.s2 ! n ! s ! c} ;
    } ;
 
   dualN : N -> N = \n -> n ** {isDual=True} ;
@@ -497,6 +480,8 @@ resource ParadigmsAra = open
                 _         => v10sound }
       } in lin V (v10fun rbT) ;
 
+  reflV v = lin V (ResAra.reflV v) ;
+
   mkFullN nsc gen spec = lin N
     { s = nsc; --NTable
       s2 = emptyNTable;

src/arabic/ResAra.gf

@@ -972,9 +972,10 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
         lemma + "ِيّ" => fixShd word  (decNisba ! s ! c) ;
         lemma + "ِي"  => fixShd lemma (dec2sg ! s ! c) ;
         _ + ("ا"|"ى") => fixShd word  (dec3sg ! s ! c) ;
+        lemma + ("ء"|"أ"|"ئ"|"ؤ") => word + dec1sgNoDoubleAlif ! s ! c ;
         lemma + "ة"   => case s of {
                             Poss => lemma + "ت" + dec1sg ! s ! c ;
-                            _    => word        + dec1sg ! s ! c
+                            _    => word        + dec1sgNoDoubleAlif ! s ! c
                           } ;
          _             => fixShd word  (dec1sg ! s ! c)
       }) ;
@@ -1026,13 +1027,20 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
           table {
             Bare => [];
             Nom => "ٌ";
-            Acc => "ً";
+            Acc => "اً";
             _Gen => "ٍ"
           };
         _ => caseTbl --think of ?axU, ?axA, (the five nouns)
 
       };
 
+    -- if a word ends in ء or ة, don't add alif for indef acc.  
+    dec1sgNoDoubleAlif : State => Case => Str = \\s,c =>
+      case <s,c> of {
+        <Indef,Acc> => "ً" ;
+        _           => dec1sg ! s ! c
+      };
+
     --indeclinables (mamnuu3 mina S-Sarf)
     indecl :  Case => Str =
       table {
@@ -1287,7 +1295,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
     mkPron : (_,_,_ : Str) -> PerGenNum -> NP = \ana,nI,I,pgn ->
      { s =
         table {
-          Nom => ana;
+          (Nom|Bare) => ana;
           Acc => BIND ++ nI; -- object suffix
           Gen => BIND ++ I;   -- possessive suffix
           Dat => I -- will only be used with preposition لِ, which already has a BIND
@@ -1354,8 +1362,18 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
       a = np.a ** {isPron=False} -- hack, sometimes we *don't* want prodrop
     } ;
 
-    refl : Case => Str = \\c => "نَفْس" + caseTbl ! c ;
+    reflPron : Case -> PerGenNum -> Str = \c,pgn ->
+      let pron : NP = pgn2pron pgn
+       in "نَفْس" + caseTbl ! c ++ pron.s ! Gen ;
 
+    reflV : Verb -> Verb = \v -> v ** {
+      s = \\vf => case vf of {
+        VPerf _ pgn   => v.s ! vf ++ reflPron Acc pgn ;
+        VImpf _ _ pgn => v.s ! vf ++ reflPron Acc pgn ;
+        VImp g n      => v.s ! vf ++ reflPron Acc (Per2 g n) ;
+        VPPart        => v.s ! vf ++ reflPron Acc (Per3 Masc Sg) ----
+        }
+      } ;
 -----------------------------------------------------------------------------
 -- IP, questions
 
@@ -1471,7 +1489,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
               <_, Past, Pos> => kataba ;
               <_, Past, Neg> => "لَمْ" ++ yaktub ;
               <_, Cond, _  > => yaktuba ;
-              <_, Fut,  Pos> => "سَ" ++ yaktubu ;
+              <_, Fut,  Pos> => glue "سَ" yaktubu ;
               <_, Fut,  Neg> => "لَنْ" ++ yaktuba
               };
             pred : ParamX.Tense -> Polarity -> Str =
@@ -1616,7 +1634,7 @@ 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 c wAhida + dec1sg ! s ! c
+              Fem => \\s,c => defArt s c wAhida + dec1sgNoDoubleAlif ! s ! c
               };
             NOrd => table {
               Masc => \\s,c => defArt s c awwal + dec1sg ! s ! c;

src/arabic/VerbAra.gf

@@ -52,7 +52,8 @@ concrete VerbAra of Verb = CatAra ** open Prelude, ResAra, ParamX in {
 
     AdvVP vp adv = insertStr adv.s vp ;
 
---    AdVVP adv vp = insertAdV adv.s vp ;
+    AdVVP adv = insertStr adv.s ;
+    AdVVPSlash adv vps = vps ** insertStr adv.s vps ;
 --
 --    ReflV2 v = insertObj (\\a => v.c2 ++ reflPron ! a) (predV v) ;
 --