No public description

PiperOrigin-RevId: 700435174

fun_langid/fun_langid.py

@@ -445,7 +445,7 @@ def _init_lexicon(self):
   ("smt", " leh/chu/ahi/iing/ung/hi./dii/diin/chi/in a/ing/hung/ang/thu/na/mah/sih/ka/ah a/in/ama/a,/ng a/hun/hule/uleh/hin/nga/va,/lou/amah/uhi/zong/ahun/in,/kaw/pat/hu a/gin/kawm/jiah/a hi/bang/hna/ngin/uhi./ahin/a ka/ng h/hian/aum/ahu/h ah/ah/kha/ach/achi/eh a/a k/uh/n a/h ch/ian/hi/tah/hi,/n ah/path/ong/khat/hat/athi/ban/sil/a di/i hi/chia/thia/thei/aah/awma/g hi/apa/a, h/, hu/h hi/a hu/ngle/h a/toh/ongl/ng k/gleh/ng u/nei/h di/a a/iah/ahi./hih/hul/na h/a, a/tun/in k/izon/hizo/uin/ihna/gam/aha/tung/na a/” ac/a ch/soi/wmah/hi h/ka h/a ah/na d/g ah/the/h ka/chih/h na/un,/ah,/ahch/jesu/zon/ite/ingi/n am/hchu/ng t/ajia/iahc/sah/bawl/um/uh a/ah k/aan/huch/nang/kei/lam/nna/lah ".strip().split("/")),
   ("iu", " ᐊᒻᒪ/ᒃᑯᑦ/ᓂᕐᒧᑦ/ᕐᒧᑦ/ᓄᓇᕗ/ᐱᓕᕆ/ᑦ ᓄᓇ/201/ᖅᑐᑦ/ᐊᒻᒪᓗ/ᓄᓇᕗᒻ/ᖃᐅᔨ/ᓇᕗᒻᒥ/ᒻᒪᓗ/ᐃᓕᓐᓂ/ᓕᓐᓂᐊ/ᕆᔨᒃᑯ/ᓕᕆᔨᒃ/, 20/ᑦ ᐊᒻ/ᓂᐊᖅᑐ/ᓘᓐᓃᑦ/ᑐᐃᓐᓇ/ᓄᓇᓕ/ᓗᑎᒃ/ᐃᓕᓐ/ᐋᓐᓂᐊ/ᒐᕙᒪ/ᖅᑎᑦᑎ/ᐱᓕᕆᐊ/ᖅᑐᖅ/ᐃᖅᑲ/ᓄᓇᕗᑦ/ᓗᒋᑦ/ᓐᓄᑦ/ᑦ ᐱᓕ/ᓐᓃᑦ/ᐃᑲᔪ/ᐋᓐᓂ/ᓪᓗᒋᑦ/ᓇᕗᑦ/ᓐᓇᖅᑐ/ᑮᓇᐅᔭ/ᔨᒃᑯᑦ/ᐃᖅᑲᓇ/ᖅᑲᓇᐃ/ᑲᓇᑕ/ᓕᕆᓂᕐ/ᔪᓐᓇᖅ/ᑲᑎᒪ/ᓐᓂᒃ/ᑮᓇᐅ/ᒃᑲᓐᓂ/ᑦ ᐃᓕ/ᖏᓐᓂᒃ/ᒐᕙᒪᒃ/ᕙᒪᒃᑯ/ᓂᒃ ᐊ/ᕙᓪᓕᐊ/ᖏᓐᓄᑦ/ᕐᓂᕐᒧ/ᖃᓄᐃ/ᓄᓇᓕᖕ/ᓐᓂᐊᕐ/ᓄᑦ ᐊ/ᖅᓯᒪᔪ/ᑦ ᖃᐅ/ᕆᓂᕐᒧ/ᐅᕗᖓ/ᓄᑦ ᐃ/ᑦ ᐃᑲ/ᐅᓪᓗ/ᐅᔪᓐᓇ/ᕐᒥᑦ/ᓐᓂᑦ/ᖃᐅᔨᓴ/ᑦ ᐊᑐ/ᐅᔪᑦ/ᐃᓄᐃᑦ/ᖅᑎᑕᐅ/ᐃᑲᔪᖅ/ᐅᑭᐅ/ᓄᐃᑦ/ᐅᕝᕙ/ᐅᖃᐅ/ᑲᓇᐃᔭ/ᑐᓂᒃ/ᕗᒻᒥ/ᑲᓇᑕᒥ/ᕙᓘᓐᓃ/ᓪᓗᑎᒃ/ᓗᑎᑦ/ᐅᕝᕙᓘ/ᕝᕙᓘᓐ/ᑦ ᒐᕙ/ᐊᖅᑐᑦ/ᑕᐅᓯᒪ/ᑦ, ᐃ/ᑦ ᑲᑎ/ᖅᑎᓪᓗ/ᑐᓕᕆᔨ/ᒧᑦ ᐊ/ᑎᑦᑎᓂ/ᑦ, ᐊ/ᕐᓂᐊᖅ/ᐅᓯᒪᔪ/ᕐᒥᒃ/, ᓄᓇ/ᐃᓄᐃ/ᓂᕐᒥᑦ/ᑦ ᐅᖃ/ᓂᕐᒥᒃ/ᒧᑦ ᐱ/ᓇᓗᓇᐃ/ᓐᓂᐊᖅ/ᖅᐳᑦ/ᐅᑭᐅᖅ/ᓄᑦ,/ᓂᒃ ᐃ/ᓐᓇᕐᓂ/ᑦ ᐃᓄ/ᒥᓂᔅᑕ/ᒪᔪᑦ/ᑐᖅᑕᐅ/ᑭᐅᖅᑕ/ᐅᖅᑕᖅ/ᓇᓗᓇ/ᑖᓐᓇ/ᐱᕙᓪᓕ/ᑎᓪᓗᒋ/ᓯᒪᔪᑦ/ᖅᑐᓂᒃ/ᒃ ᓄᓇ/ᓇᓕᖕᓂ/ᔪᓂᒃ/ᐅᖃᐅᓯ/ᖅᑕᖅᑐ/ᐊᕐᕌ/ᐅᓂᐊᖅ/ᐊᐅᓚ/ᓂᒃ ᐱ/ᑎᓪᓗᒍ/ᑎᑎᕋ/ᓄᑦ ᐱ/ᐊᕐᕌᒍ/ᓐᓂᐊᖃ/ᖏᓐᓂᑦ/ᕈᓐᓇᖅ/ᑐᓄᑦ/ᕐᓗᑎᒃ ".strip().split("/")),
   ("bmm", " tsy/agna/gna/dia/ny/amin/a ma/gny/n'ny/'ny/igny/y ma/aka/ami/min'/izy/ana/raha/tra/a mi/ara/aha/in'n/man/a ts/efa/ny f/egna/isy/a ny/tegn/ako/ny a/andr/ndra/ny t/sy m/any/a za/ary/rah/atra/aign/satr/atri/teg/misy/mis/y mi/ny m/tia/na a/ôlo/za/a am/gnan/na m/a na/tria/kara/y fa/na t/fa/y ny/mah/magn/afa/ria/ragn/y am/y za/o ma/y ta/azy/mba/na n/iky/mar/a