urdu.yml 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463
  1. general:
  2. name: Urdi
  3. roman_to_script:
  4. map:
  5. # Punctuation marks:
  6. # "%": "\u066A"; cannot transliterate the truncation character
  7. "*": "\u066D"
  8. ",": "\u060C"
  9. ";": "\u061B"
  10. "?": "\u061F"
  11. # Exceptions for specific words
  12. # Allah
  13. "Alla\u0304h": "\u0627\u0644\u0644\u0647"
  14. "alla\u0304h": "\u0627\u0644\u0644\u0647"
  15. # Qur'an
  16. "Qur\u02BCa\u0304n": "\u0642\u0631\u0622\u0646"
  17. "qur\u02BCa\u0304n": "\u0642\u0631\u0622\u0646"
  18. # aur (with spaces)
  19. " aur ": " \u0627\u0648\u0631 "
  20. "Aur ": "\u0627\u0648\u0631 "
  21. ####
  22. # Abdurrahman
  23. "\u02BBAbdurrah\u0323ma\u0301n": "\u0639\u0628\u062F\u0627\u0644\u0631\u062D\u0645\u0670\u0646"
  24. "\u02BBAbdurrah\u0323ma\u0304n": "\u0639\u0628\u062F\u0627\u0644\u0631\u062D\u0645\u0670\u0646"
  25. # Abd names
  26. "\u02BBAbdul\u02BB": "\u0639\u0628\u062F\u0627\u0644\u0639"
  27. "\u02BBAbdula": "\u0639\u0628\u062F\u0627\u0644"
  28. "\u02BBAbdulb": "\u0639\u0628\u062F\u0627\u0644\u0628"
  29. "\u02BBAbdulf": "\u0639\u0628\u062F\u0627\u0644\u0641"
  30. "\u02BBAbdulg\u0332h\u0332": "\u0639\u0628\u062F\u0627\u0644\u063A"
  31. "\u02BBAbdulh\u0323": "\u0639\u0628\u062F\u0627\u0644\u062D"
  32. "\u02BBAbdulh": "\u0639\u0628\u062F\u0627\u0644\u0647 "
  33. "\u02BBAbdulj": "\u0639\u0628\u062F\u0627\u0644\u062C"
  34. "\u02BBAbdulk\u0332h\u0332": "\u0639\u0628\u062F\u0627\u0644\u062E"
  35. "\u02BBAbdulk": "\u0639\u0628\u062F\u0627\u0644\u0643 "
  36. "\u02BBAbdulm": "\u0639\u0628\u062F\u0627\u0644\u0645"
  37. "\u02BBAbdulq": "\u0639\u0628\u062F\u0627\u0644\u0642"
  38. "\u02BBAbdulv": "\u0639\u0628\u062F\u0627\u0644\u0648"
  39. "\u02BBAbdunn": "\u0639\u0628\u062F\u0627\u0644\u0646"
  40. "\u02BBAbdurr": "\u0639\u0628\u062F\u0627\u0644\u0631"
  41. "\u02BBAbdus\u0323s\u0323": "\u0639\u0628\u062F\u0627\u0644\u0325"
  42. "\u02BBAbduss": "\u0639\u0628\u062F\u0627\u0644\u0633"
  43. "\u02BBAbdushsh": "\u0639\u0628\u062F\u0627\u0644\u0634"
  44. "\u02BBAbdutt": "\u0639\u0628\u062F\u0627\u0644\u062A"
  45. "\u02BBAbduz\u0323z\u0323": "\u0639\u0628\u062F\u0627\u0644\u0636"
  46. "\u02BBAbduz\u0324z\u0324": "\u0639\u0628\u062F\u0627\u0644\u0638"
  47. # Abu names
  48. "Abu\u0304 ": "\u0627\u0628\u0648\u0020"
  49. "Abu\u0304": "\u0627\u0628\u0648\u200C\u0627\u0644"
  50. ####
  51. #lillah
  52. "lilla\u0304h": "\u0644\u0644\u0647"
  53. #billah
  54. "billa\u0304h": "\u0628\u0644\u0644\u0647"
  55. # Rahman
  56. "Rah\u0323ma\u0304n": "\u0631\u062D\u0645\u0646"
  57. # Nuzhat
  58. "Nuzhat": "\u0646\u0632\u0647\u062A"
  59. # Uddin names
  60. "%i\u0304uddi\u0304n": "\u0649\u200C\u0627\u0644\u062F\u0651\u064A\u0646"
