Merge pull request #26 from inariksit/arabic

Arabic: compound nouns, verb paradigms, assimilations, whitespace

src/arabic/CatAra.gf

@@ -85,7 +85,6 @@ concrete CatAra of Cat = CommonX - [Utt]  ** open ResAra, Prelude, ParamX in {
     V2V, V2S, V2Q = ResAra.Verb ** {c2 : Str} ; --- AR
     V3 = ResAra.Verb ** {c2, c3 : Str} ;
     VV = ResAra.Verb ** {c2 : Str ; isAux : Bool} ; --- IL
---    VV = {s : VVForm => Str ; isAux : Bool} ;
 
     A = ResAra.Adj ;
     A2 = ResAra.Adj ** {c2 : Str} ;

src/arabic/ExtendAra.gf

@@ -16,5 +16,4 @@ concrete ExtendAra of Extend =
 
   lin
     GenNP np = { s = \\_,_,_,_ => np.s ! Gen ; d = Const ; isNum, isPron = False } ;
-
 } ;

src/arabic/MorphoAra.gf

@@ -33,7 +33,7 @@ flags optimize = all ;--noexpand;
           case g of {
             Masc => waHid;
             Fem => waHida
-          } in Al ! state + word + dec1sg ! state ! c;
+          } in defArt state waHid + word + dec1sg ! state ! c;
         n = num;
         d = state;
         isPron = False;

src/arabic/ParadigmsAra.gf

@@ -54,6 +54,8 @@ resource ParadigmsAra = open
      = mkFullN ;
    mkN : (root,sgPatt,brokenPlPatt : Str) -> Gender -> Species -> N   -- broken plural
      = brkN ;
+   mkN : N -> (attr : Str) -> N                                       -- Compound nouns
+     = \n,attr -> n ** { s = \\num,s,c => n.s ! num ! s ! c ++ attr } ; --- IL (TODO: all kinds of compounds)
 ---   mkN : (root,sgPatt : Str) -> Gender -> Species -> N                -- sound feminine plural
 ---    = sdfN ;
    } ;
@@ -278,7 +280,7 @@ resource ParadigmsAra = open
           l = dp 2 rootStr } in --last rootStr
     case <l, root.c> of {
       <"ّ",_>     => v1geminate rootStr vPerf vImpf ;
-      <"و"|"ي",_> => v1defective root vImpf ;
+      <"و"|"ي",_> => v1defective root vPerf vImpf ;
       <_,"و"|"ي"> => v1hollow root vImpf ;
       _		    => v1sound root vPerf vImpf
     };
@@ -351,10 +353,10 @@ resource ParadigmsAra = open
     \rootStr ->
     let {
       rbT = mkRoot3 rootStr ;
-    } in {
+      v8fun = case rbT.f of {
-      s = (v8sound rbT).s ;
+	("و"|"ي"|"ّ") => v8assimilated ;
-      lock_V = <>
+	_            => v8sound }
-    };
+    } in lin V (v8fun rbT) ;
 
   Preposition = Str ;
 
@@ -560,6 +562,6 @@ formV : (root : Str) -> VerbForm -> V = \s,f -> case f of {
    } ;
 
 param VerbForm =
-  FormI |  FormIII |  FormIII |  FormIV |  FormV |  FormVI | FormVIII  ;
+  FormI | FormII |  FormIII |  FormIV |  FormV |  FormVI | FormVIII  ;
 
 } ;

src/arabic/PatternsAra.gf

@@ -48,6 +48,7 @@ flags coding=utf8 ;
     eiftacal = { h = "إِ"; m1 = "ْتَ" ; m2 = "َ"   ; t = ""   } ;
     eufcil   = { h = "أُ"; m1 = "ْ"   ; m2 = "ِ"   ; t = ""   } ;
     euftucil = { h = "أُ"; m1 = "ْتُ" ; m2 = "ِ"   ; t = ""   } ;
+    euttucil = euftucil ** { h =  "تُّأُ" ; m1 = "" } ; ---- IL assimilated VIII
     afcul    = { h = "َ" ; m1 = "ْ"   ; m2 = "ُ"   ; t = ""   } ;
     faccalo  = { h = ""  ; m1 = "َ"   ; m2 = "َّ"  ; t = "ْ"  } ;
     facal    = { h = ""  ; m1 = "َ"   ; m2 = "َ"   ; t = ""   } ;
@@ -77,6 +78,8 @@ flags coding=utf8 ;
     ficlp    = { h = ""  ; m1 = "ِ"   ; m2 = "ْ"   ; t = "َة" } ;
     ftacal   = { h = ""  ; m1 = "ْتَ" ; m2 = "َ"   ; t = ""   } ;
     ftacil   = { h = ""  ; m1 = "ْتَ" ; m2 = "ِ"   ; t = ""   } ;
+    ttacal   = ftacal ** { m1 = ""  ; h = "تَّ" } ; ---- IL assimilated VIII
+    ttacil   = ftacil ** { m1 = ""  ; h = "تَّ" } ; ---- IL assimilated VIII
     fuccAl   = { h = ""  ; m1 = "ُ"   ; m2 = "َّا" ; t = ""   } ;
     fuccil   = { h = ""  ; m1 = "ُ"   ; m2 = "ِّ"  ; t = ""   } ;
     fuccilo  = { h = ""  ; m1 = "ُ"   ; m2 = "ِّ"  ; t = "ْ"  } ;

