Merge pull request #63 from inariksit/arabic

Arabic

src/arabic/LexiconAra.gf

@@ -3,7 +3,6 @@
 concrete LexiconAra of Lexicon = CatAra ** open
   ParadigmsAra,
   ResAra,
-  MorphoAra, --shouldn't open it here, only needed reg &sndf
   Prelude in {
 
 flags
@@ -19,8 +18,8 @@ flags
   ask_V2Q      = dirV2 (regV "يَسءَل") ;
 --  ask_V2Q      = dirV2 (v1 "سءل" a a) ;
   baby_N       = brkN "طفل" "فِعل" "أَفعَال" Masc Hum;
---  bad_A        = sndA "سوء" "فَيِّع" ;
+--  bad_A        = sndA "سوء" "فَيِّع" ;
-  bad_A        = degrA "سَيِّئ" "سَيِّئَة" "سَيِّئِين" ;
+  bad_A        = degrA "سَيِّئ" "سَيِّئَة" "سَيِّئِين" ;
   bank_N       = brkN "بنك" "فَعل" "فُعُول" Masc NoHum ;
   beautiful_A  = sndA "جمل" "فَعِيل" ;
   become_VA    = mkVA (v4 "صبح") ;
@@ -42,11 +41,11 @@ flags
 --  break_V2     = dirV2 (v1 "كسر" a u) ;
   broad_A      = sndA "وسع" "فاعِل" ;
   brother_N2   = mkN2 (brkN "ءخو" "فَع" "فِعلة" Masc Hum) ; --FIXME dual
-  brown_A      = sndA "بني" "فُعِّل";
+  brown_A      = sndA "بني" "فُعِّل";
   butter_N     = sdfN "سبد" "فُعلة" Fem NoHum ;
   buy_V2       = dirV2 (v8 "شري") ;
   camera_N     = sdfN "كمر" "فاعِيلا" Fem NoHum ; -- |Alö taSwIr
-  cap_N        = sdfN "قبع" "فُعَّلة" Fem NoHum ; --qalnUsö
+  cap_N        = sdfN "قبع" "فُعَّلة" Fem NoHum ; --qalnUsö
   car_N        = sdfN "سير" "فَعّالة" Fem NoHum ;
   carpet_N     = sdfN "سجد" "فَعّالة" Fem NoHum ;
   cat_N        = brkN "هرّ" "فِعّ" "فِعَلَة" Fem NoHum ;
@@ -93,13 +92,13 @@ flags
   girl_N       = brkN "بنت" "فِعل" "فَعَال" Fem Hum ;
   glove_N      = sdfN "قفز" "فُعّال" Masc NoHum ;
   gold_N       = sdfN "ذهب" "فَعَل" Masc NoHum ;
-  good_A       = sndA "جود" "فَيِّع" ; -- Hasan, HisAn
+  good_A       = sndA "جود" "فَيِّع" ; -- Hasan, HisAn
   go_V         = regV "يَذهَب" ;
 --  go_V         = v1 "ذهب" a a ;
   green_A      = clrA "خضر" ;
   harbour_N    = brkN "رفء" "مَفعَل" "مَفاعِل" Masc NoHum ; --mInA', marsaY
   hate_V2      = dirV2 (regV "كَرِه") ;
-  hat_N        = sdfN "قبع" "فُعَّلة" Fem NoHum ;
+  hat_N        = sdfN "قبع" "فُعَّلة" Fem NoHum ;
 --  have_V2      = dirV2 (v1 "ملك" a i) ;
   hear_V2      = dirV2 (regV "سَمِع") ;
 --  hear_V2      = dirV2 (v1 "سمع" i a) ;
@@ -130,14 +129,14 @@ flags
   love_N       = brkN "حبّ" "فُعّ" "فُعّ" Masc NoHum ; -- no plur
   love_V2      = dirV2 (v1 "حبّ" a i) ;
   man_N        = brkN "رجل" "فَعُل" "فِعَال" Masc Hum ;
-  married_A2   = mkA2 (sndA "زوج" "مُتَفَعِّل") "مِن" ;
+  married_A2   = mkA2 (sndA "زوج" "مُتَفَعِّل") "مِن" ;
   meat_N       = brkN "لحم" "فَعلة" "فُعُول" Masc NoHum ;
   milk_N       = brkN "حلب" "فَعِيل" "فَعِيل" Masc NoHum ; --no plur
   moon_N       = brkN "قمر" "فَعَل" "أَفعَال" Masc NoHum ;
-  mother_N2    = mkN2 (sdfN "ءم" "فُعَّ" Fem Hum) ;
+  mother_N2    = mkN2 (sdfN "ءم" "فُعَّ" Fem Hum) ;
   mountain_N   = brkN "جبل" "فَعَل" "فِعَال" Masc NoHum ;
   music_N      = mkN (reg "مُوسِيقَى" "مُوسِيقَى") Fem NoHum ; --no plur
-  narrow_A     = sndA "ضيق" "فَعِّل" ;
+  narrow_A     = sndA "ضيق" "فَعِّل" ;