  61. "%uddi\u0304n": "\u200C\u0627\u0644\u062F\u0651\u064A\u0646"
  62. # ta'lif
  63. # Ibn when it appears in the middle of a name sequence
  64. "ibn": "\u0628\u0646"
  65. # Abbreviated name elements
  66. "# Ae": "\u0627\u06D2"
  67. # Parsing "sh[dot below] as in "Ishaq [name]"
  68. "%sh\u0323%": "\u0633\u062D"
  69. # Numbers (\u06F0-06F9 for Persian/Urdu)
  70. # currently *not* valid MARC21 characters
  71. "# 0": "\u06F0"
  72. "# 1": "\u06F1"
  73. "# 2": "\u06F2"
  74. "# 3": "\u06F3"
  75. "# 4": "\u06F4"
  76. "# 5": "\u06F5"
  77. "# 6": "\u06F6"
  78. "# 7": "\u06F7"
  79. "# 8": "\u06F8"
  80. "# 9": "\u06F9"
  81. # Postpositions
  82. # Aspirates [06BE] vs. heh [062D] combinations
  83. "bh\u0323": "\u0628\u062D"
  84. "Bh": "\u0628\u06BE"
  85. "bh": "\u0628\u06BE"
  86. "ph\u0323": "\u067E\u062D"
  87. "Ph": "\u067E\u06BE"
  88. "ph": "\u067E\u06BE"
  89. "th\u0323": "\u062A\u062D"
  90. "Th": "\u062A\u06BE"
  91. "th": "\u062A\u06BE"
  92. "t\u0323h\u0323": "\u0679\u062D"
  93. "T\u0323h": "\u0679\u06BE"
  94. "t\u0323h": "\u0679\u06BE"
  95. "jh\u0323": "\u062C\u062D"
  96. "Jh": "\u062C\u06BE"
  97. "jh": "\u062C\u06BE"
  98. "ch\u0323": "\u0686\u062D"
  99. "Ch": "\u0686\u06BE"
  100. "ch": "\u0686\u06BE"
  101. "dh\u0323": "\u062F\u062D"
  102. "Dh": "\u062F\u06BE"
  103. "dh": "\u062F\u06BE"
  104. "d\u0323h\u0323": "\u0688\u062D"
  105. "D\u0323h": "\u0688\u06BE"
  106. "d\u0323h": "\u0688\u06BE"
  107. "r\u0323h\u0323": "\u0691\u062D"
  108. "R\u0323h": "\u0691\u06BE"
  109. "r\u0323h": "\u0691\u06BE"
  110. "kh\u0323": "\u06A9\u062D"
  111. "Kh": "\u06A9\u06BE"
  112. "kh": "\u06A9\u06BE"
  113. "gh\u0323": "\u06AF\u062D"
  114. "Gh": "\u06AF\u06BE"
  115. "gh": "\u06AF\u06BE"
  116. # prime = ZWNJ"
  117. "\u02B9A\u0304": "\u200C\u0622"
  118. "\u02B9a\u0304": "\u200C\u0622"
  119. "a\u0304\u02BC\u02B9": "\u0627\u0621\u200C"
  120. "i\u0304\u02B9": "\u0649\u200C"
  121. "\u02B9": "\u200C"
  122. # Izafah here
  123. "%a\u0304-yi": "\u0627\u0626\u06D2"
  124. "%u\u0304-yi": "\u0648\u0626\u06D2"
  125. "%o-yi": "\u0648\u0626\u06D2"
  126. "%e-yi": "\u06D2"
  127. "%i\u0304-yi": "\u0649"
  128. "%h-yi": "\u06C0"
  129. "%-yi": "\u06C0"
  130. "%al-i": "\u0644"
  131. "%ul-i": "\u0644"
  132. "%-i": ""
  133. # Hyphenated prefixes:
  134. "bi-": "\u0628"
  135. "al-a\u0304%": "\u0627\u0644\u0627"
  136. "ul-a\u0304%": "\u0627\u0644\u0627"
  137. "al-": "\u0627\u0644"
  138. "ul-": "\u0627\u0644"
  139. "lil-i": "\u0644\u0644"
  140. "lil-": "\u0644\u0644"
  141. # al-/ul- plus sun letters
  142. "ar-r": "\u0627\u0644\u0631"
  143. "ur-r": "\u0627\u0644\u0631"
  144. "ar-R": "\u0627\u0644\u0631"
  145. "ur-R": "\u0627\u0644\u0631"
  146. "az\u0332-z\u0332": "\u0627\u0644\u0630"
  147. "uz\u0332-z\u0332": "\u0627\u0644\u0630"
  148. "az\u0332-Z\u0332": "\u0627\u0644\u0630"
  149. "uz\u0332-Z\u0332": "\u0627\u0644\u0630"
  150. "ad-d": "\u0627\u0644\u0627"
  151. "ud-d": "\u0627\u0644\u0627"
  152. "ad-D": "\u0627\u0644\u0627"
  153. "ud-D": "\u0627\u0644\u0627"
  154. "as\u0332-s\u0332": "\u0627\u0644\u062B"
  155. "us\u0332-s\u0332": "\u0627\u0644\u062B"
  156. "as\u0332-S\u0332": "\u0627\u0644\u062B"
  157. "us\u0332-S\u0332": "\u0627\u0644\u062B"
  158. "at-t": "\u0627\u0644\u062A"
  159. "ut-t": "\u0627\u0644\u062A"
  160. "at-T": "\u0627\u0644\u062A"
  161. "ut-T": "\u0627\u0644\u062A"
  162. "an-n": "\u0627\u0644\u0646"
  163. "un-n": "\u0627\u0644\u0646"
  164. "an-N": "\u0627\u0644\u0646"
  165. "un-N": "\u0627\u0644\u0646"
  166. "al-l": "\u0627\u0644\u0644"
  167. "ul-l": "\u0627\u0644\u0644"
  168. "al-L": "\u0627\u0644\u0644"
  169. "ul-L": "\u0627\u0644\u0644"
  170. "az\u0324-z\u0324": "\u0627\u0644\u0638"
  171. "uz\u0324-z\u0324": "\u0627\u0644\u0638"
  172. "az\u0324-Z\u0324": "\u0627\u0644\u0638"
  173. "uz\u0324-Z\u0324": "\u0627\u0644\u0638"
  174. "at\u0324-t\u0324": "\u0627\u0644\u0637"
  175. "ut\u0324-t\u0324": "\u0627\u0644\u0637"
  176. "at\u0324-T\u0324": "\u0627\u0644\u0637"
  177. "ut\u0324-T\u0324": "\u0627\u0644\u0637"
  178. "az\u0323-z\u0323": "\u0627\u0644\u0636"
  179. "uz\u0323-z\u0323": "\u0627\u0644\u0636"
  180. "az\u0323-Z\u0323": "\u0627\u0644\u0636"
  181. "uz\u0323-Z\u0323": "\u0627\u0644\u0636"
  182. "as\u0323-s\u0323": "\u0627\u0644\u0635"
  183. "us\u0323-s\u0323": "\u0627\u0644\u0635"
  184. "as\u0323-S\u0323": "\u0627\u0644\u0635"
  185. "us\u0323-S\u0323": "\u0627\u0644\u0635"
  186. "ash-sh": "\u0627\u0644\u0634"
  187. "ush-sh": "\u0627\u0644\u0634"
  188. "ash-Sh": "\u0627\u0644\u0634"
  189. "ush-Sh": "\u0627\u0644\u0634"
  190. "as-s": "\u0627\u0644\u0633"
  191. "us-s": "\u0627\u0644\u0633"
  192. "as-S": "\u0627\u0644\u0633"
  193. "us-S": "\u0627\u0644\u0633"
  194. "az-z": "\u0627\u0644\u0632"
  195. "uz-z": "\u0627\u0644\u0632"
  196. "az-Z": "\u0627\u0644\u0632"
  197. "uz-Z": "\u0627\u0644\u0632"
  198. # Diphthongs here
  199. "Ae": "\u0627\u06D2"
  200. "%ai": "\u06D2"
  201. "Ai": "\u0627\u064A"
  202. "ai%": "\u0627\u064A"
  203. "ai": "\u064A"
  204. "\u02BBAu": "\u0639\u0648"
  205. "\u02BBau": "\u0639\u0648"
  206. "Au": "\u0627\u0648"
  207. "au": "\u0648"
  208. # ayn-alif combo
  209. "%\u02BBa\u0304\u02BE": "\u0639\u0627\u0621"
  210. "%\u02BBa\u0304\u02BC": "\u0639\u0627\u0621"
  211. "%\u02BBa\u0304%": "\u0639\u0627"
  212. # hamza and vowel combo
  213. # [in final position]
  214. "%u\u0304\u02BEi\u0304": "\u0648\u0626\u0649"
  215. "%u\u0304\u02BCi\u0304": "\u0648\u0626\u0649"
