Merge pull request #80 from inariksit/arabic

Arabic

src/arabic/AdjectiveAra.gf

@@ -44,4 +44,6 @@ concrete AdjectiveAra of Adjective = CatAra ** open ResAra, Prelude in {
     s = \\h,g,n,d,c => a.s ! AComp d c 
     };
 
+  -- : Ord -> AP ;       -- warmest
+  AdjOrd ord = {s = \\h,g,n,s,c => ord.s ! g ! s ! c} ;
 }

src/arabic/CatAra.gf

@@ -79,9 +79,10 @@ concrete CatAra of Cat = CommonX - [Utt]  ** open ResAra, Prelude, ParamX in {
 
     V, VS, VQ, VA = ResAra.Verb ; -- = {s : VForm => Str} ;
     V2, V2A = ResAra.Verb2 ;
-    VV = ResAra.Verb2 ** {sc : Preposition} ;
+    VV = ResAra.Verb2 ** {sc : Preposition} ; -- c2 is for verb
     V2S, V2Q = ResAra.Verb2 ;
-    V2V, V3 = ResAra.Verb3 ;
+    V3 = ResAra.Verb3 ;
+    V2V = ResAra.Verb3 ** {sc : Preposition} ; -- c3 is for verb, c2 is for dir.obj
 
     A = ResAra.Adj ;
     A2 = ResAra.Adj2 ;

src/arabic/MissingAra.gf

@@ -3,11 +3,8 @@ resource MissingAra = open GrammarAra, Prelude in {
 -- temporary definitions to enable the compilation of RGL API
 oper AdAdv : AdA -> Adv -> Adv = notYet "AdAdv" ;
 oper AdVVP : AdV -> VP -> VP = notYet "AdVVP" ;
-oper AdjOrd : Ord -> AP = notYet "AdjOrd" ;
 oper AdnCAdv : CAdv -> AdN = notYet "AdnCAdv" ;
-oper AdvCN : CN -> Adv -> CN = notYet "AdvCN" ;
 oper AdvIAdv : IAdv -> Adv -> IAdv = notYet "AdvIAdv" ;
-oper AdvS : Adv -> S -> S = notYet "AdvS" ;
 oper BaseRS : RS -> RS -> ListRS = notYet "BaseRS" ;
 oper CAdvAP : CAdv -> AP -> NP -> AP = notYet "CAdvAP" ;
 oper CleftNP : NP -> RS -> Cl = notYet "CleftNP" ;
@@ -36,8 +33,6 @@ oper Slash2V3 : V3 -> NP -> VPSlash = notYet "Slash2V3" ;
 oper SlashV2A : V2A -> AP -> VPSlash = notYet "SlashV2A" ;
 oper SlashV2Q : V2Q -> QS -> VPSlash = notYet "SlashV2Q" ;
 oper SlashV2S : V2S -> S -> VPSlash = notYet "SlashV2S" ;
-oper SlashV2V : V2V -> VP -> VPSlash = notYet "SlashV2V" ;
-oper SlashV2VNP : V2V -> NP -> VPSlash -> VPSlash = notYet "SlashV2VNP" ;
 oper SlashVS : NP -> VS -> SSlash -> ClSlash = notYet "SlashVS" ;
 oper SubjS : Subj -> S -> Adv = notYet "SubjS" ;
 oper VocNP : NP -> Voc = notYet "VocNP" ;

src/arabic/ParadigmsAra.gf

@@ -114,7 +114,7 @@ resource ParadigmsAra = open
 
   mkN2 : overload {
     mkN2 : N -> Preposition -> N2 ; -- ready-made preposition
-    mkN2 : (mother : N) -> (of_ : Str) -> N2 ; -- preposition given as a string
+    mkN2 : N -> Str -> N2 ; -- preposition given as a string
     mkN2 : N -> N2 ; -- no preposition
     mkN2 : Str -> N2 ; -- no preposition, predictable inflection
   } ;
@@ -195,21 +195,21 @@ resource ParadigmsAra = open
 
   reflV : V -> V ; -- نَفْس in the proper case and with possessive suffix, e.g.  نَفْسَكِ
 
-  v1 : Str -> Vowel -> Vowel -> V ; --Verb Form I : fa`ala, fa`ila, fa`ula
+  v1 : Str -> Vowel -> Vowel -> V ; -- Verb Form I : fa`ala, fa`ila, fa`ula
 
-  v2 : Str -> V ; --Verb Form II : fa``ala
+  v2 : Str -> V ; -- Verb Form II : fa``ala
 