   new_A        = mkA "جدد" "فَعِيل" "فُعُل" ;
   newspaper_N  = brkN "صحف" "فَعِيلة" "فُعُل" Fem NoHum ;
   oil_N        = brkN "زيت" "فَعل" "فُعُول" Masc NoHum ;
@@ -155,7 +154,7 @@ flags
   play_V2      = dirV2 (regV "لَعِب") ;
 --  play_V2      = dirV2 (v1 "لعب" i a) ;
   policeman_N  = sdmN "شرط" "فِعلِي" Masc Hum ;
-  priest_N     = brkN "قسّ" "فِعِّيل" "أَفِعّة" Masc Hum ;
+  priest_N     = brkN "قسّ" "فِعِّيل" "أَفِعّة" Masc Hum ;
   probable_AS  = mkAS (sndA "مكن" "مُفعِل") ;
   queen_N      = sdfN "ملك" "فَعِلة" Fem Hum ;
   radio_N      = mkN (sndf "راديُو") Masc NoHum ;
@@ -169,7 +168,7 @@ flags
   river_N      = brkN "نهر" "فَعل" "أَفعَال" Masc NoHum ;
   rock_N       = brkN "صخر" "فَعلة" "فُعُول" Fem NoHum ;
   roof_N       = brkN "سطح" "فَعل" "أَفعُل" Masc NoHum ;
-  rubber_N     = brkN "مطط" "فَعَّال" "فَعَّال" Masc NoHum ; -- no hum
+  rubber_N     = brkN "مطط" "فَعَّال" "فَعَّال" Masc NoHum ; -- no hum
   run_V        = regV "يَركُض" ;
 --  run_V        = v1 "ركض" a u ;
   say_VS       = mkVS (v1 "قول" a u) ; --check
@@ -187,11 +186,11 @@ flags
   shoe_N       = brkN "حذو" "فِعَاء" "أَفعِية" Masc NoHum ;
   shop_N       = brkN "تجر" "مَفعَل" "مَفاعِل" Masc NoHum ;
   short_A      = sndA "قصر" "فَعِيل" ;
-  silver_N     = brkN "فضض" "فِعَّة" "فِعَل" Fem NoHum ;
+  silver_N     = brkN "فضض" "فِعَّة" "فِعَل" Fem NoHum ;
   sister_N     = brkN "ءخو" "فُعت" "فَعَوَات" Fem Hum ; --FIXME
   sleep_V      = v1 "نوم" i a ; --check
   small_A      = sndA "صغر" "فَعِيل" ;
-  snake_N      = sdfN "حيّ" "فَعَّة" Fem NoHum ;
+  snake_N      = sdfN "حيّ" "فَعَّة" Fem NoHum ;
   sock_N       = brkN "جرب" "فَوعَل" "فَواعِل" Masc NoHum ;
   speak_V2     = dirV2 (v5 "كلم") ;
   star_N       = brkN "نجم" "فَعل" "فُعُول" Masc NoHum ; --najmö
@@ -205,7 +204,7 @@ flags
   switch8on_V2 = dirV2 (v4 "شعل") ;
   table_N      = sdfN "طول" "فاعِلة" Fem NoHum ;
   talk_V3      = mkV3 (v5 "حدث") "لِ" "عَن" ;
-  teacher_N    = sdmN "علم" "مُفَعِّل" Masc Hum ; --mucal~imö
+  teacher_N    = sdmN "علم" "مُفَعِّل" Masc Hum ; --mucal~imö
   teach_V2     = dirV2 (v2 "علم") ;
   television_N = mkN (sndf "تِلِفِزيُون") Masc NoHum ;
   thick_A      = sndA "سمك" "فَعِيل" ;
@@ -261,7 +260,7 @@ flags
   heavy_A      = sndA "ثقل" "فَعِيل" ;
   near_A       = sndA "قرب" "فَعِيل" ;
   rotten_A     = sndA "فسد" "فاعِل" ;
-  round_A      = sndA "دور" "مُفَعَّل" ;
+  round_A      = sndA "دور" "مُفَعَّل" ;
   sharp_A      = sndA "حدّ" "فاعّ" ;
   smooth_A     = sndA "نعم" "فاعِل" ;
   straight_A   = sndA "قوم" "مُستَفِيع" ;
@@ -277,7 +276,7 @@ flags
   bone_N       = brkN "عظم" "فَعلة" "فِعَال"     Fem  NoHum;
   breast_N     = brkN "صدر" "فَعل" "فُعُول"      Masc NoHum;
   cloud_N      = brkN "غيم" "فَعلة" "فُعُول"     Fem  NoHum;
-  day_N        = brkN "يوم" "فَعل" "أَفَّاع"     Masc NoHum;
+  day_N        = brkN "يوم" "فَعل" "أَفَّاع"     Masc NoHum;
   dust_N       = brkN "غبر" "فُعَال" "أَفعِلة"   Masc NoHum;
   ear_N        = brkN "ءذن" "فُعل" "أَفعَال"     Fem NoHum;
   earth_N      = brkN "ترب" "فُعلة" "فُعَل"     Fem NoHum;

