Merge pull request #226 from inariksit/persian

Persian

src/persian/AdjectivePes.gf

@@ -43,7 +43,8 @@ concrete AdjectivePes of Adjective = CatPes ** open ResPes, Prelude in {
   AdjOrd ord = {
     s = \\_ => ord.s ;
     adv = ord.s ;
-    isPre = ord.isPre
+    isPre = ord.isPre ;
+    afterPrefix = False ;
     } ;
 
   AdvAP ap adv = ap ** {

src/persian/CatPes.gf

@@ -12,7 +12,7 @@ concrete CatPes of Cat = CommonX ** open ResPes, Prelude in {
 
 ---- Sentence
 
-    Cl = ResPes.Clause ;
+    Cl = ResPes.Clause ; -- TODO: like romance does "no tiene ningún", i.e.  negative quant/det/NP forces negation in Cl and S
     ClSlash = {
       subj : Str ;
       vp : ResPes.TAnt => Polarity => Order => Str ;
@@ -72,9 +72,10 @@ concrete CatPes of Cat = CommonX ** open ResPes, Prelude in {
 -----b    Conj = {s : Str ; n : Number} ;
 -----b    DConj = {s1,s2 : Str ; n : Number} ;
     Subj = {s : Str ; compl : VVForm} ; -- subjunctive or indicative
+    VS = ResPes.Verb ** {compl : VVForm} ; -- subjunctive or indicative 
     Prep = Compl ;
 ---- Open lexical classes, e.g. Lexicon
-    V, VS, VQ = ResPes.Verb ;
+    V, VQ = ResPes.Verb ;
 
     V2, VA, V2A, V2Q, V2S = ResPes.Verb ** {c2 : Compl} ;
     V3 = ResPes.Verb ** {c2, c3 : Compl} ;

src/persian/ConjunctionPes.gf

@@ -48,7 +48,7 @@ concrete ConjunctionPes of Conjunction =
             isCmpd : CmpdStatus;
             hasAdj : Bool ;
             compl : Number => Str} ;
-    [AP] = {s1,s2 : Mod => Str ; adv : Str ; isPre : Bool} ;
+    [AP] = {s1,s2 : Mod => Str ; adv : Str ; isPre,afterPrefix : Bool} ;
     [RS] = {s1,s2 : Agr => Str ; rp : RelPron => Str} ;
 