-  v3 : Str -> V ; --Verb Form III : faa`ala
+  v3 : Str -> V ; -- Verb Form III : faa`ala
 
-  v4 : Str -> V ; --Verb Form IV : 'af`ala
+  v4 : Str -> V ; -- Verb Form IV : 'af`ala
 
-  v5 : Str -> V ; --Verb Form V : tafa``ala
+  v5 : Str -> V ; -- Verb Form V : tafa``ala
 
-  v6 : Str -> V ; --Verb Form VI : tafaa`ala
+  v6 : Str -> V ; -- Verb Form VI : tafaa`ala
 
-  v7 : Str -> V ; --Verb Form VII : infa`ala
+  v7 : Str -> V ; -- Verb Form VII : infa`ala
 
-  v8 : Str -> V ; --Verb Form VIII ifta`ala
+  v8 : Str -> V ; -- Verb Form VIII ifta`ala
 
   v10 : Str -> V ; -- Verb Form X 'istaf`ala
 
@@ -419,6 +419,7 @@ resource ParadigmsAra = open
     let root : Root3 = mkRoot3 rootStr ;
         verb : Verb  = case rootStr of {
           ? + #hamza + #weak => v4doubleweak root ;
+          ? + #weak + ?      => v4hollow root ;
           _          + #weak => v4defective root  ;
           _                  => v4sound root } ;
     in lin V verb ;
@@ -475,9 +476,10 @@ resource ParadigmsAra = open
     \rootStr ->
     let {
       rbT = mkRoot3 rootStr ;
-      v10fun = case rbT.c of {
+      v10fun : Root3 -> Verb = case rootStr of {
-                ("و"|"ي") => v10hollow ;
+                ? + #weak + ? => v10hollow ;
-                _         => v10sound }
+                ? + ? + #weak => v10defective ;
+                _             => v10sound }
       } in lin V (v10fun rbT) ;
 
   reflV v = lin V (ResAra.reflV v) ;
@@ -686,9 +688,9 @@ resource ParadigmsAra = open
   mkV2S v p = lin V2S (prepV2 v (mkPreposition p)) ;
   mkV2V = overload {
     mkV2V : V -> Str -> Str -> V2V = \v,p,q -> 
-      lin V2V (prepV3 v (mkPreposition p) (mkPreposition q)) ;
+      lin V2V (prepV3 v (mkPreposition p) (mkPreposition q) ** {sc = noPrep}) ;
     mkV2V : V -> Preposition -> Preposition -> V2V = \v,p,q ->
-      lin V2V (prepV3 v p q) ;
+      lin V2V (prepV3 v p q ** {sc = noPrep}) ;
     mkV2V : VV -> Preposition -> V2V = \vv,p ->
       lin V2V (vv ** {c2 = p ; c3 = vv.c2}) ;
   } ;

src/arabic/PatternsAra.gf

@@ -35,9 +35,13 @@ flags coding=utf8 ;
     fuci   = { h = ""  ; m1 = "ُ" ; m2 = ""; t = "ِ"  } ;
     fucu   = { h = ""  ; m1 = "ُ" ; m2 = ""; t = "ُ"  } ;
     fUc    = { h = ""  ; m1 = "ُو"; m2 = ""; t = ""   } ;
-    ufAc   = { h = "ُ" ; m1 = "َا"; m2 = ""; t = ""   } ;
     ufca   = { h = "ُ" ; m1 = "ْ" ; m2 = ""; t = "َ"  } ;
 
+    eafAc    = fAc ** { h = "أَ" } ;
+    eafac    = fac ** { h = "أَ" } ;
+    eafIc    = fIc ** { h = "أَ" } ;
+    eafic    = fic ** { h = "أَ" } ;
+
     eafAcil  = { h = "أَ"; m1 = "َا"  ; m2 = "ِ"   ; t = ""   } ;
     eafAcIl  = { h = "أَ"; m1 = "َا"  ; m2 = "ِي"  ; t = ""   } ;
     eafcilp  = { h = "أَ"; m1 = "ْ"   ; m2 = "ِ"   ; t = "َة" } ;
@@ -47,6 +51,12 @@ flags coding=utf8 ;
     eafcul   = { h = "أَ"; m1 = "ْ"   ; m2 = "ُ"   ; t = ""   } ;
     eiftacal = { h = "إِ"; m1 = "ْتَ" ; m2 = "َ"   ; t = ""   } ;
     eufcil   = { h = "أُ"; m1 = "ْ"   ; m2 = "ِ"   ; t = ""   } ;
+    eufic    = fic ** { h = "أُ" } ;
+    eufIc    = fIc ** { h = "أُ" } ;
+    ufic     = fic ** { h = "ُ" } ;
+    ufIc     = fIc ** { h = "ُ" } ;
+    ufac     = fac ** { h = "ُ" } ;
+    ufAc     = fAc ** { h = "ُ" } ;
     euftucil = { h = "أُ"; m1 = "ْتُ" ; m2 = "ِ"   ; t = ""   } ;
     euttucil = euftucil ** { h =  "اُتُّ" ; m1 = "ِ"  } ; ---- IL assimilated VIII
     afcul    = { h = "َ" ; m1 = "ْ"   ; m2 = "ُ"   ; t = ""   } ;

