bryce/language-data
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Sync language data base with jquery.uls language data

Browse Source 
This avoids the mismatch with jquery.uls database in preparation to
replace jquery.uls data with language-data module
This commit is contained in:
Santhosh Thottingal
2017-08-28 17:40:12 +05:30
parent 246f9925bd
commit bede5f48fe
3 changed files with 277 additions and 63 deletions
Download Patch File Download Diff File Expand all files Collapse all files

57
data/langdb.yaml
View File

@@ -12,10 +12,12 @@ languages:
aeb-latn: [Latn, [AF], Tûnsî]
af: [Latn, [AF], Afrikaans]
ahr: [Deva, [AS], अहिराणी]
ais: [Latn, [AS], Sakizaya]
ak: [Latn, [AF], Akan]
akz: [Latn, [AM], Albaamo innaaɬiilka]
aln: [Latn, [EU], Gegë]
am: [Ethi, [AF], አማርኛ]
ami: [Latn, [AS], Pangcah]
an: [Latn, [EU], aragonés]
ang: [Latn, [EU], Ænglisc]
anp: [Deva, [AS], अङ्गिका]
@@ -49,9 +51,8 @@ languages:
bar: [Latn, [EU], Boarisch]
bat-smg: [sgs]
bbc-latn: [Latn, [AS], Batak Toba]
# FIXME - autonym in the Batak script is unknown
bbc-batk: [Batk, [AS], Batak Toba]
bbc: [Latn, [AS], Batak Toba]
bbc-batk: [Batk, [AS], ᯅᯖᯂ᯲ ᯖᯬᯅ]
bbc: [bbc-latn]
bcc: [Arab, [AS, ME], جهلسری بلوچی]
bcl: [Latn, [AS], Bikol Central]
be-tarask: [Cyrl, [EU], беларуская (тарашкевіца)]
@@ -80,6 +81,7 @@ languages:
bto: [Latn, [AS], Iriga Bicolano]
bug: [Bugi, [AS], ᨅᨔ ᨕᨘᨁᨗ]
bxr: [Cyrl, [AS], буряад]
byn: [Ethi, [AF], ብሊን]
ca: [Latn, [EU], català]
cbk-zam: [Latn, [AS], Chavacano de Zamboanga]
cdo: [Latn, [AS], Mìng-dĕ̤ng-ngṳ̄]
@@ -100,9 +102,10 @@ languages:
cjy: [cjy-hant]
cjy-hans: [Hans, [AS], 晋语（简化字)]
cjy-hant: [Hant, [AS], 晉語]
ckb: [Arab, [ME], کوردیی ناوەندی]
ckb: [Arab, [ME], کوردی]
cnh: [Latn, [AS], Lai holh]
co: [Latn, [EU], corsu]
cop: [Copt, [AF, ME], ϯⲙⲉⲧⲣⲉⲙⲛ̀ⲭⲏⲙⲓ]
cps: [Latn, [AS], Capiceño]
cr: [Cans, [AM], ᓀᐦᐃᔭᐍᐏᐣ]
cr-cans: [cr]
@@ -117,12 +120,14 @@ languages:
cv: [Cyrl, [EU], Чӑвашла]
cy: [Latn, [EU], Cymraeg]
da: [Latn, [EU], dansk]
dag: [Latn, [AF], dagbanli]
de-at: [Latn, [EU], Österreichisches Deutsch]
de-ch: [Latn, [EU], Schweizer Hochdeutsch]
de-formal: [Latn, [EU], Deutsch (Sie-Form)]
de: [Latn, [EU], Deutsch]
din: [Latn, [AF], Thuɔŋjäŋ]
diq: [Latn, [EU, AS], Zazaki]
doi: [Deva, [AS], डोगरी]
dsb: [Latn, [EU], dolnoserbski]
dtp: [Latn, [AS], Dusun Bundu-liwan]
dty: [Deva, [AS], डोटेली]
@@ -141,6 +146,7 @@ languages:
es-formal: [Latn, [EU, AM, AF, WW], español (formal)]
