PageRenderTime 50ms CodeModel.GetById 19ms RepoModel.GetById 1ms app.codeStats 0ms

/ime/latinime/jni/src/char_utils.cpp

http://eyes-free.googlecode.com/
C++ | 899 lines | 819 code | 7 blank | 73 comment | 0 complexity | 0da86fc3dce4f8a3a2f37bbe4e674f4e MD5 | raw file
Possible License(s): GPL-3.0, Apache-2.0
  1. /*
  2. * Copyright (C) 2010 The Android Open Source Project
  3. *
  4. * Licensed under the Apache License, Version 2.0 (the "License");
  5. * you may not use this file except in compliance with the License.
  6. * You may obtain a copy of the License at
  7. *
  8. * http://www.apache.org/licenses/LICENSE-2.0
  9. *
  10. * Unless required by applicable law or agreed to in writing, software
  11. * distributed under the License is distributed on an "AS IS" BASIS,
  12. * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13. * See the License for the specific language governing permissions and
  14. * limitations under the License.
  15. */
  16. #include <stdlib.h>
  17. namespace latinime {
  18. struct LatinCapitalSmallPair {
  19. unsigned short capital;
  20. unsigned short small;
  21. };
  22. // Generated from http://unicode.org/Public/UNIDATA/UnicodeData.txt
  23. //
  24. // 1. Run the following code. Bascially taken from
  25. // Dictionary::toLowerCase(unsigned short c) in dictionary.cpp.
  26. // Then, get the list of chars where cc != ccc.
  27. //
  28. // unsigned short c, cc, ccc, ccc2;
  29. // for (c = 0; c < 0xFFFF ; c++) {
  30. // if (c < sizeof(BASE_CHARS) / sizeof(BASE_CHARS[0])) {
  31. // cc = BASE_CHARS[c];
  32. // } else {
  33. // cc = c;
  34. // }
  35. //
  36. // // tolower
  37. // int isBase = 0;
  38. // if (cc >='A' && cc <= 'Z') {
  39. // ccc = (cc | 0x20);
  40. // ccc2 = ccc;
  41. // isBase = 1;
  42. // } else if (cc > 0x7F) {
  43. // ccc = u_tolower(cc);
  44. // ccc2 = latin_tolower(cc);
  45. // } else {
  46. // ccc = cc;
  47. // ccc2 = ccc;
  48. // }
  49. // if (!isBase && cc != ccc) {
  50. // wprintf(L" 0x%04X => 0x%04X => 0x%04X %lc => %lc => %lc \n",
  51. // c, cc, ccc, c, cc, ccc);
  52. // //assert(ccc == ccc2);
  53. // }
  54. // }
  55. //
  56. // Initially, started with an empty latin_tolower() as below.
  57. //
  58. // unsigned short latin_tolower(unsigned short c) {
  59. // return c;
  60. // }
  61. //
  62. //
  63. // 2. Process the list obtained by 1 by the following perl script and apply
  64. // 'sort -u' as well. Get the SORTED_CHAR_MAP[].
  65. // Note that '$1' in the perl script is 'cc' in the above C code.
  66. //
  67. // while(<>) {
  68. // / 0x\w* => 0x(\w*) =/;
  69. // open(HDL, "grep -iw ^" . $1 . " UnicodeData.txt | ");
  70. // $line = <HDL>;
  71. // chomp $line;
  72. // @cols = split(/;/, $line);
  73. // print " { 0x$1, 0x$cols[13] }, // $cols[1]\n";
  74. // }
  75. //
  76. //
  77. // 3. Update the latin_tolower() function above with SORTED_CHAR_MAP. Enable
  78. // the assert(ccc == ccc2) above and confirm the function exits successfully.
  79. //
  80. static const struct LatinCapitalSmallPair SORTED_CHAR_MAP[] = {
  81. { 0x00C4, 0x00E4 }, // LATIN CAPITAL LETTER A WITH DIAERESIS
  82. { 0x00C5, 0x00E5 }, // LATIN CAPITAL LETTER A WITH RING ABOVE
  83. { 0x00C6, 0x00E6 }, // LATIN CAPITAL LETTER AE
  84. { 0x00D0, 0x00F0 }, // LATIN CAPITAL LETTER ETH
  85. { 0x00D5, 0x00F5 }, // LATIN CAPITAL LETTER O WITH TILDE
  86. { 0x00D6, 0x00F6 }, // LATIN CAPITAL LETTER O WITH DIAERESIS
  87. { 0x00D8, 0x00F8 }, // LATIN CAPITAL LETTER O WITH STROKE
  88. { 0x00DC, 0x00FC }, // LATIN CAPITAL LETTER U WITH DIAERESIS
  89. { 0x00DE, 0x00FE }, // LATIN CAPITAL LETTER THORN
  90. { 0x0110, 0x0111 }, // LATIN CAPITAL LETTER D WITH STROKE
  91. { 0x0126, 0x0127 }, // LATIN CAPITAL LETTER H WITH STROKE
  92. { 0x0141, 0x0142 }, // LATIN CAPITAL LETTER L WITH STROKE
  93. { 0x014A, 0x014B }, // LATIN CAPITAL LETTER ENG
  94. { 0x0152, 0x0153 }, // LATIN CAPITAL LIGATURE OE
  95. { 0x0166, 0x0167 }, // LATIN CAPITAL LETTER T WITH STROKE
  96. { 0x0181, 0x0253 }, // LATIN CAPITAL LETTER B WITH HOOK
