mirror of
https://github.com/explosion/spaCy.git
synced 2024-12-26 18:06:29 +03:00
507 lines
6.0 KiB
Python
507 lines
6.0 KiB
Python
# coding: utf8
|
|
from __future__ import unicode_literals
|
|
|
|
ID_BASE_EXCEPTIONS = """
|
|
Oedipus-kompleks
|
|
Zend-Avesta
|
|
aba-aba
|
|
abah-abah
|
|
abong-abong
|
|
abrit-abrit
|
|
abu-abu
|
|
abuk-abuk
|
|
abun-abun
|
|
acang-acang
|
|
aci-aci
|
|
adang-adang
|
|
agak-agih
|
|
agar-agar
|
|
agut-agut
|
|
ajar-ajar
|
|
aji-aji
|
|
alai-belai
|
|
alan-alan
|
|
alang-alang
|
|
alap-alap
|
|
ali-ali
|
|
alih-alih
|
|
alu-alu
|
|
alu-aluan
|
|
alun-alun
|
|
alur-alur
|
|
ambah-ambah
|
|
ambai-ambai
|
|
ambreng-ambrengan
|
|
ambu-ambu
|
|
ambung-ambung
|
|
amin-amin
|
|
amit-amit
|
|
ampai-ampai
|
|
anai-anai
|
|
ancak-ancak
|
|
ancar-ancar
|
|
andang-andang
|
|
andeng-andeng
|
|
anggung-anggip
|
|
angkal-angkal
|
|
angkul-angkul
|
|
angkut-angkut
|
|
ani-ani
|
|
aning-aning
|
|
anjang-anjang
|
|
anjing-anjing
|
|
antah-berantah
|
|
antar-antar
|
|
anter-anter
|
|
antung-antung
|
|
anyang-anyang
|
|
api-api
|
|
apit-apit
|
|
apu-apu
|
|
arem-arem
|
|
ari-ari
|
|
asuh-asuh
|
|
awar-awar
|
|
awat-awat
|
|
awik-awik
|
|
ayak-ayak
|
|
ayang-ayang
|
|
ba-bi-bu
|
|
bajang-bajang
|
|
baji-baji
|
|
balai-balai
|
|
bangun-bangun
|
|
barat-barat
|
|
barau-barau
|
|
bare-bare
|
|
bari-bari
|
|
barung-barung
|
|
basa-basi
|
|
bata-bata
|
|
batir-batir
|
|
bayan-bayan
|
|
begana-begini
|
|
belat-belit
|
|
belu-belai
|
|
bengang-bengut
|
|
benggal-benggil
|
|
bengkal-bengkil
|
|
bengkang-bengkok
|
|
bengkang-bengkong
|
|
bengkang-bengkung
|
|
berang-berang
|
|
beras-beras
|
|
berek-berek
|
|
bereng-bereng
|
|
beri-beri
|
|
bering-bering
|
|
berkali-kali
|
|
bermacam-macam
|
|
bersama-sama
|
|
bersiap-siap
|
|
bertanya-tanya
|
|
berturut-turut
|
|
betak-betak
|
|
beti-beti
|
|
biar-biar
|
|
bilang-bilang
|
|
bincang-bincut
|
|
biri-biri
|
|
bolak-balik
|
|
bolang-baling
|
|
bongkar-bangkir
|
|
boro-boro
|
|
boyo-boyo
|
|
buaya-buaya
|
|
bugi-bugi
|
|
bulang-baling
|
|
buli-buli
|
|
bulu-bulu
|
|
bulus-bulus
|
|
caci-maki
|
|
cawe-cawe
|
|
celam-celum
|
|
celangak-celinguk
|
|
celas-celus
|
|
celedang-celedok
|
|
celengkak-celengkok
|
|
cenal-cenil
|
|
cengar-cengir
|
|
cengir-cengir
|
|
cengis-cengis
|
|
ceplas-ceplos
|
|
cerai-berai
|
|
cobak-cabik
|
|
cobar-cabir
|
|
cola-cala
|
|
colang-caling
|
|
comat-comot
|
|
compang-camping
|
|
conggah-canggih
|
|
congkah-cangkih
|
|
congkah-mangkih
|
|
copak-capik
|
|
crat-crit
|
|
cumi-cumi
|
|
dag-dag
|
|
dag-dig-dug
|
|
dali-dali
|
|
dapur-dapur
|
|
dari-dari
|
|
daru-daru
|
|
dawai-dawai
|
|
dekak-dekak
|
|
desas-desus
|
|
dewi-dewi
|
|
dinul-Islam
|
|
dog-dog
|
|
dokok-dokok
|
|
dolak-dalik
|
|
dulang-dulang
|
|
engap-engap
|
|
engket-engket
|
|
erek-erek
|
|
