interscript 0.1.7 → 2.1.0b1

Sign up to get free protection for your applications and to get access to all the features.
Files changed (314) hide show
  1. checksums.yaml +4 -4
  2. data/.gitignore +11 -0
  3. data/.rspec +3 -0
  4. data/Gemfile +29 -0
  5. data/LICENSE.adoc +31 -0
  6. data/README.md +3 -0
  7. data/Rakefile +116 -0
  8. data/bin/console +14 -0
  9. data/bin/interscript +5 -0
  10. data/bin/maps_analyze_staging +168 -0
  11. data/bin/maps_debug_compilers +58 -0
  12. data/bin/maps_debug_ordering +88 -0
  13. data/bin/maps_debug_ruby_compile +24 -0
  14. data/bin/maps_debug_step_by_step +44 -0
  15. data/bin/maps_optimize_order +112 -0
  16. data/bin/maps_v1_analyze_regexps +45 -0
  17. data/bin/maps_v1_to_v2 +426 -0
  18. data/bin/setup +8 -0
  19. data/exe/interscript +6 -0
  20. data/interscript.gemspec +31 -0
  21. data/lib/interscript.rb +83 -133
  22. data/lib/interscript/command.rb +5 -5
  23. data/lib/interscript/compiler.rb +22 -0
  24. data/lib/interscript/compiler/javascript.rb +292 -0
  25. data/lib/interscript/compiler/ruby.rb +262 -0
  26. data/lib/interscript/dsl.rb +68 -0
  27. data/lib/interscript/dsl/aliases.rb +23 -0
  28. data/lib/interscript/dsl/document.rb +46 -0
  29. data/lib/interscript/dsl/group.rb +45 -0
  30. data/lib/interscript/dsl/group/parallel.rb +6 -0
  31. data/lib/interscript/dsl/items.rb +89 -0
  32. data/lib/interscript/dsl/metadata.rb +68 -0
  33. data/lib/interscript/dsl/stage.rb +6 -0
  34. data/lib/interscript/dsl/symbol_mm.rb +11 -0
  35. data/lib/interscript/dsl/tests.rb +12 -0
  36. data/lib/interscript/interpreter.rb +251 -0
  37. data/lib/interscript/node.rb +25 -0
  38. data/lib/interscript/node/alias_def.rb +15 -0
  39. data/lib/interscript/node/dependency.rb +13 -0
  40. data/lib/interscript/node/document.rb +45 -0
  41. data/lib/interscript/node/group.rb +34 -0
  42. data/lib/interscript/node/group/parallel.rb +9 -0
  43. data/lib/interscript/node/group/sequential.rb +2 -0
  44. data/lib/interscript/node/item.rb +52 -0
  45. data/lib/interscript/node/item/alias.rb +42 -0
  46. data/lib/interscript/node/item/any.rb +76 -0
  47. data/lib/interscript/node/item/capture.rb +50 -0
  48. data/lib/interscript/node/item/group.rb +51 -0
  49. data/lib/interscript/node/item/repeat.rb +40 -0
  50. data/lib/interscript/node/item/stage.rb +23 -0
  51. data/lib/interscript/node/item/string.rb +51 -0
  52. data/lib/interscript/node/metadata.rb +18 -0
  53. data/lib/interscript/node/rule.rb +6 -0
  54. data/lib/interscript/node/rule/funcall.rb +18 -0
  55. data/lib/interscript/node/rule/run.rb +15 -0
  56. data/lib/interscript/node/rule/sub.rb +68 -0
  57. data/lib/interscript/node/stage.rb +19 -0
  58. data/lib/interscript/node/tests.rb +15 -0
  59. data/lib/interscript/stdlib.rb +211 -0
  60. data/lib/interscript/utils/regexp_converter.rb +283 -0
  61. data/lib/interscript/version.rb +1 -1
  62. data/lib/interscript/visualize.rb +61 -0
  63. data/lib/interscript/visualize/group.html.erb +59 -0
  64. data/lib/interscript/visualize/json.rb +57 -0
  65. data/lib/interscript/visualize/map.html.erb +46 -0
  66. data/lib/interscript/visualize/nodes.rb +89 -0
  67. data/requirements.txt +1 -0
  68. metadata +78 -416
  69. data/README.adoc +0 -298
  70. data/lib/g2pwrapper.py +0 -34
  71. data/lib/interscript/fs.rb +0 -69
  72. data/lib/interscript/mapping.rb +0 -142
  73. data/lib/interscript/opal.rb +0 -57
  74. data/lib/interscript/opal/entrypoint.rb +0 -12
  75. data/lib/interscript/opal/map_translate.rb +0 -7
  76. data/lib/interscript/opal/maps.js.erb +0 -10
  77. data/lib/model-7 +0 -0
  78. data/lib/tha-pt-b-7 +0 -0
  79. data/maps/acadsin-zho-Hani-Latn-2002.yaml +0 -38916
  80. data/maps/alalc-amh-Ethi-Latn-1997.yaml +0 -513
  81. data/maps/alalc-amh-Ethi-Latn-2011.yaml +0 -138
  82. data/maps/alalc-ara-Arab-Latn-1997.yaml +0 -1287
  83. data/maps/alalc-asm-Deva-Latn-1997.yaml +0 -165
  84. data/maps/alalc-asm-Deva-Latn-2012.yaml +0 -40
  85. data/maps/alalc-aze-Cyrl-Latn-1997.yaml +0 -145
  86. data/maps/alalc-bel-Cyrl-Latn-1997.yaml +0 -129
  87. data/maps/alalc-ben-Beng-Latn-2017.yaml +0 -130
  88. data/maps/alalc-bul-Cyrl-Latn-1997.yaml +0 -98
  89. data/maps/alalc-ell-Grek-Latn-1997.yaml +0 -628
  90. data/maps/alalc-ell-Grek-Latn-2010.yaml +0 -626
  91. data/maps/alalc-guj-Gujr-Latn-1997.yaml +0 -266
  92. data/maps/alalc-guj-Gujr-Latn-2011.yaml +0 -64
  93. data/maps/alalc-hin-Deva-Latn-1997.yaml +0 -211
  94. data/maps/alalc-hin-Deva-Latn-2011.yaml +0 -47
  95. data/maps/alalc-kat-Geok-Latn-1997.yaml +0 -111