src/arabic/ResAra.gf

@@ -76,12 +76,11 @@ resource ResAra = PatternsAra ** open  Prelude, Predef, ParamX  in {
 
     -- takes a weak pattern and a triliteral root and makes
     -- a word, deducing which root consonant is weak
-    mkWeak : Pattern -> Str -> Str = \pat,root ->
+    mkWeak : Pattern -> Root3 -> Str = \pat,fcl ->
-      let fcl = mkRoot3 root in 
+      case <fcl.f,fcl.c,fcl.l> of {
-      case root of {
+       <_,_,("و"|"ي"|"ّ")> => mkDefective   pat fcl;
-        _ + ("و"|"ي"|"ّ") => mkDefective   pat fcl;
+       <_,("و"|"ي"),_>    => mkHollow      pat fcl;
-        _ + ("و"|"ي") + _ => mkHollow      pat fcl;
+       <("و"|"ي"),_,_>    => mkAssimilated pat fcl
-        ("و"|"ي") + _     => mkAssimilated pat fcl
       };
 
     mkBilit : Pattern -> Root2 -> Str = \p,fcl ->
@@ -95,7 +94,7 @@ resource ResAra = PatternsAra ** open  Prelude, Predef, ParamX  in {
         w@_ + "ف" + x@_ + "ع" + y@_ =>
           let pat = { h = w ; m1 = x; m2 = ""; t = y} in
           case <length rS : Ints 100> of {
-            6 | 5 => mkWeak pat rS ; --3=>
+            6 | 5 => mkWeak pat (mkRoot3 rS) ; --3=>
             4 | 3 => mkBilit pat (mkRoot2 rS) ; --2=>
             _ => rS ---- AR error "expected 3--6"
           }
@@ -247,18 +246,30 @@ oper
     } ;
 
   --macro for defective verbs:
-  verbDef: (_,_,_,_,_,_,_,_,_,_,_ : Str) -> Vowel -> Verb =
+  verbDef : DefForms -> Vowel -> Verb =
-    \rama,ramay,rumi,rumu,rumiy,armi,armu,urma,Irmi,Irmu,ppart,vowel ->
+    \vforms,vImpf ->
     let {
+       rama  = vforms ! 0 ; -- VPerf Act (Per3 Masc Sg)
+       ramay = vforms ! 1 ; -- VPerf Act (Per3 Fem  Pl)
+       rumi  = vforms ! 2 ; -- VPerf Pas (Per3 _    Sg)
+       rumu  = vforms ! 3 ; -- VPerf Pas (Per3 Masc Pl)
+       rumiy = vforms ! 4 ; -- VPerf Pas (Per3 Fem  Pl)
+       armi  = vforms ! 5 ; -- VImpf _ Act (Per1 _ _ | Per2/3 Fem _ | Per2/3 Masc Sg)
+       armu  = vforms ! 6 ; -- VImpf _ Act (Per2/3 Masc Pl)
+       urma  = vforms ! 7 ; -- VImpf _ Pas
+       Irmi  = vforms ! 8 ; -- VImp Masc Sg | VImp Fem _
+       Irmu  = vforms ! 9 ; -- VImp Masc Pl
+       ppart = vforms ! 10 ; -- VPPart
+
        patPerf = patDefPerf rama ramay rumi rumu rumiy ;
        patImpfAct = patDefImpfAct armi armu ;
        patImp = patDefImp Irmi Irmu
     } in
     { s = table {
         VPerf v pgn => patPerf ! v ! pgn + (suffixPerfDef v) ! pgn ;
-	    VImpf m Act pgn => prefixImpf ! pgn + patImpfAct ! pgn + (suffixImpfDef Act vowel) ! m  ! pgn ;
+        VImpf m Act pgn => prefixImpf ! pgn + patImpfAct ! pgn + (suffixImpfDef Act vImpf) ! m ! pgn ;
-	    VImpf m Pas pgn => prefixImpf ! pgn + urma + (suffixImpfDef Pas vowel) ! m  ! pgn ;
+        VImpf m Pas pgn => prefixImpf ! pgn + urma + (suffixImpfDef Pas vImpf) ! m  ! pgn ;
-	    VImp g n =>  patImp ! g ! n + (suffixImpfDef Act vowel) ! Jus ! (Per2 g n) ;
+        VImp g n =>  patImp ! g ! n + (suffixImpfDef Act vImpf) ! Jus ! (Per2 g n) ;
         VPPart => ppart
         }
     } ;
@@ -464,23 +475,59 @@ patGem3 : Vowel => Pattern =
     i => fcil                --no such verb probably exists
   } ;
 