src/arabic/OrthoAra.gf

@@ -12,7 +12,8 @@ flags coding=utf8 ;
   -- Shadda: https://www.unicode.org/L2/L2017/17253-arabic-ordering.pdf
     fixShd : Str -> Str -> Str = \word,suffix ->
       case <word,suffix> of {
-        <x + "ّ", v@#vow + y> => x + v + "ّ" + y ;
+       -- <x + "ّ", v@#vow + y> => x + v + "ّ" + y ;
+        <x + v@#vow, "ّ" + y> => x + "ّ" + v + y ;
         _                     => word + suffix
       } ;
 

src/arabic/ParadigmsAra.gf

@@ -39,23 +39,31 @@ resource ParadigmsAra = open
 
   Preposition : Type ;
 
+  Gender : Type ;
+  masc : Gender ;
+  fem : Gender ;
+
+  Species : Type ;
+  hum : Species ;
+  nohum : Species ;
+
+  Vowel : Type ;
+  va : Vowel ;
+  vi : Vowel ;
+  vu : Vowel ;
+
 --2 Nouns
 
 -- Overloaded operator for main cases
 
- mkN = overload {
+
-   mkN : (sg : Str) -> N                                     -- non-human regular nouns
+ mkN : overload {
-     = smartN ;
+   mkN : (sg : Str) -> N ; -- non-human regular nouns
-   mkN : Species -> N -> N
+   mkN : Species -> N -> N ;
-     = \p,n -> n ** {h = p} ;
+   mkN : (sg,pl : Str) -> Gender -> Species -> N ;
-   mkN : (sg,pl : Str) -> Gender -> Species -> N
+   mkN : NTable -> Gender -> Species -> N ;  -- loan words, irregular
-     = \sg,pl -> mkFullN (reg sg pl) ;
+   mkN : (root,sgPatt,brokenPlPatt : Str) -> Gender -> Species -> N ; -- broken plural
-   mkN : NTable -> Gender -> Species -> N                             -- loan words, irregular
+   mkN : N -> (attr : Str) -> N ; -- Compound nouns
-     = 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 ;
    } ;
@@ -156,13 +164,11 @@ resource ParadigmsAra = open
 
 -- Overloaded operations
 
-  mkV = overload {
+  mkV : overload {
-    mkV : (imperfect : Str) -> V
+    mkV : (imperfect : Str) -> V ;
-      = regV ;
+    mkV : (root : Str) -> (perf,impf : Vowel) -> V ; -- verb form I ; vowel = a|i|u
-    mkV : (root : Str) -> (perf,impf : Vowel) -> V  -- verb form I ; vowel = a|i|u
+    mkV : (root : Str) -> VerbForm -> V ;  -- FormI .. FormX (no VII, IX) ; default vowels a u for I
-      = v1 ;
+    mkV : V -> (particle : Str) -> V -- V with a non-inflecting particle/phrasal verb
-    mkV : (root : Str) -> VerbForm -> V             -- FormI .. FormX (no VII, IX) ; default vowels a u for I
-      = formV ;
     } ;
 
 -- The verb in the imperfect tense gives the most information
@@ -261,6 +267,48 @@ resource ParadigmsAra = open
 -- The definitions should not bother the user of the API. So they are
 -- hidden from the document.
 
