Skip to content
Snippets Groups Projects
Commit cd7e633b authored by Adam Radziszewski's avatar Adam Radziszewski
Browse files

add takipi rules in CCL

parent 34b1f117
Branches
No related merge requests found
Hide whitespace changes
Inline Side-by-side
examples/takipi_rules.ccl 0 → 100644
+ 692
− 0
View file @ cd7e633b
tag_rules(
rule("10",
and(
equal(class[0],{adj,conj,pred,qub,subst}),
equal(class[-1],{prep}),
inter(cas[-1],cas[0]),
not(
in(orth[-1],["Niby","niby"]),
agrpp(0, 1, {sg,n,cas})
),
not(
and(
equal(class[1],{ppron3}),
inter(cas[1],{gen}),
agrpp(0, 2, {sg,n,cas})
)
)
),
delete(not(
equal(class[0],{subst})
))
);
rule("40",
and(
equal(orth[0],["wszystkim"]),
in(orth[-1],["przede","Przede"])
),
delete(not(
and(
equal(class[0],{subst}),
equal(cas[0],{inst}),
equal(gnd[0],{n})
)
))
);
rule("41",
and(
in(orth[0],["ze","Ze","ZE"]),
in(orth[1],["wszech","WSZECH"])
),
delete(not(equal(cas[0],{gen})))
);
rule("42",
and(
in(orth[0],["W","w"]),
in(orth[1],["stanie","STANIE"])
),
delete(not(equal(cas[0],{loc})))
);
rule("70",
and(
in(class[0],{adj,ger,subst}),
inter(cas[0],{gen}),
in(orth[-1],["dużo","mało","mnóstwo","parę","niewiele","wiele"])
),
delete(not(
equal(cas[0],{gen})
))
);
rule("96",
and(
in(orth[0],["z","Z","ze","Ze"]),
or(
in(orth[1],["dala","daleka","bliska","godzinę"]),
and(
inter(class[1],{num}),
not(inter(cas[1],{gen,inst}))
)
)
),
delete(not(
equal(class[0],{qub})
))
);
rule("97",
and(
in(orth[0],["z","Z","ze","Ze"]),
or(
inter(cas[1],{gen,inst}),
regex(orth[1], "\\p{Lu}.*"),
regex(orth[0], ".*[0-9].*")
)
),
delete(not(
equal(class[0],{prep})
))
);
rule("98",
and(
in(orth[0],["z","Z","ze","Ze"]),
in(orth[1], ["tak", "Tak"])
),
delete(not(
equal(class[0],{prep})
))
);
rule("99",
and(
in(orth[0],["niedawna","dawna"]),
in({prep},class[-1])
),
delete(not(equal(class[0],{qub})))
);
rule("100",
and(
equal(class[-1],{prep}),
not(
and(
not(
regex(orth[-1], "\\p{Lu}.*")
),
regex(orth[0], "\\p{Lu}.*")
),
in(orth[-1],["niby","Niby"]),
inter(class[0],{ppron3,num}),
in(orth[0],["tysiąc","tego","niedawna"]),
and(
in(orth[-1],["po","Po"]),
equal(cas[0],{dat})
),
and(
in(orth[-1],["o","O"]),
in(orth[0],["rany", "RANY"])
),
equal(class[0],{prep}),
and(
in(orth[-1],["mimo","Mimo","pomimo","Pomimo"]),
equal(orth[0],["to"])
)
),
in(cas[0],{nom,gen,acc,dat,loc,inst,voc})
),
delete(
not(
equal(cas[0],{}),
equal(cas[0],{gen}),
in(cas[0],cas[-1])
))
);
rule("101",
and(
equal(class[-1],{prep}),
equal(class[0],{ppron3}),
not(
inter(class[1],{adj,ger,ign,num,numcol,pact,ppas,subst,xxs,xxx}),
and(
inter(class[1],{adv,qub}),
inter(class[2],{adj,ger,ign,num,numcol,pact,ppas,subst,xxs,xxx})
)
)
),
delete(not(
equal(cas[0],{}),
in(cas[0],cas[-1])
))
);
rule("103",
and(
equal(orth[-1],["w"]),
equal(orth[0],["tysiąc"]),