  96. data/maps/alalc-kat-Geor-Latn-1997.yaml +0 -150
  97. data/maps/alalc-kor-Hang-Latn-1997.yaml +0 -98
  98. data/maps/alalc-mal-Mlym-Latn-1997.yaml +0 -303
  99. data/maps/alalc-mal-Mlym-Latn-2012.yaml +0 -73
  100. data/maps/alalc-mar-Deva-Latn-1997.yaml +0 -189
  101. data/maps/alalc-mar-Deva-Latn-2011.yaml +0 -45
  102. data/maps/alalc-mkd-Cyrl-Latn-1997.yaml +0 -114
  103. data/maps/alalc-mkd-Cyrl-Latn-2013.yaml +0 -103
  104. data/maps/alalc-mon-Cyrl-Latn-1997.yaml +0 -220
  105. data/maps/alalc-pan-Guru-Latn-1997.yaml +0 -256
  106. data/maps/alalc-pan-Guru-Latn-2011.yaml +0 -78
  107. data/maps/alalc-per-Arab-Latn-1997.yaml +0 -375
  108. data/maps/alalc-pli-Deva-Latn-2012.yaml +0 -144
  109. data/maps/alalc-pra-Deva-Latn-2012.yaml +0 -47
  110. data/maps/alalc-rus-Cyrl-Latn-1997.yaml +0 -225
  111. data/maps/alalc-rus-Cyrl-Latn-2012.yaml +0 -162
  112. data/maps/alalc-san-Deva-Latn-2012.yaml +0 -172
  113. data/maps/alalc-sin-Sinh-Latn-1997.yaml +0 -292
  114. data/maps/alalc-sin-Sinh-Latn-2011.yaml +0 -71
  115. data/maps/alalc-srp-Cyrl-Latn-1997.yaml +0 -118
  116. data/maps/alalc-srp-Cyrl-Latn-2013.yaml +0 -135
  117. data/maps/alalc-tam-Taml-Latn-1997.yaml +0 -62
  118. data/maps/alalc-tam-Taml-Latn-2011.yaml +0 -58
  119. data/maps/alalc-ukr-Cyrl-Latn-1997.yaml +0 -145
  120. data/maps/alalc-ukr-Cyrl-Latn-2011.yaml +0 -16
  121. data/maps/apcbg-bul-Cyrl-Latn-1995.yaml +0 -283
  122. data/maps/bas-rus-Cyrl-Latn-2017-bss.yaml +0 -174
  123. data/maps/bas-rus-Cyrl-Latn-2017-oss.yaml +0 -169
  124. data/maps/bgn-jpn-Hrkt-Latn-1962.yaml +0 -292
  125. data/maps/bgn-kor-Hang-Latn-1943.yaml +0 -35
  126. data/maps/bgn-kor-Kore-Latn-1943.yaml +0 -31
  127. data/maps/bgna-bul-Cyrl-Latn-2006.yaml +0 -208
  128. data/maps/bgna-bul-Cyrl-Latn-2009.yaml +0 -208
  129. data/maps/bgnpcgn-amh-Ethi-Latn-1967.yaml +0 -532
  130. data/maps/bgnpcgn-ara-Arab-Latn-1956.yaml +0 -596
  131. data/maps/bgnpcgn-arm-Armn-Latn-1981.yaml +0 -108
  132. data/maps/bgnpcgn-aze-Cyrl-Latn-1993.yaml +0 -104
  133. data/maps/bgnpcgn-bak-Cyrl-Latn-2007.yaml +0 -188
  134. data/maps/bgnpcgn-bel-Cyrl-Latn-1979.yaml +0 -289
  135. data/maps/bgnpcgn-bul-Cyrl-Latn-1952.yaml +0 -119
  136. data/maps/bgnpcgn-bul-Cyrl-Latn-2013.yaml +0 -42
  137. data/maps/bgnpcgn-che-Cyrl-Latn-2008.yaml +0 -184
  138. data/maps/bgnpcgn-ell-Grek-Latn-1962.yaml +0 -705
  139. data/maps/bgnpcgn-ell-Grek-Latn-1996.yaml +0 -23
  140. data/maps/bgnpcgn-fas-Arab-Latn-1956.yaml +0 -96
  141. data/maps/bgnpcgn-jpn-Hrkt-Latn-1976.yaml +0 -257
  142. data/maps/bgnpcgn-kat-Geor-Latn-1981.yaml +0 -131
  143. data/maps/bgnpcgn-kat-Geor-Latn-2009.yaml +0 -42
  144. data/maps/bgnpcgn-kor-Hang-Latn-kn-1945.yaml +0 -253
  145. data/maps/bgnpcgn-kor-Hang-Latn-rok-2011.yaml +0 -48
  146. data/maps/bgnpcgn-kor-Kore-Latn-rok-2011.yaml +0 -48
  147. data/maps/bgnpcgn-mkd-Cyrl-Latn-1981.yaml +0 -163
  148. data/maps/bgnpcgn-mkd-Cyrl-Latn-2013.yaml +0 -190
  149. data/maps/bgnpcgn-mon-Cyrl-Latn-1964.yaml +0 -223
  150. data/maps/bgnpcgn-nep-Deva-Latn-2011.yaml +0 -230
  151. data/maps/bgnpcgn-per-Arab-Latn-1958.yaml +0 -336
  152. data/maps/bgnpcgn-prs-Arab-Latn-2007.yaml +0 -639
  153. data/maps/bgnpcgn-prs-Arab-Latn-yaghoubi.yaml +0 -459
  154. data/maps/bgnpcgn-rue-Cyrl-Latn-2016.yaml +0 -168
  155. data/maps/bgnpcgn-rus-Cyrl-Latn-1947.yaml +0 -318
  156. data/maps/bgnpcgn-srp-Cyrl-Latn-2005.yaml +0 -170
  157. data/maps/bgnpcgn-tat-Cyrl-Latn-2007.yaml +0 -220
  158. data/maps/bgnpcgn-tgk-Cyrl-Latn-1994.yaml +0 -240
  159. data/maps/bgnpcgn-ukr-Cyrl-Latn-1965.yaml +0 -166
  160. data/maps/bgnpcgn-ukr-Cyrl-Latn-2019.yaml +0 -119
  161. data/maps/bgnpcgn-uzb-Cyrl-Latn-1979.yaml +0 -127
  162. data/maps/bgnpcgn-uzb-Cyrl-Latn-2000.yaml +0 -82
  163. data/maps/bgnpcgn-zho-Hans-Latn-1979.yaml +0 -7456
  164. data/maps/bis-asm-Beng-Latn-13194-1991.yaml +0 -159
  165. data/maps/bis-ben-Beng-Latn-13194-1991.yaml +0 -156
  166. data/maps/bis-dev-Deva-Latn-13194-1991.yaml +0 -184
  167. data/maps/bis-guj-Gujr-Latn-13194-1991.yaml +0 -181
  168. data/maps/bis-knd-Knda-Latn-13194-1991.yaml +0 -173
  169. data/maps/bis-mlm-Mlym-Latn-13194-1991.yaml +0 -176
  170. data/maps/bis-ori-Orya-Latn-13194-1991.yaml +0 -160
  171. data/maps/bis-pnj-Guru-Latn-13194-1991.yaml +0 -175
  172. data/maps/bis-tel-Telu-Latn-13194-1991.yaml +0 -170
  173. data/maps/bis-tml-Taml-Latn-13194-1991.yaml +0 -155
  174. data/maps/by-bel-Cyrl-Latn-1998.yaml +0 -172
  175. data/maps/by-bel-Cyrl-Latn-2007.yaml +0 -115
  176. data/maps/din-grc-Grek-Latn-31634-2011-t1.yaml +0 -899