src/arabic/ResAra.gf

@@ -649,6 +649,28 @@ v4sound : Root3 -> Verb =
   } in
   verb eaqnac euqnic uqnic uqnac eaqnic muqnac;
 
+v4hollow : Root3 -> Verb =
+  \rwd ->
+  let {
+    earad = mkHollow eafac rwd ; -- VPerf Act (Per3 Fem Pl) etc.
+    earAd = mkHollow eafAc rwd ; -- VPerf Act
+    eurid = mkHollow eufic rwd ; -- VPerf Pas (Per3 Fem Pl) etc.
+    eurId = mkHollow eufIc rwd ; -- VPerf Pas
+
+    urid = mkHollow ufic rwd ; -- VImpf Act (Per2/Per3 Fem Pl)
+    urId = mkHollow ufIc rwd ; -- VImpf Act
+    urad = mkHollow ufac rwd ; -- VImpf Pas (Per2/Per3 Fem Pl)
+    urAd = mkHollow ufAc rwd ; -- VImpf Pas
+
+    earid = mkHollow eafic rwd ; -- VImp (Sg Masc / Pl Fem)
+    earId = mkHollow eafIc rwd ; -- VImp (Pl Masc / Sg Fem)
+
+    ppart = "م" + urAd ;
+
+  } in verbHollow (toDefForms
+                      earAd earad eurId eurid
+                      urId urid urAd urad
+                      earId earid ppart) ;
 
  v4DefForms : Root3 -> DefForms = \cTy ->
   let {
@@ -771,7 +793,7 @@ v8hollow : Root3 -> Verb = -- IL
   }  in verbHollow (toDefForms
                      iHtAj iHtaj uHtIj uHtij aHtAj aHtaj
                      uHtAj uHtaj iHtAj iHtaj ppart) ;
-v10sound : Root3 -> Verb = ---- IL 10s -- to be checked
+v10sound : Root3 -> Verb = -- IL 10s -- to be checked
   \qtl ->
   let {
     _staqtal = "ستَ" + mkStrong fcal qtl ;
@@ -785,7 +807,7 @@ v10sound : Root3 -> Verb = ---- IL 10s -- to be checked
   } in
   verb istaqtal ustuqtil astaqtil astaqtal istaqtil mustaqtal ;
 
-v10hollow : Root3 -> Verb = ---- IL 10h -- to be checked
+v10hollow : Root3 -> Verb = -- IL 10h -- to be checked
   \xwf ->
   let {
     _staxaf = "سْتَ" + mkHollow fac xwf ;
@@ -804,10 +826,33 @@ v10hollow : Root3 -> Verb = ---- IL 10h -- to be checked
     ustaxAf = "ُ" + _staxAf ; -- VImpf Pas _
     ppart = "مُ" + _staxIf -- PPart ("weird anomalies" here too?)
 
-  }  in verbHollow (toDefForms
+  } in verbHollow (toDefForms
                      istaxAf istaxaf ustuxIf ustuxif astaxIf astaxif
                      ustaxAf ustaxaf istaxif istaxIf ppart) ;
 