equal(class[1],{num})
),
delete(not(
equal(cas[0],{}),
equal(cas[0],{nom})
))
);
rule("105",
and(
equal(class[0],{subst}),
not(
equal(orth[0],["tysiąc"]),
and(
equal(class[-1],{prep}),
regex(orth[0], "\\p{Lu}.*"),
not(
regex(orth[-1], "\\p{Lu}.*")
)
)
),
llook(-1,begin,$Prep,equal(class[$Prep],{prep})),
not(
inter(base[$Prep],["niby"])
),
only($Prep + 1,-1,$A,and(
equal(class[$A],{adj}),
not(
inter(base[$A],["który","jakiś","jaki"])
)
))
),
delete(not(
equal(cas[0],{gen}),
in(cas[0],cas[-1])
))
);
rule("130",
and(
and(
equal(class[-1],{num}),
not(
inter(base[-1],["oba","półtora","wiele"])
)
),
and(
in(class[0],{subst}),
not(
inter(base[0],["jeden","procent","deka","gram","kilo","para","wolta"])
),
agrpp(0, -1, {nmb,gnd,cas})
)
),
delete(equal(nmb[0],{sg}))
);
rule("140",
and(
inter(class[0],{fin}),
equal(class[-1],{fin}),
not(
and(
in(orth[-1],["jest","znaczy"]),
equal(orth[-2],["to"])
),
and(
equal(orth[-1],["wydaje"]),
equal(orth[0],["może"])
)
)
),
delete(equal(class[0],{fin}))
);
rule("141",
and(
inter(class[0],{fin}),
equal(class[-1],{qub}),
equal(class[-2],{fin})
),
delete(equal(class[0],{fin}))
);
rule("142",
and(
inter(class[0],{fin}),
equal(class[-1],{qub}),
equal(class[-2],{qub}),
equal(class[-3],{fin})
),
delete(equal(class[0],{fin}))
);
rule("150",
and(
inter(class[0],{fin}),
equal(class[1],{fin})
),
delete(equal(class[0],{fin}))
);
rule("151",
and(
inter(class[0],{fin}),
equal(class[1],{qub}),
equal(class[2],{fin})
),
delete(equal(class[0],{fin}))
);
rule("152",
and(
inter(class[0],{fin}),
equal(class[1],{qub}),
equal(class[2],{qub}),
equal(class[3],{fin})
),
delete(equal(class[0],{fin}))
);
rule("160",
and(
equal(class[-1],{prep}),
not(
in(orth[-1],["Niby","niby"])
),
inter(class[0],{fin,praet})
),
delete(in(class[0],{fin,praet}))
);
rule("170",
and(
equal(class[-1],{prep}),
equal(class[0],{adj}),
not(
inter(base[0],["który"]),
and(
equal(orth[-1],["po"]),
in(orth[0],["pierwsze","drugie","trzecie","czwarte","piąte"])
),
and(
inter(base[0],["niektóry","inny"]),
inter(nmb[0],{pl})
)
),
or(
equal(class[1],{adj}),
and(
equal(class[1],{subst}),
in(cas[1],{nom,acc,dat,loc,inst,voc}),
not(
in(base[1],["tysiąc","milion","miliard"])
)
)
),
agrpp(0, 1, {nmb,gnd,cas})
),
delete(not(
agrpp(0, 1, {nmb,gnd,cas})
))
);
rule("175",
and(
equal(orth[-1],["po"]),
in(orth[0],["pierwsze","drugie","trzecie","czwarte","piąte"])
),
delete(
not(
and(
equal(class[0],{adj}),
equal(nmb[0],{sg}),
equal(cas[0],{acc}),
equal(gnd[0],{n})
)
))
);
rule("180",
and(
equal(orth[0],["po"]),
equal(class[1],{adjp})
),
delete(not(
equal(cas[0],{loc})
))
);
rule("181",
and(
in(orth[0],["prawie","Prawie","PRAWIE"]),
not(
and(
in({prep},class[-1]),
inter(cas[-1],cas[0])
)
)
),
delete(not(
equal(class[0],{qub})
))
);
rule("190",
and(
