_cyrillic_base.yml 4.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241
  1. general:
  2. name: Cyrillic base
  3. notes: copied from Ukrainian .cfg file.
  4. roman_to_script:
  5. ignore:
  6. - "At head of title"
  7. - "Colophon"
  8. - "II"
  9. - "III"
  10. - "IV"
  11. - "IX"
  12. - "LI"
  13. - "LII"
  14. - "LIII"
  15. - "LIV"
  16. - "LIX"
  17. - "LV"
  18. - "LVI"
  19. - "LVII"
  20. - "LVIII"
  21. - "LX"
  22. - "LXI"
  23. - "LXII"
  24. - "LXIII"
  25. - "LXIV"
  26. - "LXIX"
  27. - "LXV"
  28. - "LXVI"
  29. - "LXVII"
  30. - "LXVIII"
  31. - "LXX"
  32. - "LXXI"
  33. - "LXXII"
  34. - "LXXIII"
  35. - "LXXIV"
  36. - "LXXIX"
  37. - "LXXV"
  38. - "LXXVI"
  39. - "LXXVII"
  40. - "LXXVIII"
  41. - "LXXX"
  42. - "LXXXI"
  43. - "LXXXII"
  44. - "LXXXIII"
  45. - "LXXXIV"
  46. - "LXXXIX"
  47. - "LXXXV"
  48. - "LXXXVI"
  49. - "LXXXVII"
  50. - "LXXXVIII"
  51. - "VI"
  52. - "VII"
  53. - "VIII"
  54. - "XI"
  55. - "XII"
  56. - "XIII"
  57. - "XIV"
  58. - "XIX"
  59. - "XL"
  60. - "XLI"
  61. - "XLII"
  62. - "XLIII"
  63. - "XLIV"
  64. - "XLIX"
  65. - "XLV"
  66. - "XLVI"
  67. - "XLVII"
  68. - "XLVIII"
  69. - "XV"
  70. - "XVI"
  71. - "XVII"
  72. - "XVIII"
  73. - "XX"
  74. - "XXI"
  75. - "XXII"
  76. - "XXIII"
  77. - "XXIV"
  78. - "XXIX"
  79. - "XXV"
  80. - "XXVI"
  81. - "XXVII"
  82. - "XXVIII"
  83. - "XXX"
  84. - "XXXI"
  85. - "XXXII"
  86. - "XXXIII"
  87. - "XXXIV"
  88. - "XXXIX"
  89. - "XXXV"
  90. - "XXXVI"
  91. - "XXXVII"
  92. - "XXXVIII"
  93. - "and one other"
  94. - "and two others"
  95. - "and three others"
  96. - "and four others"
  97. - "and five others"
  98. - "and six others"
  99. - "and seven others"
  100. - "and eight others"
  101. - "and nine others"
  102. - "and ten others"
  103. - "and eleven others"
  104. - "and twelve others"
  105. - "and thirteen others"
  106. - "and fourteen others"
  107. - "and fifteen others"
  108. - "and sixteen others"
  109. - "and seventeen others"
  110. - "and eighteen others"
  111. - "and nineteen others"
  112. - "and others"
  113. - "et al."