  97. { 0x0182, 0x0183 }, // LATIN CAPITAL LETTER B WITH TOPBAR
  98. { 0x0184, 0x0185 }, // LATIN CAPITAL LETTER TONE SIX
  99. { 0x0186, 0x0254 }, // LATIN CAPITAL LETTER OPEN O
  100. { 0x0187, 0x0188 }, // LATIN CAPITAL LETTER C WITH HOOK
  101. { 0x0189, 0x0256 }, // LATIN CAPITAL LETTER AFRICAN D
  102. { 0x018A, 0x0257 }, // LATIN CAPITAL LETTER D WITH HOOK
  103. { 0x018B, 0x018C }, // LATIN CAPITAL LETTER D WITH TOPBAR
  104. { 0x018E, 0x01DD }, // LATIN CAPITAL LETTER REVERSED E
  105. { 0x018F, 0x0259 }, // LATIN CAPITAL LETTER SCHWA
  106. { 0x0190, 0x025B }, // LATIN CAPITAL LETTER OPEN E
  107. { 0x0191, 0x0192 }, // LATIN CAPITAL LETTER F WITH HOOK
  108. { 0x0193, 0x0260 }, // LATIN CAPITAL LETTER G WITH HOOK
  109. { 0x0194, 0x0263 }, // LATIN CAPITAL LETTER GAMMA
  110. { 0x0196, 0x0269 }, // LATIN CAPITAL LETTER IOTA
  111. { 0x0197, 0x0268 }, // LATIN CAPITAL LETTER I WITH STROKE
  112. { 0x0198, 0x0199 }, // LATIN CAPITAL LETTER K WITH HOOK
  113. { 0x019C, 0x026F }, // LATIN CAPITAL LETTER TURNED M
  114. { 0x019D, 0x0272 }, // LATIN CAPITAL LETTER N WITH LEFT HOOK
  115. { 0x019F, 0x0275 }, // LATIN CAPITAL LETTER O WITH MIDDLE TILDE
  116. { 0x01A2, 0x01A3 }, // LATIN CAPITAL LETTER OI
  117. { 0x01A4, 0x01A5 }, // LATIN CAPITAL LETTER P WITH HOOK
  118. { 0x01A6, 0x0280 }, // LATIN LETTER YR
  119. { 0x01A7, 0x01A8 }, // LATIN CAPITAL LETTER TONE TWO
  120. { 0x01A9, 0x0283 }, // LATIN CAPITAL LETTER ESH
  121. { 0x01AC, 0x01AD }, // LATIN CAPITAL LETTER T WITH HOOK
  122. { 0x01AE, 0x0288 }, // LATIN CAPITAL LETTER T WITH RETROFLEX HOOK
  123. { 0x01B1, 0x028A }, // LATIN CAPITAL LETTER UPSILON
  124. { 0x01B2, 0x028B }, // LATIN CAPITAL LETTER V WITH HOOK
  125. { 0x01B3, 0x01B4 }, // LATIN CAPITAL LETTER Y WITH HOOK
  126. { 0x01B5, 0x01B6 }, // LATIN CAPITAL LETTER Z WITH STROKE
  127. { 0x01B7, 0x0292 }, // LATIN CAPITAL LETTER EZH
  128. { 0x01B8, 0x01B9 }, // LATIN CAPITAL LETTER EZH REVERSED
  129. { 0x01BC, 0x01BD }, // LATIN CAPITAL LETTER TONE FIVE
  130. { 0x01E4, 0x01E5 }, // LATIN CAPITAL LETTER G WITH STROKE
  131. { 0x01EA, 0x01EB }, // LATIN CAPITAL LETTER O WITH OGONEK
  132. { 0x01F6, 0x0195 }, // LATIN CAPITAL LETTER HWAIR
  133. { 0x01F7, 0x01BF }, // LATIN CAPITAL LETTER WYNN
  134. { 0x021C, 0x021D }, // LATIN CAPITAL LETTER YOGH
  135. { 0x0220, 0x019E }, // LATIN CAPITAL LETTER N WITH LONG RIGHT LEG
  136. { 0x0222, 0x0223 }, // LATIN CAPITAL LETTER OU
  137. { 0x0224, 0x0225 }, // LATIN CAPITAL LETTER Z WITH HOOK
  138. { 0x0226, 0x0227 }, // LATIN CAPITAL LETTER A WITH DOT ABOVE
  139. { 0x022E, 0x022F }, // LATIN CAPITAL LETTER O WITH DOT ABOVE
  140. { 0x023A, 0x2C65 }, // LATIN CAPITAL LETTER A WITH STROKE
  141. { 0x023B, 0x023C }, // LATIN CAPITAL LETTER C WITH STROKE
  142. { 0x023D, 0x019A }, // LATIN CAPITAL LETTER L WITH BAR
  143. { 0x023E, 0x2C66 }, // LATIN CAPITAL LETTER T WITH DIAGONAL STROKE
  144. { 0x0241, 0x0242 }, // LATIN CAPITAL LETTER GLOTTAL STOP
  145. { 0x0243, 0x0180 }, // LATIN CAPITAL LETTER B WITH STROKE
  146. { 0x0244, 0x0289 }, // LATIN CAPITAL LETTER U BAR
  147. { 0x0245, 0x028C }, // LATIN CAPITAL LETTER TURNED V
  148. { 0x0246, 0x0247 }, // LATIN CAPITAL LETTER E WITH STROKE
  149. { 0x0248, 0x0249 }, // LATIN CAPITAL LETTER J WITH STROKE
  150. { 0x024A, 0x024B }, // LATIN CAPITAL LETTER SMALL Q WITH HOOK TAIL
  151. { 0x024C, 0x024D }, // LATIN CAPITAL LETTER R WITH STROKE
  152. { 0x024E, 0x024F }, // LATIN CAPITAL LETTER Y WITH STROKE
  153. { 0x0370, 0x0371 }, // GREEK CAPITAL LETTER HETA
  154. { 0x0372, 0x0373 }, // GREEK CAPITAL LETTER ARCHAIC SAMPI
  155. { 0x0376, 0x0377 }, // GREEK CAPITAL LETTER PAMPHYLIAN DIGAMMA
  156. { 0x0391, 0x03B1 }, // GREEK CAPITAL LETTER ALPHA
  157. { 0x0392, 0x03B2 }, // GREEK CAPITAL LETTER BETA
  158. { 0x0393, 0x03B3 }, // GREEK CAPITAL LETTER GAMMA
  159. { 0x0394, 0x03B4 }, // GREEK CAPITAL LETTER DELTA
  160. { 0x0395, 0x03B5 }, // GREEK CAPITAL LETTER EPSILON
  161. { 0x0396, 0x03B6 }, // GREEK CAPITAL LETTER ZETA
  162. { 0x0397, 0x03B7 }, // GREEK CAPITAL LETTER ETA