 oper

src/persian/LexiconPes.gf

@@ -76,7 +76,7 @@ concrete LexiconPes of Lexicon = CatPes **
   enemy_N = mkN02 "دشمن" animate;
   factory_N = mkN01 "کارخانه" inanimate;
   father_N2 = mkN2 (mkN02 "پدر" animate) [];
-  fear_VS = mkV_1 "ترسیدن";
+  fear_VS = mkVS (mkV_1 "ترسیدن") ;
   find_V2 = mkV2 (compoundV "پیدا" doVerb) "را";
   fish_N = mkN01 "ماهی" animate;
   floor_N = mkN01 "زمین" inanimate; -- Note: floor in persian can have 3 different translations
@@ -97,7 +97,7 @@ concrete LexiconPes of Lexicon = CatPes **
   hat_N = mkN01 "کلاه" inanimate;
   hear_V2 = mkV2 (mkV "شنیدن" "شنو") "را" ;
   hill_N = mkN01 "تپه" inanimate;
-  hope_VS = compoundV "امیدوار" beVerb;
+  hope_VS = mkVS (compoundV "امیدوار" beVerb) ;
   horse_N = mkN01 "اسب" animate;
   hot_A = mkA "داغ" ["داغ داغ"] ;
   house_N = mkN01 "خانه" inanimate;
@@ -106,7 +106,7 @@ concrete LexiconPes of Lexicon = CatPes **
   iron_N = mkN01 "آهن" inanimate;
   king_N = mkN "پادشاه" "پادشاهان" animate;
   know_V2 = mkV2 (mkV "شناختن" "شناس") "را";
-  know_VS = (mkV_1 "دانستن") ; -- danestan -> dan ; needs explicit mkV_1
+  know_VS = mkVS (mkV_1 "دانستن") ; -- danestan -> dan ; needs explicit mkV_1
   know_VQ = (mkV_1 "دانستن") ;
   lake_N = mkN01 "دریاچه" inanimate;
   lamp_N = mkN01 "چراغ" inanimate; -- also "لامپ", but they have different usage
@@ -157,7 +157,7 @@ concrete LexiconPes of Lexicon = CatPes **
   roof_N = mkN01 "بام" inanimate; -- has  variant "سقف"
   rubber_N = mkN01 "پاککن" inanimate; -- also "لاستیک"
   run_V = mkV_1 "دویدن" ;
-  say_VS = mkV "گفتن" "گوی" ;
+  say_VS = mkVS (mkV "گفتن" "گوی") ;
   school_N = mkN "مدرسه" "مدارس" inanimate;
   science_N = mkN "علم" "علوم" inanimate; -- also "دانش"
   sea_N = mkN01 "دریا" inanimate;
@@ -215,7 +215,7 @@ concrete LexiconPes of Lexicon = CatPes **
   woman_N = mkN02 "زن" animate;
   wonder_VQ = compoundV "متعجب" beVerb ;
   wood_N = mkN01 "چوب" inanimate;
-  write_V2 = mkV2 (mkV "نوشتن" "نویس") "را" ;
+  write_V2 = mkV2 (mkV "نوشتن" "نویس") ; -- would need ra with a definite object, but better rule of thumb to go without ra /NM
   yellow_A = mkA "زرد" ;
   young_A = mkA "جوان""جوانانه" ;
   do_V2 = mkV2 (compoundV "انجام" giveVerb) "را";

src/persian/MorphoPes.gf

@@ -150,7 +150,8 @@ oper
   Adjective : Type = {
     s : Mod => Str ;
     adv : Str ;
-    isPre : Bool
+    isPre : Bool ;        -- as attributive
+    afterPrefix : Bool ;  -- as predicative, does it go between the prefix and the light verb
     } ;
 
  mkAdj : Str -> Str -> Adjective = \adj,adv -> {
@@ -159,7 +160,7 @@ oper
                Clitic => mkEnclic adj ;
                Poss => mkPossStem adj
              } ;
-   adv = adv ; isPre = False
+   adv = adv ; isPre = False ; afterPrefix = True ;
    };
 
 ------------------------------------------------------------------

src/persian/ParadigmsPes.gf

@@ -88,6 +88,7 @@ oper
     mkA : Str -> A ; -- Regular adjective, same form for adjective and adverb.
     mkA : (adj,adv : Str) -> A -- Different forms for adjective and adverb.
   } ;
+  prefixA : A -> A ; -- Adjective that comes before the noun
 
   mkA2 : (married,to : Str) -> A2 -- Takes string and complementiser, returns A2.
     = \a,c -> lin A2 (mkAdj a a ** {c2 = c}) ;
@@ -148,9 +149,9 @@ oper
 
   mkVS = overload {
    mkVS : Str -> VS -- predictable verb with sentence complement
-    = \s -> lin VS (regV s) ;
+    = \s -> lin VS (regV s ** {compl=subjunctive}) ;
    mkVS : V -> VS -- VS out of a verb
-    = \v -> lin VS v
+    = \v -> lin VS (v ** {compl=subjunctive})
   } ;
 
   mkVV = overload {
@@ -303,6 +304,8 @@ oper
       = \a,c -> lin A2 (mkAdj a a ** {c2 = c})
    } ;
 
+   prefixA a = a ** {isPre=True};
+
    preA : (adj,adv : Str) -> A = \adj,adv ->
      lin A ((mkAdj adj adv) ** {isPre=True}) ;
 

src/persian/ResPes.gf

@@ -67,16 +67,23 @@ resource ResPes = MorphoPes ** open Prelude,Predef in {
  -----------------------
 param
   VVType = NoVV | FullVV | DefVV ;
-  VVForm = Indic | Subj ;
+  VVForm = Indic | Subj ; ---| SubjPast ; -- TODO extend this to VV, VS and Subj
-  VVTense = VVPres | VVPerf | VVPast ; -- VVPast Anteriority ???
+  VVTense = VVPres | VVPast VVForm ;
   TAnt = TA Tense Anteriority ;
 
 oper
 
-  -- TODO: all forms
+  -- VVPast Subj is another possibility, used in constructions such as
-  ta2vvt : TAnt -> VVTense = \ta -> case ta of {
+  -- قاتل نمی توانسته آنجا بوده باشد، چون او آن زمان در پاریس بوده
-    TA Pres Anter => VVPerf ;
+  -- The form is created in complVV, but not currently used in other functions. /IL
-    TA Past _     => VVPast ;
+  ta2vvt : TAnt -> VVType -> VVTense = \ta,vvtype -> case ta of {
+    TA Pres Anter |
+    TA Past _     => VVPast Indic ;
+    TA Cond Simul => VVPres ;
+    TA Cond Anter =>
+       case vvtype of {
+          DefVV => VVPast Indic ;
+          _     => VVPres } ;
     _ => VVPres } ;
 
   VV : Type = Verb ** {
@@ -95,21 +102,19 @@ oper
       } ;
 
   showVPH = overload {
-    showVPH : VerbForm -> Agr -> VPH -> Str = showVPH' VVPres ;
+    showVPH : VerbForm -> Agr -> VPH -> Str = showVPH' False VVPres ;
-    showVPH : VVTense -> VerbForm -> Agr -> VPH -> Str = showVPH'
+    showVPH : VVTense -> VerbForm -> Agr -> VPH -> Str = showVPH' False
   } ;
 
-  showVPH' : VVTense -> VerbForm -> Agr -> VPH -> Str =
+  showVPH' : Bool -> VVTense -> VerbForm -> Agr -> VPH -> Str =
-    \ant,vf,agr,vp -> vp.ad ++ vp.comp ! agr ++ vp.obj
+    \showImpPref,ant,vf,agr,vp ->
-                   ++ vp.prefix ++ vp.s ! vf
+       let impPref = case showImpPref of {
+         True => vp.s ! ImpPrefix Pos ;
+         False => [] }
+        in vp.ad ++ vp.comp ! agr ++ vp.obj
+        ++ vp.prefix ++ impPref ++ vp.s ! vf
         ++ vp.vComp ! agr ! ant ++ vp.embComp ;
 
-  -- A hack: we reuse the obj field for the VP complement in
-  -- SlashV2V and this is needed to get the right word order for complVV.
-  showVPHvv : VerbForm -> Agr -> VPH -> Str = \vf,agr,vp ->
-      vp.comp ! agr ++ vp.prefix ++ vp.s ! vf  -- vp.ad is missing on purpose! we add it in insertVV.
-   ++ vp.obj ++ vp.vComp ! agr ! VVPres ++ vp.embComp ;
-
   Compl : Type = {s : Str ; ra : Str ; mod : Mod} ;
 
   VPHSlash : Type = VPH ** {
@@ -131,11 +136,13 @@ oper
    predVc : (Verb ** {c2 : Compl}) -> VPHSlash = \verb ->
     predV verb ** vs verb.c2 ;
 
