msyds/gf-rgl
1
0
Fork 0
forked from GitHub/gf-rgl
Code Pull Requests Activity

Merge pull request #157 from inariksit/persian

Browse Source 
Persian
This commit is contained in:
Inari Listenmaa
2019-02-04 16:59:09 +01:00
committed by GitHub
parent 271feb0295 7cb2298b55
commit f2cba3fb4d
8 changed files with 199 additions and 194 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
src/persian/CatPes.gf
View File

@@ -59,9 +59,9 @@ concrete CatPes of Cat = CommonX - [Adv] ** open ResPes, Prelude in {
Pron = {s : Str ; ps : Str ; a : AgrPes}; Pron = {s : Str ; ps : Str ; a : AgrPes};
Det = ResPes.Determiner ; Det = ResPes.Determiner ;
Predet = {s : Str} ; Predet = {s : Str} ;
Num = {s : Str ; n : Number} ; Num = {s : Str ; n : Number ; isNum : Bool} ;
Card = {s : Str; n : Number} ; Card = {s : Str; n : Number} ;
Ord = {s : Str; n : Number} ; Ord = {s : Str; n : Number ; isNum : Bool} ;
Quant = {s: Number => Str ; a:AgrPes ; fromPron : Bool}; Quant = {s: Number => Str ; a:AgrPes ; fromPron : Bool};
Art = {s : Str} ; Art = {s : Str} ;
@@ -82,15 +82,15 @@ concrete CatPes of Cat = CommonX - [Adv] ** open ResPes, Prelude in {
V2, V2A, V2Q, V2S = ResPes.Verb ** {c2 : Compl} ; V2, V2A, V2Q, V2S = ResPes.Verb ** {c2 : Compl} ;
V3 = ResPes.Verb ** {c2, c3 : Str} ; V3 = ResPes.Verb ** {c2, c3 : Str} ;
VV = ResPes.Verb ** { isAux : Bool} ; VV = ResPes.Verb ** {isAux : Bool} ;
V2V = ResPes.Verb ** {c1 : Str ; c2 : Str ; isAux : Bool} ; V2V = ResPes.Verb ** {c1 : Str ; c2 : Str ; isAux : Bool} ;
A = ResPes.Adjective ; --- {s : Gender => Number => Case => Str} ; A = ResPes.Adjective ; --- {s : Gender => Number => Case => Str} ;
A2 = ResPes.Adjective ** { c2 : Str} ; A2 = ResPes.Adjective ** { c2 : Str} ;
N = {s : Ezafa => Number => Str ; animacy : Animacy ; definitness : Bool} ; N = ResPes.Noun ;
N2 = {s : Ezafa => Number => Str ; animacy : Animacy ; definitness : Bool} ** {c : Str}; N2 = ResPes.Noun ** {c : Str};
N3 = {s : Ezafa => Number => Str ; animacy : Animacy ; definitness : Bool} ** {c2 : Str ; c3 : Str } ; N3 = ResPes.Noun ** {c2 : Str ; c3 : Str} ;
PN = {s : Str ; animacy : Animacy} ; PN = {s : Str ; animacy : Animacy} ;
} }

17
src/persian/ExtendPes.gf Normal file
View File

@@ -0,0 +1,17 @@
--# -path=.:../common:../abstract:../prelude
concrete ExtendPes of Extend =
CatPes ** ExtendFunctor - [
GenNP
]
with (Grammar=GrammarPes)
** open Prelude, ResPes in {
lin
-- NP -> Quant ; -- this man's
GenNP np = np ** {
s = \\num => np.s ! NPC bEzafa ; -- the possessed will get aEzafa, possesser is just unmarked; https://sites.la.utexas.edu/persian_online_resources/language-specific-grammar/ezfe/
fromPron = True -- not necessarily from Pron, but it should come after the noun, and if fromPron=True, then DetCN places determiner after cn.
} ;
}

6
src/persian/LangPes.gf
View File

@@ -1,8 +1,8 @@
--# -path=.:../abstract:../common:../hindustani --# -path=.:../abstract:../common
concrete LangPes of Lang = concrete LangPes of Lang =
GrammarPes, GrammarPes
LexiconPes ,LexiconPes
** { ** {
flags startcat = Phr ; unlexer=unwords ; lexer=words ; flags startcat = Phr ; unlexer=unwords ; lexer=words ;

