persian.yml 7.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311
  1. ---
  2. general:
  3. name: Persian
  4. case_sensitive: false
  5. roman_to_script:
  6. map:
  7. # Punctuation marks:
  8. "# %": "\u066A ; cannot transliterate the truncation character"
  9. "*": "\u066D"
  10. ",": "\u060C"
  11. ";": "\u061B"
  12. "?": "\u061F"
  13. # Exceptions for specific words
  14. # Allah
  15. "Alla\u0304h": "\u0627\u0644\u0644\u0647"
  16. # Qur'an
  17. "Qur\u02BCa\u0304n": "\u0642\u0631\u0622\u0646"
  18. # lillah
  19. "lilla\u0304h": "\u0644\u0644\u0647"
  20. # billah
  21. "billa\u0304h": "\u0628\u0644\u0644\u0647"
  22. # Rahman
  23. "Rah\u0323ma\u0304n": "\u0631\u062D\u0645\u0646"
  24. # ta'lif
  25. # Ibn when it appears in the middle of a name sequence
  26. "ibn": "\u0628\u0646"
  27. # Parsing "sh[dot below] as in "Ishaq [name]"
  28. "%sh\u0323%": "\u0633\u062D"
  29. # Edition statements with Latin number
  30. "Cha\u0304p-i 1": "\u0686\u0627\u067E 1"
  31. "Cha\u0304p-i 2": "\u0686\u0627\u067E 2"
  32. "Cha\u0304p-i 3": "\u0686\u0627\u067E 3"
  33. "Cha\u0304p-i 4": "\u0686\u0627\u067E 4"
  34. "Cha\u0304p-i 5": "\u0686\u0627\u067E 5"
  35. "Cha\u0304p-i 6": "\u0686\u0627\u067E 6"
  36. "Cha\u0304p-i 7": "\u0686\u0627\u067E 7"
  37. "Cha\u0304p-i 8": "\u0686\u0627\u067E 8"
  38. "Cha\u0304p-i 9": "\u0686\u0627\u067E 9"
  39. # Numbers (Extended Arabic \u06F0-06F9 for Persian)
  40. # currently *not* valid MARC21 characters
  41. "# 0": "\u06F0"
  42. "# 1": "\u06F1"
  43. "# 2": "\u06F2"
  44. "# 3": "\u06F3"
  45. "# 4": "\u06F4"
  46. "# 5": "\u06F5"
  47. "# 6": "\u06F6"
  48. "# 7": "\u06F7"
  49. "# 8": "\u06F8"
  50. "# 9": "\u06F9"
  51. # Use Basic Arabic \u0660-0669, instead:
  52. "0": "\u0660"
  53. "1": "\u0661"
  54. "2": "\u0662"
  55. "3": "\u0663"
  56. "4": "\u0664"
  57. "5": "\u0665"
  58. "6": "\u0666"
  59. "7": "\u0667"
  60. "8": "\u0668"
  61. "9": "\u0669"
  62. "# prime ": " ZWNJ"
  63. "\u02B9a\u0304": "\u200C\u0622"
  64. "\u02B9a": "\u200C\u0627"
  65. "%\u02B9i\u0304": "\u200C\u0627\u0649"
  66. "\u02B9i\u0304": "\u200C\u0627\u064A"
  67. "i\u0304\u02B9a\u0304": "\u0649\u200C\u0622"
  68. "i\u0304\u02B9a": "\u0649\u200C\u0627"
  69. "i\u0304\u02B9u\u0304": "\u0649\u200C\u0622"
  70. "i\u0304\u02B9u": "\u0649\u200C\u0627"
  71. "i\u0304\u02B9": "\u0649\u200C"
  72. "\u02B9": "\u200C"
  73. # Vowel and vowel/consonant combinations
  74. # and hyphenated suffixes:
  75. # izafah here
  76. "%h-\u02BEi": "\u06C0"
  77. "%h-\u02BCi": "\u06C0 "
  78. "%-\u02BEi": "\u06C0"
  79. "%-\u02BCi": "\u06C0"
  80. "%h-yi": "\u0647\u200C\u0649"
  81. "%-yi": "\u0649"
  82. "%yi": "\u0649"
  83. "%\u02BEi\u0304": "\u0649"
  84. "%\u02BCi\u0304": "\u0649"
  85. "%i\u0304-i": "\u0649"