equal(class[0],{prep}),
in(class[1],{adj,ppas,pact,subst,depr,ger,num,ppron12,ppron3,numcol}),
not(
and(
inter(cas[1],{gen}),
not(
equal(cas[2],{})
)
),
and(
in(orth[0],["Po","po"]),
equal(cas[1],{dat})
),
and(
in(orth[-1],["Z","z"]),
equal(orth[0],["nad"])
),
and(
in(orth[-1],["Od","od"]),
equal(orth[0],["przed"])
),
and(
in(orth[0],["Ze","ze","Pomiędzy","pomiędzy"]),
inter(class[1],{num})
),
and(
in(base[1],["tysiąc","milion","miliard"]),
inter(cas[1],{nom})
),
and(
not(
regex(orth[0], "\\p{Lu}.*")
),
regex(orth[1], "\\p{Lu}.*")
)
)
),
delete(not(
in(cas[0],cas[1])
))
);
rule("200",
and(
inter(cas[0],{acc}),
not(
inter(class[0],{prep})
),
equal(orth[-1],["jako"])
),
delete(not(
in(cas[0],{gen,acc,nom}),
equal(cas[0],{})
))
);
rule("210",
and(
inter(base[0],["który"]),
equal(orth[-1],[","]),
not(
equal(cas[-2],{}),
inter(cas[-2],{gen})
),
agrpp(-2, 0, {nmb,gnd})
),
delete(not(
agrpp(-2, 0, {nmb,gnd})
))
);
rule("250",
and(
inter(base[2],["który"]),
equal(orth[1],[","]),
not(
equal(cas[0],{}),
inter(cas[0],{gen}),
and(
not(
agrpp(2, 0, {nmb,gnd})
),
inter( gnd[0],gnd[1])
)
),
agrpp(2, 0, {nmb,gnd})
),
delete(not(
agrpp(2, 0, {nmb,gnd})
))
);
rule("270",
and(
in(class[-1],{subst,ger}),
in(class[0],{adj,pact,ppas}),
equal(orth[1],[","]),
inter(base[2],["który"]),
not(
inter(base[0],["ten"])
),
agrpp(-1, 0, {nmb,gnd,cas})
),
delete(not(
agrpp(-1, 0, {nmb,gnd,cas})
))
);
rule("280",
and(
equal(class[-1],{}),
equal(class[0],{adj}),
in(class[1],{subst,ger}),
agrpp(0, 1, {nmb,gnd,cas})
),
delete(not(
in(gnd[0],gnd[1])
))
);
rule("300",
and(
inter(acm[-1],{rec}),
inter(gnd[0],{m1}),
not(
inter(class[0],{num}),
in(orth[0],["jeden"])
)
),
delete(and(
equal(gnd[0],{m1}),
equal(cas[0],{nom})
))
);
rule("320",
and(
equal(orth[0],["obok"]),
equal(class[1],{interp}),
in(orth[1],[",","."])
),
delete(equal(class[0],{prep}))
);
rule("330",
and(
equal(class[-1],{prep}),
not(
equal(class[0],{prep})
),
inter(class[0],{prep}),
not(
in(orth[-1],["Niby","niby"]),
and(
in(orth[-1],["Na","na"]),
equal(orth[1],["dzień"])
),
and(
equal(orth[0],["ponad"]),
in(orth[-1],["O","o","Przez","przez","Na","na"])
),
and(
in(orth[0],["z"]),
in(orth[-1],["Przed","przed"])
),
and(
in(orth[0],["przed"]),
in(orth[-1],["Od","od"])
)
)
),
delete(equal(class[0],{prep}))
);
rule("340",
and(
equal(class[0],{adj}),
in(class[1],{subst,ger}),
agrpp(0, 1, {nmb,gnd,cas}),
not(
in(base[0],["jeden","czyj","nasz","swój","twój","mój","godny","niegodny"]),
and(
inter(base[0],["który","jaki"]),
or(
equal(class[-1],{interp}),
equal(class[-2],{interp})
)
),
inter(class[-1],{prep,ign}),
and(
equal(orth[-1],["-"]),
inter(class[-2],{adja})
),
agrpp(-1, 0, {nmb,gnd,cas})
)
),
delete(not(
agrpp(0, 1, {nmb,gnd,cas})
))
)
)
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment