Merge pull request #81 from inariksit/arabic

Arabic

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") datPrep ;
+  easy_A2V     = mkA2 (sndA "سهل" "فَعل") liPrep ;
   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") datPrep (mkPrep "عَن") ;
+  talk_V3      = mkV3 (v5 "حدث") liPrep (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/NounAra.gf

@@ -6,7 +6,12 @@ lin
 
   DetCN det cn = let {
     cas : Case -> Case = if_then_else Case det.is1sg Bare ;
-    number = case cn.isDual of {True => Dl ; _ => sizeToNumber det.n} ;
+    number = case cn.isDual of {
+                True => 
+                  case sizeToNumber det.n of {
+                         Sg => Sg ;
+                         _  => Dl } ;
+                False => sizeToNumber det.n } ;
     determiner : Case -> Str = \c ->
       det.s ! cn.h ! (detGender cn.g det.n) ! c ;
     noun : Case -> Str = \c -> 

src/arabic/QuestionAra.gf

@@ -28,7 +28,7 @@ concrete QuestionAra of Question = CatAra ** open ResAra, ParamX, Prelude, VerbA
 ---- IL guessed
    -- : IComp -> NP -> QCl
    QuestIComp ic np =
-     let vp = UseComp (CompNP np) ;
+     let vp = UseComp (CompNP np) ; -- puts NP in nominative
          ip : ResAra.IP = np ** { -- NP's s is already present in VP, we only want its agr
             s = \\_,_,_,_ => ic.s ! pgn2gn np.a.pgn } ;
       in QuestVP ip vp ;

src/arabic/RelativeAra.gf

@@ -13,7 +13,7 @@ concrete RelativeAra of Relative = CatAra **
      s = \\t,p,agr,c => 
        let 
          npS : Case => Str = \\_ => rp.s ! agr2ragr agr c ;
-         np : NP = agrNP agr ** {s = npS} ;
+         np : ResAra.NP = agrNP agr ** {s = npS} ;
          cl = predVP np vp ;
        in
        cl.s ! t ! p ! Nominal

src/arabic/ResAra.gf

@@ -127,7 +127,15 @@ resource ResAra = PatternsAra ** open  Prelude, Predef, OrthoAra, ParamX  in {
     } ;
 
     noPrep : Preposition = mkPreposition [] Nom ;
-    datPrep : Preposition = mkPreposition ("لِ" ++ BIND) Dat ;
+    liPrep : Preposition = mkPreposition (
+      pre { #pronSuffAndOther => "لِ" ;
+            #pronSuff         => "لَ" ;
+            _                 => "لِ" 
+          }  ++ BIND) Dat ;
+    biPrep : Preposition = mkPreposition ("بِ"++BIND) ;
+
+    pronSuff : pattern Str = #("كَ"|"كِ"|"كُمَا"|"كُمْ"|"كُنَّ"|"هُ"|"ها"|"هُمَا"|"هُمْ"|"هُنَّ") ;
+    pronSuffAndOther : pattern Str = #( "كَم" ) ; -- TODO list words that begin like pron.suff. but aren't
 