-  passV : Verb -> VPH = \v -> predV v ** {
+  passV : Verb -> VPH = \v -> passVP (predV v) ;
+
+  passVP : VPH -> VPH = \vp -> vp ** {
     s = becomeVerb.s ;
-    prefix = case v.passive of {
+    prefix = case vp.passive of {
-                    Add => v.s ! PerfStem ++ v.prefix ;
+                    Add => vp.s ! PerfStem ++ vp.prefix ;
-                    Replace => v.prefix
+                    Replace => vp.prefix
              } ;
   } ;
 -- ---------------------
@@ -159,7 +166,6 @@ oper
   insertVV : VV -> VPH -> VPH = \vv,vp -> predV vv ** {
     vComp = \\a,t => vp.vComp ! a ! t ++ complVV vv vp ! a ! t ;
     vvtype = case vv.isDef of {True => DefVV ; _ => FullVV} ;
-    ad = vp.ad  -- because complVV doesn't include ad! for word order.
   } ;
 
   embComp : Str -> VPH -> VPH = \str,vp -> vp ** {
@@ -181,15 +187,15 @@ oper
     \\agr,ant => if_then_Str vv.isAux conjThat [] ++
       case <ant,vv.isDef,vv.compl> of {
        -- Auxiliaries with defective inflection: complement inflects in tense
-        <VVPast,True,_> => showVPHvv (VPast Pos agr) agr vp ;
+        <VVPast Indic,True,>  => showVPH' True VVPres (VPast Pos agr) agr vp ;
---        <VVPast Anter> => showVPH PerfStem agr vp ++ pluperfAux Pos agr ; -- TODO do we need this?
+        <VVPast Indic,_,_>    => showVPH (VPast Pos agr) agr vp ;
-        <VVPerf,True,_> => showVPHvv PerfStem agr vp ++ subjAux Pos agr ;
+        <VVPast Subj>         => showVPH PerfStem agr vp ++ subjAux Pos agr ;
 
         -- Auxiliaries that take indicative (full or defective inflection)
-        <VVPres,_,Indic> => showVPHvv (VAor Pos agr) agr vp ;
+        <VVPres,_,Indic> => showVPH (VAor Pos agr) agr vp ;
 
        -- Default: complement in subjunctive
-        _ => showVPHvv (VSubj Pos agr) agr vp ---- TODO more forms ?
+        _ => showVPH (VSubj Pos agr) agr vp
     } ;
 
   insertAdV : Str -> VPH -> VPH = \ad,vp -> vp ** {
@@ -205,16 +211,15 @@ oper
   SlClause : Type = {quest : Order => Str ; subj : Str ; vp : TAnt => Polarity => Order => Str} ;
 ---- AR 18/9/2017 intermediate SClause to preserve SOV in e.g. QuestionPes.QuestSlash
 
- -- TODO: check the VV forms with defective verbs
   clTable : VPH -> (Agr => TAnt => Polarity => Str) = \vp ->
     \\agr,vt,pol => vp.prefix ++ case vt of {
       TA Pres Simul => vp.s ! ImpPrefix pol ++ vp.s ! VAor pol agr ; -- for reg. verbs, VAor pol is invariant and negation comes in ImpPrefix.
       TA Pres Anter => vp.s ! VPerf pol agr ;
-      TA Past Simul => vp.s ! VPast pol agr ;
+      TA Past Simul => vp.s ! VPast pol agr ; -- Past Simul: simple past
-      TA Past Anter =>
+      TA Past Anter | TA Cond _ =>   -- Past Anter & Cond _: continuous past
         case vp.vvtype of {
-           DefVV => vp.s ! ImpPrefix pol ++ vp.s ! VAor pol agr ;
+          DefVV => vp.s ! VPast pol agr ;
-           _ => vp.s ! PerfStem ++ pluperfAux pol agr } ;
+          _ => vp.s ! ImpPrefix pol ++ vp.s ! VPast Pos agr } ;
       TA Fut  Simul =>
          case vp.vvtype of {
            DefVV => vp.s ! ImpPrefix pol ++ vp.s ! VAor pol agr ;
@@ -223,13 +228,8 @@ oper
       TA Fut  Anter =>
          case vp.vvtype of {
            DefVV => vp.s ! VPerf pol agr ;
-           _ => "خواسته" ++ pluperfAux pol agr ++ vp.s ! PastStem
+           _ => futAux pol agr ++ vp.s ! PastStem
-         } ; -- verb form need to be confirmed
+         }
-      TA Cond Simul => vp.s ! VSubj pol agr ;
-      TA Cond Anter =>
-         case vp.vvtype of {
-           DefVV => vp.s ! VSubj pol agr ;
-           _ => vp.s ! PerfStem ++ subjAux pol agr } -- verb form to be confirmed
   } ;
 
   mkClause : NP -> VPH -> Clause = \np,vp ->
@@ -243,7 +243,7 @@ oper
     subj = np2str np ;
     vp = \\ta,p,ord =>
       let vps = clTable vp ! np.a ! ta ! p ;
-          vvt = ta2vvt ta ;
+          vvt = ta2vvt ta vp.vvtype ;
        in case vp.vvtype of {
             DefVV
               => vps ++ vp.ad ++ vp.comp ! np.a ++ vp.obj
@@ -257,9 +257,8 @@ oper
     s = \\ta,p,ord =>
       let vps = clTable vp ! agr ! ta ! p ;
           quest = case ord of { ODir => [] ; OQuest => "آیا" } ;
-          vvt = ta2vvt ta ;
+          vvt = ta2vvt ta vp.vvtype ;
        in quest ++ subj ++ vp.ad ++ vp.comp ! agr ++ vp.obj
---       in quest ++ vp.ad ++ subj ++ vp.comp ! agr ++ vp.obj -- TODO check which word order is better /IL
        ++ vps ++ vp.vComp ! agr ! vvt ++ vp.embComp
   };
 

src/persian/StructuralPes.gf

@@ -26,7 +26,7 @@ concrete StructuralPes of Structural = CatPes **
   either7or_DConj = sd2 "یا" "یا" ** {n = Sg} ;
 --  everybody_NP =  R.indeclNP "هر کwی";
   every_Det = mkDet "هر" Sg ;
---  everything_NP = R.indeclNP ["هر XE"]));
+  everything_NP = DetCN (mkDet "همه" Sg) (UseN (mkN "چیز")) ;
   everywhere_Adv = ss ["هر جا"] ;
   few_Det = mkDet ["تعداد کمی"] Pl True; -- check
   for_Prep = mkPrep "برای" Ezafe ;

src/persian/SymbolPes.gf

@@ -24,8 +24,8 @@ concrete SymbolPes of Symbol = CatPes ** open Prelude, ResPes in {
     } ;
 
   SymbS sy = {s = \\_ => sy.s} ;
-  SymbNum sy = { s = sy.s ; n = Pl } ;
+  SymbNum sy = {s = sy.s ; n = Pl} ;
-  SymbOrd sy = { s = sy.s ++ "wN" ; n = Pl; isNum,isPre=False} ;
+  SymbOrd sy = {s = sy.s ; n = Sg ; isNum,isPre=False} ;
 
 lincat
 

src/persian/VerbPes.gf

@@ -13,9 +13,13 @@ concrete VerbPes of Verb = CatPes ** open ResPes,Prelude in {
     ComplSlash = complSlash ;
 
     ComplVV = insertVV ;
-    ComplVS v s  = embComp (conjThat ++ s.s ! Indic) (predV v) ;
+    ComplVS v s  = embComp (conjThat ++ s.s ! v.compl) (predV v) ;
     ComplVQ v q  = embComp (conjThat ++ q.s)         (predV v) ;
-    ComplVA v ap = insertObj (appComp v.c2 ap.s)     (predV v) ; -- check form of adjective
+    ComplVA v ap = let adjStr = appComp v.c2 ap.s  in
+      case ap.afterPrefix of {
+        True  => predV (v ** {prefix = v.prefix ++ adjStr}) ;
+        False => insertObj adjStr (predV v) -- check form of adjective
+      } ;
 
     SlashVV vv vps = vps ** ComplVV vv vps ;
     SlashV2S v s  = predVc v ** embComp (conjThat ++ s.s ! Indic) (predV v) ;