# world?
es: [Latn, [EU, AM, AF, WW], español]
es-ni: [Latn, [AM], español nicaragüense]
esu: [Latn, [AM], "Yup'ik"]
et: [Latn, [EU], eesti]
eu: [Latn, [EU], euskara]
@@ -161,6 +167,7 @@ languages:
fur: [Latn, [EU], furlan]
fy: [Latn, [EU], Frysk]
ga: [Latn, [EU], Gaeilge]
gaa: [Latn, [AF], Ga]
gag: [Latn, [EU], Gagauz]
gah: [Latn, [AS], Alekano]
gan-hans: [Hans, [AS], 赣语（简体）]
@@ -173,13 +180,14 @@ languages:
glk: [Arab, [AS, ME], گیلکی]
gn: [Latn, [AM], "Avañe'ẽ"]
gom: [gom-deva]
gom-deva: [Deva, [AS], गोवा कोंकणी]
gom-latn: [Latn, [AS], Gova Konknni]
gom-deva: [Deva, [AS], गोंयची कोंकणी]
gom-latn: [Latn, [AS], Gõychi Konknni]
got: [Goth, [EU], 𐌲𐌿𐍄𐌹𐍃𐌺]
grc: [Grek, [EU], Ἀρχαία ἑλληνικὴ]
gsw: [Latn, [EU], Alemannisch]
gu: [Gujr, [AS], ગુજરાતી]
guc: [Latn, [AM], Wayúu]
gum: [Latn, [AM], Namtrik]
gur: [Latn, [AF], Gurenɛ]
gv: [Latn, [EU], Gaelg]
# CLDR uses ha-latn and ha-arab. Latin is more common and is used in Wikipedia.
@@ -189,6 +197,8 @@ languages:
hak: [Latn, [AS], Hak-kâ-fa]
haw: [Latn, [AM, PA], Hawai`i]
he: [Hebr, [ME], עברית]
hak-hans: [Hans, [AS], 客家语（简体）]
hak-hant: [Hant, [AS], 客家語（繁體）]
hi: [Deva, [AS], हिन्दी]
hif: [Latn, [PA, AS], Fiji Hindi]
hif-deva: [Deva, [AS], फ़ीजी हिन्दी]
@@ -196,6 +206,7 @@ languages:
hil: [Latn, [AS], Ilonggo]
hne: [Deva, [AS], छत्तीसगढ़ी]
ho: [Latn, [PA], Hiri Motu]
hoc: [Wara, [AS], 𑢹𑣉𑣉]
hr: [Latn, [EU], hrvatski]
hrx: [Latn, [AM], Hunsrik]
hsb: [Latn, [EU], hornjoserbsce]
@@ -218,8 +229,7 @@ languages:
io: [Latn, [WW], Ido]
is: [Latn, [EU], íslenska]
it: [Latn, [EU], italiano]
# For variants ike-* is used
iu: [Cans, [AM], ᐃᓄᒃᑎᑐᑦ]
iu: [ike-cans]
ja: [Jpan, [AS], 日本語]
jam: [Latn, [AM], Patois]
jbo: [Latn, [WW], lojban]
@@ -269,7 +279,7 @@ languages:
krl: [Latn, [EU], Karjala]
ks-arab: [Arab, [AS], کٲشُر]
ks-deva: [Deva, [AS], कॉशुर]
ks: [Arab, [AS], کٲشُر]
ks: [ks-arab]
ksf: [Latn, [AF], Bafia]
ksh: [Latn, [EU], Ripoarisch]