-v1defective : Root3 -> Vowel -> Verb =
+DefForms : Type = Predef.Ints 10 => Str ;
-	\rmy,vowel -> 
+toDefForms : (x1,_,_,_,_,_,_,_,_,_,x11 : Str) -> DefForms =
+  \rama,ramay,rumi,rumu,rumiy,armi,armu,urma,eirmi,eirmu,marmiy ->
+  table {
+      0 => rama ;
+      1 => ramay ;
+      2 => rumi ;
+      3 => rumu ;
+      4 => rumiy ;
+      5 => armi ;
+      6 => armu ;
+      7 => urma ;
+      8 => eirmi ;
+      9 => eirmu ;
+      10 => marmiy
+      } ;
+
+
+def1Forms_vPerfA : Root3 -> Vowel -> DefForms = \rmy,vImpf ->
  let {
    rama = mkDefective faca rmy ;
    ramay = mkStrong facalo rmy ;
    rumi = mkDefective fuci rmy ;
    rumu = mkDefective fucu rmy ;
    rumiy = mkStrong fucilo rmy ;
-	rmi = mkDefective (patDef1 ! vowel)  rmy ;
+   rmi = mkDefective (patDef1 ! vImpf)  rmy ;
    armi = "َ" + rmi ;
-	rmu = mkDefective (patDef2 ! vowel) rmy ;	   
+   rmu = mkDefective (patDef2 ! vImpf) rmy ;
    armu = "َ" + rmu ;
    urma = mkDefective ufca rmy ;
-    eirmi = prefixImp ! vowel + rmi;
+   eirmi = prefixImp ! vImpf + rmi;
-    eirmu = prefixImp ! vowel + rmu;
+   eirmu = prefixImp ! vImpf + rmu;
    marmiy = mkStrong mafcil rmy
-  } in verbDef rama ramay rumi rumu rumiy armi armu urma eirmi eirmu marmiy vowel ;
+ } in toDefForms rama ramay rumi rumu rumiy armi armu urma eirmi eirmu marmiy ;
+
+v1defective : Root3 -> Vowel -> Vowel -> Verb = \rmy,vPerf,vImpf ->
+  case vPerf of {
+    i => v1defective_i rmy vImpf ;
+    _ => v1defective_a rmy vImpf
+  } ;
+
+v1defective_a : Root3 -> Vowel -> Verb = \rmy,vImpf ->
+  let vforms = def1Forms_vPerfA rmy vImpf
+   in verbDef vforms vImpf ;
+
+v1defective_i : Root3 -> Vowel -> Verb = \bqy,vImpf -> -- IL (conjugation 1d4)
+  let vforms_a = def1Forms_vPerfA bqy vImpf ;
+      baqI  = mkDefective facIl bqy ;
+      baqiy = mkDefective facil bqy ;
+      vforms_i = table { 0 => baqI ;
+                         1 => baqiy ;
+                         x => vforms_a ! x } ;
+   in verbDef vforms_i vImpf ;
 
 patDef1 : Vowel => Pattern =
   table {
@@ -519,7 +566,7 @@ v2defective : Root3 -> Verb = \gny ->
     ugannu = "ُ" + gannu;
     uganna = "ُ" + ganna;
     mugannaY = "مُ" + ganna + "ى"
-  } in verbDef ganna gannay gunni gunnu gunniy uganni ugannu uganna ganni gannu mugannaY i;
+  } in verbDef (toDefForms ganna gannay gunni gunnu gunniy uganni ugannu uganna ganni gannu mugannaY) i;
 
 v3sound : Root3 -> Verb =
   \tbc ->
@@ -561,7 +608,7 @@ v4defective : Root3 -> Verb = \cTy ->
     eacTi = "أَ" + cTi;
     eacTu = "أَ" + cTu;
     mucTaY = "م" + ucTa +"ى"
-  } in verbDef eacTa eacTay eucTi eucTu eucTiy ucTi ucTu ucTa eacTi eacTu mucTaY i;
+  } in verbDef (toDefForms eacTa eacTay eucTi eucTu eucTiy ucTi ucTu ucTa eacTi eacTu mucTaY) i;
 
 v5sound : Root3 -> Verb =
   \nfs ->
@@ -596,6 +643,20 @@ v8sound : Root3 -> Verb =
     murtabaT =  "م" + urtabaT
   } in verb eirtabaT eurtubiT artabiT urtabaT eirtabiT murtabaT;
 
+v8assimilated : Root3 -> Verb = --- IL 8a1
+  \wfq ->
+  let {
+    ttafiq = mkWeak ttacil wfq ;
+    ttafaq = mkWeak ttacal wfq ;
+    eittafaq = "إِ" + ttafaq ;
+    euttufiq = mkWeak euttucil wfq ;
+    attafiq = "َ" + ttafiq ;
+    uttafaq  = "ُ" + ttafaq ;
+    eittafiq = "إِ" + ttafiq ;
+    muttafaq =  "م" + uttafaq
+  } in verb eittafaq euttufiq attafiq uttafaq eittafiq muttafaq;
+
+
 patV1Perf : Vowel => Pattern =
   table {
     a => facal ; --katab
@@ -715,7 +776,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
 
   -- indeclinable nominal word (mamnuu3 mina S-Sarf)
   indeclN : Str -> State => Case => Str =
-    \aHmar -> \\s,c => Al!s + aHmar + indecl!c;
+    \aHmar -> \\s,c => defArt s aHmar + indecl!c;
 
     -- takes 2 words, singular and broken plural, and gives the
     -- complete noun inflection table
@@ -752,32 +813,32 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
     sing : Str -> State => Case => Str =
       \word ->
       case word of {
-         lemma + "ِي" => \\s,c => Al ! s + lemma + dec2sg ! s ! c ;
+         lemma + "ِي"   => \\s,c => defArt s lemma + dec2sg ! s ! c ;
-        _ + ("ا"|"ى") => \\s,c => Al ! s + word + dec3sg ! s ! c ;
+          _ + ("ا"|"ى") => \\s,c => defArt s word + dec3sg ! s ! c ;
-        _                 => \\s,c => Al ! s + word + dec1sg ! s ! c 
+          _             => \\s,c => defArt s 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 =
       \caSaA ->
       case caSaA of {
-        lemma + ("ا"|"ى") => \\s,c => Al ! s + lemma + "ي" + dl ! s ! c ;
+        lemma + ("ا"|"ى") => \\s,c => defArt s lemma + "ي" + dl ! s ! c ;
-        _                 => \\s,c => Al ! s + caSaA + dl ! s ! c 
+        _                 => \\s,c => defArt s caSaA + dl ! s ! c
       };
 
     -- takes a singular word and gives the corresponding sound
     --plural feminine table
     plurF : Str -> State => Case => Str =
       \kalima ->
-      \\s,c => Al ! s + (mkAt kalima) + f_pl ! s ! c ;
+      \\s,c => defArt s (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 => Al ! s + mucallim + m_pl ! s ! c ;
+      \\s,c => defArt s mucallim + m_pl ! s ! c ;
-
 
     -- to add the Al prefix for Definite words
     Al : State => Str =
@@ -786,7 +847,28 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
         _   => ""
       };
 
