Merge pull request #143 from inariksit/arabic

Arabic

src/arabic/ParadigmsAra.gf

@@ -201,36 +201,39 @@ resource ParadigmsAra = open
 
   mkV : overload {
     mkV : (imperfect : Str) -> V ; -- The verb in Per3 Sg Masc imperfect tense gives the most information
-    mkV : (root : Str) -> (perf,impf : Vowel) -> (masdar : Str) -> V ; -- verb form I ; vowel = a|i|u ; unpredictable masdar given as an argument.
+    mkV : (root : Str) -> (perf,impf : Vowel) -> (masdar : Str) -> V ; -- Verb form I. Vowel is one of {va,vi,vu}. Unpredictable masdar given as an argument.
-    mkV : (root : Str) -> (perf,impf : Vowel) -> V ; -- verb form I ; vowel = a|i|u ; dummy masdar inserted. This function is here only to keep compatibility for the old API; for new grammars, use the constructor with masdar as an argument.
+    mkV : (root : Str) -> (perf,impf : Vowel) -> V ; -- Like above, but dummy masdar inserted. This function is here only to keep compatibility for the old API; for new grammars, use the constructor with masdar as an argument.
-    mkV : (root : Str) -> VerbForm -> V ;  -- FormI .. FormX (no IX) ; default vowels a u for I, and dummy masdar. Forms II-X have predictable masdar.
+    mkV : (root,masdar : Str) -> VerbForm -> V ;  -- FormI…FormXI (no IX). XI is quadriliteral. For FormI, default vowels are va and vu, and dummy masdar is inserted. Forms II-XI have predictable masdar, so this constructor works fine.
+    mkV : (root : Str) -> VerbForm -> V ;  -- Like above, but dummy masdar inserted for FormI verbs. No difference for FormII-FormXI, because they have predictable masdar.
     mkV : V -> (particle : Str) -> V -- V with a non-inflecting particle/phrasal verb
     } ;
 
   reflV : V -> V ; -- نَفْس in the proper case and with possessive suffix, e.g.  نَفْسَكِ
 
   v1 = overload {
-    v1 : Str -> (perf,impf : Vowel) -> (masdar : Str) -> V -- Verb Form I : fa`ala, fa`ila, fa`ula. Verbal noun (masdar) given as the third argument.
+    v1 : (root : Str) -> (perf,impf : Vowel) -> (masdar : Str) -> V -- Verb Form I: fa`ala, fa`ila, fa`ula. Verbal noun (masdar) given as the third argument.
      = v1masdar ;
-    v1 : Str -> (perf,impf : Vowel) -> V -- To keep compatibility for the old API; dummy masdar inserted.
+    v1 : (root : Str) -> (perf,impf : Vowel) -> V -- To keep compatibility for the old API; dummy masdar inserted.
      = v1dummymasdar ;
   } ;
 
-  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
+  v10 : Str -> V ; -- Verb Form X: 'istaf`ala
+
+  v11 : Str -> V ; -- Verb Form XI (quadriliteral): fa`laba
 
 --3 Two-place verbs
 
@@ -474,6 +477,7 @@ resource ParadigmsAra = open
     \rootStr ->
     let root : Root3 = mkRoot3 rootStr ;
         verb : Verb  = case rootStr of {
+          #weak + ? + ?      => v4assimilated root ;
           ? + #hamza + #weak => v4doubleweak root ;
           ? + #weak + ?      => v4hollow root ;
           _          + #weak => v4defective root  ;
@@ -538,6 +542,17 @@ resource ParadigmsAra = open
           _               => v10sound fcl }
       } in lin V verb ;
 
+  v11 =
+    \rootStr ->
+    let fcl : Root3 = case rootStr of {
+          f@? + c@? + l@? + b@? => {f=f ; c=c+"ْ"+l ; l=b} ;
+          _ => Predef.error "v11: implement quadriliterals properly"
+        } ;
+        verb : Verb  = case rootStr of {
+          _ => v11sound fcl -- TODO more cases?
+        } ;
+    in lin V verb ;
+
   reflV v = lin V (ResAra.reflV v) ;
 
   mkFullN nsc gen spec = lin N
@@ -823,10 +838,11 @@ formV : (root : Str) -> VerbForm -> V = \s,f -> case f of {
    FormVI   => v6 s ;
    FormVII  => v7 s ;
    FormVIII => v8 s ;
-   FormX   => v10 s
+   FormX    => v10 s ;
+   FormXI   => v11 s
    } ;
 
 param VerbForm =
-  FormI | FormII | FormIII | FormIV | FormV | FormVI | FormVII | FormVIII | FormX ;
+  FormI | FormII | FormIII | FormIV | FormV | FormVI | FormVII | FormVIII | FormX | FormXI ;
 