  114. - "date of publication not identified"
  115. - "Place of publication not identified"
  116. - "publisher not identified"
  117. map:
  118. "A": "\u0410"
  119. "a": "\u0430"
  120. "B": "\u0411"
  121. "b": "\u0431"
  122. "V": "\u0412"
  123. "v": "\u0432"
  124. "D": "\u0414"
  125. "d": "\u0434"
  126. "E": "\u0415"
  127. "e": "\u0435"
  128. # this conversion shouldn't be needed, but does no harm
  129. "Z": "\u0417"
  130. "z": "\u0437"
  131. "I\u0306": "\u0419"
  132. # this conversion shouldn't be needed, but does no harm
  133. "I\uFE20U\uFE21": "\u042E"
  134. # this conversion shouldn't be needed, but does no harm
  135. "I\uFE20u\uFE21": "\u042E"
  136. "I\uFE20A\uFE21": "\u042F"
  137. # this conversion shouldn't be needed, but does no harm
  138. "I\uFE20a\uFE21": "\u042F"
  139. "i\u0306": "\u0439"
  140. "i\uFE20u\uFE21": "\u044E"
  141. "i\uFE20a\uFE21": "\u044F"
  142. # this conversion shouldn't be needed, but does no harm
  143. "KH": "\u0425"
  144. "Kh": "\u0425"
  145. "K": "\u041A"
  146. "kh": "\u0445"
  147. "k": "\u043A"
  148. "L": "\u041B"
  149. "l": "\u043B"
  150. "M": "\u041C"
  151. "m": "\u043C"
  152. "N": "\u041D"
  153. "n": "\u043D"
  154. "O": "\u041E"
  155. "o": "\u043E"
  156. "P": "\u041F"
  157. "p": "\u043F"
  158. "R": "\u0420"
  159. "r": "\u0440"
  160. # this conversion shouldn't be needed, but does no harm
  161. # this conversion shouldn't be needed, but does no harm
  162. "SH": "\u0428"
  163. "Sh": "\u0428"
  164. "S": "\u0421"
  165. "sh": "\u0448"
  166. "s": "\u0441"
  167. # this conversion shouldn't be needed, but does no harm
  168. "T": "\u0422"
  169. "t": "\u0442"
  170. "U": "\u0423"
  171. "u": "\u0443"
  172. "F": "\u0424"
  173. "f": "\u0444"
  174. # this conversion shouldn't be needed, but does no harm
  175. "CH": "\u0427"
  176. "Ch": "\u0427"
  177. "ch": "\u0447"
  178. # this conversion shouldn't be needed, but does no harm
  179. "\uFE20": ""
  180. # this conversion shouldn't be needed, but does no harm
  181. "\uFE21": ""
  182. # this conversion is ambiguous - \u042C is also theoretically possible
  183. "\u02B9": "\u044C"
  184. script_to_roman:
  185. map:
  186. "\u0404": "I\uFE20E\uFE21"
  187. "\u0407": "I\u0308"
  188. "\u0410": "A"
  189. "\u0411": "B"
  190. "\u0412": "V"
  191. "\u0414": "D"
  192. "\u0415": "E"
  193. "\u0417": "Z"
  194. "\u0419": "I\u0306"
  195. "\u041A": "K"
  196. "\u041B": "L"
  197. "\u041C": "M"
  198. "\u041D": "N"
  199. "\u041E": "O"
  200. "\u041F": "P"
  201. "\u0420": "R"
  202. "\u0421": "S"
  203. "\u0422": "T"
  204. "\u0423": "U"
  205. "\u0424": "F"
  206. "\u0425": "Kh"
  207. "\u0427": "Ch"
  208. "\u0428": "Sh"
  209. "\u0429": "Shch"
  210. "\u042C": "\u02B9"
  211. "\u042E": "I\uFE20U\uFE21"
  212. "\u042F": "I\uFE20A\uFE21"
  213. "\u0430": "a"
  214. "\u0431": "b"
  215. "\u0432": "v"
  216. "\u0433": "h"
  217. "\u0434": "d"
  218. "\u0435": "e"
  219. "\u0437": "z"
  220. "\u0439": "i\u0306"
  221. "\u043A": "k"
  222. "\u043B": "l"
  223. "\u043C": "m"
  224. "\u043D": "n"
  225. "\u043E": "o"
  226. "\u043F": "p"
  227. "\u0440": "r"
  228. "\u0441": "s"
  229. "\u0442": "t"
  230. "\u0443": "u"
  231. "\u0444": "f"
  232. "\u0445": "kh"
  233. "\u0447": "ch"
  234. "\u0448": "sh"
  235. "\u0449": "shch"
  236. "\u044C": "\u02B9"
  237. "\u044E": "i\uFE20u\uFE21"
  238. "\u044F": "i\uFE20a\uFE21"