flip-plop
|
|
gaba-gaba
|
|
gada-gada
|
|
gading-gading
|
|
gado-gado
|
|
gala-gala
|
|
galing-galing
|
|
galu-galu
|
|
gana-gini
|
|
gapah-gopoh
|
|
gara-gara
|
|
gayang-gayang
|
|
gebyah-uyah
|
|
gedana-gedini
|
|
gedebak-gedebuk
|
|
gedebar-gedebur
|
|
geli-geli
|
|
gembar-gembor
|
|
gembrang-gembreng
|
|
gempul-gempul
|
|
genjang-genjot
|
|
gerbas-gerbus
|
|
gereng-gereng
|
|
gerot-gerot
|
|
getak-getuk
|
|
getem-getem
|
|
geti-geti
|
|
gili-gili
|
|
girap-girap
|
|
giring-giring
|
|
golak-galik
|
|
gondas-gandes
|
|
gonta-ganti
|
|
gorong-gorong
|
|
gresek-gresek
|
|
gudu-gudu
|
|
gulang-gulang
|
|
halai-balai
|
|
hasta-wara
|
|
hila-hila
|
|
hiru-biru
|
|
hiru-hara
|
|
hubar-habir
|
|
hubaya-hubaya
|
|
hula-hula
|
|
hura-hura
|
|
huru-hara
|
|
ibar-ibar
|
|
icak-icak
|
|
ikan-ikan
|
|
ilam-ilam
|
|
ilat-ilatan
|
|
iming-iming
|
|
imut-imut
|
|
incang-incut
|
|
ingat-ingat
|
|
inggang-inggung
|
|
ira-ira
|
|
irah-irahan
|
|
isak-isak
|
|
jali-jali
|
|
jarum-jarum
|
|
jegal-jegalan
|
|
jelur-jelir
|
|
jentik-jentik
|
|
jongkar-jangkir
|
|
jongkat-jangkit
|
|
kabang-kabang
|
|
kabu-kabu
|
|
kacang-kacang
|
|
kacang-kacangan
|
|
kala-kala
|
|
kalau-kalau
|
|
kalut-malut
|
|
kanak-kanak
|
|
kareseh-peseh
|
|
kasak-kusuk
|
|
katang-katang
|
|
kawa-kawa
|
|
kecuh-kecah
|
|
kelak-kelik
|
|
kelak-keluk
|
|
kelang-kelok
|
|
kelap-kelip
|
|
kelemak-kelemek
|
|
kelik-kelik
|
|
kelontang-kelantung
|
|
kelusuh-kelasah
|
|
kencrang-kencring
|
|
kerang-keroh
|
|
kercap-kercip
|
|
kercap-kercup
|
|
keriang-keriut
|
|
kernyat-kernyut
|
|
keropas-kerapis
|
|
keruntang-pungkang
|
|
kesap-kesip
|
|
kesu-kesi
|
|
kesuh-kesih
|
|
kesuk-kesik
|
|
ketakar-keteker
|
|
ketap-ketap
|
|
ketir-ketir
|
|
kiang-kiut
|
|
kibang-kibut
|
|
kicang-kecoh
|
|
kicang-kicu
|
|
kida-kida
|
|
kijang-kijang
|
|
kira-kira
|
|
kiu-kiu
|
|
kluntang-klantung
|
|
kocah-kacih
|
|
kocar-kacir
|
|
kodok-kodok
|
|
kolang-kaling
|
|
kole-kole
|
|
koleh-koleh
|
|
komat-kamit
|
|
kontal-kantil
|
|
kontang-kanting
|
|
kopat-kapit
|
|
kosak-kasik
|
|
kotak-katik
|
|
kucar-kacir
|
|
kuda-kuda
|
|
kulik-kulik
|
|
kunang-kunang
|
|
kunar-kunar
|
|
kupat-kapit
|
|
kupu-kupu
|
|
kura-kura
|
|
kusat-mesat
|
|
kutat-kutet
|
|
kuwung-kuwung
|
|
laba-laba
|
|
labi-labi
|
|
laguh-lagah
|
|
lalu-lalang
|
|
lamat-lamat
|
|
lancar-lancar
|
|
langak-longok
|
|
langit-langit
|
|
lapat-lapat
|
|
legak-legok
|
|
legup-legup
|
|
leha-leha
|
|
lekak-lekuk
|
|
lekap-lekup
|
|
lekuh-lekih
|
|
lekum-lekum
|
|
lekup-lekap
|
|
lenggak-lenggok
|
|
lentam-lentum
|
|
lentang-lentok
|
|
lete-lete
|
|
lidah-lidah
|
|
liuk-liuk
|
|
liung-liung
|
|
lobi-lobi
|
|
longak-longok
|
|
lontang-lanting
|
|
lontang-lantung
|
|
lumba-lumba
|
|
lumi-lumi
|
|
luntang-lantung
|
|
lupa-lupa
|
|
mahi-mahi
|
|
mali-mali
|
|
