AzerbaijaniRomanization.cfg 5.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285
  1. # Original table by Randall K. Barry, 11 March 2021
  2. [General]
  3. Name=Azerbaijani
  4. # UNICODE CYRILLIC CHARACTER RANGE: U+0400-U+052F
  5. # This configuration file maps all valid Unicode used in the Azerbaijani
  6. # and Russian languages written in the Cyrillic script
  7. [RomanToScript]
  8. FieldsIncluded=100 110 111 130 240 245 246 250 260 264 440 490 600 610 611 630 651 700
  9. 710 711 730 740 800 830
  10. SubfieldsAlwaysExcluded=uvxy0123456789
  11. OtherSubfieldsExcludedByTag=100/e 110/e 111/j 246/i 260/c 264/c 650/a 700/e 700/i 710/e
  12. 710/i 711/i 711/j 730/i
  13. # RDA boilerplate phrases not transliterated:
  14. Place of publication not identified=Place of publication not identified
  15. publisher not identified=publisher not identified
  16. date of publication not identified=date of publication not identified
  17. At head of title=At head of title
  18. Colophon=Colophon
  19. III=III
  20. IV=IV
  21. VII=VII
  22. VIII=VIII
  23. IX=IX
  24. X=X
  25. XI=XI
  26. XII=XII
  27. XIV=XIV
  28. XV=XV
  29. XVI=XVI
  30. XVII=XVII
  31. XX=XX
  32. XXI=XXI
  33. XXII=XXII
  34. XXIV=XXIV
  35. XXV=XXV
  36. XXVI=XXVI
  37. XXVII=XXVII
  38. XXIX=XXIX
  39. XXX=XXX
  40. XXXI=XXXI
  41. XXXII=XXXII
  42. XXXIV=XXXIV
  43. XXXV=XXXV
  44. XXXVI=XXXVI
  45. XXXVII=XXXVII
  46. XL=XL
  47. LX=LX
  48. LXX=LXX
  49. LXXX=LXXX
  50. # COMMON COMBINING CHARACTERS (always follow a base letter):
  51. # combining grave U+0300
  52. # combining acute U+0301
  53. # combining circumflex U+0302
  54. # combining macron U+0304
  55. # combining breve U+0306
  56. # combining dot above U+0307
  57. # combining diaeresis U+0308
  58. # combining caron (hachek) U+030C
  59. # combining candrabindu U+0310
  60. # combining dot below U+0323
  61. # combining comma below U+0326 (Romanian, Latvian, Livonian)
  62. # combining cedilla U+0327 (French, Turkish, Azeri)
  63. # combining ogonek (hook) U+0328 (Polish, Lithuanian)
  64. # combining left ligature U+FE20 (Cyrillic transliteration)
  65. # combining right ligature U+FE21 (Cyrillic transliteration)
  66. # soft sign/prime (spacing) U+02B9(Cyrillic transliteration)
  67. # hard sign/double prime (spacing) U+02BA (Cyrillic transliteration)
  68. AU+0306=U+04D8
  69. aU+0306=U+04D9
  70. A=U+0410
  71. a=U+0430
  72. B=U+0411
  73. b=U+0431
  74. V=U+0412
  75. v=U+0432
  76. Gh=U+0492
  77. gh=U+0493
  78. GU+0307=U+049C
  79. gU+0307=U+049D
  80. G=U+0413
  81. g=U+0433
  82. D=U+0414
  83. d=U+0434
  84. EU+0307=U+042D
  85. EU+0308=U+0401
  86. EU+0328=U+0466
  87. E=U+0415
  88. eU+0307=U+044D
  89. eU+0308=U+0451
  90. eU+0328=U+0467
  91. e=U+0435
  92. ZH=U+0416
  93. Zh=U+0416
  94. Z=U+0417
  95. zh=U+0436
  96. z=U+0437
  97. IU+0304=U+0406
  98. IU+0306=U+0419
  99. IU+0310=U+0408
  100. IU+FE20EU+FE21=U+0462
  101. IU+FE20eU+FE21=U+0462
  102. IU+FE20UU+FE21=U+042E
  103. IU+FE20uU+FE21=U+042E
  104. IU+FE20AU+FE21=U+042F
  105. IU+FE20aU+FE21=U+042F
  106. I=U+0418
  107. iU+0304=U+0456
  108. iU+0306=U+0439
  109. iU+0310=U+0458
  110. iU+FE20eU+FE21=U+0463
  111. iU+FE20uU+FE21=U+044E
  112. iU+FE20aU+FE21=U+044F
  113. i=U+0438
  114. J=U+04B8
  115. j=U+04B9
  116. KH=U+0425
  117. Kh=U+0425
  118. K=U+041A
  119. kh=U+0445
  120. k=U+043A
  121. L=U+041B
  122. l=U+043B
  123. M=U+041C
  124. m=U+043C
  125. N=U+041D