  216. "%\u02BEi\u0304": "\u0626\u0649"
  217. "%\u02BCi\u0304": "\u0626\u0649"
  218. "%\u02BEe": "\u0626\u06D2"
  219. "%\u02BCe": "\u0626\u06D2"
  220. "%\u02BEu\u0304": "\u0624"
  221. "%\u02BCu\u0304": "\u0624"
  222. "%\u02BEo": "\u0624"
  223. "%\u02BCo": "\u0624"
  224. # [in medial position]
  225. "a\u02BEa": "\u0623"
  226. "a\u02BCa": "\u0623"
  227. "a\u0304\u02BEa": "\u0627\u0621"
  228. "a\u0304\u02BCa": "\u0627\u0621"
  229. "a\u02BEa\u0304": "\u0622"
  230. "a\u02BCa\u0304": "\u0622"
  231. "o\u02BEi\u0304": "\u0648\u0626\u064A"
  232. "o\u02BCi\u0304": "\u0648\u0626\u064A"
  233. "o\u02BEi": "\u0648\u0626"
  234. "o\u02BCi": "\u0648\u0626"
  235. "\u02BEi\u0304": "\u0626\u064A"
  236. "\u02BCi\u0304": "\u0626\u064A"
  237. "\u02BEi": "\u0626"
  238. "\u02BCi": "\u0626"
  239. "\u02BEe": "\u0626\u064A"
  240. "\u02BCe": "\u0626\u064A"
  241. "\u02BEu\u0304": "\u0624"
  242. "\u02BCu\u0304": "\u0624"
  243. "u\u0304\u02BE": "\u0624"
  244. "u\u0304\u02BC": "\u0624"
  245. "\u02BEo": "\u0624"
  246. "\u02BCo": "\u0624"
  247. "o\u02BE": "\u0624"
  248. "o\u02BC": "\u0624"
  249. "au\u02BE": "\u0624"
  250. "au\u02BC": "\u0624"
  251. "\u02BEa": "\u0626"
  252. "\u02BCa": "\u0626"
  253. "%i\u0304": "\u0649"
  254. "%a\u0301": "\u0649\u0670"
  255. # A
  256. "\u02BBA\u0304": "\u0639\u0627"
  257. "\u02BBa\u0304": "\u0639\u0627"
  258. "\u02BBA%": "\u0639"
  259. "\u02BBa": "\u0639"
  260. "A\u02BB": "\u0627\u0639"
  261. "a\u02BB%": "\u0627\u0639"
  262. "a\u02BB": "\u0639"
  263. "A\u0304%": "\u0622"
  264. "a\u0304%": "\u0622"
  265. "a\u0304": "\u0627"
  266. "a\u0301": "\u0649"
  267. "ayy": "\u064A\u0651"
  268. "A%": "\u0627"
  269. "a%": "\u0627"
  270. "A": ""
  271. "a": ""
  272. # E
  273. "%e": "\u06D2"
  274. "E%": "\u0627\u064A"
  275. "e%": "\u0627\u064A"
  276. "e": "\u064A"
  277. # I
  278. "\u02BBI\u0304": "\u0639\u064A"
  279. "\u02BBi\u0304": "\u0639\u064A"
  280. "I\u02BB": "\u0627\u0639"
  281. "i\u02BB": "\u0639"
  282. "\u02BBI": "\u0639"
  283. "I\u0304%": "\u0627\u064A"
  284. "i\u0304%": "\u0627\u064A"
  285. "i\u0304y": "\u064A"
  286. "i\u0304": "\u064A"
  287. "iyy": "\u064A\u0651"
  288. "I%": "\u0627"
  289. "i%": "\u0627"
  290. "I": "\u0627"
  291. "i": ""
  292. # O
  293. "O%": "\u0627\u0648"
  294. "o": "\u0648"
  295. # U
  296. "\u02BBu\u0304": "\u0639\u0648"
  297. "\u02BBU": "\u0639"
  298. "\u02BBu": "\u0639"
  299. "U\u0304%": "\u0627\u0648"
  300. "u\u0304%": "\u0627\u0648"
  301. "u\u0304": "\u0648"
  302. "U%": "\u0627"
  303. "u%": "\u0627"
  304. "U": ""
  305. "u": ""
  306. # Consonants:
  307. "B": "\u0628"
  308. "bb": "\u0628\u0651"
  309. "b": "\u0628"
  310. "P": "\u067E"
  311. "pp": "\u067E\u0651"
  312. "p": "\u067E"