88
src/persian/NounPes.gf
View File

@@ -3,13 +3,13 @@ concrete NounPes of Noun = CatPes ** open ResPes, Prelude in {
flags optimize=all_subs ; flags optimize=all_subs ;
lin lin
DetCN det cn = { DetCN det cn = { s = \\_ =>
s = \\_ => case <det.isNum,det.fromPron> of { case <det.isNum,det.fromPron> of {
<False,True> => cn.s ! aEzafa ! det.n ++ det.s ; -- det.n ; <False,True> => cn.s ! aEzafa ! det.n ++ det.s ;
<False,False> => det.s ++ cn.s ! bEzafa ! det.n ; -- det.n ; <False,False> => det.s ++ cn.s ! bEzafa ! det.n ;
<True,True> => cn.s ! aEzafa ! Sg ++ det.s ; <True,True> => cn.s ! aEzafa ! Sg ++ det.s ; -- noun modified by a number is invariably singular
<True,False> => det.s ++ cn.s ! bEzafa ! Sg <True,False> => det.s ++ cn.s ! bEzafa ! Sg
}; };
a = agrPesP3 det.n ; a = agrPesP3 det.n ;
animacy = cn.animacy animacy = cn.animacy
} ; } ;
@@ -17,42 +17,34 @@ concrete NounPes of Noun = CatPes ** open ResPes, Prelude in {
UsePN pn = {s = \\_ => pn.s ; a = agrPesP3 Sg ; animacy = pn.animacy } ; UsePN pn = {s = \\_ => pn.s ; a = agrPesP3 Sg ; animacy = pn.animacy } ;
UsePron p = {s = \\_ => p.s ; a = p.a ; animacy = Animate} ; UsePron p = {s = \\_ => p.s ; a = p.a ; animacy = Animate} ;
PredetNP pred np = { PredetNP pred np = np ** {
s = \\ez => pred.s ++ np.s ! ez ; s = \\ez => pred.s ++ np.s ! ez
a = np.a;
animacy = np.animacy
} ; } ;
PPartNP np v2 = { PPartNP np v2 = np ** {
s = \\ez => np.s ! ez ++ partNP (v2.s ! Root1) ; s = \\ez => np.s ! ez ++ partNP (v2.s ! Root1)
a = np.a ;
animacy = np.animacy
} ; } ;
RelNP np rs = { RelNP np rs = np ** {
s = \\ez => np.s ! ez ++ rs.s ! np.a ; s = \\ez => np.s ! ez ++ rs.s ! np.a
a = np.a ;
animacy = np.animacy
} ; } ;
AdvNP np adv = { AdvNP np adv = np ** {
s = \\ez => np.s ! NPC aEzafa ++ adv.s ; s = \\ez => np.s ! NPC aEzafa ++ adv.s
a = np.a ;
animacy = np.animacy
} ; } ;
DetQuantOrd quant num ord = { DetQuantOrd quant num ord = {
s = quant.s ! num.n ++ num.s ++ ord.s ; s = quant.s ! num.n ++ num.s ++ ord.s ;
isNum = True; isNum = orB num.isNum ord.isNum ;
fromPron = quant.fromPron ; fromPron = quant.fromPron ;
n = num.n n = num.n
} ; } ;
DetQuant quant num = { DetQuant quant num = {
s = quant.s ! num.n ++ num.s; s = quant.s ! num.n ++ num.s;
isNum = True ; -- this does not work in case of 'these women' but works in case of 'five women' isNum = num.isNum;
fromPron = quant.fromPron ; fromPron = quant.fromPron ;
n = num.n n = num.n
} ; } ;
DetNP det = { DetNP det = {
@@ -63,20 +55,20 @@ concrete NounPes of Noun = CatPes ** open ResPes, Prelude in {
PossPron p = {s = \\_ => p.ps ; a = p.a ; fromPron = True} ; PossPron p = {s = \\_ => p.ps ; a = p.a ; fromPron = True} ;
NumSg = {s = [] ; n = Sg} ; NumSg = {s = [] ; n = Sg ; isNum = False} ;
NumPl = {s = [] ; n = Pl} ; NumPl = {s = [] ; n = Pl ; isNum = False} ;
-- from here -- from here
NumCard n = n ** {hasCard = True} ; NumCard n = n ** {isNum = True} ;