an/na,/ka/zah/tsar/sara/ny,/nagn/maha/no/aby/y fi/agn/y ra/ave/egny/zaho/ampi/nana/y an/ia m/mand/olo/iako/mag/a fa/nah/hita/igna/asa/y ka/tsa/ady/y di/y na/atao/aho/tam/raik/fa m/mam/kar/tiak/y ts/mal/tan/ny n/ka m/raka/ra a/aty/aiky/a ta/koà/zegn/ika/sy a/ny r/agny/aza/tami/na/a ti/, ma/a di/azo/gno/ita/mia/lagn/io/oko ".strip().split("/")),
-  ("lhu", " awˬ/ve/teˇ/hui/leh/awˇ/hta/lie/ta‸/hta‸/yoˬ/hk'/ve y/ˇ ve/maˇ/e yo/yaw/yoˬ./yawˇ/g'a/wˬ h/tuˬ/chi/hk'a/piˇ/haw/sha/gaˬ/k'o/wˇ h/chaw/cha/naw/nga/taˍ/k'a/nyi/k'aw/ˬ hk/hpa/ˬ ve/te/ngaˬ/lehˬ/paˍ/k'oˆ/ˍ ve/e aw/ve a/taw/e te/ˇ hp/nawˬ/che/ˬ hu/lo/eh‸/yaˇ/mad/'oˆ/laˇ/shu/caw/eˇ h/aˬ h/aw l/ve h/hpaˍ/ˇ ht/e ch/ve t/ehˬ/law/ve,/ve c/uiˬ/da‸/teh/gal/'aw/ˇ hu/ngal/, ya/ehˬ,/hˬ,/awl/eh,/leh,/ehˇ/aˇ h/peu/ˇ g'/geh/'aˇ/shi/w lo/euˬ/awd/wˬ l/ui h/ˬ ht/‸ k'/‸ ve/maw/la/hpe/yol/ˬ la/oˬ./ˆ ve/i ht/e le/cheh/she/ˬ le/a ve/ˬ te/ˬ sh/hpeh/hehˇ/hka/aˇ v/teh‸/aw‸/ma/tawˇ/yuˬ/aˍ v/suh/wˬ v/g'aˇ/e ht/pa/l li/ˇ le/a‸ h/a‸ k/yol./hk'e/g'u/ˇ ma/ya‸/ve l/ˇ hk/wˇ v/miˬ/eˇ g/i ve/iˇ v/bon ".strip().split("/")),
+  ("lhu", "lie/awl/ve/ted/awd/mad/qha/chi/lad/nga/tha/hawd/mud/gal/kal/naw/har/taw/qhaw/ngal/khaw/yad/haw/lir/thar/mil/nawl/caw/d ve/hawr/tawd/sha/heu/taf/had/cawl/eul/l li/e li/awt/l ve/dkha/awq/awr/chaw/ve l/law/qhad/qot/daw/shi/yol/xeul/qhal/d li/awdk/wdkh/qhe/shif/she/awf/yul/qo/veu/nid/shel/hawq/hie/te/gad/ve t/eof/kheu/siq/iel/ied/cad/l ma/khie/bur/dawd/tedx/xad/yaw/mie/f ve/thad/thor/mid/cha/dqha/qaw/lal/l la/ath/atho/ni/yie/keul/l th/ve y/al l/dar/horm/orma/ad c/meu/i te/hied/rma/tul/aw l/a ve/wd l/yawd/e aw/e ta/tedk/ad l/mal/ma/cal/hi t/e yo/d ca/paf/e ve/d ma/ad v/maw/hif/shie/eol/lheu/l ca/ced/l qh/edkh/t ve/ca/chie/r th/e la/kha/q ve/phu/ve a/r qh/jad/-aw".strip().split("/")),