+
+  Preposition = Str ;
+
+  Gender = ResAra.Gender ;
+  masc = ResAra.Masc ;
+  fem = ResAra.Fem ;
+
+  Species = ResAra.Species ;
+  hum = ResAra.Hum ;
+  nohum = ResAra.NoHum ;
+
+  Vowel = ResAra.Vowel ;
+  va = ResAra.a ;
+  vu = ResAra.u ;
+  vi = ResAra.i ;
+
+ mkN = overload {
+   mkN : (sg : Str) -> N                                     -- non-human regular nouns
+     = smartN ;
+   mkN : Species -> N -> N
+     = \p,n -> n ** {h = p} ;
+   mkN : (sg,pl : Str) -> Gender -> Species -> N
+     = \sg,pl -> mkFullN (reg sg pl) ;
+   mkN : NTable -> Gender -> Species -> N                             -- loan words, irregular
+     = 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)
+   } ;
+
+  mkV = overload {
+    mkV : (imperfect : Str) -> V
+      = regV ;
+    mkV : (root : Str) -> (perf,impf : Vowel) -> V  -- verb form I ; vowel = a|i|u
+      = v1 ;
+    mkV : (root : Str) -> VerbForm -> V             -- FormI .. FormX (no VII, IX) ; default vowels a u for I
+      = formV ;
+    mkV : V -> (particle : Str) -> V = \v,p -> 
+      v ** { s = \\vf => v.s ! vf ++ p } ;
+    } ;
+
   regV : Str -> V = \wo ->
     let rau : Str * Vowel * Vowel =
     case wo of {
@@ -282,7 +330,7 @@ resource ParadigmsAra = open
       lock_V = <>
     } ;
 
-  va : Vowel = ResAra.a ;
+
 
   v1' : Str ->  Vowel -> Vowel -> Verb =
     \rootStr,vPerf,vImpf ->
@@ -378,8 +426,6 @@ resource ParadigmsAra = open
                 _         => v10sound }
       } in lin V (v10fun rbT) ;
 
-  Preposition = Str ;
-
   mkFullN nsc gen spec =
     { s = nsc; --NTable
       g = gen;

src/arabic/PatternsAra.gf

@@ -48,7 +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 =  "اُتُّ" ; m1 = "ِ"  } ; ---- IL assimilated VIII
+    euttucil = euftucil ** { h =  "اُتُّ" ; m1 = "ِ"  } ; ---- IL assimilated VIII
     afcul    = { h = "َ" ; m1 = "ْ"   ; m2 = "ُ"   ; t = ""   } ;
     faccalo  = { h = ""  ; m1 = "َ"   ; m2 = "َّ"  ; t = "ْ"  } ;
     facal    = { h = ""  ; m1 = "َ"   ; m2 = "َ"   ; t = ""   } ;
@@ -78,8 +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
+    ttacal   = ftacal ** { m1 = ""  ; h = "تَّ" } ; ---- IL assimilated VIII
-    ttacil   = ftacil ** { 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

@@ -904,8 +904,7 @@ patHollowImp : (_,_ :Str) -> Gender => Number => Str =\xaf,xAf ->
       case st of {
         Def =>
           case stem of {
-            s@#sun + v@#vow + x => al + s + v + "ّ" + x ; -- vowel before shadda
+            s@#sun + x          => fixShd (al + s) ("ّ" + x) ;
-            s@#sun + x          => al + s + "ّ" + x;
             x                   => al + x } ;
         _   => stem
       };

src/arabic/StructuralAra.gf

@@ -99,12 +99,12 @@ concrete StructuralAra of Structural = CatAra **
   which_IQuant = {
     s = \\s,c => case <c,s> of {
              <Bare,_>    => "أيّ" ;
-             <Nom,Indef> => "أيٌّ" ;
+             <Nom,Indef> => "أيٌّ" ;
-             <Nom,_>     => "أيُّ" ;
+             <Nom,_>     => "أيُّ" ;
              <Acc,Indef> => "أيّاً" ;
-             <Acc,_>     => "أيَّ" ;
+             <Acc,_>     => "أيَّ" ;
-             <Gen,Indef> => "أيٍّ" ;
+             <Gen,Indef> => "أيٍّ" ;
-             <Gen,_>     => "أيِّ"
+             <Gen,_>     => "أيِّ"
              }
     } ;
   whoSg_IP = mkIP "مَنْ" Sg ;