  163. { 0x0398, 0x03B8 }, // GREEK CAPITAL LETTER THETA
  164. { 0x0399, 0x03B9 }, // GREEK CAPITAL LETTER IOTA
  165. { 0x039A, 0x03BA }, // GREEK CAPITAL LETTER KAPPA
  166. { 0x039B, 0x03BB }, // GREEK CAPITAL LETTER LAMDA
  167. { 0x039C, 0x03BC }, // GREEK CAPITAL LETTER MU
  168. { 0x039D, 0x03BD }, // GREEK CAPITAL LETTER NU
  169. { 0x039E, 0x03BE }, // GREEK CAPITAL LETTER XI
  170. { 0x039F, 0x03BF }, // GREEK CAPITAL LETTER OMICRON
  171. { 0x03A0, 0x03C0 }, // GREEK CAPITAL LETTER PI
  172. { 0x03A1, 0x03C1 }, // GREEK CAPITAL LETTER RHO
  173. { 0x03A3, 0x03C3 }, // GREEK CAPITAL LETTER SIGMA
  174. { 0x03A4, 0x03C4 }, // GREEK CAPITAL LETTER TAU
  175. { 0x03A5, 0x03C5 }, // GREEK CAPITAL LETTER UPSILON
  176. { 0x03A6, 0x03C6 }, // GREEK CAPITAL LETTER PHI
  177. { 0x03A7, 0x03C7 }, // GREEK CAPITAL LETTER CHI
  178. { 0x03A8, 0x03C8 }, // GREEK CAPITAL LETTER PSI
  179. { 0x03A9, 0x03C9 }, // GREEK CAPITAL LETTER OMEGA
  180. { 0x03CF, 0x03D7 }, // GREEK CAPITAL KAI SYMBOL
  181. { 0x03D8, 0x03D9 }, // GREEK LETTER ARCHAIC KOPPA
  182. { 0x03DA, 0x03DB }, // GREEK LETTER STIGMA
  183. { 0x03DC, 0x03DD }, // GREEK LETTER DIGAMMA
  184. { 0x03DE, 0x03DF }, // GREEK LETTER KOPPA
  185. { 0x03E0, 0x03E1 }, // GREEK LETTER SAMPI
  186. { 0x03E2, 0x03E3 }, // COPTIC CAPITAL LETTER SHEI
  187. { 0x03E4, 0x03E5 }, // COPTIC CAPITAL LETTER FEI
  188. { 0x03E6, 0x03E7 }, // COPTIC CAPITAL LETTER KHEI
  189. { 0x03E8, 0x03E9 }, // COPTIC CAPITAL LETTER HORI
  190. { 0x03EA, 0x03EB }, // COPTIC CAPITAL LETTER GANGIA
  191. { 0x03EC, 0x03ED }, // COPTIC CAPITAL LETTER SHIMA
  192. { 0x03EE, 0x03EF }, // COPTIC CAPITAL LETTER DEI
  193. { 0x03F7, 0x03F8 }, // GREEK CAPITAL LETTER SHO
  194. { 0x03FA, 0x03FB }, // GREEK CAPITAL LETTER SAN
  195. { 0x03FD, 0x037B }, // GREEK CAPITAL REVERSED LUNATE SIGMA SYMBOL
  196. { 0x03FE, 0x037C }, // GREEK CAPITAL DOTTED LUNATE SIGMA SYMBOL
  197. { 0x03FF, 0x037D }, // GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL
  198. { 0x0402, 0x0452 }, // CYRILLIC CAPITAL LETTER DJE
  199. { 0x0404, 0x0454 }, // CYRILLIC CAPITAL LETTER UKRAINIAN IE
  200. { 0x0405, 0x0455 }, // CYRILLIC CAPITAL LETTER DZE
  201. { 0x0406, 0x0456 }, // CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
  202. { 0x0408, 0x0458 }, // CYRILLIC CAPITAL LETTER JE
  203. { 0x0409, 0x0459 }, // CYRILLIC CAPITAL LETTER LJE
  204. { 0x040A, 0x045A }, // CYRILLIC CAPITAL LETTER NJE
  205. { 0x040B, 0x045B }, // CYRILLIC CAPITAL LETTER TSHE
  206. { 0x040F, 0x045F }, // CYRILLIC CAPITAL LETTER DZHE
  207. { 0x0410, 0x0430 }, // CYRILLIC CAPITAL LETTER A
  208. { 0x0411, 0x0431 }, // CYRILLIC CAPITAL LETTER BE
  209. { 0x0412, 0x0432 }, // CYRILLIC CAPITAL LETTER VE
  210. { 0x0413, 0x0433 }, // CYRILLIC CAPITAL LETTER GHE
  211. { 0x0414, 0x0434 }, // CYRILLIC CAPITAL LETTER DE
  212. { 0x0415, 0x0435 }, // CYRILLIC CAPITAL LETTER IE
  213. { 0x0416, 0x0436 }, // CYRILLIC CAPITAL LETTER ZHE
  214. { 0x0417, 0x0437 }, // CYRILLIC CAPITAL LETTER ZE
  215. { 0x0418, 0x0438 }, // CYRILLIC CAPITAL LETTER I
  216. { 0x041A, 0x043A }, // CYRILLIC CAPITAL LETTER KA
  217. { 0x041B, 0x043B }, // CYRILLIC CAPITAL LETTER EL
  218. { 0x041C, 0x043C }, // CYRILLIC CAPITAL LETTER EM
  219. { 0x041D, 0x043D }, // CYRILLIC CAPITAL LETTER EN
  220. { 0x041E, 0x043E }, // CYRILLIC CAPITAL LETTER O
  221. { 0x041F, 0x043F }, // CYRILLIC CAPITAL LETTER PE
  222. { 0x0420, 0x0440 }, // CYRILLIC CAPITAL LETTER ER
  223. { 0x0421, 0x0441 }, // CYRILLIC CAPITAL LETTER ES
  224. { 0x0422, 0x0442 }, // CYRILLIC CAPITAL LETTER TE
  225. { 0x0423, 0x0443 }, // CYRILLIC CAPITAL LETTER U
  226. { 0x0424, 0x0444 }, // CYRILLIC CAPITAL LETTER EF
  227. { 0x0425, 0x0445 }, // CYRILLIC CAPITAL LETTER HA
  228. { 0x0426, 0x0446 }, // CYRILLIC CAPITAL LETTER TSE
  229. { 0x0427, 0x0447 }, // CYRILLIC CAPITAL LETTER CHE
  230. { 0x0428, 0x0448 }, // CYRILLIC CAPITAL LETTER SHA
  231. { 0x0429, 0x0449 }, // CYRILLIC CAPITAL LETTER SHCHA
  232. { 0x042A, 0x044A }, // CYRILLIC CAPITAL LETTER HARD SIGN
  233. { 0x042B, 0x044B }, // CYRILLIC CAPITAL LETTER YERU