  313. "T\u0323": "\u0679"
  314. "t\u0323t\u0323": "\u0679\u0651"
  315. "t\u0323": "\u0679"
  316. "T\u0324": "\u0637"
  317. "t\u0324t\u0324": "\u0637\u0651"
  318. "t\u0324": "\u0637"
  319. "T": "\u062A"
  320. "tt": "\u062A\u0651"
  321. "t": "\u062A"
  322. "Sh": "\u0634"
  323. "shsh": "\u0634\u0651"
  324. "sh": "\u0634"
  325. "S\u0323": "\u0635"
  326. "s\u0323s\u0323": "\u0635\u0651"
  327. "s\u0323": "\u0635"
  328. "S\u0332": "\u062B"
  329. "s\u0332s\u0332": "\u062B\u0651"
  330. "s\u0332": "\u062B"
  331. "S": "\u0633"
  332. "ss": "\u0633\u0651"
  333. "s": "\u0633"
  334. "J": "\u062C"
  335. "jj": "\u062C\u0651"
  336. "j": "\u062C"
  337. "C": "\u0686"
  338. "cc": "\u0686\u0651"
  339. "c": "\u0686"
  340. "H\u0323": "\u062D"
  341. "h\u0323h\u0323": "\u062D\u0651"
  342. "h\u0323": "\u062D"
  343. "H": "\u0647"
  344. "hh": "\u0647\u0651"
  345. "h": "\u0647"
  346. "K\u0332h\u0332": "\u062E"
  347. "k\u0332h\u0332k\u0332h\u0332": "\u062E\u0651"
  348. "k\u0332h\u0332": "\u062E"
  349. "K": "\u06A9"
  350. "kk": "\u06A9\u0651"
  351. "k": "\u06A9"
  352. "D\u0323": "\u0688"
  353. "d\u0323d\u0323": "\u0688\u0651"
  354. "d\u0323": "\u0688"
  355. "D": "\u062F"
  356. "dd": "\u062F\u0651"
  357. "d": "\u062F"
  358. "Z\u0324": "\u0638"
  359. "z\u0324z\u0324": "\u0638\u0651"
  360. "z\u0324": "\u0638"
  361. "Z\u0323": "\u0636"
  362. "z\u0323z\u0323": "\u0636\u0651"
  363. "z\u0323": "\u0636"
  364. "Z\u0332": "\u0630"
  365. "z\u0332z\u0332": "\u0630\u0651"
  366. "z\u0332": "\u0630"
  367. "zz": "\u0632\u0651"
  368. "Zh": "\u0698"
  369. "zhzh": "\u0698\u0651"
  370. "zh": "\u0698"
  371. "Z": "\u0632"
  372. "z": "\u0632"
  373. "R\u0323": "\u0691"
  374. "r\u0323r\u0323": "\u0691\u0651"
  375. "r\u0323": "\u0691"
  376. "R": "\u0631"
  377. "rr": "\u0631\u0651"
  378. "r": "\u0631"
  379. "G\u0332h\u0332": "\u063A"
  380. "g\u0332h\u0332g\u0332h\u0332": "\u063A\u0651"
  381. "g\u0332h\u0332": "\u063A"
  382. "G": "\u06AF"
  383. "gg": "\u06AF\u0651"
  384. "g": "\u06AF"
  385. "F": "\u0641"
  386. "ff": "\u0641\u0651"
  387. "f": "\u0641"
  388. "Q": "\u0642"
  389. "qq": "\u0642\u0651"
  390. "q": "\u0642"
  391. "L": "\u0644"
  392. "ll": "\u0644\u0651"
  393. "l": "\u0644"
  394. "M": "\u0645"
  395. "mm": "\u0645\u0651"
  396. "m": "\u0645"
  397. "N\u0332": "\u06BA"
  398. "n\u0332n\u0332": "\u06BA\u0651"
  399. "n\u0332": "\u06BA"
  400. "N": "\u0646"
  401. "nn": "\u0646\u0651"
  402. "n": "\u0646"
  403. "V": "\u0648"
  404. "vv": "\u0648\u0651"
  405. "v": "\u0648"
  406. "Y": "\u064A"
  407. "yy": "\u064A\u0651"
  408. "y": "\u064A"
  409. # ain (alone)
  410. "\u02BB": "\u0639"
  411. # hamza (alone in final position)
  412. "%\u02BE": "\u0621"
  413. "%\u02BC": "\u0621"