-    --declection 1 (strong ending) of the singular or broken plural words
+    defArt : State -> Str -> Str = \st,stem ->
+      let arra = "الرَّ" ;
+          arri = "الرِّ" ;
+          arr  = "الرِّ" ;
+          atta = "التَ" ;
+          atti = "التِّ";
+          att  = "التّ" ;
+          al = "ال" ;
+      in case st of { -- TODO rest of the assimilations
+           Def => case stem of {
+                     ra@"رَ" + x => arra + x ;
+                     ri@"رِ" + x => arri + x ;
+                     r@"ر"  + x => arr + x ; -- no vowel specified
+                     ta@"تَ" + x => atta + x ;
+                     ti@"تِ" + x => atti + x ;
+                     t@"ت"  + x => att + x ; -- no vowel specified
+                     _          => al + stem
+                   } ;
+           _   => "" + stem
+      };
+
+    --declension 1 (strong ending) of the singular or broken plural words
     dec1sg : State => Case => Str =
       table {
         Indef =>
@@ -1140,16 +1222,16 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
             NCard => table {
               Masc => \\s,c => (sing wAhid) ! s ! c ;
               --all fem are first declension:
-              Fem => \\s,c => Al ! s + wAhida + dec1sg ! s ! c 
+              Fem => \\s,c => defArt s wAhida + dec1sg ! s ! c
               };
             NOrd => table {
-              Masc => \\s,c => Al ! s + awwal + dec1sg ! s ! c;
+              Masc => \\s,c => defArt s awwal + dec1sg ! s ! c;
               Fem => \\s,c => (sing Ula) ! s ! c
               }
             };
           ten => table {
-            NCard => \\_,s,c => Al ! s + wAhid + m_pl ! Indef ! c;
+            NCard => \\_,s,c => defArt s wAhid + m_pl ! Indef ! c;
-            NOrd => \\_,s,c => Al ! s + awwal + m_pl ! Indef ! c
+            NOrd => \\_,s,c => defArt s awwal + m_pl ! Indef ! c
             }
           }
       };

src/arabic/SentenceAra.gf

@@ -4,6 +4,7 @@ concrete SentenceAra of Sentence = CatAra ** open
   ResAra,
   Prelude,
   ResAra,
+  TenseX,
   ParamX,
   CommonX in {
 
@@ -111,31 +112,22 @@ concrete SentenceAra of Sentence = CatAra ** open
 --    EmbedVP vp = {s = infVP False vp (agrP3 Sg)} ; --- agr
 --
 
-    --FIXME, all tenses
+    UseCl t p cl =
-    UseCl  t ap cl = 
+      {s = case <t.t,t.a> of { --- IL guessed tenses
-      {s =  cl.s ! ResAra.Pres ! ap.p ! Verbal 
+             <(Pres|Cond),Simul>  => cl.s ! ResAra.Pres ! p.p ! Verbal ;
-{-         case t of {
+             <Fut        ,_    >  => cl.s ! ResAra.Fut ! p.p ! Verbal ;
-           TPres => cl.s ! ResAra.Pres ! p.p ! Verbal ;
+             <_          ,_    >  => cl.s ! ResAra.Past ! p.p ! Verbal
-           TCond => cl.s ! ResAra.Pres ! p.p ! Verbal ;
-           TPast => cl.s ! ResAra.Past ! p.p ! Verbal ;
-           TFut => cl.s ! ResAra.Fut ! p.p ! Verbal 
            }
--}      };
+      };
 
-    --FIXME, all tenses
+    UseQCl t p qcl =
-    UseQCl t ap qcl = --{s = cl.s ! t ! p ! Verbal } ;
+      {s = \\q =>
-      {s = 
+     case <t.t,t.a> of { --- IL guessed tenses
-         table {
+           <(Pres|Cond),Simul>  => qcl.s ! ResAra.Pres ! p.p ! q ;
-           QDir => qcl.s ! ResAra.Pres ! ap.p ! QDir;
+           <Fut        ,_    >  => qcl.s ! ResAra.Fut ! p.p ! q ;
-           QIndir =>  qcl.s ! ResAra.Pres ! ap.p ! QIndir
+           <_          ,_    >  => qcl.s ! ResAra.Past ! p.p ! q
          }
-{-         case t of {
+      };
-           TPres => "هَل" ++ qcl.s ! ResAra.Pres ! p.p ! q ;
-           TCond => "هَل" ++ qcl.s ! ResAra.Pres ! p.p ! q ;
-           TPast => "هَل" ++ qcl.s ! ResAra.Past ! p.p ! q ;
-           TFut  => "هَل" ++ qcl.s ! ResAra.Fut ! p.p ! q 
-         } 
--}      };
 
 --    UseRCl t a p cl = {s = \\r => t.s ++ a.s ++ p.s ++ cl.s ! t.t ! a.a ! p.p ! r} ;
 

src/arabic/SymbolAra.gf

@@ -46,4 +46,3 @@ oper
    { pgn = Per3 Masc n ; isPron = False } ;
 
 }
-