     Adj  : Type = {s : AForm => Str} ;
     Adj2 : Type = Adj ** {c2 : Preposition} ;
@@ -1341,9 +1349,9 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
      { s =
         table {
           (Nom|Bare) => ana;
-          Acc => BIND ++ nI; -- object suffix
+          Acc => nI ; -- object suffix
-          Gen => BIND ++ I;   -- possessive suffix
+          Gen => I ;  -- possessive suffix
-          Dat => I -- will only be used with preposition لِ, which already has a BIND
+          Dat => I -- will only be used with preposition لِ
         };
       a = {pgn = pgn; isPron = True };
       empty = []
@@ -1545,24 +1553,40 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
               <True, Pres, Pos> => vp.pred.s ! gn ! Nom; --xabar marfooc
               _ => vp.pred.s ! gn ! Acc --xabar kaana wa laysa manSoob
               };
-            sc : Preposition = case o of {Subord => {s=[]; c=Acc} ; Dir => vp.sc} ;
+            sc : Preposition = case o of {Subord => {s=[]; c=Acc} ; _ => vp.sc} ;
             subj = np.empty ++ sc.s
                 ++ case vp.isPred of {
                       False => (proDrop np).s ! sc.c ; -- prodrop if it's not predicative
                       True  =>           np.s ! sc.c
                    } ;
+          } in wordOrder o
+                  vp.obj.a.isPron np.a.isPron
+                  (vStr t p) 
+                  subj 
+
+                  (case <vp.isPred,vp.obj.a.isPron> of {
+                     <False,True> => BIND ++ vp.obj.s ;
+                     _            =>         vp.obj.s })
+
+                  (pred t p) 
+                  vp.s2
 
-          } in
-          case o of {
-            Verbal => vStr t p ++ case vp.obj.a.isPron of {
-                            True  => vp.obj.s ++ subj ; -- obj. clitic attaches directly to the verb
-                            False => subj ++ vp.obj.s }
-                   ++ vp.s2 ++ pred t p ;
-            VOS => vStr t p ++ vp.obj.s ++ vp.s2 ++ pred t p ++ subj ;
-            Nominal|Subord => subj ++ vStr t p ++ vp.obj.s ++ vp.s2 ++ pred t p
-          }
       } ;
 
+    -- seems complicated, but this is to share code between PredVP and ClSlash-using funs
+    wordOrder : Order -> (objIsPron,subjIsPron : Bool) -> (verb,subj,obj,pred,adv : Str) -> Str =
+      \o,objIsPron,subjIsPron,verb,subj,obj,pred,adv ->
+          case o of {
+            VOS    => verb ++ obj ++ pred ++ adv ++ subj ;
+            Verbal => verb ++ 
+              case objIsPron of {
+                    True  => obj ++ subj ; -- obj. clitic attaches directly to the verb
+                    False => subj ++ obj } ++ adv ++ pred ;
+            Nominal => subj ++ verb ++ obj ++ adv ++ pred ;
+            Subord  => if_then_Str subjIsPron BIND [] -- in subord. clause, subj. pronoun binds to the main verb
+                    ++ subj ++ verb ++ obj ++ adv ++ pred 
+          } ;
+
     -- in verbal sentences, the verb agrees with the subject
     -- in Gender but not in number
     verbalAgr : PerGenNum -> PerGenNum = \pgn ->
@@ -1580,9 +1604,13 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
 
     Obj : Type = {
       s : Str ;
-      a : Agr -- default Agr in a VP without real Obj is Per3 Masc Sg
+      a : Agr -- default Agr in a VP without real Obj is Per3 Masc Sg.
-      };
+      };      -- need isPron for word order in predVP, and pgn for ImpersCl
 
+    Subj : Type = {s : Case => Str ; isPron : Bool} ;
+
+    np2subj : NP -> Subj = \np -> np ** {isPron = np.a.isPron} ;
+    subj2np : Subj -> NP = \su -> su ** {a = {pgn = emptyNP.a.pgn ; isPron = su.isPron} ; empty=[]} ;
     emptyObj : Obj = emptyNP ** {s=[]} ;
 
     insertObj : NP -> VPSlash -> VP = \np,vp -> vp **
@@ -1608,7 +1636,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
 -- Slash categories
 