# CLDR uses ku-latn and ku-arab. Latin is more common and is used in Wikipedia.
@@ -283,6 +293,7 @@ languages:
lad: [lad-latn]
lad-latn: [Latn, [ME, EU, AM], Ladino]
lad-hebr: [Hebr, [ME, EU, AM], לאדינו]
lag: [Latn, [AF], Kilaangi]
lb: [Latn, [EU], Lëtzebuergesch]
lbe: [Cyrl, [EU], лакку]
lez: [Cyrl, [EU], лезги]
@@ -291,7 +302,7 @@ languages:
li: [Latn, [EU], Limburgs]
lij: [Latn, [EU], Ligure]
liv: [Latn, [EU], Līvõ kēļ]
lki: [Arab, [AS, ME], لەکی‎]
lki: [Arab, [AS, ME], لەکی]
lkt: [Latn, [AM], Lakȟótiyapi]
lld: [Latn, [EU], Ladin]
lmo: [Latn, [EU], lumbaart]
@@ -358,19 +369,22 @@ languages:
nn: [Latn, [EU], norsk (nynorsk)]
# There's also nb for Bokmål and nn for Nynorsk
"no": [Latn, [EU], norsk]
nod: [Thai, [AS], คำเมือง]
nov: [Latn, [WW], Novial]
nqo: [Nkoo, [AF], ߒߞߏ]
nrm: [Latn, [EU], Nouormand]
nso: [Latn, [AF], Sesotho sa Leboa]
nv: [Latn, [AM], Diné bizaad]
ny: [Latn, [AF], Chi-Chewa]
nys: [Latn, [PA], Nyungar]
oc: [Latn, [EU], occitan]
olo: [Latn, [AS, EU], Livvin-Karjalan]
olo: [Latn, [AS, EU], Livvinkarjala]
om: [Latn, [AF], Oromoo]
or: [Orya, [AS], ଓଡ଼ିଆ]
os: [Cyrl, [EU], Ирон]
# Bug: 60815
ota: [Arab, [AS, EU], لسان عثمانى]
ovd: [Latn, [EU], övdalsk]
pa: [pa-guru]
pa-guru: [Guru, [AS], ਪੰਜਾਬੀ]
pag: [Latn, [AS], Pangasinan]
@@ -448,11 +462,14 @@ languages:
shn: [Mymr, [AS], လိၵ်ႈတႆး]
si: [Sinh, [AS], සිංහල]
simple: [Latn, [WW], Simple English]
sjo: [Mong, [AS], ᠰᡞᠪᡝ ᡤᡞᠰᡠᠨ]
sk: [Latn, [EU], slovenčina]
sl: [Latn, [EU], slovenščina]
sli: [Latn, [EU], Schläsch]
slr: [Latn, [AS], Salırça]
sly: [Latn, [AS], Bahasa Selayar]
skr-arab: [Arab, [AS], سرائیکی]
skr: [skr-arab]
syc: [Syrc, [ME], ܣܘܪܝܝܐ]
sm: [Latn, [PA], Gagana Samoa]
sma: [Latn, [EU], åarjelsaemien]
@@ -472,6 +489,7 @@ languages:
ss: [Latn, [AF], SiSwati]
st: [Latn, [AF], Sesotho]
stq: [Latn, [EU], Seeltersk]
sty: [Cyrl, [EU, AS], себертатар]
su: [Latn, [AS], Basa Sunda]
sv: [Latn, [EU], svenska]
sw: [Latn, [AF], Kiswahili]
@@ -479,19 +497,22 @@ languages:
sxu: [Latn, [EU], Säggssch]
szl: [Latn, [EU], ślůnski]
ta: [Taml, [AS], தமிழ்]
tay: [Latn, [AS], Tayal]
tcy: [Knda, [AS], ತುಳು]
te: [Telu, [AS], తెలుగు]
tet: [Latn, [AS, PA], tetun]
tg-cyrl: [Cyrl, [AS], тоҷикӣ]
tg-latn: [Latn, [AS], tojikī]
tg: [Cyrl, [AS], тоҷикӣ]