 } ;

src/arabic/PatternsAra.gf

@@ -48,6 +48,7 @@ flags coding=utf8 ;
     eafcal   = { h = "أَ"; m1 = "ْ"   ; m2 = "َ"   ; t = ""   } ;
     eafcAl   = { h = "أَ"; m1 = "ْ"   ; m2 = "َا"  ; t = ""   } ;
     eifcAl   = { h = "إِ"; m1 = "ْ"   ; m2 = "َا"  ; t = ""   } ;
+    eIfcAl   = eifcAl ** {h = "إِي"} ;
     eafcil   = { h = "أَ"; m1 = "ْ"   ; m2 = "ِ"   ; t = ""   } ;
     eafcul   = { h = "أَ"; m1 = "ْ"   ; m2 = "ُ"   ; t = ""   } ;
     eiftacal = { h = "إِ"; m1 = "ْتَ" ; m2 = "َ"   ; t = ""   } ;

src/arabic/ResAra.gf

@@ -41,7 +41,6 @@ resource ResAra = PatternsAra ** open  Prelude, Predef, OrthoAra, ParamX  in {
     Root2   : Type = Root ** {c : Str} ;
     Root3   : Type = Root2 ** {l : Str} ;
 
--- AR 7/12/2009 changed this to avoid duplication of consonants
     mkRoot3 : Str -> Root3 = \fcl -> case fcl of {
       f@? + c@? + l => {f = f ; c = c ; l = l} ;
       _ => error ("mkRoot3: too short root" ++ fcl)
@@ -237,13 +236,13 @@ oper
         })
     } ;
   verb' : SoundForms -> Verb = \vforms ->
-   let katab = vforms ! 0 ;
+   let katab = vforms ! 0 ; -- VPerf Act
-       kutib = vforms ! 1 ;
+       kutib = vforms ! 1 ; -- VPerf Pas
-       aktub = vforms ! 2 ;
+       aktub = vforms ! 2 ; -- VImpf _ Act
-       uktab = vforms ! 3 ;
+       uktab = vforms ! 3 ; -- VImpf _ Pas
-       euktub = vforms ! 4 ;
+       euktub = vforms ! 4 ; -- VImp
-       maktUb = vforms ! 5 ;
+       maktUb = vforms ! 5 ; -- VPPart
-       katb   = vforms ! 6 ;
+       katb   = vforms ! 6 ; -- Masdar
     in verb katab kutib aktub uktab euktub maktUb katb ;
   --affixes of sound verbs
 
@@ -709,11 +708,11 @@ v2defective : Root3 -> Verb = \gny ->
     ugannu = "ُ" + gannu;
     uganna = "ُ" + ganna;
     mugannaY = "مُ" + ganna + "ى";
-    tagniya = "TODO" ;
+    tagniyat = "تَ" + mkStrong fcil (gny ** {l="ي"}) + "َة" ;
   } in verbDef (toDefForms
           ganna gannay gunni gunnu gunniy -- VPerf
           uganni ugannu uganna            -- VImpf
-          ganni gannu mugannaY tagniya) i ;
+          ganni gannu mugannaY tagniyat) i ;
 
 v3sound : Root3 -> Verb =
   \tbc ->
@@ -727,8 +726,7 @@ v3sound : Root3 -> Verb =
     mutAbacAt = mutAbac + "َاَة"
   } in verb tAbac twbic utAbic utAbac tAbic mutAbac mutAbacAt ;
 
-v4sound : Root3 -> Verb =
+v4soundForms : Root3 -> SoundForms = \qnc ->
-  \qnc ->
   let {
     eaqnac = mkStrong eafcal qnc;
     euqnic = mkStrong eufcil qnc;
@@ -738,7 +736,17 @@ v4sound : Root3 -> Verb =
     muqnac = "م" + uqnac;
     eiqnAc = mkStrong eifcAl qnc
   } in
-  verb eaqnac euqnic uqnic uqnac eaqnic muqnac eiqnAc;
+  toSoundForms eaqnac euqnic uqnic uqnac eaqnic muqnac eiqnAc;
+
+v4sound : Root3 -> Verb = \qnc ->
+  verb' (v4soundForms qnc) ;
+
+-- TODO: other differences
+v4assimilated : Root3 -> Verb = \wqf ->
+  let eIqAf = mkStrong eIfcAl (wqf ** {f=""}) ;
+      vforms_snd = v4soundForms wqf ;
+      vforms_ass = table {6 => eIqAf ; n => vforms_snd ! n}
+  in verb' vforms_ass ;
 