  234. { 0x042C, 0x044C }, // CYRILLIC CAPITAL LETTER SOFT SIGN
  235. { 0x042D, 0x044D }, // CYRILLIC CAPITAL LETTER E
  236. { 0x042E, 0x044E }, // CYRILLIC CAPITAL LETTER YU
  237. { 0x042F, 0x044F }, // CYRILLIC CAPITAL LETTER YA
  238. { 0x0460, 0x0461 }, // CYRILLIC CAPITAL LETTER OMEGA
  239. { 0x0462, 0x0463 }, // CYRILLIC CAPITAL LETTER YAT
  240. { 0x0464, 0x0465 }, // CYRILLIC CAPITAL LETTER IOTIFIED E
  241. { 0x0466, 0x0467 }, // CYRILLIC CAPITAL LETTER LITTLE YUS
  242. { 0x0468, 0x0469 }, // CYRILLIC CAPITAL LETTER IOTIFIED LITTLE YUS
  243. { 0x046A, 0x046B }, // CYRILLIC CAPITAL LETTER BIG YUS
  244. { 0x046C, 0x046D }, // CYRILLIC CAPITAL LETTER IOTIFIED BIG YUS
  245. { 0x046E, 0x046F }, // CYRILLIC CAPITAL LETTER KSI
  246. { 0x0470, 0x0471 }, // CYRILLIC CAPITAL LETTER PSI
  247. { 0x0472, 0x0473 }, // CYRILLIC CAPITAL LETTER FITA
  248. { 0x0474, 0x0475 }, // CYRILLIC CAPITAL LETTER IZHITSA
  249. { 0x0478, 0x0479 }, // CYRILLIC CAPITAL LETTER UK
  250. { 0x047A, 0x047B }, // CYRILLIC CAPITAL LETTER ROUND OMEGA
  251. { 0x047C, 0x047D }, // CYRILLIC CAPITAL LETTER OMEGA WITH TITLO
  252. { 0x047E, 0x047F }, // CYRILLIC CAPITAL LETTER OT
  253. { 0x0480, 0x0481 }, // CYRILLIC CAPITAL LETTER KOPPA
  254. { 0x048A, 0x048B }, // CYRILLIC CAPITAL LETTER SHORT I WITH TAIL
  255. { 0x048C, 0x048D }, // CYRILLIC CAPITAL LETTER SEMISOFT SIGN
  256. { 0x048E, 0x048F }, // CYRILLIC CAPITAL LETTER ER WITH TICK
  257. { 0x0490, 0x0491 }, // CYRILLIC CAPITAL LETTER GHE WITH UPTURN
  258. { 0x0492, 0x0493 }, // CYRILLIC CAPITAL LETTER GHE WITH STROKE
  259. { 0x0494, 0x0495 }, // CYRILLIC CAPITAL LETTER GHE WITH MIDDLE HOOK
  260. { 0x0496, 0x0497 }, // CYRILLIC CAPITAL LETTER ZHE WITH DESCENDER
  261. { 0x0498, 0x0499 }, // CYRILLIC CAPITAL LETTER ZE WITH DESCENDER
  262. { 0x049A, 0x049B }, // CYRILLIC CAPITAL LETTER KA WITH DESCENDER
  263. { 0x049C, 0x049D }, // CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE
  264. { 0x049E, 0x049F }, // CYRILLIC CAPITAL LETTER KA WITH STROKE
  265. { 0x04A0, 0x04A1 }, // CYRILLIC CAPITAL LETTER BASHKIR KA
  266. { 0x04A2, 0x04A3 }, // CYRILLIC CAPITAL LETTER EN WITH DESCENDER
  267. { 0x04A4, 0x04A5 }, // CYRILLIC CAPITAL LIGATURE EN GHE
  268. { 0x04A6, 0x04A7 }, // CYRILLIC CAPITAL LETTER PE WITH MIDDLE HOOK
  269. { 0x04A8, 0x04A9 }, // CYRILLIC CAPITAL LETTER ABKHASIAN HA
  270. { 0x04AA, 0x04AB }, // CYRILLIC CAPITAL LETTER ES WITH DESCENDER
  271. { 0x04AC, 0x04AD }, // CYRILLIC CAPITAL LETTER TE WITH DESCENDER
  272. { 0x04AE, 0x04AF }, // CYRILLIC CAPITAL LETTER STRAIGHT U
  273. { 0x04B0, 0x04B1 }, // CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE
  274. { 0x04B2, 0x04B3 }, // CYRILLIC CAPITAL LETTER HA WITH DESCENDER
  275. { 0x04B4, 0x04B5 }, // CYRILLIC CAPITAL LIGATURE TE TSE
  276. { 0x04B6, 0x04B7 }, // CYRILLIC CAPITAL LETTER CHE WITH DESCENDER
  277. { 0x04B8, 0x04B9 }, // CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE
  278. { 0x04BA, 0x04BB }, // CYRILLIC CAPITAL LETTER SHHA
  279. { 0x04BC, 0x04BD }, // CYRILLIC CAPITAL LETTER ABKHASIAN CHE
  280. { 0x04BE, 0x04BF }, // CYRILLIC CAPITAL LETTER ABKHASIAN CHE WITH DESCENDER
  281. { 0x04C0, 0x04CF }, // CYRILLIC LETTER PALOCHKA
  282. { 0x04C3, 0x04C4 }, // CYRILLIC CAPITAL LETTER KA WITH HOOK
  283. { 0x04C5, 0x04C6 }, // CYRILLIC CAPITAL LETTER EL WITH TAIL
  284. { 0x04C7, 0x04C8 }, // CYRILLIC CAPITAL LETTER EN WITH HOOK
  285. { 0x04C9, 0x04CA }, // CYRILLIC CAPITAL LETTER EN WITH TAIL
  286. { 0x04CB, 0x04CC }, // CYRILLIC CAPITAL LETTER KHAKASSIAN CHE
  287. { 0x04CD, 0x04CE }, // CYRILLIC CAPITAL LETTER EM WITH TAIL
  288. { 0x04D4, 0x04D5 }, // CYRILLIC CAPITAL LIGATURE A IE
  289. { 0x04D8, 0x04D9 }, // CYRILLIC CAPITAL LETTER SCHWA
  290. { 0x04E0, 0x04E1 }, // CYRILLIC CAPITAL LETTER ABKHASIAN DZE
  291. { 0x04E8, 0x04E9 }, // CYRILLIC CAPITAL LETTER BARRED O
  292. { 0x04F6, 0x04F7 }, // CYRILLIC CAPITAL LETTER GHE WITH DESCENDER
  293. { 0x04FA, 0x04FB }, // CYRILLIC CAPITAL LETTER GHE WITH STROKE AND HOOK