NumDigits n = {s = n.s ! NCard ; n = n.n} ; NumDigits n = n ** {s = n.s ! NCard; isNum = True} ;
OrdDigits n = {s = n.s ! NOrd; n = n.n} ; OrdDigits n = n ** {s = n.s ! NOrd ; isNum = True} ;
NumNumeral numeral = {s = numeral.s ! NCard; n = numeral.n} ; NumNumeral n = n ** {s = n.s ! NCard; isNum = True} ;
OrdNumeral numeral = {s = numeral.s ! NOrd ; n = numeral.n} ; OrdNumeral n = n ** {s = n.s ! NOrd ; isNum = True} ;
-- to here -- to here
AdNum adn num = {s = adn.s ++ num.s ; n = num.n} ; AdNum adn num = num ** {s = adn.s ++ num.s} ;
OrdSuperl a = {s = a.s ! bEzafa ++ taryn; n = Sg} ; -- check the form of adjective OrdSuperl a = {s = a.s ! bEzafa ++ taryn; n = Sg ; isNum=False} ; -- check the form of adjective
DefArt = {s = \\_ => [] ; a = defaultAgrPes ; fromPron = False} ; DefArt = {s = \\_ => [] ; a = defaultAgrPes ; fromPron = False} ;
IndefArt = {s = table { Sg => IndefArticle ; Pl => []} ; a =defaultAgrPes ; fromPron = False} ; IndefArt = {s = table { Sg => IndefArticle ; Pl => []} ; a =defaultAgrPes ; fromPron = False} ;
@@ -111,7 +103,7 @@ concrete NounPes of Noun = CatPes ** open ResPes, Prelude in {
s = \\ez,n => f.s ! ez ! n ++ f.c ++ x.s ! NPC ez ; s = \\ez,n => f.s ! ez ! n ++ f.c ++ x.s ! NPC ez ;
animacy = f.animacy; animacy = f.animacy;
definitness = True definitness = True
}; };
ComplN3 f x = { ComplN3 f x = {
s = \\ez,n => f.s ! ez ! n ++ f.c2 ++ x.s ! NPC ez ; s = \\ez,n => f.s ! ez ! n ++ f.c2 ++ x.s ! NPC ez ;
c = f.c3; c = f.c3;
@@ -119,22 +111,18 @@ concrete NounPes of Noun = CatPes ** open ResPes, Prelude in {
definitness = True; definitness = True;
} ; } ;
AdjCN ap cn = { AdjCN ap cn = cn ** {
s = \\ez,n => cn.s ! aEzafa ! n ++ ap.s ! ez; -- check the form of adjective and also cn.s!ez!n changed from cn.s!aEzafa!n to have correct enclicitic form other wise it creats wrong enclictic form of old man s = \\ez,n => cn.s ! aEzafa ! n ++ ap.s ! ez -- check the form of adjective and also cn.s!ez!n changed from cn.s!aEzafa!n to have correct enclicitic form other wise it creats wrong enclictic form of old man
animacy = cn.animacy ;
definitness = cn.definitness
} ; } ;
RelCN cn rs = { RelCN cn rs = cn ** {
s = \\ez,n => cn.s ! enClic ! n ++ rs.s ! agrPesP3 n ; s = \\ez,n => cn.s ! enClic ! n ++ rs.s ! agrPesP3 n ;
animacy = cn.animacy ;
definitness = cn.definitness
} ; } ;
AdvCN cn ad = {s = \\ez,n => cn.s ! aEzafa ! n ++ ad.s ; animacy = cn.animacy ; definitness = cn.definitness} ; AdvCN cn ad = cn ** {s = \\ez,n => cn.s ! aEzafa ! n ++ ad.s} ;
SentCN cn sc = {s = \\ez,n => cn.s ! ez ! n ++ sc.s ; animacy = cn.animacy ; definitness = cn.definitness} ; SentCN cn sc = cn ** {s = \\ez,n => cn.s ! ez ! n ++ sc.s} ;
ApposCN cn np = {s = \\ez,n => cn.s ! ez ! n ++ np.s ! NPC aEzafa ; animacy = cn.animacy ; definitness = True} ; -- ezafa form of city to be used ApposCN cn np = cn ** {s = \\ez,n => cn.s ! ez ! n ++ np.s ! NPC aEzafa ; definitness = True} ; -- ezafa form of city to be used
} }