     VPSlash : Type = VP ** {c2 : Preposition ; agrObj : PerGenNum => Str} ;
-    ClSlash : Type = VPSlash ** {subj : NP} ;
+    ClSlash : Type = VPSlash ** {subj : Subj} ;
 
     slashV2 : Verb2 -> VPSlash = \v ->
       predV v ** {c2 = v.c2 ; agrObj = \\_ => []} ;
@@ -1617,14 +1645,15 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
     -- but keep the structure as VP, because later on
     -- we might need different word orders for the ClSlash.
     predVPSlash : NP -> VPSlash -> ClSlash = \np,v -> v ** {
-      subj = np
+      subj = np2subj np ;
+      s = \\_pgn,vf => v.s ! np.a.pgn ! vf -- so we can throw away subject's pgn
       } ;
 
     complClSlash = overload {
       complClSlash : NP -> ClSlash -> Cl = \obj,cls ->
-        predVP cls.subj (insertObj obj cls) ;
+        predVP (subj2np cls.subj) (insertObj obj cls) ;
       complClSlash :       ClSlash -> Cl = \cls ->
-        predVP cls.subj (insertObj emptyNP cls) -- Empty subject and object
+        predVP (subj2np cls.subj) (insertObj emptyNP cls) -- Empty subject and object
       } ;
 
     Cl  : Type = {s : Tense => Polarity => Order => Str} ;

src/arabic/SentenceAra.gf

@@ -33,9 +33,9 @@ concrete SentenceAra of Sentence = CatAra ** open
     ImpVP vp = {
       s = \\p,g,n =>
         case p of {
-          Pos => vp.s ! (Per2 g n) ! VPImp  ++ vp.obj.s  ++ vp.s2 ;
+          Pos =>          vp.s ! Per2 g n ! VPImp ;
-          Neg => "لا" ++ vp.s ! (Per2 g n) ! (VPImpf Jus) ++ vp.obj.s ++ vp.s2
+          Neg => "لَا" ++ vp.s ! Per2 g n ! VPImpf Jus
-        }
+        } ++ vp.obj.s  ++ vp.pred.s ! {g=g;n=n} ! Acc ++ vp.s2 
       };
 
 --

src/arabic/StructuralAra.gf

@@ -18,8 +18,8 @@ concrete StructuralAra of Structural = CatAra **
   between_Prep = mkPrep "بَيْنَ" ;
 --  both7and_DConj = sd2 "بْته" "َند" ** {n = Pl} ;
 --  but_PConj = ss "بُت" ;
-  by8agent_Prep = mkPrep "بِ" ;
+  by8agent_Prep,
-  by8means_Prep = mkPrep "بِ" ;
+  by8means_Prep = biPrep ;
   can_VV = mkVV (mkV "طوع" FormX) ;
 --  can8know_VV = {
 --    s = table VVForm [["بي َبلي تْ"] ; "عَن" ; "عُْلد" ;
@@ -34,6 +34,7 @@ concrete StructuralAra of Structural = CatAra **
 --  everywhere_Adv = ss "ثريوهري" ;
   few_Det = mkDet "بَعض" Pl Const ;
 --  first_Ord = ss "فِرست" ;
+  for_Prep = liPrep ;
   from_Prep = mkPrep "مِنَ" ;
   he_Pron = ResAra.he_Pron ;
   here_Adv = ss "هُنا" ;
@@ -68,7 +69,7 @@ concrete StructuralAra of Structural = CatAra **
 --  otherwise_PConj = ss "ْتهروِسي" ;
   part_Prep = mkPrep "مِنَ" ;
 --  please_Voc = ss "ةلَسي" ;
-  possess_Prep = mkPrep "ل" ;
+  possess_Prep = liPrep ;
 --  quite_Adv = ss "قُِتي" ;
   she_Pron = ResAra.she_Pron ;
 --  so_AdA = ss "سْ" ;

src/arabic/VerbAra.gf

@@ -56,7 +56,7 @@ concrete VerbAra of Verb = CatAra ** open Prelude, ResAra, ParamX in {
       } ;
 
     -- : VA -> AP -> VP ;  -- they become red
-    ComplVA v ap = predV v ** {comp = CompAP ap} ;
+    ComplVA v ap = predV v ** {pred = CompAP ap} ;
 
 --    ComplV2A v np ap =
 --      insertObj (\\_ => v.c2 ++ np.s ! Acc ++ ap.s ! np.a) (predV v) ;