Add languages used in Wikimedia MinT

Autonyms from Ethnologue, except Tamasheq (taq), whose Tifinagh spelling is taken from the Russian Wikipedia.
2023-05-09 18:23:59 +03:00
parent d8b9138988
commit 4198119b5c
2 changed files with 197 additions and 9 deletions
--- a/data/langdb.yaml
+++ b/data/langdb.yaml
@@ -7,6 +7,7 @@ languages:
  ace: [Latn, [AS, PA], Acèh]
  acf: [Latn, [AM], kwéyòl]
  acm: [Arab, [ME], عراقي]
+  acq: [Arab, [ME, AF], تعزية-عدنية]
  ady: [Cyrl, [EU, ME], адыгабзэ]
  ady-cyrl: [ady]
  ady-latn: [Latn, [EU, ME], Adygabze]
@@ -28,6 +29,7 @@ languages:
  ale-cyrl: [Cyrl, [AS], унаӈам тунуу]
  aln: [Latn, [EU], Gegë]
  alt: [Cyrl, [EU, AS], алтай тил]
+  als: [sq]
  am: [Ethi, [AF], አማርኛ]
  ami: [Latn, [AS], Pangcah]
  an: [Latn, [EU], aragonés]
@@ -84,6 +86,7 @@ languages:
  be-tarask: [Cyrl, [EU], беларуская (тарашкевіца)]
  be-x-old: [be-tarask]
  be: [Cyrl, [EU], беларуская]
+  bem: [Latn, [AF], IciBemba]
  bew: [Latn, [AS], Bahasa Betawi]
  bfa: [Latn, [AF], Bari]
  bft: [Arab, [AS], بلتی]
@@ -147,6 +150,7 @@ languages:
  cjy-hant: [Hant, [AS], 晉語]
  ckb: [Arab, [ME], کوردی]
  ckt: [Cyrl, [AS], ԓыгъоравэтԓьэн]
+  cjk: [Latn, [AF], cokwe]
  cnh: [Latn, [AS], Lai holh]
  cnr: [cnr-latn]
  cnr-cyrl: [Cyrl, [EU], црногорски]
@@ -176,6 +180,10 @@ languages:
  de-formal: [Latn, [EU], Deutsch (Sie-Form)]
  de: [Latn, [EU], Deutsch]
  dga: [Latn, [AF], Dagaare]
+ # In ISO 639, it refers to Rek Dinka. It may
+ # be changed like that here, too, but for
+ # now, redirect it to macro Dinka.
+  dik: [din]
  din: [Latn, [AF], Thuɔŋjäŋ]
  diq: [Latn, [EU, AS], Zazaki]
  doi: [Deva, [AS], डोगरी]
@@ -183,6 +191,7 @@ languages:
  dtp: [Latn, [AS], Dusun Bundu-liwan]
  dty: [Deva, [AS], डोटेली]
  dv: [Thaa, [AS], ދިވެހިބަސް]
+  dyu: [Latn, [AF], Julakan]
  dz: [Tibt, [AS], ཇོང་ཁ]
  ee: [Latn, [AF], eʋegbe]
  efi: [Latn, [AF], efịk]
@@ -224,6 +233,10 @@ languages:
  frr: [Latn, [EU], Nordfriisk]
  fuf: [Latn, [AF], Fuuta Jalon]
  fur: [Latn, [EU], furlan]
+ # In ISO 639, it refers to Nigerian Fula. It may
+ # be changed like that here, too, but for
+ # now, redirect it to macro Fula.
+  fuv: [ff]
  fvr: [Latn, [AF], poor’íŋ belé’ŋ]
  fy: [Latn, [EU], Frysk]
  ga: [Latn, [EU], Gaeilge]
@@ -233,6 +246,10 @@ languages:
  gan-hans: [Hans, [AS], 赣语（简体）]
  gan-hant: [gan]
  gan: [Hant, [AS], 贛語]
+ # In ISO 639, it refers to West Central Oromo. It may
+ # be changed like that here, too, but for
+ # now, redirect it to macro Oromo.
+  gaz: [om]
  gbm: [Deva, [AS], गढ़वळि]
  gbz: [Latn, [AS], Dari-e Mazdeyasnā]
  gcf: [Latn, [AM], Guadeloupean Creole French]
@@ -320,6 +337,7 @@ languages:
 # Can also be Tfng, but the Wikipedia is mostly Latn
  kab: [Latn, [AF, EU], Taqbaylit]
  kac: [Latn, [AS], Jinghpaw]
+  kam: [Latn, [AF], kĩkamba]
  kbd-cyrl: [kbd]
  kbd-latn: [Latn, [EU], Qabardjajəbza]
  kbd: [Cyrl, [EU, ME], адыгэбзэ]
@@ -330,6 +348,7 @@ languages:
  ken: [Latn, [AF], kɛ́nyáŋ]
  kg: [Latn, [AF], Kongo]
  kgp: [Latn, [AM], Kaingáng]
+  khk: [mn]
  khw: [Arab, [ME, AS], کھوار]
  ki: [Latn, [AF], Gĩkũyũ]
  kiu: [Latn, [EU, ME], Kırmancki]
@@ -345,7 +364,13 @@ languages:
  kk-tr: [kk-latn]
  kl: [Latn, [AM, EU], kalaallisut]
  km: [Khmr, [AS], ភាសាខ្មែរ]
+  kmb: [Latn, [AF], kimbundu]
+  kmr: [ku-latn]
  kn: [Knda, [AS], ಕನ್ನಡ]
+ # In ISO 639, it refers to Central Kanuri. It may
+ # be changed like that here, too, but for
+ # now, redirect it to macro Kanuri.