  294. { 0x04FC, 0x04FD }, // CYRILLIC CAPITAL LETTER HA WITH HOOK
  295. { 0x04FE, 0x04FF }, // CYRILLIC CAPITAL LETTER HA WITH STROKE
  296. { 0x0500, 0x0501 }, // CYRILLIC CAPITAL LETTER KOMI DE
  297. { 0x0502, 0x0503 }, // CYRILLIC CAPITAL LETTER KOMI DJE
  298. { 0x0504, 0x0505 }, // CYRILLIC CAPITAL LETTER KOMI ZJE
  299. { 0x0506, 0x0507 }, // CYRILLIC CAPITAL LETTER KOMI DZJE
  300. { 0x0508, 0x0509 }, // CYRILLIC CAPITAL LETTER KOMI LJE
  301. { 0x050A, 0x050B }, // CYRILLIC CAPITAL LETTER KOMI NJE
  302. { 0x050C, 0x050D }, // CYRILLIC CAPITAL LETTER KOMI SJE
  303. { 0x050E, 0x050F }, // CYRILLIC CAPITAL LETTER KOMI TJE
  304. { 0x0510, 0x0511 }, // CYRILLIC CAPITAL LETTER REVERSED ZE
  305. { 0x0512, 0x0513 }, // CYRILLIC CAPITAL LETTER EL WITH HOOK
  306. { 0x0514, 0x0515 }, // CYRILLIC CAPITAL LETTER LHA
  307. { 0x0516, 0x0517 }, // CYRILLIC CAPITAL LETTER RHA
  308. { 0x0518, 0x0519 }, // CYRILLIC CAPITAL LETTER YAE
  309. { 0x051A, 0x051B }, // CYRILLIC CAPITAL LETTER QA
  310. { 0x051C, 0x051D }, // CYRILLIC CAPITAL LETTER WE
  311. { 0x051E, 0x051F }, // CYRILLIC CAPITAL LETTER ALEUT KA
  312. { 0x0520, 0x0521 }, // CYRILLIC CAPITAL LETTER EL WITH MIDDLE HOOK
  313. { 0x0522, 0x0523 }, // CYRILLIC CAPITAL LETTER EN WITH MIDDLE HOOK
  314. { 0x0524, 0x0525 }, // CYRILLIC CAPITAL LETTER PE WITH DESCENDER
  315. { 0x0531, 0x0561 }, // ARMENIAN CAPITAL LETTER AYB
  316. { 0x0532, 0x0562 }, // ARMENIAN CAPITAL LETTER BEN
  317. { 0x0533, 0x0563 }, // ARMENIAN CAPITAL LETTER GIM
  318. { 0x0534, 0x0564 }, // ARMENIAN CAPITAL LETTER DA
  319. { 0x0535, 0x0565 }, // ARMENIAN CAPITAL LETTER ECH
  320. { 0x0536, 0x0566 }, // ARMENIAN CAPITAL LETTER ZA
  321. { 0x0537, 0x0567 }, // ARMENIAN CAPITAL LETTER EH
  322. { 0x0538, 0x0568 }, // ARMENIAN CAPITAL LETTER ET
  323. { 0x0539, 0x0569 }, // ARMENIAN CAPITAL LETTER TO
  324. { 0x053A, 0x056A }, // ARMENIAN CAPITAL LETTER ZHE
  325. { 0x053B, 0x056B }, // ARMENIAN CAPITAL LETTER INI
  326. { 0x053C, 0x056C }, // ARMENIAN CAPITAL LETTER LIWN
  327. { 0x053D, 0x056D }, // ARMENIAN CAPITAL LETTER XEH
  328. { 0x053E, 0x056E }, // ARMENIAN CAPITAL LETTER CA
  329. { 0x053F, 0x056F }, // ARMENIAN CAPITAL LETTER KEN
  330. { 0x0540, 0x0570 }, // ARMENIAN CAPITAL LETTER HO
  331. { 0x0541, 0x0571 }, // ARMENIAN CAPITAL LETTER JA
  332. { 0x0542, 0x0572 }, // ARMENIAN CAPITAL LETTER GHAD
  333. { 0x0543, 0x0573 }, // ARMENIAN CAPITAL LETTER CHEH
  334. { 0x0544, 0x0574 }, // ARMENIAN CAPITAL LETTER MEN
  335. { 0x0545, 0x0575 }, // ARMENIAN CAPITAL LETTER YI
  336. { 0x0546, 0x0576 }, // ARMENIAN CAPITAL LETTER NOW
  337. { 0x0547, 0x0577 }, // ARMENIAN CAPITAL LETTER SHA
  338. { 0x0548, 0x0578 }, // ARMENIAN CAPITAL LETTER VO
  339. { 0x0549, 0x0579 }, // ARMENIAN CAPITAL LETTER CHA
  340. { 0x054A, 0x057A }, // ARMENIAN CAPITAL LETTER PEH
  341. { 0x054B, 0x057B }, // ARMENIAN CAPITAL LETTER JHEH
  342. { 0x054C, 0x057C }, // ARMENIAN CAPITAL LETTER RA
  343. { 0x054D, 0x057D }, // ARMENIAN CAPITAL LETTER SEH
  344. { 0x054E, 0x057E }, // ARMENIAN CAPITAL LETTER VEW
  345. { 0x054F, 0x057F }, // ARMENIAN CAPITAL LETTER TIWN
  346. { 0x0550, 0x0580 }, // ARMENIAN CAPITAL LETTER REH
  347. { 0x0551, 0x0581 }, // ARMENIAN CAPITAL LETTER CO
  348. { 0x0552, 0x0582 }, // ARMENIAN CAPITAL LETTER YIWN
  349. { 0x0553, 0x0583 }, // ARMENIAN CAPITAL LETTER PIWR
  350. { 0x0554, 0x0584 }, // ARMENIAN CAPITAL LETTER KEH
  351. { 0x0555, 0x0585 }, // ARMENIAN CAPITAL LETTER OH
  352. { 0x0556, 0x0586 }, // ARMENIAN CAPITAL LETTER FEH
  353. { 0x10A0, 0x2D00 }, // GEORGIAN CAPITAL LETTER AN
  354. { 0x10A1, 0x2D01 }, // GEORGIAN CAPITAL LETTER BAN
  355. { 0x10A2, 0x2D02 }, // GEORGIAN CAPITAL LETTER GAN
  356. { 0x10A3, 0x2D03 }, // GEORGIAN CAPITAL LETTER DON
  357. { 0x10A4, 0x2D04 }, // GEORGIAN CAPITAL LETTER EN
  358. { 0x10A5, 0x2D05 }, // GEORGIAN CAPITAL LETTER VIN
  359. { 0x10A6, 0x2D06 }, // GEORGIAN CAPITAL LETTER ZEN
  360. { 0x10A7, 0x2D07 }, // GEORGIAN CAPITAL LETTER TAN
  361. { 0x10A8, 0x2D08 }, // GEORGIAN CAPITAL LETTER IN
  362. { 0x10A9, 0x2D09 }, // GEORGIAN CAPITAL LETTER KAN