   ("djk", " den/wan/fu/di/san/be/i a/tak/anga/i de/taki/nga/sama/ama/ang/a a/ani/abi/aki/sam/a de/taa/ala/n sa/kon/i ta/sani/i en/i fu/esi/eke/man/i e/ini/a sa/de/en/gi/en s/meke/e a/u a/moo/i di/a be/wi/bun/mek/i sa/fu d/isi/mi/yeho/ehof/hofa/u de/na/gadu/a b/gad/ati/du/yeh/iti/ten/i u/a di/i be/e de/soi/da/u sa/mu/n fu/e e/a s/fa/saa/a wa/ni d/ofa/efu/ibi/en b/seef/eefi/be e/masa/a ta/yee/a t/asaa/u fu/adu/mas/sesi/a o/libi/fu a/enke/nke/enk/i an/ya/lib/baka/i ye/a ga/eses/na a/e ta/a w/n ta/a d/te/uku/ma d/aini/ain/en t/an s/n a/ooko/asi/efi/go/eepi/i wa/a fu/i ya/epi/dee/son/sa/ni a/iki/tan/i fa/wani/ki a/yeep/aki:/a u/a m/an f/aka/ikin/bak/piki/kai ".strip().split("/")),
   ("cuk", " mala/alad/emar/bab/mar-/degi/dumm/umma/dule/mar/mmad/deg/ad,/dul/sog/be/esús/jesú/bab-/, de/-dum/an/bema/lema/bem/gu,/ega/ar b/bmar/gwen/ibma/anma/ulem/nue/nmar/ab-d/b-du/imak/suli/ar/gine/anm/marg/d, a/id,/egi/ar a/na/jes/, be/egin/ed,/gi,/lad,/noni/mad-/dum/ine,/ne,/adi/igar/bela/ule-/sogd/ga s/b du/d, d/a so/ab d/, ba/, an/geb/-dul/ogde/ibm/sunn/unna/arga/amar/bel/le,/r be/gan/gi b/urba/a be/mad/agi/bali/igi/ela/bur/i be/we-/ima/amal/egi,/imal/ar d/alid/-bur/arma/, ar/aksa/ar,/adga/burb/an-/yob/nued/ulid/sús/dgi/arba/rmal/, du/smal/i an/lad./argi/oged/rgi/dak/nmak/ar-i/e an/oggu/ar i/gusa/mer/an b/usad/ar n/sogg/unma/abin/wen/ade/rga/gar-/bin/rgwe/sun/negg/d, b/lad/-neg/-nue/niki/e, a ".strip().split("/")),
   ("mni", " অসি/্না/নবা/অমসু/অমস/মসুং/সুং/অদু/গী ম/অমা/বা ম/হায়/ম্না/য়াম্/শিং/িবা/া য়া/য়াম/াম্ন/বা অ/া লৈ/া পা/া অম/ংদা/না ম/ম্বা/পাম/ন্না/্বা/ইবা/দগী/না প/মদা/মতম/বা প/থবক/ুনা/বা য়/না ল/ংগী/কপা/ংবা/্তা/য়না/্নবা/বগী/িং অ/না হ/দুনা/িগী/ন্নব/হায়ব/ৌবা/াবা/ইনা/ওইবা/া মী/রিবা/ওইন/মগী/ওইব/কয়া/িনা/ংনা/লৈব/শিংগ/মসি/পাম্/িংগী/িন্ন/লৈবা/ওইনা/া মত/দা ম/ক্তা/্ত্র/শিংদ/ম্মী/না খ/অমদ/া হা/বা ল/সিগী/া ওই/মহাক/ুম্ব/দা ল/য়াও/ান্ন/ঐগী/গী প/বদা/না অ/সিনা/হাক্/প্র/ী মত/হেন/বা হ/বা থ/ায়না/হায়ন/গী অ/ক্কী/াগী/শিংন/ং অস/না ত/্কী/মখোয়/না শ/িদা/খিবা/মফম/রগা/গী ল/বা চ/না চ/লাই/ক্না/ুং ম/দবা/না থ/িংনা/িংদা/েন্ন/া লা/ঐখোয়/সিদা/ল্লি/াম্ম/মদি/খোয়গ/য়গী/লৈর/া অস/া চা/না ফ/ানা/োয়গী/ী ওই/া চৎ/া পু/গুম্/েন্ত/মখল/া ৱা/নবগী/মরু/া মর/মরম/া ঙম/বা ফ/না ক/মখো/জিন্/ী মর/স্ত/দা অ ".strip().split("/")),