 v4hollow : Root3 -> Verb =
   \rwd ->
@@ -757,7 +765,7 @@ v4hollow : Root3 -> Verb =
     earId = mkHollow eafIc rwd ; -- VImp (Pl Masc / Sg Fem)
 
     ppart = "م" + urAd ;
-    eirAdat = mkWeak eifcAl rwd + "َة" ;
+    eirAdat = mkHollow eifcAl rwd + "َة" ;
 
   } in verbHollow (toDefForms
                       earAd earad eurId eurid -- VPerf
@@ -780,7 +788,7 @@ v4hollow : Root3 -> Verb =
     eacTi = "أَ" + _cTi; -- VImp (Masc Sg / Fem _)
     eacTu = "أَ" + _cTu; -- VImp Masc Pl
     mucTaY = "م" + ucTa +"ى" ;
-    eicTA' = "TODO" ;
+    eicTA' = mkStrong eifcAl (cTy ** {l="ء"}) ;
   } in toDefForms eacTa eacTay eucTi eucTu eucTiy -- VPerf
                   ucTi ucTu ucTa                  -- VImpf
                   eacTi eacTu mucTaY eicTA' ;
@@ -790,7 +798,12 @@ v4defective : Root3 -> Verb = \cTy ->
 
 v4doubleweak : Root3 -> Verb = \r'y ->
   let ry = r'y ** {c = ""} ;
-      vforms : DefForms = \\x => rmSukun (v4DefForms ry ! x) ; -- only remove the first sukun
+      r' = ry  ** {l = "ء"} ;
+      eirA'at = mkStrong eifcAl r' + "َة" ;
+      vforms : DefForms = table {
+        11 => rmSukun eirA'at ;
+        n  => rmSukun (v4DefForms ry ! n) -- only remove the first sukun
+      } ;
    in verbDoubleDef vforms i ; -- sukun in suffixes is removed in verbDoubleDef
 
 v5sound : Root3 -> Verb =
@@ -928,7 +941,7 @@ v8hollow : Root3 -> Verb = -- IL
     -- iHtaj again          -- VImp Sg Masc / Pl Fem
     -- iHtAj again          -- VImp Pl Masc / Sg Fem
     ppart = "مُ" + _HtAj ;  -- PPart
-    iHtiyAj = "TODO" ;
+    iHtiyAj = "اِ" + mkStrong fticAl (Hwj ** {c="ي"}) ;
 
   }  in verbHollow (toDefForms
                      iHtAj iHtaj uHtIj uHtij aHtAj aHtaj
@@ -968,7 +981,7 @@ v10hollow : Root3 -> Verb = -- IL 10h -- to be checked
     ustaxaf = "ُ" + _staxaf ; -- VImpf Pas (Per2/Per3 Fem Pl)
     ustaxAf = "ُ" + _staxAf ; -- VImpf Pas _
     ppart = "مُ" + _staxIf ; -- PPart ("weird anomalies" here too?)
-    istixAfat = "TODO ficAlp" ;
+    istixAfat = "اِسْتِ" + mkHollow fAc xwf + "َة" ;
 
   } in verbHollow (toDefForms
                      istaxAf istaxaf ustuxIf ustuxif astaxIf astaxif
@@ -1028,6 +1041,18 @@ v10geminate : Str -> Verb = \fcl ->
          n@(6|7) => "ُسْتَ" + vforms ! n -- ???
     }) ;
 
+v11sound : Root3 -> Verb = \fclb ->
+  let faclabat = mkStrong facalp fclb ;
+      faclib = mkStrong facil fclb ;
+      faclab = mkStrong facal fclb ;
+      vforms = table {
+        2 => "ُ" + faclib ; -- VImpf Act
+        3 => "ُ" + faclab ; -- VImpf Pas
+        5 => "مُ" + faclab ; -- VPPart
+        n => v1soundForms fclb a i faclabat ! n
+      } ;
+   in verb' vforms ;
+
 patV1Perf : Vowel => Pattern =
   table {
     a => facal ; --katab