+  knc: [kr]
  knn: [Deva, [AS], महाराष्ट्रीय कोंकणी]
  ko-kp: [Kore, [AS], 조선말]
  ko: [Kore, [AS], 한국어]
@@ -396,11 +421,14 @@ languages:
  lt: [Latn, [EU], lietuvių]
  lrc: [Arab, [AS, ME], لۊری شومالی]
  ltg: [Latn, [EU], latgaļu]
+  lua: [Latn, [AF], ciluba]
  lud: [Latn, [EU], lüüdi]
+  luo: [Latn, [AF], dholuo]
  lus: [Latn, [AS], Mizo ţawng]
  lut: [Latn, [AM], dxʷləšucid]
  luz: [Arab, [ME], لئری دوٙمینی]
  lv: [Latn, [EU], latviešu]
+  lvs: [lv]
  lzh: [Hant, [AS], 文言]
 # Also Geor, but the incubator is in Latn
  lzz: [Latn, [EU, ME], Lazuri]
@@ -476,6 +504,7 @@ languages:
  nod-thai: [Thai, [AS], คำเมือง]
  nog: [Cyrl, [EU], ногайша]
  nov: [Latn, [WW], Novial]
+  npi: [ne]
  nqo: [Nkoo, [AF], ߒߞߏ]
  nr: [Latn, [AF], isiNdebele seSewula]
  nrf-gg: [Latn, [EU], Guernésiais]
@@ -495,6 +524,7 @@ languages:
  om: [Latn, [AF], Oromoo]
  ood: [Latn, [AM], "ʼOʼodham ha-ñeʼokĭ"]
  or: [Orya, [AS], ଓଡ଼ିଆ]
+  ory: [or]
  os: [Cyrl, [EU], ирон]
  osi: [Latn, [AS], Using]
  ota: [Arab, [AS, EU], لسان عثمانى]
@@ -506,10 +536,15 @@ languages:
  pap: [Latn, [AM], Papiamentu]
  pap-aw: [Latn, [AM], Papiamento]
  pbb: [Latn, [AM], Nasa Yuwe]
+ # In ISO 639, it refers to Southern Pashto. It may
+ # be changed like that here, too, but for
+ # now, redirect it to macro Pashto.
+  pbt: [ps]
  pcd: [Latn, [EU], Picard]
  pcm: [Latn, [AF], Naijá]
  pdc: [Latn, [EU, AM], Deitsch]
  pdt: [Latn, [EU, AM], Plautdietsch]
+  pes: [fa]
  pfl: [Latn, [EU], Pälzisch]
  pi: [Deva, [AS], पालि]
  pih: [Latn, [PA], Norfuk / Pitkern]
@@ -517,6 +552,7 @@ languages:
  pjt: [Latn, [PA], Pitjantjatjara]
  pko: [Latn, [AF], Pökoot]
  pl: [Latn, [EU], polski]
+  plt: [mg]
  pms: [Latn, [EU], Piemontèis]
  pnb: [Arab, [AS, ME], پنجابی]
  pnt: [Grek, [EU], Ποντιακά]
@@ -531,6 +567,10 @@ languages:
  qu: [Latn, [AM], Runa Simi]
  quc: [Latn, [AM], "K'iche'"]
  qug: [Latn, [AM], Runa shimi]
+ # In ISO 639, it refers to Chanka Quechua. It may
+ # be changed like that here, too, but for
+ # now, redirect it to macro Quechua.
+  quy: [qu]
  qwh: [Latn, [AM], anqash qichwa]
  rag: [Latn, [AF], Lologooli]
  rap: [Latn, [PA, AM], arero rapa nui]
@@ -636,11 +676,17 @@ languages:
  su: [Latn, [AS], Sunda]
  sv: [Latn, [EU], svenska]
  sw: [Latn, [AF], Kiswahili]
+  swh: [sw]
  swb: [Latn, [AF], Shikomoro]
  sxu: [Latn, [EU], Säggssch]
  szl: [Latn, [EU], ślůnski]
  szy: [Latn, [AS], Sakizaya]
  ta: [Taml, [AS], தமிழ்]
+ # This is initially added for Wikimiedia MinT, which outputs Tifinagh.
+ # Perhaps some day it can be changed to Latin as default.
+  taq-latn: [Latn, [AF], təmajəq]
+  taq-tfng: [Latn, [AF], ⵜⴰⵎⴰⵌⴰⵆ]
+  taq: [taq-tfng]
  tay: [Latn, [AS], Tayal]
  tcy: [Knda, [AS], ತುಳು]
  tdd: [Tale, [AS], ᥖᥭᥰᥖᥬᥳᥑᥨᥒᥰ]
@@ -689,11 +735,13 @@ languages:
  ug-latn: [Latn, [AS], uyghurche]
  ug-cyrl: [Cyrl, [AS], уйғурчә]
  uk: [Cyrl, [EU], українська]
+  umb: [Latn, [AF], umbundu]
  umu: [Latn, [AM], Huluníixsuwaakan]
  ur: [Arab, [AS, ME], اردو]
-  uz: [Latn, [AS], oʻzbekcha]
  uz-cyrl: [Cyrl, [AS], ўзбекча]
  uz-latn: [uz]
+  uz: [Latn, [AS], oʻzbekcha]
+  uzn: [uz]
  ve: [Latn, [AF], Tshivenda]
  vai: [Vaii, [AF], ꕙꔤ]
  vec: [Latn, [EU, AM], vèneto]
@@ -741,6 +789,7 @@ languages:
  zh-tw: [Hant, [AS], 中文（台灣）]
  zh-yue: [yue]
  zh-cdo: [cdo]
+  zsm: [ms]
  zu: [Latn, [AF], isiZulu]
  zun: [Latn, [AM], "Shiwi'ma"]