@@ -1700,6 +1700,11 @@ def _init_lexicon(self):
   ("pcd", " ch'/est/ech/din/ins/in/del/dins/une/eune/icar/pica/onme/pi/pic/int/et/con/che/des/'est/ion/ed/t pi/mint/et p/e pi/ch'e/h'es/conm/un/card/e di/eun/s ch/ns l/in f/st e/e co/n fr/nt d/s l'/nme/t de/t eu/) ch/part/ne c/gion/régi/ns c/égio/n pi/par/ale/e et/fri/son/frin/rins/insé/pi d/ne p/du/ardi/nmeu/meun/ée d/arte/épar/on p/dépa/cart/dép/sit/i di/situ/itué/uée/e de/rdie/l'r/sonm/tuée/e in/me e/' es/l'ré/nsé/pou/h' e/'rég/el s/l'/l so/sin/h' d/t un/artm/il/die./rtmi/tmin/' dé/) c/ène/ché/oait/ait/(in/éd/chés/hés/te,/é )/sé )/lle/, ch/ille/e ch/aveu/nche/ches/veuc/euc/ave/anch/nne/ère/, si/st u/d'/e ed/e à/ache/in 1/s in/inch/fran/e, s/ch’/hes/qui/e si/les/dec/rte,/éto/dech/e d' ".strip().split("/")),
   ("stl", " de/van/een/et/n de/in/en/laan/aand/is/n et/an d/en d/mit/en v/in d/die/eur/ver/den/wod/ries/dat/ste/uut/bin/frie/ten/nder/ien/veu/en e/veur/erla/eder/n va/rlaa/aans/e ge/nede/in e/ling/en i/odde/tot/de g/d'r/ing/fri/wodd/en./ente/stel/emie/en b/mien/gemi/t de/an e/ient/gem/i'j/derl/sch/dörp/n in/and/gen/llin/ken/elli/is e/en o/ned/het/der/de s/. de/bi'j/en,/r de/was/deur/op/dden/tell/s ee/en w/e st/eens/de n/oord/nte/dör/nao/deu/de b/eer/aor/ok/sta/ingw/de v/ende/gwar/ngwa/nde/as/ter/per/ven/ster/e ve/onde/oost/. et/elik/mar/de k/oold/mee/en a/e va/ande/e fr/west/ere/n en/nse/ied/n ve/scho/de o/nen/over/eken/ers/e in/diel/an/örp/s de/aeme/de d/naem/anse/en s/stao/gro/ond ".strip().split("/")),
   ("twd", " van/de/en/in/n de/met/wör/aand/den/oet/is/nen/det/uur/ver/dee/an d/ste/n t/laan/nder/der/bie/t de/ook/as/vuur/wat/op/vuu/en d/was/oar/et/ne/n va/hef/noa/and/oond/te/eant/r de/ande/in d/duur/wörd/. de/n en/ear/duu/en v/ter/e va/eent/an/nog/et d/zee/ene/e st/doar/n n/onde/ken/in t/hee/lle/twe/doa/n in/e ge/de s/ten/ing/e de/an