  177. data/maps/din-hin-Deva-Latn-33904-2018.yaml +0 -100
  178. data/maps/din-kat-Geor-Latn-32707-2010.yaml +0 -145
  179. data/maps/din-mar-Deva-Latn-33904-2018.yaml +0 -84
  180. data/maps/din-nep-Deva-Latn-33904-2018.yaml +0 -119
  181. data/maps/din-pli-Deva-Latn-33904-2018.yaml +0 -75
  182. data/maps/din-pra-Deva-Latn-33904-2018.yaml +0 -63
  183. data/maps/din-san-Deva-Latn-33904-2018.yaml +0 -338
  184. data/maps/din-tam-Taml-Latn-33903-2016.yaml +0 -213
  185. data/maps/dos-nep-Deva-Latn-1997.yaml +0 -47
  186. data/maps/elot-ell-Grek-Latn-743-1982-tl.yaml +0 -684
  187. data/maps/elot-ell-Grek-Latn-743-1982-ts.yaml +0 -680
  188. data/maps/elot-ell-Grek-Latn-743-2001-tl.yaml +0 -19
  189. data/maps/elot-ell-Grek-Latn-743-2001-ts.yaml +0 -31
  190. data/maps/ggg-kat-Geor-Latn-2002.yaml +0 -92
  191. data/maps/gki-bel-Cyrl-Latn-1992.yaml +0 -33
  192. data/maps/gki-bel-Cyrl-Latn-2000.yaml +0 -201
  193. data/maps/gost-rus-Cyrl-Latn-16876-71-1983.yaml +0 -190
  194. data/maps/gost-rus-Cyrl-Latn-7.79-2000-2002.yaml +0 -157
  195. data/maps/hk-yue-Hani-Latn-1888.yaml +0 -38497
  196. data/maps/icao-bel-Cyrl-Latn-9303.yaml +0 -136
  197. data/maps/icao-bul-Cyrl-Latn-9303.yaml +0 -118
  198. data/maps/icao-fas-Arab-Latn-9303.yaml +0 -103
  199. data/maps/icao-heb-Hebr-Latn-9303.yaml +0 -151
  200. data/maps/icao-mkd-Cyrl-Latn-9303.yaml +0 -117
  201. data/maps/icao-rus-Cyrl-Latn-9303.yaml +0 -117
  202. data/maps/icao-srp-Cyrl-Latn-9303.yaml +0 -117
  203. data/maps/icao-ukr-Cyrl-Latn-9303.yaml +0 -119
  204. data/maps/iso-ara-Arab-Latn-233-1984.yaml +0 -323
  205. data/maps/iso-asm-Beng-Latn-15919-2001.yaml +0 -75
  206. data/maps/iso-ben-Beng-Latn-15919-2001.yaml +0 -175
  207. data/maps/iso-ell-Grek-Latn-843-1997-t1.yaml +0 -613
  208. data/maps/iso-ell-Grek-Latn-843-1997-t2.yaml +0 -44
  209. data/maps/iso-guj-Gujr-Latn-15919-2001.yaml +0 -220
  210. data/maps/iso-hin-Deva-Latn-15919-2001.yaml +0 -87
  211. data/maps/iso-inc-Deva-Latn-15919-2001.yaml +0 -61
  212. data/maps/iso-jpn-Hrkt-Latn-3602-1989.yaml +0 -66
  213. data/maps/iso-kan-Knda-Latn-15919-2001.yaml +0 -220
  214. data/maps/iso-kat-Geor-Latn-9984-1996.yaml +0 -145
  215. data/maps/iso-kor-Hang-Latn-1996-method1.yaml +0 -240
  216. data/maps/iso-kor-Hang-Latn-1996-method2.yaml +0 -226
  217. data/maps/iso-mal-Mlym-Latn-15919-2001.yaml +0 -281
  218. data/maps/iso-mar-Deva-Latn-15919-2001.yaml +0 -75
  219. data/maps/iso-nep-Deva-Latn-15919-2001.yaml +0 -87
  220. data/maps/iso-ori-Orya-Latn-15919-2001.yaml +0 -193
  221. data/maps/iso-pan-Guru-Latn-15919-2001.yaml +0 -222
  222. data/maps/iso-pli-Beng-Latn-15919-2001.yaml +0 -73
  223. data/maps/iso-pli-Deva-Latn-15919-2001.yaml +0 -74
  224. data/maps/iso-pli-Sinh-Latn-15919-2001.yaml +0 -219
  225. data/maps/iso-pli-Thai-Latn-15919-2001.yaml +0 -55
  226. data/maps/iso-pra-Deva-Latn-15919-2001.yaml +0 -59
  227. data/maps/iso-prs-Arab-Latn-233-3-1999.yaml +0 -366
  228. data/maps/iso-rus-Cyrl-Latn-9-1995.yaml +0 -271
  229. data/maps/iso-san-Deva-Latn-15919-2001.yaml +0 -220
  230. data/maps/iso-tam-Taml-Latn-15919-2001.yaml +0 -159
  231. data/maps/iso-tel-Telu-Latn-15919-2001.yaml +0 -220
  232. data/maps/iso-tha-Thai-Latn-11940-1998.yaml +0 -109
  233. data/maps/kp-kor-Hang-Latn-2002.yaml +0 -909
  234. data/maps/lshk-yue-Hani-Latn-jyutping-1993.yaml +0 -44820
  235. data/maps/mext-jpn-Hrkt-Latn-1954.yaml +0 -411
  236. data/maps/mns-mon-Cyrl-Latn-5217-2012.yaml +0 -163
  237. data/maps/mns-mon-Latn-Cyrl-5217-2012.yaml +0 -200
  238. data/maps/moct-kor-Hang-Latn-2000.yaml +0 -807
  239. data/maps/mofa-jpn-Hrkt-Latn-1989.yaml +0 -541
  240. data/maps/mvd-bel-Cyrl-Latn-2008.yaml +0 -225
  241. data/maps/mvd-bel-Cyrl-Latn-2010.yaml +0 -63
  242. data/maps/mvd-rus-Cyrl-Latn-2008.yaml +0 -109
  243. data/maps/mvd-rus-Cyrl-Latn-2010.yaml +0 -37
  244. data/maps/odni-ara-Arab-Latn-2015.yaml +0 -425
  245. data/maps/odni-aze-Cyrl-Latn-2015.yaml +0 -144
  246. data/maps/odni-bel-Cyrl-Latn-2015.yaml +0 -148
  247. data/maps/odni-bul-Cyrl-Latn-2015.yaml +0 -96
  248. data/maps/odni-che-Cyrl-Latn-2015.yaml +0 -169
  249. data/maps/odni-fas-Arab-Latn-2015.yaml +0 -406
  250. data/maps/odni-hin-Deva-Latn-2015.yaml +0 -258
  251. data/maps/odni-kat-Geor-Latn-2015.yaml +0 -87
  252. data/maps/odni-kaz-Cyrl-Latn-2015.yaml +0 -148
  253. data/maps/odni-kir-Cyrl-Latn-2015.yaml +0 -136
  254. data/maps/odni-kor-Hang-Latn-2015.yaml +0 -375
  255. data/maps/odni-mkd-Cyrl-Latn-2015.yaml +0 -122
  256. data/maps/odni-per-Arab-Latn-2015.yaml +0 -228
  257. data/maps/odni-rus-Cyrl-Latn-2015.yaml +0 -77