+v10defective : Root3 -> Verb = -- IL
+  \lqy ->
+  let {
+    _stalqa = "سْتَ" + mkDefective fca lqy ;
+    _stalqu = "سْتَ" + mkDefective fcu lqy ;
+    _stalqi = "سْتَ" + mkDefective fci lqy ;
+    _stulqi = "سْتُ" + mkDefective fci lqy ;
+
+    istalqa  = "اِ" + _stalqa ;               -- VPerf Act (Per3 Masc Sg)
+    istalqay = "اِسْتَ" + mkStrong fcal lqy ; -- VPerf Act (Per3 Fem Pl)
+    ustulqi = "اُ" + _stulqi; -- VPerf Pas (Per3 _ _)
+
+    astalqu = "َ" + _stalqu ; -- VImpf Act (Per2/3 Masc Pl)
+    astalqi = "َ" + _stalqi ; -- VImpf Act _
+    ustalqa = "ُ" + _stalqa ; -- VImpf Pas _
+    istalqi = "اِ" + _stalqi; -- VImp (Masc Sg / Fem _)
+    istalqu = "اِ" + _stalqu; -- VImp Masc Pl
+    mustalqin = "مُ" + _stalqi + "ت" ;
+
+  } in verbDef (toDefForms
+                  istalqa istalqay ustulqi ustulqi ustulqi
+                  astalqi astalqu ustalqa istalqi istalqu mustalqin) i ;
+
 patV1Perf : Vowel => Pattern =
   table {
     a => facal ; --katab
@@ -1317,20 +1362,23 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
 
     agrNP : Agr -> NP = \agr -> emptyNP ** {a = agr} ;
 
-    i_Pron : NP = mkPron "أَنَا" "نِي" "ي" (Per1 Sing) ;
+    i_Pron  : NP = mkPron "أَنَا" "نِي" "ي" (Per1 Sing) ;
-    youSgMasc_Pron : NP = mkPron "أَنتَ" "كَ" "كَ" (Per2 Masc Sg) ;
-    youSgFem_Pron : NP = mkPron "أَنتِ" "كِ" "كِ" (Per2 Fem Sg) ;
-    youDlMasc_Pron : NP = mkPron "أَنتُمَا" "كُمَا" "كُمَا" (Per2 Masc Dl) ;
-    youDlFem_Pron : NP = mkPron "أَنتُمَا" "كُمَا" "كُمَا" (Per2 Fem Dl) ;
-    he_Pron : NP = mkPron "هُوَ" "هُ" "هُ" (Per3 Masc Sg) ;
-    she_Pron : NP = mkPron "هِيَ" "ها" "ها" (Per3 Fem Sg) ;
     we_Pron : NP = mkPron "نَحنُ" "نا" "نا" (Per1 Plur) ;
-    youPlMasc_Pron : NP = mkPron "أَنتُمْ" "كُمْ" "كُمْ" (Per2 Masc Sg) ;
+
-    youPlFem_Pron : NP = mkPron "أَنتُنَّ" "كُنَّ" "كُنَّ" (Per2 Fem Sg) ;
+    youSgMasc_Pron : NP = mkPron "أَنتَ"    "كَ"    "كَ"    (Per2 Masc Sg) ;
-    theyMasc_Pron : NP = mkPron "هُمْ" "هُمْ" "هُمْ" (Per3 Masc Pl) ;
+    youSgFem_Pron  : NP = mkPron "أَنتِ"    "كِ"    "كِ"    (Per2 Fem  Sg) ;
-    theyFem_Pron : NP = mkPron "هُنَّ" "هُنَّ" "هُنَّ" (Per3 Fem Pl) ;
+    youDlMasc_Pron : NP = mkPron "أَنتُمَا" "كُمَا" "كُمَا" (Per2 Masc Dl) ;
+    youDlFem_Pron  : NP = mkPron "أَنتُمَا" "كُمَا" "كُمَا" (Per2 Fem  Dl) ;
+    youPlMasc_Pron : NP = mkPron "أَنتُمْ"  "كُمْ"  "كُمْ"  (Per2 Masc Pl) ;
+    youPlFem_Pron  : NP = mkPron "أَنتُنَّ" "كُنَّ" "كُنَّ" (Per2 Fem  Pl) ;
+
+    he_Pron         : NP = mkPron "هُوَ"  "هُ"    "هُ"    (Per3 Masc Sg) ;
+    she_Pron        : NP = mkPron "هِيَ"  "ها"    "ها"    (Per3 Fem  Sg) ;
     theyDlMasc_Pron : NP = mkPron "هُمَا" "هُمَا" "هُمَا" (Per3 Masc Dl) ;
-    theyDlFem_Pron : NP = mkPron "هُمَا" "هُمَا" "هُمَا" (Per3 Fem Dl) ;
+    theyDlFem_Pron  : NP = mkPron "هُمَا" "هُمَا" "هُمَا" (Per3 Fem  Dl) ;
+    theyMasc_Pron   : NP = mkPron "هُمْ"  "هُمْ"  "هُمْ"  (Per3 Masc Pl) ;
+    theyFem_Pron    : NP = mkPron "هُنَّ" "هُنَّ" "هُنَّ" (Per3 Fem  Pl) ;
+
 