  86. "%i\u0304": "\u0649"
  87. "%ayy-i": "\u0649"
  88. "%ay": "\u0649"
  89. "%al-i": "\u0644"
  90. "%-i": ""
  91. # Hyphenated prefixes:
  92. "wa-": "\u0648"
  93. "bi-": "\u0628"
  94. "al-": "\u0627\u0644"
  95. "lil-": "\u0644\u0644"
  96. "li-": "\u0644"
  97. "ka-": "\u0643"
  98. # ayn combo
  99. "%\u02BBa\u0304%": "\u0639\u0627"
  100. # alif combos
  101. # [final position]
  102. "%a\u0304\u02BE": "\u0627\u0621"
  103. "%a\u0304\u02BC": "\u0627\u0621"
  104. "%a\u0304\u02BEi\u0304": "\u0627\u0626\u0649"
  105. "%a\u0304\u02BCi\u0304": "\u0627\u0626\u0649"
  106. # [initial position]
  107. "A\u0304\u02BEi\u0304%": "\u0622\u0626\u064A"
  108. "A\u0304\u02BCi\u0304%": "\u0622\u0626\u064A"
  109. "a\u0304\u02BEi\u0304%": "\u0622\u0626\u064A"
  110. "a\u0304\u02BCi\u0304%": "\u0622\u0626\u064A"
  111. # [medial position]
  112. "a\u0304\u02BEi\u0304": "\u0627\u0626\u064A"
  113. "a\u0304\u02BCi\u0304": "\u0627\u0626\u064A"
  114. "a\u0304\u02BEi": "\u0627\u0626"
  115. "a\u0304\u02BCi": "\u0627\u0626"
  116. "a\u02BEi\u0304": "\u0626\u064A"
  117. "a\u02BCi\u0304": "\u0626\u064A"
  118. "u\u0304\u02BEi\u0304": "\u0648\u0626\u0649"
  119. "u\u0304\u02BCi\u0304": "\u0648\u0626\u0649"
  120. # A
  121. "\u02BBA\u0304": "\u0639\u0627"
  122. "\u02BBa\u0304": "\u0639\u0627"
  123. "\u02BBA%": "\u0639"
  124. "\u02BBa": "\u0639"
  125. "A\u02BB%": "\u0627\u0639"
  126. "a\u02BB%": "\u0627\u0639"
  127. "A\u02BB": "\u0623\u0639"
  128. "a\u02BB": "\u0639"
  129. "a\u02BE": "\u0623"
  130. "a\u02BC": "\u0623"
  131. "\u02BEa": "\u0623"
  132. "\u02BCa": "\u0623"
  133. "a\u0304\u02BE": "\u0621"
  134. "a\u0304\u02BC": "\u0621"
  135. "A\u0304%": "\u0622"
  136. "a\u0304%": "\u0622"
  137. # previously an alif:
  138. "A\u0304": "\u0622"
  139. "a\u0304": "\u0627"
  140. "A\u0301": "\u0649"
  141. "a\u0301": "\u0649"
  142. "ayy": "\u064A"
  143. "A%": "\u0627"
  144. "a%": "\u0627"
  145. "A": "\u0627"
  146. "a": ""
  147. # I
  148. "\u02BBI\u0304": "\u0639\u064A"
  149. "\u02BBi\u0304": "\u0639\u064A"
  150. "I\u02BB%": "\u0627\u0639"
  151. "i\u02BB%": "\u0627\u0639"
  152. "i\u02BB": "\u0639"
  153. "\u02BBI%": "\u0639"
  154. "I\u02BE%": "\u0627\u0626"
  155. "I\u02BC%": "\u0627\u0626"
  156. "i\u02BE": "\u0626"
  157. "i\u02BC": "\u0626"
  158. "\u02BEi\u0304": "\u0626\u0649"
  159. "\u02BCi\u0304": "\u0626\u0649"
  160. "\u02BEi": "\u0626"
  161. "\u02BCi": "\u0626"
  162. "I\u0304%": "\u0627\u064A"
  163. "i\u0304%": "\u0627\u064A"
  164. "i\u0304y": "\u064A"
  165. "I\u0304": "\u0627\u0649"
  166. "i\u0304": "\u064A"
  167. "I%": "\u0627"
  168. "i%": "\u0627"
  169. "I": "\u0627"
  170. "i": ""
  171. # U
  172. "\u02BEu\u0304": "\u0626\u0648"
  173. "\u02BCu\u0304": "\u0626\u0648"