t/ers/een/leu/mea/este/zin/s de/de v/of/en e/e be/awer/e ve/t n/wal/oawe/, en/e in/en t/de g/mear/mer/de b/dr/e he/elik/e me/woar/ölle/en o/eaid/stoa/gro/töt/t t/völ/wean/en a/um/n me/ent/zeen/chte/oon/e en/is n/ze/ante/ant/, de/vers/oaw/nde/an n/joar/t in/n he/ere/de m/en s/en n/dit/heu/en./n be/joa/en b/acht/tied/, me/gen/eur/en w ".strip().split("/")),
+  ("bca", "nox/zix/zil/lil/anl/mel/inl/ainl/yin/lap/yon/mot/het/l no/gop/sua/mal/yaf/anx/onx/yind/hhe/hol/yan/yap/zei/ded/kex/gua/l zi/yanl/net/yonx/nl n/ainx/gvp/zai/gol/x zi/uinl/not/inx/ient/gan/ganx/xinl/de/tel/sex/gai/nga/mox/gel/yonp/dien/die/bai/gain/aid/jia/ind/uanl/pia/x no/iain/nl z/ox y/x yi/il y/zain/l ya/sei/ionl/han/zul/gaid/cux/ngv/ent/sit/onp/ait/ngua/cel/sain/l ng/dgai/il n/cai/xin/nl y/ox z/l la/sui/lai/t no/iai/gei/jil/qion/nal/baip/xiai/nl h/x ng/huan/aip/x ji/xit/jien/sai/x ya/t zi/indg/p no/l go/hanl/el n/ix y/ngai/nl d/ngo/nx n/ngot/got/lait/x xi/p zi/ienl/nge/il z/einl/nl l/l de/l xi/bei/l yi/l yo/qio/ye/xia/ox s/nl g/inp/jiai/onl/x la/zux/ndga/ngu ".strip().split("/")),
+  ("bfs", "ded/nid/aot/bel/hel/zex/zil/bei/no/cux/nil/sua/het/at/yef/bao/bal/gou/nge/erl/mel/x ni/dao/det/baot/bet/nel/nga/sei/zid/daot/jie/aox/mux/pia/de/maox/oux/svz/lei/erf/bif/erd/tel/hal/sex/erx/nao/ger/jia/gux/nivx/sal/dul/x de/eix/gerd/goux/eit/lao/el n/xia/eid/nal/bex/nger/niv/gel/laoz/huo/ber/ngel/f de/xif/eif/seit/uox/ga/you/leid/ngao/hhe/zou/mer/gaot/wei/zet/zao/ux n/t ni/ye/l no/gerf/ierl/f no/youd/f ni/gvf/mix/cer/l de/bier/zaop/x be/hhex/aod/zei/hao/l ye/daf/x he/il n/bia/der/aof/uix/eil/bie/zi/t ze/l zi/iax/x zi/duo/ngv/yao/ot z/huox/qie/qier/l be/fvf/lil/jier/nei/gei/if n/sit/gao/naot/herl/d be/el z/duox/weix/l ni/il d/t de/t zi/ax".strip().split("/")),