  258. data/maps/odni-srp-Cyrl-Latn-2015.yaml +0 -129
  259. data/maps/odni-tat-Cyrl-Latn-2015.yaml +0 -142
  260. data/maps/odni-tgk-Cyrl-Latn-2015.yaml +0 -148
  261. data/maps/odni-uig-Cyrl-Latn-2015.yaml +0 -138
  262. data/maps/odni-ukr-Cyrl-Latn-2015.yaml +0 -157
  263. data/maps/odni-urd-Arab-Latn-2015.yaml +0 -221
  264. data/maps/odni-uzb-Cyrl-Latn-2015.yaml +0 -166
  265. data/maps/royin-tha-Thai-Latn-1939-generic.yaml +0 -90
  266. data/maps/royin-tha-Thai-Latn-1968.yaml +0 -183
  267. data/maps/royin-tha-Thai-Latn-1999-chained.yaml +0 -180
  268. data/maps/royin-tha-Thai-Latn-1999.yaml +0 -80
  269. data/maps/sac-zho-Hans-Latn-1979.yaml +0 -24763
  270. data/maps/sasm-mon-Mong-Latn-general-1978.yaml +0 -389
  271. data/maps/sasm-mon-Mong-Latn-phonetic-1978.yaml +0 -354
  272. data/maps/ses-ara-Arab-Latn-1930.yaml +0 -283
  273. data/maps/stategeocadastre-ukr-Cyrl-Latn-1993.yaml +0 -222
  274. data/maps/ua-ukr-Cyrl-Latn-1996.yaml +0 -197
  275. data/maps/ua-ukr-Cyrl-Latn-2007.yaml +0 -75
  276. data/maps/ua-ukr-Cyrl-Latn-2010.yaml +0 -192
  277. data/maps/un-amh-Ethi-Latn-2016.yaml +0 -602
  278. data/maps/un-ara-Arab-Latn-1971.yaml +0 -139
  279. data/maps/un-ara-Arab-Latn-1972.yaml +0 -159
  280. data/maps/un-ara-Arab-Latn-2017.yaml +0 -420
  281. data/maps/un-bel-Cyrl-Latn-2007.yaml +0 -114
  282. data/maps/un-ben-Beng-Latn-2016.yaml +0 -534
  283. data/maps/un-ell-Grek-Latn-1987-phonetic.yaml +0 -780
  284. data/maps/un-ell-Grek-Latn-1987-tl.yaml +0 -31
  285. data/maps/un-ell-Grek-Latn-1987-ts.yaml +0 -19
  286. data/maps/un-hin-Deva-Latn-2016.yaml +0 -222
  287. data/maps/un-mar-Deva-Latn-2016.yaml +0 -91
  288. data/maps/un-mon-Mong-Latn-general-2013.yaml +0 -264
  289. data/maps/un-mon-Mong-Latn-phonetic-2013.yaml +0 -264
  290. data/maps/un-nep-Deva-Latn-1972.yaml +0 -350
  291. data/maps/un-nep-Deva-Latn-2013.yaml +0 -74
  292. data/maps/un-rus-Cyrl-Latn-1987.yaml +0 -166
  293. data/maps/un-ukr-Cyrl-Latn-1998.yaml +0 -53
  294. data/maps/un-ukr-Cyrl-Latn-2012.yaml +0 -162
  295. data/maps/var-hin-Deva-Latn-hunterian-1872.yaml +0 -221
  296. data/maps/var-jpn-Hrkt-Latn-hepburn-1886.yaml +0 -406
  297. data/maps/var-jpn-Hrkt-Latn-hepburn-1954.yaml +0 -386
  298. data/maps/var-kor-Hang-Hang-jamo.yaml +0 -11193
  299. data/maps/var-kor-Hang-Latn-mr-1939.yaml +0 -1054
  300. data/maps/var-kor-Kore-Hang-2013.yaml +0 -59754
  301. data/maps/var-kor-Kore-Latn-mr-1939.yaml +0 -36
  302. data/maps/var-mar-Deva-Latn-hunterian-1872.yaml +0 -43
  303. data/maps/var-mon-Mong-Latn-1930.yaml +0 -102
  304. data/maps/var-mon-Mong-Latn-lessing.yaml +0 -272
  305. data/maps/var-mon-Mong-Latn-vpmc.yaml +0 -274
  306. data/maps/var-pra-Deva-Latn-iast-1912.yaml +0 -30
  307. data/maps/var-san-Deva-Latn-iast-1912.yaml +0 -149
  308. data/maps/var-tha-Thai-Thai-phonemic.yaml +0 -59
  309. data/maps/var-tha-Thai-Zsym-ipa.yaml +0 -301
  310. data/maps/var-zho-Hani-Latn-wd-1979.yaml +0 -38912
  311. data/spec/interscript/filenames_spec.rb +0 -384
  312. data/spec/interscript/mapping_spec.rb +0 -42
  313. data/spec/interscript_spec.rb +0 -29
  314. data/spec/spec_helper.rb +0 -3
@@ -1,130 +0,0 @@
1
- ---
2
- authority_id: alalc
3
- id: 2017
4
- language: iso-639-2:ben
5
- source_script: Beng
6
- destination_script: Latn
7
- name: Bengali Romanization, 2017
8
- url: https://www.loc.gov/catdir/cpso/romanization/bengali.pdf
9
- creation_date: 2017
10
- description: |
11
- ALA-LC Romanization table for Bengali
12
-
13
- notes:
14
-
15
- - Only the vowel forms that appear at the beginning of a syllable are
16
- listed; the forms used for vowels following a consonant can be found in
17
- grammars; no distinction between the two is made in transliteration.
18
-
19
- - |
20
- The vowel a is implicit after all consonants and consonant clusters
21
- and is supplied in transliteration, with the following exceptions:
22
-
23
- a) when another vowel is indicated by its appropriate sign; and
24
- b) when the absence of any vowel is indicated by the subscript symbol ( ্ )
25
- called hasanta or birāma.
26
-
27
- - ব is used both as a labial and as a semivowel. When it occurs as the
28
- second or subsequent consonant of a consonant cluster, it is
29
- transliterated va. When ব is doubled, it is transliterated bba.
30
-
31
- - Candrabindu before guttural, palatal, cerebral, and dental occlusives
32
- is transliterated n̐. Before labials, sibilants, semivowels, the
33
- aspirate, vowels, and in final position it is transliterated m̐.
34
-
35
- - When doubled, abagraha is transliterated by two apostrophes ( ’’ ).
36
-
37
- tests:
38
- - source: "র্ক"
39
- expected: "rka"
40
- - source: "গ্র"
41
- expected: "gra"
42
- - source: "ত্য"
43
- expected: "tya"
44
-
45
- map:
46
- inherit: "un-ben-Beng-Latn-2016"
47
-
48
- characters:
49
-
50
- # I. Vowels and Diphthongs (see Note 1)
51
-
52
- 'অ': 'a'
53
- 'ৠ': 'ṝ'
54
- 'আ': 'ā'
55
- 'ঌ': 'ḹ'
56
- 'ই': 'I'
57
- 'এ': 'e'
58
- 'ঈ': 'ī'
59
- 'ঐ': 'ai'
60
- 'উ': 'u'
61
- 'ও': 'o'
62
- 'ঊ': 'ū'
63
- 'ঔ': 'au'
64
- 'ঋ': 'ṛ'
65
-
66
- # II. Consonants (see Note 2)
67
- # Gutturals
68
- 'ক': 'ka'
69
- 'খ': 'kha'
70
- 'গ': 'ga'
71
- 'ঘ': 'gha'
72
- 'ঙ': 'ṅa'
73
-
74
- # Palatals
75
- 'চ': 'ca'
76
- 'ছ': 'cha'
77
- 'জ': 'ja'
78
- 'ঝ': 'jha'
79
- 'ঞ': 'ña'
80
-
81
- # Cerebrals
82
- 'ট': 'ṭa'
83
- 'ঠ': 'ṭha'
84
- 'ড': 'ḍa'
85
- 'ড়': 'ṛa'
86
- 'ঢ': 'ḍha'
87
- 'ঢ়': 'ṛha'
88
- 'ণ': 'ṇa'
89
-
90
- # Dentals
91
- 'ত': 'ta'
92
- 'ৎ': 't'
93
- 'থ': 'tha'
94
- 'দ': 'da'
95
- 'ধ': 'dha'
96
- 'ন': 'na'
97
-
98
- # Labials
99
- 'প': 'pa'
100
- 'ফ': 'pha'
101
- 'ব': 'ba' # see Note 3
102
- 'ভ': 'bha'
103
- 'ম': 'ma'
104
-
105
- # Semivowels
106
- 'য': 'ya'
107
- 'য়': 'ẏa'
108
- 'র': 'ra'
109
- 'ল': 'la'
110
- 'ব': 'ba' # see Note 3
111
-
112
- # Sibilants
113
- 'শ': 'śa'
114
- 'ষ': 'sha'
115
- 'স': 'sa'
116
-
117
- # Aspirate
118
- 'হ': 'ha'
119
-
120
- # Anusvāra
121
- ' ং': 'ṃ'
122
-
123
- # Bisarga
124
- ' ঃ': 'ḥ'
125
-
126
- # Candrabindu (anunāsika) see note 4
127
- '\u0981': 'n̐' # ঁ : n̐, m̐
128
-
129
- # Abagraha (see Note 5)
130
- 'ऽ': '’' # (apostrophe)
@@ -1,98 +0,0 @@
1
- ---
2
- authority_id: alalc
3
- id: 1997
4
- language: iso-639-2:bul
5
- source_script: Cyrl
6
- destination_script: Latn
7
- name: ALA-LC Romanization Table -- Bulgarian (1997)
8
- alias:
9
- ogc11122:
10
- code: bul_Cyrl2Latn_ALA_1997
11
- description: Bulgarian ALA-Library of Congress Bulgarian 1997 System
12
- url: http://www.rechtertie.nl/databases/judd/downloads/Bulgarian.pdf
13
- creation_date: 1997
14
-
15
- notes:
16
- - |
17
- The letter Ъ in the initial and medial position is a vowel in Bulgarian and is romanized as ŭ.
18
- It appears in the final position chiefly in texts published before 1945, in which it serves as
19
- a hard sign and is romanized as ʺ.
20
- - The letters Ѣ and Ѫ are found chiefly in texts published before the orthographic reform of 1945.
21
-
22
- tests:
23
- # missing
24
-
25
- map:
26
- rules:
27
- # note[1]
28
- - pattern: (?<=)\u042a(?=\b)
29
- result: "\u02BA"
30
- - pattern: (?<=)u044a(?=\b)
31
- result: "\u02BA"
32
-
33
- characters:
34
- '\u0410': 'A'
35
- '\u0411': 'B'
36
- '\u0412': 'V'
37
- '\u0413': 'G'
38
- '\u0414': 'D'
39
- '\u0415': 'E'
40
- '\u0416': 'Zh'
41
- '\u0417': 'Z'
42
- '\u0418': 'I'
43
- '\u0419': "I\u0306" # Ĭ
44
- '\u041a': 'K'
45
- '\u041b': 'L'
46
- '\u041c': 'M'
47
- '\u041d': 'N'
48
- '\u041e': 'O'
49
- '\u041f': 'P'
50
- '\u0420': 'R'
51
- '\u0421': 'S'
52
- '\u0422': 'T'
53
- '\u0423': 'U'
54
- '\u0424': 'F'
55
- '\u0425': 'Kh'
56
- '\u0426': "T\u0361S"
57
- '\u0427': 'Ch'
58
- '\u0428': 'Sh'
59
- '\u0429': 'Sht'
60
- '\u042a': "U\u0306" # or ʺ check note[1]
61
- '\u042c': '\u02B9'
62
- '\u0462': "I\u0361E" # Ѣ check note [2]
63
- '\u042e': "I\u0361U"
64
- '\u042f': "I\u0361A"
65
- '\u046A': "U\u0310" # Ѫ check note [2]
66
-
67
- '\u0430': 'a'
68
- '\u0431': 'b'
69
- '\u0432': 'v'
70
- '\u0433': 'g'
71
- '\u0434': 'd'
72
- '\u0435': 'e'
73
- '\u0436': 'zh'
74
- '\u0437': 'z'
75
- '\u0438': 'i'
76
- '\u0439': "i\u0306" # ĭ
77
- '\u043a': 'k'
78
- '\u043b': 'l'
79
- '\u043c': 'm'
80
- '\u043d': 'n'
81
- '\u043e': 'o'
82
- '\u043f': 'p'
83
- '\u0440': 'r'
84
- '\u0441': 's'
85
- '\u0442': 't'
86
- '\u0443': 'u'
87
- '\u0444': 'f'
88
- '\u0445': 'kh'
89
- '\u0446': "t\u0361s"
90
- '\u0447': 'ch'
91
- '\u0448': 'sh'
92
- '\u0449': 'sht'
93
- '\u044a': "u\u0306" # or ʺ check note[1]
94
- '\u044c': '\u02B9'
95
- '\u0463': "i\u0361e" # ѣ check note [2]
96
- '\u044e': "i\u0361u"
97
- '\u044f': "i\u0361a"
98
- '\u046B': "u\u0310" # ѫ check note [2]
@@ -1,628 +0,0 @@
1
- ---
2
- authority_id: alalc
3
- id: 1997
4
- language: iso-639-2:ell
5
- source_script: Grek
6
- destination_script: Latn
7
- name: Greek Romanization, 1997
8
- alias:
9
- ogc11122:
10
- code: ell_Grek2Latn_ALA_1997
11
- description: Greek ALA-Library of Congress 1997 System
12
- url: http://catdir.loc.gov/catdir/cpso/romanization/greek.pdf
13
- creation_date: 1997
14
- description: |
15
- ALA-LC Romanization table for Greek
16
-
17
- note:
18
- - Applies to texts after 1453 (Modern Greek)
19
- - This table presupposes monotonic accentuation; rough breathings are accordingly not addressed.
20
- - The diphthong definition for upsilon is taken from the 2010 version
21
- - Generalised gramma digraph rule to capitals
22
-
23
- tests:
24
-
25
- - source: |
26
- Ένα πράμα μόνον με παρακίνησε κι εμένα να γράψω ότι τούτην την πατρίδα την έχομεν όλοι μαζί, και σοφοί κι αμαθείς και πλούσιοι και φτωχοί και πολιτικοί και στρατιωτικοί και οι πλέον μικρότεροι άνθρωποι· όσοι αγωνιστήκαμεν, αναλόγως ο καθείς, έχομεν να ζήσομεν εδώ. Το λοιπόν δουλέψαμεν όλοι μαζί, να την φυλάμεν κι όλοι μαζί και να μην λέγει ούτε ο δυνατός «εγώ» ούτε ο αδύνατος. Ξέρετε πότε να λέγει ο καθείς «εγώ»; Όταν αγωνιστεί μόνος του και φκιάσει ή χαλάσει, να λέγει «εγώ»· όταν όμως αγωνίζονται πολλοί και φκιάνουν, τότε να λένε «εμείς». Είμαστε εις το «εμείς» κι όχι εις το «εγώ». Και εις το εξής να μάθομεν γνώση, αν θέλομεν να φκιάσομεν χωριόν, να ζήσομεν όλοι μαζί.
27
-
28
- Γιάννης Μακρυγιάννης.
29
-
30
- expected: |
31
- Ena prama monon me parakinēse ki emena na grapsō oti toutēn tēn patrida tēn echomen oloi mazi, kai sophoi ki amatheis kai plousioi kai phtōchoi kai politikoi kai stratiōtikoi kai oi pleon mikroteroi anthrōpoi; osoi agōnistēkamen, analogōs o katheis, echomen na zēsomen edō. To loipon doulepsamen oloi mazi, na tēn phylamen ki oloi mazi kai na mēn legei oute o dynatos «egō» oute o adynatos. Xerete pote na legei o katheis «egō»? Otan agōnistei monos tou kai phkiasei ē chalasei, na legei «egō»; otan omōs agōnizontai polloi kai phkianoun, tote na lene «emeis». Eimaste eis to «emeis» ki ochi eis to «egō». Kai eis to exēs na mathomen gnōsē, an thelomen na phkiasomen chōrion, na zēsomen oloi mazi.
32
-
33
- Giannēs Makrygiannēs.
34
-
35
-
36
- - source: ΑΘΗΝΑ
37
- expected: ATHĒNA
38
- - source: μπαμπάκι
39
- expected: bampaki
40
- - source: νταντά
41
- expected: ḏanta
42
- - source: γκέγκε
43
- expected: nkenke
44
- - source: Γκαμπόν
45
- expected: Nkampon
46
- - source: Μάγχη
47
- expected: Manchē
48
- - source: κογξ
49
- expected: konx
50
- - source: υιός
51
- expected: uios
52
- - source: Υιός
53
- expected: Uios
54
- - source: νεράντζι
55
- expected: nerantzi
56
- - source: Γοίθιος
57
- expected: Goithios
58
- - source: μπέικον
59
- expected: beikon
60
- - source: μπέϊκον
61
- expected: beikon
62
- - source: βόλεϊ
63
- expected: volei
64
- - source: αθεΐα
65
- expected: atheia
66
- - source: Εϊγιαφιάτλαγιοκουτλ
67
- expected: Eigiaphiatlagiokoutl
68
- - source: Εΐτζι
69
- expected: Eitzi
70
- - source: Μυρτώο
71
- expected: Myrtōo
72
- - source: αέρας
73
- expected: aeras
74
- - source: γαυ γαυ
75
- expected: gau gau
76
- - source: Ταΰγετος
77
- expected: Taygetos
78
- - source: σπρέυ
79
- expected: sprey
80
-
81
- - source: Αθήνα
82
- expected: Athēna
83
- - source: Άγιον Όρος
84
- expected: Agion Oros
85
- - source: Άγραφα
86
- expected: Agrapha
87
- - source: Αγρίνιο
88
- expected: Agrinio
89
- - source: Αίγινα
90
- expected: Aigina
91
- - source: Αίγιο
92
- expected: Aigio
93
- - source: Αλεξανδρούπολη
94
- expected: Alexandroupolē
95
- - source: Αλεποχώρι
96
- expected: Alepochōri
97
- - source: Αμοργός
98
- expected: Amorgos
99
- - source: Άμφισσα
100
- expected: Amphissa
101
- - source: Αράχωβα
102
- expected: Arachōva
103
- - source: Άργος
104
- expected: Argos
105
- - source: Αρκαδία
106
- expected: Arkadia
107
- - source: Άρτα
108
- expected: Arta
109
- - source: Βελούχι
110
- expected: Velouchi
111
- - source: Βέροια
112
- expected: Veroia
113
- - source: Βοιωτία
114
- expected: Voiōtia
115
- - source: Βόλος
116
- expected: Volos
117
- - source: Βόνιτσα
118
- expected: Vonitsa
119
- - source: Γαλαξίδι
120
- expected: Galaxidi
121
- - source: Γαλάτσι
122
- expected: Galatsi
123
- - source: Γιαννιτσά
124
- expected: Giannitsa
125
- - source: Γλυφάδα
126
- expected: Glyphada
127
- - source: Γρανίτσα
128
- expected: Granitsa
129
- - source: Γρεβενά
130
- expected: Grevena
131
- - source: Γύθειο
132
- expected: Gytheio
133
- - source: Διόνυσος
134
- expected: Dionysos
135
- - source: Δίστομο
136
- expected: Distomo
137
- - source: Δολιανά
138
- expected: Doliana
139
- - source: Δράμα
140
- expected: Drama
141
- - source: Δωδεκάνησα
142
- expected: Dōdekanēsa
143
- - source: Έδεσσα
144
- expected: Edessa
145
- - source: Ελευσίνα
146
- expected: Eleusina
147
- - source: Επίδαυρος
148
- expected: Epidauros
149
- - source: Επτάνησα
150
- expected: Eptanēsa
151
- - source: Ερμούπολη
152
- expected: Ermoupolē
153
- - source: Εύβοια
154
- expected: Euvoia
155
- - source: Ζάκυνθος
156
- expected: Zakynthos
157
- - source: Ήπειρος
158
- expected: Ēpeiros
159
- - source: Ηράκλειο
160
- expected: Ērakleio
161
- - source: Θάσος
162
- expected: Thasos
163
- - source: Θεσσαλονίκη
164
- expected: Thessalonikē
165
- - source: Θεσσαλία
166
- expected: Thessalia
167
- - source: Θεσπρωτία
168
- expected: Thesprōtia
169
- - source: Θήβα
170
- expected: Thēva
171
- - source: Θράκη
172
- expected: Thrakē
173
- - source: Ιθάκη
174
- expected: Ithakē
175
- - source: Ίος
176
- expected: Ios
177
- - source: Ιωάννινα
178
- expected: Iōannina
179
- - source: Καβάλα
180
- expected: Kavala
181
- - source: Καλάβρυτα
182
- expected: Kalavryta
183
- - source: Καλαμάτα
184
- expected: Kalamata
185
- - source: Καλαμπάκα
186
- expected: Kalampaka
187
- - source: Καλύβια
188
- expected: Kalyvia
189
- - source: Κάλυμνος
190
- expected: Kalymnos
191
- - source: Καρδίτσα
192
- expected: Karditsa
193
- - source: Καρπενήσι
194
- expected: Karpenēsi
195
- - source: Κάρυστος
196
- expected: Karystos
197
- - source: Καστελλόριζο
198
- expected: Kastellorizo
199
- - source: Καστοριά
200
- expected: Kastoria
201
- - source: Κατερίνη
202
- expected: Katerinē
203
- - source: Κάτω Αχαΐα
204
- expected: Katō Achaia
205
- - source: Κερατέα
206
- expected: Keratea
207
- - source: Κέρκυρα
208
- expected: Kerkyra
209
- - source: Κεφαλλονιά
210
- expected: Kephallonia
211
- - source: Κηφισιά
212
- expected: Kēphisia
213
- - source: Κιλκίς
214
- expected: Kilkis
215
- - source: Κοζάνη
216
- expected: Kozanē
217
- - source: Κολωνός
218
- expected: Kolōnos
219
- - source: Κομοτηνή
220
- expected: Komotēnē
221
- - source: Κόρινθος
222
- expected: Korinthos
223
- - source: Κορώνη
224
- expected: Korōnē
225
- - source: Κρανίδι
226
- expected: Kranidi
227
- - source: Κρέστενα
228
- expected: Krestena
229
- - source: Κρήτη
230
- expected: Krētē
231
- - source: Κύθηρα
232
- expected: Kythēra
233
- - source: Κυκλάδες
234
- expected: Kyklades
235
- - source: Κύμη
236
- expected: Kymē
237
- - source: Κυψέλη
238
- expected: Kypselē
239
- - source: Κως
240
- expected: Kōs
241
- - source: Λαγκαδάς
242
- expected: Lankadas
243
- - source: Λαμία
244
- expected: Lamia
245
- - source: Λάρισα
246
- expected: Larisa
247
- - source: Λαύριο
248
- expected: Laurio
249
- - source: Λέρος
250
- expected: Leros
251
- - source: Λέσβος
252
- expected: Lesvos
253
- - source: Λευκάδα
254
- expected: Leukada
255
- - source: Λήμνος
256
- expected: Lēmnos
257
- - source: Λιβαδειά
258
- expected: Livadeia
259
- - source: Μακεδονία
260
- expected: Makedonia
261
- - source: Μάνη
262
- expected: Manē
263
- - source: Μαραθώνας
264
- expected: Marathōnas
265
- - source: Μαρκόπουλο
266
- expected: Markopoulo
267
- - source: Μαρούσι
268
- expected: Marousi
269
- - source: Μέγαρα
270
- expected: Megara
271
- - source: Μεσολόγγι
272
- expected: Mesolongi
273
- - source: Μεταξουργείο
274
- expected: Metaxourgeio
275
- - source: Μέτσοβο
276
- expected: Metsovo
277
- - source: Μήλος
278
- expected: Mēlos
279
- - source: Μύκονος
280
- expected: Mykonos
281
- - source: Μυστράς
282
- expected: Mystras
283
- - source: Μυτιλήνη
284
- expected: Mytilēnē
285
- - source: Νάξος
286
- expected: Naxos
287
- - source: Νάουσα
288
- expected: Naousa
289
- - source: Ναύπακτος
290
- expected: Naupaktos
291
- - source: Ναύπλιο
292
- expected: Nauplio
293
- - source: Νέα Σμύρνη
294
- expected: Nea Smyrnē
295
- - source: Νίσυρος
296
- expected: Nisyros
297
- - source: Ξάνθη
298
- expected: Xanthē
299
- - source: Όλυμπος
300
- expected: Olympos
301
- - source: Παγκράτι
302
- expected: Pankrati
303
- - source: Παπάγου
304
- expected: Papagou
305
- - source: Πάρος
306
- expected: Paros
307
- - source: Πασαλιμάνι
308
- expected: Pasalimani
309
- - source: Πατήσια
310
- expected: Patēsia
311
- - source: Πάτμος
312
- expected: Patmos
313
- - source: Πάτρα
314
- expected: Patra
315
- - source: Πειραιάς
316
- expected: Peiraias
317
- - source: Πελοπόννησος
318
- expected: Peloponnēsos
319
- - source: Περιστέρι
320
- expected: Peristeri
321
- - source: Πεύκη
322
- expected: Peukē
323
- - source: Πήλιο
324
- expected: Pēlio
325
- - source: Πολύγυρος
326
- expected: Polygyros
327
- - source: Πόρος
328
- expected: Poros
329
- - source: Πρέβεζα
330
- expected: Preveza
331
- - source: Πτολεμαΐδα
332
- expected: Ptolemaida
333
- - source: Πύλος
334
- expected: Pylos
335
- - source: Πύργος
336
- expected: Pyrgos
337
- - source: Ρέθυμνο
338
- expected: Rethymno
339
- - source: Ρόδος
340
- expected: Rodos
341
- - source: Ρούμελη
342
- expected: Roumelē
343
- - source: Σαλαμίνα
344
- expected: Salamina
345
- - source: Σαμοθράκη
346
- expected: Samothrakē
347
- - source: Σάμος
348
- expected: Samos
349
- - source: Σαντορίνη
350
- expected: Santorinē
351
- - source: Σέρρες
352
- expected: Serres
353
- - source: Σίκινος
354
- expected: Sikinos
355
- - source: Σίφνος
356
- expected: Siphnos
357
- - source: Σκιάθος
358
- expected: Skiathos
359
- - source: Σκόπελος
360
- expected: Skopelos
361
- - source: Σούλι
362
- expected: Souli
363
- - source: Σπάρτη
364
- expected: Spartē
365
- - source: Στερεά Ελλάδα
366
- expected: Sterea Ellada
367
- - source: Στύρα
368
- expected: Styra
369
- - source: Σύμη
370
- expected: Symē
371
- - source: Σύρος
372
- expected: Syros
373
- - source: Σφακιά
374
- expected: Sphakia
375
- - source: Τήλος
376
- expected: Tēlos
377
- - source: Τήνος
378
- expected: Tēnos
379
- - source: Τρίκαλα
380
- expected: Trikala
381
- - source: Τρίπολη
382
- expected: Tripolē
383
- - source: Τσακωνιά
384
- expected: Tsakōnia
385
- - source: Ύδρα
386
- expected: Ydra
387
- - source: Φάληρο
388
- expected: Phalēro
389
- - source: Φλώρινα
390
- expected: Phlōrina
391
- - source: Φολέγανδρος
392
- expected: Pholegandros
393
- - source: Χάλκη
394
- expected: Chalkē
395
- - source: Χαλκίδα
396
- expected: Chalkida
397
- - source: Χαλάνδρι
398
- expected: Chalandri
399
- - source: Χαλκιδική
400
- expected: Chalkidikē
401
- - source: Χανιά
402
- expected: Chania
403
- - source: Χίος
404
- expected: Chios
405
- - source: Ψαρά
406
- expected: Psara
407
- - source: Αβάνα
408
- expected: Avana
409
- - source: Αγγλία
410
- expected: Anglia
411
- - source: Αϊβαλί
412
- expected: Aivali
413
- - source: Αλεξάνδρεια
414
- expected: Alexandreia
415
- - source: Άμστερνταμ
416
- expected: Amsterntam
417
- - source: Βαυαρία
418
- expected: Vauaria
419
- - source: Βενετία
420
- expected: Venetia
421
- - source: Βερολίνο
422
- expected: Verolino
423
- - source: Βερόνα
424
- expected: Verona
425
- - source: Βιέννη
426
- expected: Viennē
427
- - source: Γένοβα
428
- expected: Genova
429
- - source: Δουβλίνο
430
- expected: Douvlino
431
- - source: Καλαβρία
432
- expected: Kalavria
433
- - source: Καλιφόρνια
434
- expected: Kaliphornia
435
- - source: Καύκασος
436
- expected: Kaukasos
437
- - source: Κονγκό
438
- expected: Konnko
439
- - source: Κορσική
440
- expected: Korsikē
441
- - source: Κουρδιστάν
442
- expected: Kourdistan
443
- - source: Κωνσταντινούπολη
444
- expected: Kōnstantinoupolē
445
- - source: Κατεχόμενη Κύπρος
446
- expected: Katechomenē Kypros
447
- - source: Λαπωνία
448
- expected: Lapōnia
449
- - source: Λευκωσία
450
- expected: Leukōsia
451
- - source: Λιβόρνο
452
- expected: Livorno
453
- - source: Λονδίνο
454
- expected: Londino
455
- - source: Λυών
456
- expected: Lyōn
457
- - source: Μάλαγα
458
- expected: Malaga
459
- - source: Μασσαλία
460
- expected: Massalia
461
- - source: Μικρονησία
462
- expected: Mikronēsia
463
- - source: Μιλάνο
464
- expected: Milano
465
- - source: Μόσχα
466
- expected: Moscha
467
- - source: Μπολόνια
468
- expected: Bolonia
469
- - source: Νάπολη
470
- expected: Napolē
471
- - source: Νταγκεστάν
472
- expected: Ḏankestan
473
- - source: Νέα Υόρκη
474
- expected: Nea Yorkē
475
- - source: Οξφόρδη
476
- expected: Oxphordē
477
- - source: Ουαλία
478
- expected: Oualia
479
- - source: Παρίσι
480
- expected: Parisi
481
- - source: Πάφος
482
- expected: Paphos
483
- - source: Πολυνησία
484
- expected: Polynēsia
485
- - source: Ρώμη
486
- expected: Rōmē
487
- - source: Σαμάρεια
488
- expected: Samareia
489
- - source: Σικελία
490
- expected: Sikelia
491
- - source: Σκανδιναβία
492
- expected: Skandinavia
493
- - source: Σκόπια
494
- expected: Skopia
495
- - source: Σκωτία
496
- expected: Skōtia
497
- - source: Σμύρνη
498
- expected: Smyrnē
499
- - source: Ταϊτή
500
- expected: Taitē
501
- - source: Ταταρστάν
502
- expected: Tatarstan
503
- - source: Τζαμάικα
504
- expected: Tzamaika
505
- - source: Τηλλυρία
506
- expected: Tēllyria
507
- - source: Τιρόλο
508
- expected: Tirolo
509
- - source: Τορίνο
510
- expected: Torino
511
- - source: Φανάρι
512
- expected: Phanari
513
- - source: Φλωρεντία
514
- expected: Phlōrentia
515
- - source: Χαβάη
516
- expected: Chavaē
517
- - source: Χονγκ Κονγκ
518
- expected: Chonnk Konnk
519
-
520
- map:
521
- # https://en.wikipedia.org/wiki/Romanization_of_Greek
522
- rules:
523
- - pattern: (?<=[ΑαΕεΟοΗηΩω])\u03A5 # Υ (after Α, Ε, Ο, Η, Ω)
524
- result: U
525
- - pattern: (?<=[ΑαΕεΟοΗηΩω])\u03C5 # υ (after Α, Ε, Ο, Η, Ω)
526
- result: u
527
- - pattern: (?<=[ΑαΕεΟοΗηΩω])\u03CD # ύ (after Α, Ε, Ο, Η, Ω)
528
- result: u
529
- - pattern: \u03A5(?=[Ιιί]) # Υ (before Ι)
530
- result: U
531
- - pattern: \u03C5(?=[Ιιί]) # Υ (before Ι)
532
- result: u
533
- - pattern: \u0393(?=[γΓκΚξΞχΧ]) # Γ (before Γ, Κ, Ξ, Χ)
534
- result: N
535
- - pattern: \u03B3(?=[γΓκΚξΞχΧ]) # γ (before Γ, Κ, Ξ, Χ)
536
- result: n
537
- - pattern: (?<=\b)\u039D\u03A4 # ΝΤ
538
- result: Ḏ
539
- - pattern: (?<=\b)\u039D\u03C4 # Ντ
540
- result: Ḏ
541
- - pattern: (?<=\b)\u03BD\u03C4 # ντ
542
- result: ḏ
543
- - pattern: (?<=\b)\u039C\u03A0 # ΜΠ
544
- result: B
545
- - pattern: (?<=\b)\u039C\u03C0 # Μπ
546
- result: B
547
- - pattern: (?<=\b)\u03BC\u03C0 # μπ
548
- result: b
549
- - pattern: \u037E # ;
550
- result: "?"
551
- - pattern: \u003B # ;
552
- result: "?"
553
-
554
- characters:
555
- "\u0027": ""
556
- "\u0386": "A" # Ά
557
- "\u0391": "A" # Α
558
- "\u0392": "V" # Β
559
- "\u0393": "G" # Γ
560
- "\u0394": "D" # Δ
561
- "\u0395": "E" # Ε
562
- "\u0396": "Z" # Ζ
563
- "\u0397": "Ē" # Η
564
- "\u0398": "Th" # Θ
565
- "\u0399": "I" # Ι
566
- "\u039A": "K" # Κ
567
- "\u039B": "L" # Λ
568
- "\u039C": "M" # Μ
569
- "\u039D": "N" # Ν
570
- "\u039E": "X" # Ξ
571
- "\u039F": "O" # Ο
572
- "\u03A0": "P" # Π
573
- "\u03A1": "R" # Ρ
574
- "\u03A3": "S" # Σ
575
- "\u03A4": "T" # Τ
576
- "\u03A5": "Y" # Υ
577
- "\u03A6": "Ph" # Φ
578
- "\u03A7": "Ch" # Χ
579
- "\u03A8": "Ps" # Ψ
580
- "\u03A9": "Ō" # Ω
581
- "\u0388": "E" # Έ
582
- "\u0389": "Ē" # Ή
583
- "\u038A": "I" # Ί
584
- "\u038C": "O" # Ό
585
- "\u038E": "Y" # Ύ
586
- "\u038F": "Ō" # Ώ
587
- "\u03AA": "I" # Ϊ
588
- "\u03AB": "Y" # Ϋ
589
-
590
- "\u03AC": "a" # ά
591
- "\u03B1": "a" # α
592
- "\u03B2": "v" # β
593
- "\u03B3": "g" # γ
594
- "\u03B4": "d" # δ
595
- "\u03B5": "e" # ε
596
- "\u03B6": "z" # ζ
597
- "\u03B7": "ē" # η
598
- "\u03B8": "th" # θ
599
- "\u03B9": "i" # ι
600
- "\u03BA": "k" # κ
601
- "\u03BB": "l" # λ
602
- "\u03BC": "m" # μ
603
- "\u03BD": "n" # ν
604
- "\u03BE": "x" # ξ
605
- "\u03BF": "o" # ο
606
- "\u03C0": "p" # π
607
- "\u03C1": "r" # ρ
608
- "\u03C3": "s" # σ
609
- "\u03C2": "s" # ς
610
- "\u03C4": "t" # τ
611
- "\u03C5": "y" # υ
612
- "\u03C6": "ph" # φ
613
- "\u03C7": "ch" # χ
614
- "\u03C8": "ps" # ψ
615
- "\u03C9": "ō" # ω
616
- "\u03AD": "e" # έ
617
- "\u03AE": "ē" # ή
618
- "\u03AF": "i" # ί
619
- "\u03CC": "o" # ό
620
- "\u03CD": "y" # ύ
621
- "\u03CE": "ō" # ώ
622
- "\u03CA": "i" # ϊ
623
- "\u03CB": "y" # ϋ
624
- "\u0390": "i" # ΐ
625
- "\u03B0": "y" # ΰ
626
-
627
- "\u0387": ";" # ·
628
- "\u00B7": ";" # ·