  363. { 0x10AA, 0x2D0A }, // GEORGIAN CAPITAL LETTER LAS
  364. { 0x10AB, 0x2D0B }, // GEORGIAN CAPITAL LETTER MAN
  365. { 0x10AC, 0x2D0C }, // GEORGIAN CAPITAL LETTER NAR
  366. { 0x10AD, 0x2D0D }, // GEORGIAN CAPITAL LETTER ON
  367. { 0x10AE, 0x2D0E }, // GEORGIAN CAPITAL LETTER PAR
  368. { 0x10AF, 0x2D0F }, // GEORGIAN CAPITAL LETTER ZHAR
  369. { 0x10B0, 0x2D10 }, // GEORGIAN CAPITAL LETTER RAE
  370. { 0x10B1, 0x2D11 }, // GEORGIAN CAPITAL LETTER SAN
  371. { 0x10B2, 0x2D12 }, // GEORGIAN CAPITAL LETTER TAR
  372. { 0x10B3, 0x2D13 }, // GEORGIAN CAPITAL LETTER UN
  373. { 0x10B4, 0x2D14 }, // GEORGIAN CAPITAL LETTER PHAR
  374. { 0x10B5, 0x2D15 }, // GEORGIAN CAPITAL LETTER KHAR
  375. { 0x10B6, 0x2D16 }, // GEORGIAN CAPITAL LETTER GHAN
  376. { 0x10B7, 0x2D17 }, // GEORGIAN CAPITAL LETTER QAR
  377. { 0x10B8, 0x2D18 }, // GEORGIAN CAPITAL LETTER SHIN
  378. { 0x10B9, 0x2D19 }, // GEORGIAN CAPITAL LETTER CHIN
  379. { 0x10BA, 0x2D1A }, // GEORGIAN CAPITAL LETTER CAN
  380. { 0x10BB, 0x2D1B }, // GEORGIAN CAPITAL LETTER JIL
  381. { 0x10BC, 0x2D1C }, // GEORGIAN CAPITAL LETTER CIL
  382. { 0x10BD, 0x2D1D }, // GEORGIAN CAPITAL LETTER CHAR
  383. { 0x10BE, 0x2D1E }, // GEORGIAN CAPITAL LETTER XAN
  384. { 0x10BF, 0x2D1F }, // GEORGIAN CAPITAL LETTER JHAN
  385. { 0x10C0, 0x2D20 }, // GEORGIAN CAPITAL LETTER HAE
  386. { 0x10C1, 0x2D21 }, // GEORGIAN CAPITAL LETTER HE
  387. { 0x10C2, 0x2D22 }, // GEORGIAN CAPITAL LETTER HIE
  388. { 0x10C3, 0x2D23 }, // GEORGIAN CAPITAL LETTER WE
  389. { 0x10C4, 0x2D24 }, // GEORGIAN CAPITAL LETTER HAR
  390. { 0x10C5, 0x2D25 }, // GEORGIAN CAPITAL LETTER HOE
  391. { 0x1E00, 0x1E01 }, // LATIN CAPITAL LETTER A WITH RING BELOW
  392. { 0x1E02, 0x1E03 }, // LATIN CAPITAL LETTER B WITH DOT ABOVE
  393. { 0x1E04, 0x1E05 }, // LATIN CAPITAL LETTER B WITH DOT BELOW
  394. { 0x1E06, 0x1E07 }, // LATIN CAPITAL LETTER B WITH LINE BELOW
  395. { 0x1E08, 0x1E09 }, // LATIN CAPITAL LETTER C WITH CEDILLA AND ACUTE
  396. { 0x1E0A, 0x1E0B }, // LATIN CAPITAL LETTER D WITH DOT ABOVE
  397. { 0x1E0C, 0x1E0D }, // LATIN CAPITAL LETTER D WITH DOT BELOW
  398. { 0x1E0E, 0x1E0F }, // LATIN CAPITAL LETTER D WITH LINE BELOW
  399. { 0x1E10, 0x1E11 }, // LATIN CAPITAL LETTER D WITH CEDILLA
  400. { 0x1E12, 0x1E13 }, // LATIN CAPITAL LETTER D WITH CIRCUMFLEX BELOW
  401. { 0x1E14, 0x1E15 }, // LATIN CAPITAL LETTER E WITH MACRON AND GRAVE
  402. { 0x1E16, 0x1E17 }, // LATIN CAPITAL LETTER E WITH MACRON AND ACUTE
  403. { 0x1E18, 0x1E19 }, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX BELOW
  404. { 0x1E1A, 0x1E1B }, // LATIN CAPITAL LETTER E WITH TILDE BELOW
  405. { 0x1E1C, 0x1E1D }, // LATIN CAPITAL LETTER E WITH CEDILLA AND BREVE
  406. { 0x1E1E, 0x1E1F }, // LATIN CAPITAL LETTER F WITH DOT ABOVE
  407. { 0x1E20, 0x1E21 }, // LATIN CAPITAL LETTER G WITH MACRON
  408. { 0x1E22, 0x1E23 }, // LATIN CAPITAL LETTER H WITH DOT ABOVE
  409. { 0x1E24, 0x1E25 }, // LATIN CAPITAL LETTER H WITH DOT BELOW
  410. { 0x1E26, 0x1E27 }, // LATIN CAPITAL LETTER H WITH DIAERESIS
  411. { 0x1E28, 0x1E29 }, // LATIN CAPITAL LETTER H WITH CEDILLA
  412. { 0x1E2A, 0x1E2B }, // LATIN CAPITAL LETTER H WITH BREVE BELOW
  413. { 0x1E2C, 0x1E2D }, // LATIN CAPITAL LETTER I WITH TILDE BELOW
  414. { 0x1E2E, 0x1E2F }, // LATIN CAPITAL LETTER I WITH DIAERESIS AND ACUTE
  415. { 0x1E30, 0x1E31 }, // LATIN CAPITAL LETTER K WITH ACUTE
  416. { 0x1E32, 0x1E33 }, // LATIN CAPITAL LETTER K WITH DOT BELOW
  417. { 0x1E34, 0x1E35 }, // LATIN CAPITAL LETTER K WITH LINE BELOW
  418. { 0x1E36, 0x1E37 }, // LATIN CAPITAL LETTER L WITH DOT BELOW
  419. { 0x1E38, 0x1E39 }, // LATIN CAPITAL LETTER L WITH DOT BELOW AND MACRON
  420. { 0x1E3A, 0x1E3B }, // LATIN CAPITAL LETTER L WITH LINE BELOW
  421. { 0x1E3C, 0x1E3D }, // LATIN CAPITAL LETTER L WITH CIRCUMFLEX BELOW
  422. { 0x1E3E, 0x1E3F }, // LATIN CAPITAL LETTER M WITH ACUTE
  423. { 0x1E40, 0x1E41 }, // LATIN CAPITAL LETTER M WITH DOT ABOVE
  424. { 0x1E42, 0x1E43 }, // LATIN CAPITAL LETTER M WITH DOT BELOW
  425. { 0x1E44, 0x1E45 }, // LATIN CAPITAL LETTER N WITH DOT ABOVE
  426. { 0x1E46, 0x1E47 }, // LATIN CAPITAL LETTER N WITH DOT BELOW
  427. { 0x1E48, 0x1E49 }, // LATIN CAPITAL LETTER N WITH LINE BELOW
  428. { 0x1E4A, 0x1E4B }, // LATIN CAPITAL LETTER N WITH CIRCUMFLEX BELOW
  429. { 0x1E4C, 0x1E4D }, // LATIN CAPITAL LETTER O WITH TILDE AND ACUTE
  430. { 0x1E4E, 0x1E4F }, // LATIN CAPITAL LETTER O WITH TILDE AND DIAERESIS
  431. { 0x1E50, 0x1E51 }, // LATIN CAPITAL LETTER O WITH MACRON AND GRAVE
  432. { 0x1E52, 0x1E53 }, // LATIN CAPITAL LETTER O WITH MACRON AND ACUTE
  433. { 0x1E54, 0x1E55 }, // LATIN CAPITAL LETTER P WITH ACUTE
  434. { 0x1E56, 0x1E57 }, // LATIN CAPITAL LETTER P WITH DOT ABOVE
  435. { 0x1E58, 0x1E59 }, // LATIN CAPITAL LETTER R WITH DOT ABOVE
  436. { 0x1E5A, 0x1E5B }, // LATIN CAPITAL LETTER R WITH DOT BELOW
  437. { 0x1E5C, 0x1E5D }, // LATIN CAPITAL LETTER R WITH DOT BELOW AND MACRON
  438. { 0x1E5E, 0x1E5F }, // LATIN CAPITAL LETTER R WITH LINE BELOW
  439. { 0x1E60, 0x1E61 }, // LATIN CAPITAL LETTER S WITH DOT ABOVE
  440. { 0x1E62, 0x1E63 }, // LATIN CAPITAL LETTER S WITH DOT BELOW
  441. { 0x1E64, 0x1E65 }, // LATIN CAPITAL LETTER S WITH ACUTE AND DOT ABOVE
  442. { 0x1E66, 0x1E67 }, // LATIN CAPITAL LETTER S WITH CARON AND DOT ABOVE
  443. { 0x1E68, 0x1E69 }, // LATIN CAPITAL LETTER S WITH DOT BELOW AND DOT ABOVE
  444. { 0x1E6A, 0x1E6B }, // LATIN CAPITAL LETTER T WITH DOT ABOVE
  445. { 0x1E6C, 0x1E6D }, // LATIN CAPITAL LETTER T WITH DOT BELOW
  446. { 0x1E6E, 0x1E6F }, // LATIN CAPITAL LETTER T WITH LINE BELOW
  447. { 0x1E70, 0x1E71 }, // LATIN CAPITAL LETTER T WITH CIRCUMFLEX BELOW
  448. { 0x1E72, 0x1E73 }, // LATIN CAPITAL LETTER U WITH DIAERESIS BELOW
  449. { 0x1E74, 0x1E75 }, // LATIN CAPITAL LETTER U WITH TILDE BELOW
  450. { 0x1E76, 0x1E77 }, // LATIN CAPITAL LETTER U WITH CIRCUMFLEX BELOW
  451. { 0x1E78, 0x1E79 }, // LATIN CAPITAL LETTER U WITH TILDE AND ACUTE
  452. { 0x1E7A, 0x1E7B }, // LATIN CAPITAL LETTER U WITH MACRON AND DIAERESIS
  453. { 0x1E7C, 0x1E7D }, // LATIN CAPITAL LETTER V WITH TILDE
  454. { 0x1E7E, 0x1E7F }, // LATIN CAPITAL LETTER V WITH DOT BELOW
  455. { 0x1E80, 0x1E81 }, // LATIN CAPITAL LETTER W WITH GRAVE
  456. { 0x1E82, 0x1E83 }, // LATIN CAPITAL LETTER W WITH ACUTE
  457. { 0x1E84, 0x1E85 }, // LATIN CAPITAL LETTER W WITH DIAERESIS
  458. { 0x1E86, 0x1E87 }, // LATIN CAPITAL LETTER W WITH DOT ABOVE
  459. { 0x1E88, 0x1E89 }, // LATIN CAPITAL LETTER W WITH DOT BELOW
  460. { 0x1E8A, 0x1E8B }, // LATIN CAPITAL LETTER X WITH DOT ABOVE
  461. { 0x1E8C, 0x1E8D }, // LATIN CAPITAL LETTER X WITH DIAERESIS
  462. { 0x1E8E, 0x1E8F }, // LATIN CAPITAL LETTER Y WITH DOT ABOVE
  463. { 0x1E90, 0x1E91 }, // LATIN CAPITAL LETTER Z WITH CIRCUMFLEX
  464. { 0x1E92, 0x1E93 }, // LATIN CAPITAL LETTER Z WITH DOT BELOW
  465. { 0x1E94, 0x1E95 }, // LATIN CAPITAL LETTER Z WITH LINE BELOW
  466. { 0x1E9E, 0x00DF }, // LATIN CAPITAL LETTER SHARP S
  467. { 0x1EA0, 0x1EA1 }, // LATIN CAPITAL LETTER A WITH DOT BELOW
  468. { 0x1EA2, 0x1EA3 }, // LATIN CAPITAL LETTER A WITH HOOK ABOVE
  469. { 0x1EA4, 0x1EA5 }, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND ACUTE
  470. { 0x1EA6, 0x1EA7 }, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND GRAVE
  471. { 0x1EA8, 0x1EA9 }, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND HOOK ABOVE
  472. { 0x1EAA, 0x1EAB }, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND TILDE
  473. { 0x1EAC, 0x1EAD }, // LATIN CAPITAL LETTER A WITH CIRCUMFLEX AND DOT BELOW
  474. { 0x1EAE, 0x1EAF }, // LATIN CAPITAL LETTER A WITH BREVE AND ACUTE
  475. { 0x1EB0, 0x1EB1 }, // LATIN CAPITAL LETTER A WITH BREVE AND GRAVE
  476. { 0x1EB2, 0x1EB3 }, // LATIN CAPITAL LETTER A WITH BREVE AND HOOK ABOVE
  477. { 0x1EB4, 0x1EB5 }, // LATIN CAPITAL LETTER A WITH BREVE AND TILDE
  478. { 0x1EB6, 0x1EB7 }, // LATIN CAPITAL LETTER A WITH BREVE AND DOT BELOW
  479. { 0x1EB8, 0x1EB9 }, // LATIN CAPITAL LETTER E WITH DOT BELOW
  480. { 0x1EBA, 0x1EBB }, // LATIN CAPITAL LETTER E WITH HOOK ABOVE
  481. { 0x1EBC, 0x1EBD }, // LATIN CAPITAL LETTER E WITH TILDE