+  ("hea", "dai/ngx/gha/mong/ail/mon/ngl/dail/angb/ngb/aib/iang/ongl/ait/angx/lol/wil/ongx/angl/dax/non/nong/ngd/hab/nen/b da/aix/ngt/ghab/hot/ax/ghe/dol/lai/angd/nian/nia/dot/ngs/angt/hang/mai/il d/angs/laib/dio/enx/ngf/ghax/hax/bil/ib d/nang/gid/b gh/dia/hna/nai/jef/nenx/dab/dli/dad/jus/il n/x mo/ib/vang/l gh/aid/bib/ob/eix/l da/van/jox/vut/yan/yang/b di/xid/naib/nan/angf/ib h/xang/ongt/ban/bang/dan/dang/zai/daib/zaid/dei/sei/mang/dla/hnai/ghet/t gh/x da/dian/ongd/hli/x gh/ib g/het/iot/x di/d da/diot/gb d/gx d/jit/t da/xan/ab d/lul/ib n/il g/lei/seix/l ni/fang/fan/nix/det/bon/bong/ob d/gl d/ix d/ais/xit/it d/not/vob/it n/ghan/ax d/bub/jan/jang/lian/mais/des/b ni/bit/l na".strip().split("/")),
+  ("kmc", "ngc/lis/eec/nya/angc/enc/gue/guee/ueec/aoh/map/enl/ngl/ail/yac/mao/maoh/aeng/suh/bai/yao/yan/aol/bail/aoc/wee/nai/yaoc/naih/nyi/dah/aih/dos/yenc/ngs/eex/aox/xiv/nuv/weex/ngh/wox/nyac/buh/yuh/nye/anl/enp/ongc/l la/eis/lan/angs/xen/c li/yuv/c ny/nyen/xan/yang/ngx/eix/lang/ngp/lai/tou/dei/touk/nae/ngk/yen/sai/ouk/nup/lao/l ny/anc/xon/lail/mee/oux/xang/dao/ban/duc/nyao/yaoh/jan/man/mang/c ma/eel/aenl/il/xih/h ny/wan/lia/don/eep/mae/sen/ngv/eeng/ags/angl/xic/yanc/c ba/c la/jen/l na/l li/s ny/h ya/ben/jiu/angh/liu/son/c sa/engl/ens/san/c da/h ma/maen/x ma/gl l/yav/il n/jang/h li/h ba/lie/uih/angx/meec/yee/xuh/gua/daol/inc/jah/dong/dou/jil/c gu/c ja/jav".strip().split("/")),
+  ("lic", "ueng/bhae/gha/neix/hauu/kun/loms/nyue/meuu/dhat/duix/dhes/uuda/ruus/goms/hes/ngan/vaeu/yous/ras/hou/max/na/fa/hloo/da/engs/engx/aeng/eeng/uuen/oeng/zuu/uens/aeis/haen/ieng/aeus/luue/hlu/pas/dha/aens/dhae/bais/aeix/toen/ienx/hluu/bai/ongx/auus/uuek/haei/iens/gong/long/oeis/hla/nga/gae/dhie/zuen/hwan/dhi/ongs/dzue/hwou/wous/hwoe/enyx/hien/guu/lae/bhe/bho/hwa/kuu/gon/hle/hlen/leny/gie/huen/ghae/aenx/ieux/bhi/gwae/hwo/gho/noms/ueis/bha/ngae/gaen/kuue/hae/die/dhe/jie/aeux/coe/laeu/dho/ieus/bhu/heen/dza/hlo/jien/vue/uuem/waen/woek/dzie/loe/ghw/toe/bhie/muen/dhu/hieu/loei/uenx/ngxg/dzu/laen/gaei/qie/duis/tuue/zuug/haeu/iems/cuen/tae/ueix/mue/hloe/hang/nyae/qien/tuu/ensd/ghoe/hoei/zien/tun/dong".strip().split("/")),
 ]