111
src/persian/ParadigmsPes.gf
View File

@@ -1,6 +1,6 @@
--# -path=.:../abstract:../../prelude:../common --# -path=.:../abstract:../prelude:../common
-- --
----1 Pnbu Lexical Paradigms ----1 Persian Lexical Paradigms
resource ParadigmsPes = open resource ParadigmsPes = open
Predef, Predef,
@@ -23,14 +23,26 @@ oper
singular = Sg ; plural = Pl; singular = Sg ; plural = Pl;
animate = Animate ; inanimate = Inanimate ; --i animate = Animate ; inanimate = Inanimate ; --i
mkN01 : Str -> Animacy -> Noun ; mkN01 : (sg : Str) -> Animacy -> Noun ; -- Takes singular form and animacy, forms plural with ها
mkN01 str ani = mkN str (str ++ "ها") ani; mkN01 sg ani =
mkN02 : Str -> Animacy -> Noun ; let pl = zwnj sg "ها" ; -- Using zero-width non-joiner, defined in ResPes
mkN02 str ani = case (last str) of { in MorphoPes.mkN sg pl ani ;
"ه" => mkN str ((init str) + "گان") ani ;
("ا"|"و") => mkN str (str + "یان") ani ; mkN02 : (sg : Str) -> Animacy -> Noun ; -- Takes singular form and animacy, pattern matches singular and forms plural with either گان, یان or ان
_ => mkN str (str+"ان") ani mkN02 str ani = case last str of {
"ه" => MorphoPes.mkN str (init str + "گان") ani ;
("ا"|"و") => MorphoPes.mkN str (str + "یان") ani ;
_ => MorphoPes.mkN str (str + "ان") ani
}; };
mkN = overload {
mkN : (sg : Str) -> N -- Takes singular form, returns an inanimate noun with ها as the plural form
= \sg -> mkN01 sg inanimate ;
mkN : (sg,pl : Str) -> N -- Takes singular and plural form, returns an inanimate noun
= \sg,pl -> MorphoPes.mkN sg pl inanimate ;
mkN : (sg,pl : Str) -> Animacy -> N -- Worst-case constructor: takes singular and plural forms and animacy
= \sg,pl,ani -> MorphoPes.mkN sg pl ani
} ;
{- {-
--2 Nouns --2 Nouns
@@ -44,21 +56,20 @@ oper
-} -}
-- Compound Nouns -- Compound Nouns
mkCmpdNoun1 : Str -> N -> N mkCmpdNoun1 : Str -> N -> N
= \s,noun -> {s =\\ez,n => s ++ noun.s ! ez ! n ; animacy = noun.animacy ; definitness = noun.definitness ; lock_N = <>}; = \s,noun -> noun ** {s =\\ez,n => s ++ noun.s ! ez ! n};
mkCmpdNoun2 : N -> Str -> N mkCmpdNoun2 : N -> Str -> N
= \noun,s -> {s =\\ez,n => noun.s ! ez ! n ++ s ; animacy = noun.animacy ; definitness =noun.definitness ; lock_N = <>}; = \noun,s -> noun ** {s =\\ez,n => noun.s ! ez ! n ++ s};
-- Proper names -- Proper names
mkPN : Str -> Animacy -> PN = mkPN : Str -> Animacy -> PN
= \str,ani -> lin PN {s = str ; animacy = ani} ;
\str,ani -> {s = str ; animacy = ani ; lock_PN = <>} ;
-- Personal Pronouns -- Personal Pronouns
personalPN : Str -> Number -> PPerson -> Pron = personalPN : Str -> Number -> PPerson -> Pron
\str,nn,p -> {s = str ; a = AgPes nn p ; ps = str ; lock_Pron = <>}; = \str,nn,p -> lin Pron {s = str ; a = AgPes nn p ; ps = str};
{- {-
-- Demonstration Pronouns -- Demonstration Pronouns
demoPN : Str -> Str -> Str -> Quant = demoPN : Str -> Str -> Str -> Quant =
@@ -66,11 +77,11 @@ oper
-- Determiner -- Determiner
-} -}
mkDet = overload { mkDet = overload {