masing-masing
|
|
mata-mata
|
|
megal-megol
|
|
megap-megap
|
|
megrek-megrek
|
|
melak-melak
|
|
menanti-nanti
|
|
mencla-mencle
|
|
mentul-mentul
|
|
mercak-mercik
|
|
mondar-mandir
|
|
morat-marit
|
|
nasi-nasi
|
|
neka-neka
|
|
nong-nong
|
|
ogah-agih
|
|
ogah-ogah
|
|
ogak-agik
|
|
ogak-ogak
|
|
olak-alik
|
|
olak-olak
|
|
olang-aling
|
|
oleh-oleh
|
|
olong-olong
|
|
ombang-ambing
|
|
onde-onde
|
|
ondel-ondel
|
|
oneng-oneng
|
|
ongol-ongol
|
|
ontran-ontran
|
|
onyah-anyih
|
|
onyak-anyik
|
|
opak-apik
|
|
orak-arik
|
|
orang-aring
|
|
orat-oret
|
|
orok-orok
|
|
orong-orong
|
|
otak-atik
|
|
otak-otak
|
|
palas-palas
|
|
palato-alveolar
|
|
pao-pao
|
|
parang-parang
|
|
pasu-pasu
|
|
paya-paya
|
|
perak-perak
|
|
peras-perus
|
|
pertama-tama
|
|
petantang-petenteng
|
|
petatang-peteteng
|
|
pete-pete
|
|
pina-pina
|
|
plintat-plintut
|
|
plonga-plongo
|
|
pontang-panting
|
|
porak-parik
|
|
porak-peranda
|
|
porak-poranda
|
|
pukang-pukang
|
|
pungak-pinguk
|
|
pura-pura
|
|
puruk-parak
|
|
puvi-puvi
|
|
raba-rubu
|
|
rama-rama
|
|
randa-rondo
|
|
rango-rango
|
|
rep-repan
|
|
riang-riang
|
|
ribu-ribu
|
|
rigi-rigi
|
|
rintik-rintik
|
|
robak-rabik
|
|
robat-rabit
|
|
rombang-rambing
|
|
romol-romol
|
|
rondah-rondih
|
|
ropak-rapik
|
|
ruba-ruba
|
|
rumbai-rumbai
|
|
rundu-rundu
|
|
runggu-rangga
|
|
runtang-runtung
|
|
sabu-sabu
|
|
sama-sama
|
|
samo-samo
|
|
sampai-sampai
|
|
sarit-sarit
|
|
sebaik-baiknya
|
|
segan-segan
|
|
sekali-kali
|
|
sekonyong-konyong
|
|
sekurang-kurangnya
|
|
selak-seluk
|
|
selama-lamanya
|
|
selang-seli
|
|
selang-seling
|
|
selentang-selenting
|
|
seluk-beluk
|
|
sema-sema
|
|
semang-semang
|
|
semata-mata
|
|
sengar-sengir
|
|
seolah-olah
|
|
sepala-pala
|
|
serbah-serbih
|
|
serembah-serembih
|
|
serta-merta
|
|
setidak-tidaknya
|
|
sia-sia
|
|
slintat-slintut
|
|
sorak-sorai
|
|
sudu-sudu
|
|
sulat-sulit
|
|
sumpah-sumpah
|
|
tabar-tabar
|
|
takang-takik
|
|
tawa-tawa
|
|
tawak-tawak
|
|
tawang-tawang
|
|
tawar-tawar
|
|
teka-teki
|
|
temut-temut
|
|
teringat-ingat
|
|
tiba-tiba
|
|
timah-timah
|
|
timu-timu
|
|
tuang-tuang
|
|
tuli-tuli
|
|
tumpang-tindih
|
|
tupai-tupai
|
|
ubrak-abrik
|
|
ubun-ubun
|
|
ubur-ubur
|
|
uci-uci
|
|
ugal-ugalan
|
|
uget-uget
|
|
uir-uir
|
|
ula-ula
|
|
ulang-alik
|
|
ulang-aling
|
|
ulap-ulap
|
|
ular-ular
|
|
ular-ularan
|
|
ulu-ulu
|
|
ulung-ulung
|
|
umang-umang
|
|
umbang-ambing
|
|
umbul-umbul
|
|
umbut-umbut
|
|
uncang-uncit
|
|
unduk-unduk
|
|
undung-undung
|
|
undur-undur
|
|
unek-unek
|
|
ungah-angih
|
|
unggang-anggit
|
|
unggat-unggit
|
|
untang-anting
|
|
ura-ura
|
|
uran-uran
|
|
user-user
|
|
utak-atik
|
|
utar-utar
|
|
uter-uter
|
|
wanti-wanti
|
|
wara-wiri
|
|
warna-warni
|
|
wira-wiri
|
|
wora-wari
|
|
yang-yang
|
|
zig-zag
|
|
""".split()
|