  174. "U\u02BE%": "\u0627\u0624"
  175. "U\u02BC%": "\u0627\u0624"
  176. "u\u02BE": "\u0624"
  177. "u\u02BC": "\u0624"
  178. "U\u0304w%": "\u0627\u0628"
  179. "u\u0304w": "\u0628"
  180. "U\u0304": "\u0627\u0648"
  181. "u\u0304": "\u0648"
  182. "U%": "\u0627"
  183. "u%": "\u0627"
  184. "U": "\u0627"
  185. "u": ""
  186. # Consonants:
  187. "B": "\u0628"
  188. "bb": "\u0628"
  189. "b": "\u0628"
  190. "P": "\u067E"
  191. "pp": "\u067E"
  192. "p": "\u067E"
  193. "T\u0323": "\u0637"
  194. "t\u0323t\u0323": "\u0637"
  195. "t\u0323": "\u0637"
  196. "T": "\u062A"
  197. "tt": "\u062A"
  198. "t": "\u062A"
  199. "Sh": "\u0634"
  200. "shsh": "\u0634"
  201. "sh": "\u0634"
  202. "S\u0323": "\u0635"
  203. "s\u0323s\u0323": "\u0635"
  204. "s\u0323": "\u0635"
  205. "S\u0332": "\u062B"
  206. "ss\u0332": "\u062B"
  207. "s\u0332": "\u062B"
  208. "S": "\u0633"
  209. "ss": "\u0633"
  210. "s": "\u0633"
  211. "J": "\u062C"
  212. "jj": "\u062C"
  213. "j": "\u062C"
  214. "Ch": "\u0686"
  215. "chch": "\u0686"
  216. "ch": "\u0686"
  217. "H\u0323": "\u062D"
  218. "h\u0323h\u0323": "\u062D"
  219. "h\u0323": "\u062D"
  220. "H": "\u0647"
  221. "hh": "\u0647"
  222. "h": "\u0647"
  223. "Kh": "\u062E"
  224. "khkh": "\u062E"
  225. "kh": "\u062E"
  226. "K": "\u06A9"
  227. "kk": "\u06A9"
  228. "k": "\u06A9"
  229. # Arabic d with dot below:
  230. "D\u0323": "\u0636"
  231. "d\u0323d\u0323": "\u0636"
  232. "d\u0323": "\u0636"
  233. "D": "\u062F"
  234. "dd": "\u062F"
  235. "d": "\u062F"
  236. "Zh": "\u0698"
  237. "zhzh": "\u0698"
  238. "zh": "\u0698"
  239. "Z\u0323": "\u0638"
  240. "z\u0323z\u0323": "\u0638"
  241. "z\u0323": "\u0638"
  242. "Z\u0324": "\u0636"
  243. "z\u0324z\u0324": "\u0636"
  244. "z\u0324": "\u0636"
  245. "Z\u0332": "\u0630"
  246. "zz\u0332": "\u0630"
  247. "z\u0332": "\u0630"
  248. "Z": "\u0632"
  249. "zz": "\u0632"
  250. "z": "\u0632"
  251. "R": "\u0631"
  252. "rr": "\u0631"
  253. "r": "\u0631"
  254. "Gh": "\u063A"
  255. "ghgh": "\u063A"
  256. "gh": "\u063A"
  257. "G": "\u06AF"
  258. "g": "\u06AF"
  259. "F": "\u0641"
  260. "ff": "\u0641"
  261. "f": "\u0641"
  262. "Q": "\u0642"
  263. "qq": "\u0642"
  264. "q": "\u0642"
  265. "L": "\u0644"
  266. "ll": "\u0644"
  267. "l": "\u0644"
  268. "M": "\u0645"
  269. "mm": "\u0645"
  270. "m": "\u0645"
  271. "N": "\u0646"
  272. "nn": "\u0646"
  273. "n": "\u0646"
  274. "V": "\u0648"
  275. "vv": "\u0648"
  276. "v": "\u0648"
  277. "W": "\u0648"
  278. "ww": "\u0648"
  279. "w": "\u0648"
  280. "Y": "\u064A"
  281. "y": "\u064A"
  282. # non-Arabic consonants:
  283. # ain (alone)
  284. "\u02BB": "\u0639"
  285. # hamza (alone in final position)
  286. "%\u02BE": "\u0621"
  287. "%\u02BC": "\u0621"