mkDet : Str -> Number -> Det = mkDet : Str -> Number -> Det
\s1,n -> makeDet s1 n False ** { lock_Det = <>}; = \s1,n -> lin Det (makeDet s1 n False);
mkDet : Str -> Number -> Bool -> Det = mkDet : Str -> Number -> Bool -> Det
\s1,n,b -> makeDet s1 n b ** { lock_Det = <>}; = \s1,n,b -> lin Det (makeDet s1 n b)
}; };
{- {-
-- Intergative pronouns -- Intergative pronouns
mkIP : (x1,x2,x3,x4:Str) -> Number -> Gender -> IP = mkIP : (x1,x2,x3,x4:Str) -> Number -> Gender -> IP =
@@ -82,13 +93,13 @@ oper
--2 Adjectives --2 Adjectives
mkA = overload { mkA = overload {
mkA : Str-> A mkA : Str -> A
= \str -> mkAdj str str ** { lock_A = <>} ; = \str -> lin A (mkAdj str str);
mkA : Str-> Str -> A mkA : Str-> Str -> A
= \str,adv -> mkAdj str adv ** { lock_A = <>} ; = \str,adv -> lin A (mkAdj str adv);
mkA : Str -> Str -> A2 mkA : Str -> Str -> A2
= \a,c -> mkAdj a a ** { c2 = c ; lock_A2 = <>} ; = \a,c -> lin A2 (mkAdj a a ** {c2 = c})
} ; } ;
--2 Verbs --2 Verbs
mkV : Str -> Str -> V mkV : Str -> Str -> V
@@ -113,26 +124,28 @@ oper
} ; } ;
mkV3 : V -> Str -> Str -> V3; mkV3 : V -> Str -> Str -> V3;
mkV3 v p q = v ** { c2 = p ; c3 = q ; lock_V3 = <>} ; mkV3 v p q = lin V3 (v ** {c2 = p ; c3 = q}) ;
mkV2V : V -> Str -> Str -> Bool -> V2V ; mkV2V : V -> Str -> Str -> Bool -> V2V ;
mkV2V v s1 s2 b = v ** {isAux = b ; c1 = s1 ; c2 = s2 ; lock_V2V = <>} ; mkV2V v s1 s2 b = lin V2V (v ** {isAux = b ; c1 = s1 ; c2 = s2}) ;
-- compund verbs -- compund verbs
compoundV = overload { compoundV = overload {
compoundV : Str -> V -> V = \s,v -> {s = \\vf => s ++ v.s ! vf ; lock_V = <>} ; compoundV : Str -> V -> V
compoundV : Str -> V2 -> V = \s,v -> {s = \\vf => s ++ v.s ! vf ; lock_V = <>} ; = \s,v -> v ** {s = \\vf => s ++ v.s ! vf} ;
}; compoundV : Str -> V2 -> V
= \s,v -> lin V {s = \\vf => s ++ v.s ! vf} ;
};
invarV : Str -> V -- for verbs like " بایستن " ("must"), which don't inflect invarV : Str -> V -- for verbs like " بایستن " ("must"), which don't inflect
= \s -> lin V {s = \\_ => s} ; = \s -> lin V {s = \\_ => s} ;
----2 Adverbs ----2 Adverbs
mkAdv : Str -> Adv = \str -> {s = str ; lock_Adv = <>}; mkAdv : Str -> Adv = \str -> lin Adv {s = str} ;
----2 Prepositions ----2 Prepositions
mkPrep : Str -> Prep ; mkPrep : Str -> Prep ;
mkPrep str = {s = str ; lock_Prep = <>}; mkPrep str = lin Prep {s = str};
{- {-
--3 Determiners and quantifiers --3 Determiners and quantifiers
@@ -186,23 +199,5 @@ oper
-- --
-- V0 : Type ; -- V0 : Type ;
-- AS, A2S, AV, A2V : Type ; -- AS, A2S, AV, A2V : Type ;
-- --}
----.
----2 Definitions of paradigms
----
---- The definitions should not bother the user of the API. So they are
---- hidden from the document.
--
-- Gender = MorphoHin.Gender ;
-- Number = MorphoHin.Number ;
-- Case = MorphoHin.Case ;
-- human = Masc ;
-- nonhuman = Neutr ;
-- masculine = Masc ;
-- feminine = Fem ;
-- singular = Sg ;
-- plural = Pl ;
-- nominative = Nom ;
-- genitive = Gen ;
-}
} }