  126. n=U+043D
  127. OU+0307=U+04E8
  128. O=U+041E
  129. oU+0307=U+04E9
  130. o=U+043E
  131. P=U+041F
  132. p=U+043F
  133. R=U+0420
  134. r=U+0440
  135. SHCH=U+0429
  136. Shch=U+0429
  137. SH=U+0428
  138. Sh=U+0428
  139. S=U+0421
  140. shch=U+0449
  141. sh=U+0448
  142. s=U+0441
  143. TU+FE20SU+FE21=U+0426
  144. TU+FE20sU+FE21=U+0426
  145. T=U+0422
  146. tU+FE20sU+FE21=U+0446
  147. t=U+0442
  148. UU+0307=U+04AE
  149. U=U+0423
  150. uU+0307=U+04AF
  151. u=U+0443
  152. FU+0307=U+0472
  153. F=U+0424
  154. fU+0307=U+0473
  155. f=U+0444
  156. CH=U+0427
  157. Ch=U+0427
  158. ch=U+0447
  159. HU+0307=U+04BA
  160. hU+0307=U+04BB
  161. YU+0307=U+0474
  162. Y=U+042B
  163. yU+0307=U+0475
  164. y=U+044B
  165. # this conversion shouldn't be needed, but does no harm
  166. U+FE20=
  167. # this conversion shouldn't be needed, but does no harm
  168. U+FE21=
  169. # this conversion is ambiguous - U+042C is also theoretically possible
  170. U+0027=U+044C
  171. # this conversion is ambiguous - U+042C is also theoretically possible
  172. U+02B9=U+044C
  173. # this conversion is ambiguous - U+044C is also theoretically possible
  174. U+02BA=U+044A
  175. [ScriptToRoman]
  176. FieldsIncluded=100 110 111 130 240 245 246 250 260 264 440 490 600 610 611 630 651 700 710 711 730 740 800 830
  177. SubfieldsAlwaysExcluded=uvxy0123456789
  178. OtherSubfieldsExcludedByTag=100/e 110/e 111/j 246/i 260/c 264/c 650/a 700/e 700/i 710/e 710/i 711/i 711/j 730/i
  179. # RDA boilerplate phrases not transliterated:
  180. Place of publication not identified=Place of publication not identified
  181. publisher not identified=publisher not identified
  182. U+0401=EU+0308
  183. U+0406=IU+0304
  184. U+0408=IU+0310
  185. U+0410=A
  186. U+0411=B
  187. U+0412=V
  188. U+0413=G
  189. U+0414=D
  190. U+0415=E
  191. U+0416=Zh
  192. U+0417=Z
  193. U+0418=I
  194. U+0419=IU+0306
  195. U+041A=K
  196. U+041B=L
  197. U+041C=M
  198. U+041D=N
  199. U+041E=O
  200. U+041F=P
  201. U+0420=R
  202. U+0421=S
  203. U+0422=T
  204. U+0423=U
  205. U+0424=F
  206. U+0425=Kh
  207. U+0426=TU+FE20SU+FE21
  208. U+0427=Ch
  209. U+0428=Sh
  210. U+0429=Shch
  211. U+042A=U+02BA
  212. U+042B=Y
  213. U+042C=U+02B9
  214. U+042D=EU+0307
  215. U+042E=IU+FE20UU+FE21
  216. U+042F=IU+FE20AU+FE21
  217. U+0430=a
  218. U+0431=b
  219. U+0432=v
  220. U+0433=g
  221. U+0434=d
  222. U+0435=e
  223. U+0436=zh
  224. U+0437=z
  225. U+0438=i
  226. U+0439=iU+0306
  227. U+043A=k
  228. U+043B=l
  229. U+043C=m
  230. U+043D=n
  231. U+043E=o
  232. U+043F=p
  233. U+0440=r
  234. U+0441=s
  235. U+0442=t
  236. U+0443=u
  237. U+0444=f
  238. U+0445=kh
  239. U+0446=tU+FE20sU+FE21
  240. U+0447=ch
  241. U+0448=sh
  242. U+0449=shch
  243. U+044A=U+02BA
  244. U+044B=y
  245. U+044C=U+02B9
  246. U+044D=eU+0307
  247. U+044E=iU+FE20uU+FE21
  248. U+044F=iU+FE20aU+FE21
  249. U+0451=eU+0308
  250. U+0456=iU+0304
  251. U+0458=iU+0310
  252. U+0462=IU+FE20EU+FE21
  253. U+0463=iU+FE20eU+FE21
  254. U+0466=EU+0328
  255. U+0467=eU+0328
  256. U+0472=FU+0307
  257. U+0473=fU+0307
  258. U+0474=YU+0307
  259. U+0475=yU+0307
  260. U+0492=Gh
  261. U+0493=gh
  262. U+049C=GU+0307
  263. U+049D=gU+0307
  264. U+04AE=UU+0307
  265. U+04AF=uU+0307
  266. U+04B8=J
  267. U+04B9=j
  268. U+04BA=HU+0307
  269. U+04BB=hU+0307
  270. U+04D8=AU+0306
  271. U+04D9=aU+0306
  272. U+04E8=OU+0307
  273. U+04E9=oU+0307
  274.