     -- Used e.g. to encode the subject as an object clitic
     -- or to find a possessive suffix corresponding to the NP.
@@ -1538,7 +1586,10 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
     emptyObj : Obj = emptyNP ** {s=[]} ;
 
     insertObj : NP -> VPSlash -> VP = \np,vp -> vp **
-      { obj = {s = vp.obj.s ++ vp.c2.s ++ np.s ! vp.c2.c ; a = np.a} };
+      { obj = {s = vp.obj.s -- old object, if there was one
+                ++ vp.c2.s ++ np.s ! vp.c2.c -- new object
+                ++ vp.agrObj ! np.a.pgn ; -- only used for SlashV2V
+               a = np.a} };
 
     insertPred : Comp -> VP -> VP = \p,vp -> vp **
       { pred = p;
@@ -1556,11 +1607,11 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
 -----------------------------------------------------------------------------
 -- Slash categories
 
-    VPSlash : Type = VP ** {c2 : Preposition} ;
+    VPSlash : Type = VP ** {c2 : Preposition ; agrObj : PerGenNum => Str} ;
     ClSlash : Type = VPSlash ** {subj : NP} ;
 
     slashV2 : Verb2 -> VPSlash = \v ->
-      predV v ** {c2 = v.c2} ;
+      predV v ** {c2 = v.c2 ; agrObj = \\_ => []} ;
 
     -- Add subject string, fix agreement to the subject,
     -- but keep the structure as VP, because later on

src/arabic/SentenceAra.gf

@@ -86,4 +86,6 @@ concrete SentenceAra of Sentence = CatAra ** open
     UseRCl t p cl = {s = \\agr,c => t.s ++ p.s ++ cl.s ! t.t ! p.p ! agr ! c} ;
 
     UseSlash t p cl = UseCl t p (complClSlash cl) ;
+
+    AdvS adv s = s ** {s = \\o => adv.s ++ s.s ! o} ;
 }

src/arabic/StructuralAra.gf

@@ -120,7 +120,7 @@ concrete StructuralAra of Structural = CatAra **
   yes_Utt = {s = \\_ => "نَعَم"} ;
   youSg_Pron = youSgMasc_Pron ;
   youPl_Pron = youPlMasc_Pron ;
-  youPol_Pron = youSgFem_Pron ; -- arbitrary?
+  youPol_Pron = youPlFem_Pron ; -- arbitrary?
 
   have_V2      = dirV2 (regV "يَملِك") ;
 

src/arabic/VerbAra.gf

@@ -7,12 +7,31 @@ concrete VerbAra of Verb = CatAra ** open Prelude, ResAra, ParamX in {
 
     SlashVV vv vps = vps ** predV vv ; ----IL
 
-    -- TODO: --c3 is for verb, c2 is for dir.obj
+    -- : V2V -> VP -> VPSlash ;  -- beg (her) to go
-    --SlashV2V : V2V -> VP -> VPSlash ;  -- beg (her) to go
+    SlashV2V v2v vp = let v2vVP = predV v2v in -- IL
-    --SlashV2VNP : V2V -> NP -> VPSlash -> VPSlash ; -- beg me to buy
+      vp ** {
+        s = v2vVP.s ;
+        agrObj = \\pgn => v2v.c3.s  -- أَنْ
+                       ++ vp.s ! pgn ! VPImpf Cnj ;
+        isPred = False ;
+        c2 = v2v.c2 ; -- for the direct object
+        sc = v2v.sc
+      } ;
+
+    -- : V2V -> NP -> VPSlash -> VPSlash ; -- beg me to buy
+    SlashV2VNP v2v np vps =  let v2vVP = predV v2v in -- IL
+      vps ** {
+        s = \\pgn,vpf => v2vVP.s ! pgn ! vpf
+                      ++ v2v.c2.s ++ np.s ! v2v.c2.c
+                      ++ v2v.c3.s  -- أَنْ
+                      ++ vps.s ! pgn ! VPImpf Cnj ;
+        isPred = False ;
+                    -- preposition for the direct object comes from VP
+        sc = v2v.sc
+      } ;
 
     SlashV2a = slashV2 ;
-    Slash3V3 v np = insertObj np (slashV2 v) ** {c2 = v.c3};
+    Slash3V3 v np = insertObj np (slashV2 v) ** {c2 = v.c3 ; agrObj = \\_ => []};
 
     ComplSlash vp np = insertObj np vp ;