4
src/persian/ResPes.gf
View File

@@ -62,6 +62,10 @@ resource ResPes = ParamX ** open Prelude,Predef in {
Neg => CNeg b Neg => CNeg b
} ; } ;
-- Zero-width non-joiner, used for certain morphemes
-- See https://en.wikipedia.org/wiki/Persian_alphabet#Word_boundaries
ZWNJ : Str = "‌" ;
zwnj : Str -> Str -> Str = \s1,s2 -> s1 + ZWNJ + s2 ;
----------------------- -----------------------
--- Verb Phrase --- Verb Phrase
----------------------- -----------------------

6
src/persian/SymbolPes.gf
View File

@@ -16,19 +16,19 @@ concrete SymbolPes of Symbol = CatPes ** open Prelude, ResPes in {
animacy = cn.animacy animacy = cn.animacy
} ; } ;
CNSymbNP det cn xs = { CNSymbNP det cn xs = {
s = \\ez => det.s ++ cn.s ! aEzafa ! det.n ++ xs.s ; s = \\ez => det.s ++ cn.s ! aEzafa ! det.n ++ xs.s ;
a = agrPesP3 det.n ; a = agrPesP3 det.n ;
animacy = cn.animacy animacy = cn.animacy
} ; } ;
CNNumNP cn i = { CNNumNP cn i = {
s = \\ez => cn.s ! aEzafa ! Sg ++ i.s ; s = \\ez => cn.s ! aEzafa ! Sg ++ i.s ;
a = agrPesP3 Sg ; a = agrPesP3 Sg ;
animacy = cn.animacy animacy = cn.animacy
} ; } ;
SymbS sy = sy ; SymbS sy = sy ;
SymbNum sy = { s = sy.s ; n = Pl } ; SymbNum sy = { s = sy.s ; n = Pl } ;
SymbOrd sy = { s = sy.s ++ "wN" ; n = Pl} ; SymbOrd sy = { s = sy.s ++ "wN" ; n = Pl; isNum=False} ;
lincat lincat

5
src/persian/VerbPes.gf
View File

@@ -41,9 +41,10 @@ concrete VerbPes of Verb = CatPes ** open ResPes in {
ReflVP v = insertObjPre (\\a => reflPron ! a) v ; ReflVP v = insertObjPre (\\a => reflPron ! a) v ;
PassV2 v = predV v ; -- need to be fixed PassV2 v = predV v ; -- need to be fixed
CompAP ap ={s = \\_ => ap.s ! bEzafa} ; -- check form of adjective CompAP ap ={s = \\_ => ap.s ! bEzafa} ; -- check form of adjective
CompNP np = {s = \\a => np.s ! NPC bEzafa} ;
CompAdv adv = {s = \\_ => adv.s } ; CompAdv adv = {s = \\_ => adv.s } ;
CompCN cn = {s = \\a => cn.s ! bEzafa ! giveNumber a } ; CompCN cn = {s = \\a => cn.s ! bEzafa ! giveNumber a } ;
-- IL 2019-01-31 changed bEzafa to enClic according to
-- https://sites.la.utexas.edu/persian_online_resources/nouns/noun-in-a-predicative-position/
CompNP np = {s = \\a => np.s ! NPC enClic} ;
} }