tg: [tg-cyrl]
th: [Thai, [AS], ไทย]
ti: [Ethi, [AF], ትግርኛ]
tig: [Ethi, [AF], ትግረ]
tk: [Latn, [AS], Türkmençe]
tkr: [Cyrl, [AS], ЦӀаьхна миз]
tkr: [Cyrl, [AS], ЦӀаӀхна миз]
tl: [Latn, [AS], Tagalog]
# A very complicated case. Names.php is Cyrl. In TWN they argue about Cyrl, Latn, and Arab. I can't find reliable external sources. --Amir
tly: [Cyrl, [EU, AS, ME], толышә зывон]
tmr: [Hebr, [ME, EU, AM], ארמית בבלית]
tn: [Latn, [AF], Setswana]
to: [Latn, [PA], lea faka-Tonga]
tokipona: [Latn, [WW], Toki Pona]
@@ -499,6 +520,7 @@ languages:
tr: [Latn, [EU, ME], Türkçe]
trp: [Latn, [AS], Kokborok (Tripuri)]
tru: [Latn, [AS], Ṫuroyo]
trv: [Latn, [AS], Sediq Taroko]
ts: [Latn, [AF], Xitsonga]
tsd: [Grek, [EU], Τσακωνικά]
tt: [Cyrl, [EU], татарча]
@@ -539,6 +561,7 @@ languages:
xal: [Cyrl, [EU], хальмг]
xh: [Latn, [AF], isiXhosa]
xmf: [Geor, [EU], მარგალური]
xsy: [Latn, [AS], SaiSiyat]
ydd: [Hebr, [AS, EU], Eastern Yiddish]
yi: [Hebr, [ME, EU, AM], ייִדיש]
yo: [Latn, [AF], Yorùbá]
@@ -548,10 +571,10 @@ languages:
yue: [Hant, [AS], 粵語]
za: [Latn, [AS], Vahcuengh]
zea: [Latn, [EU], Zeêuws]
zgh: [Tfng, [AF], ⵜⴰⵎⴰⵣⵉⵖⵜ]
zgh: [Tfng, [AF], ⵜⴰⵎⴰⵣⵉⵖⵜ ⵜⴰⵏⴰⵡⴰⵢⵜ]
# world? (may apply to many varieties of Chinese)
zh: [Hans, [AS], 中文]
zh-classical: [Hant, [AS], 文言]
zh-classical: [lzh]
zh-cn: [Hans, [AS], 中文（中国大陆）]
zh-hans: [Hans, [AS], 中文（简体）]
zh-hant: [Hant, [AS], 中文（繁體）]
@@ -588,7 +611,7 @@ scriptgroups:
# Maybe Thaana can be moved here from SouthAsian.
# Maybe it can be unified with African.
MiddleEastern: [Hebr, Syrc]
African: [Ethi, Nkoo, Tfng]
African: [Ethi, Nkoo, Tfng, Copt]
# India, Nepal, Bangladesh, Sri-Lanka, Bhutan, Maldives.
#
# Thaana (Thaa, the script of Maldives) is here, even though it's RTL,
@@ -598,7 +621,7 @@ scriptgroups:
# Tibetan (Tibt) is here, even though it's classified as "Central Asian" by
# Unicode, because linguistically and geographically it's closely related to
# the Brahmic family.
SouthAsian: [Beng, Deva, Gujr, Guru, Knda, Mlym, Orya, Saur, Sinh, Taml, Telu, Tibt, Thaa]
SouthAsian: [Beng, Deva, Gujr, Guru, Knda, Mlym, Orya, Saur, Sinh, Taml, Telu, Tibt, Thaa, Wara]
Cyrillic: [Cyrl]
CJK: [Hans, Hant, Kana, Kore, Jpan, Yiii]
SouthEastAsian: [Bali, Batk, Bugi, Java, Khmr, Laoo, Mymr, Thai]