  482. { 0x1EBE, 0x1EBF }, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND ACUTE
  483. { 0x1EC0, 0x1EC1 }, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND GRAVE
  484. { 0x1EC2, 0x1EC3 }, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND HOOK ABOVE
  485. { 0x1EC4, 0x1EC5 }, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND TILDE
  486. { 0x1EC6, 0x1EC7 }, // LATIN CAPITAL LETTER E WITH CIRCUMFLEX AND DOT BELOW
  487. { 0x1EC8, 0x1EC9 }, // LATIN CAPITAL LETTER I WITH HOOK ABOVE
  488. { 0x1ECA, 0x1ECB }, // LATIN CAPITAL LETTER I WITH DOT BELOW
  489. { 0x1ECC, 0x1ECD }, // LATIN CAPITAL LETTER O WITH DOT BELOW
  490. { 0x1ECE, 0x1ECF }, // LATIN CAPITAL LETTER O WITH HOOK ABOVE
  491. { 0x1ED0, 0x1ED1 }, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND ACUTE
  492. { 0x1ED2, 0x1ED3 }, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND GRAVE
  493. { 0x1ED4, 0x1ED5 }, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND HOOK ABOVE
  494. { 0x1ED6, 0x1ED7 }, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND TILDE
  495. { 0x1ED8, 0x1ED9 }, // LATIN CAPITAL LETTER O WITH CIRCUMFLEX AND DOT BELOW
  496. { 0x1EDA, 0x1EDB }, // LATIN CAPITAL LETTER O WITH HORN AND ACUTE
  497. { 0x1EDC, 0x1EDD }, // LATIN CAPITAL LETTER O WITH HORN AND GRAVE
  498. { 0x1EDE, 0x1EDF }, // LATIN CAPITAL LETTER O WITH HORN AND HOOK ABOVE
  499. { 0x1EE0, 0x1EE1 }, // LATIN CAPITAL LETTER O WITH HORN AND TILDE
  500. { 0x1EE2, 0x1EE3 }, // LATIN CAPITAL LETTER O WITH HORN AND DOT BELOW
  501. { 0x1EE4, 0x1EE5 }, // LATIN CAPITAL LETTER U WITH DOT BELOW
  502. { 0x1EE6, 0x1EE7 }, // LATIN CAPITAL LETTER U WITH HOOK ABOVE
  503. { 0x1EE8, 0x1EE9 }, // LATIN CAPITAL LETTER U WITH HORN AND ACUTE
  504. { 0x1EEA, 0x1EEB }, // LATIN CAPITAL LETTER U WITH HORN AND GRAVE
  505. { 0x1EEC, 0x1EED }, // LATIN CAPITAL LETTER U WITH HORN AND HOOK ABOVE
  506. { 0x1EEE, 0x1EEF }, // LATIN CAPITAL LETTER U WITH HORN AND TILDE
  507. { 0x1EF0, 0x1EF1 }, // LATIN CAPITAL LETTER U WITH HORN AND DOT BELOW
  508. { 0x1EF2, 0x1EF3 }, // LATIN CAPITAL LETTER Y WITH GRAVE
  509. { 0x1EF4, 0x1EF5 }, // LATIN CAPITAL LETTER Y WITH DOT BELOW
  510. { 0x1EF6, 0x1EF7 }, // LATIN CAPITAL LETTER Y WITH HOOK ABOVE
  511. { 0x1EF8, 0x1EF9 }, // LATIN CAPITAL LETTER Y WITH TILDE
  512. { 0x1EFA, 0x1EFB }, // LATIN CAPITAL LETTER MIDDLE-WELSH LL
  513. { 0x1EFC, 0x1EFD }, // LATIN CAPITAL LETTER MIDDLE-WELSH V
  514. { 0x1EFE, 0x1EFF }, // LATIN CAPITAL LETTER Y WITH LOOP
  515. { 0x1F08, 0x1F00 }, // GREEK CAPITAL LETTER ALPHA WITH PSILI
  516. { 0x1F09, 0x1F01 }, // GREEK CAPITAL LETTER ALPHA WITH DASIA
  517. { 0x1F0A, 0x1F02 }, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND VARIA
  518. { 0x1F0B, 0x1F03 }, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND VARIA
  519. { 0x1F0C, 0x1F04 }, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND OXIA
  520. { 0x1F0D, 0x1F05 }, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND OXIA
  521. { 0x1F0E, 0x1F06 }, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PERISPOMENI
  522. { 0x1F0F, 0x1F07 }, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PERISPOMENI
  523. { 0x1F18, 0x1F10 }, // GREEK CAPITAL LETTER EPSILON WITH PSILI
  524. { 0x1F19, 0x1F11 }, // GREEK CAPITAL LETTER EPSILON WITH DASIA
  525. { 0x1F1A, 0x1F12 }, // GREEK CAPITAL LETTER EPSILON WITH PSILI AND VARIA
  526. { 0x1F1B, 0x1F13 }, // GREEK CAPITAL LETTER EPSILON WITH DASIA AND VARIA
  527. { 0x1F1C, 0x1F14 }, // GREEK CAPITAL LETTER EPSILON WITH PSILI AND OXIA
  528. { 0x1F1D, 0x1F15 }, // GREEK CAPITAL LETTER EPSILON WITH DASIA AND OXIA
  529. { 0x1F28, 0x1F20 }, // GREEK CAPITAL LETTER ETA WITH PSILI
  530. { 0x1F29, 0x1F21 }, // GREEK CAPITAL LETTER ETA WITH DASIA
  531. { 0x1F2A, 0x1F22 }, // GREEK CAPITAL LETTER ETA WITH PSILI AND VARIA
  532. { 0x1F2B, 0x1F23 }, // GREEK CAPITAL LETTER ETA WITH DASIA AND VARIA
  533. { 0x1F2C, 0x1F24 }, // GREEK CAPITAL LETTER ETA WITH PSILI AND OXIA
  534. { 0x1F2D, 0x1F25 }, // GREEK CAPITAL LETTER ETA WITH DASIA AND OXIA
  535. { 0x1F2E, 0x1F26 }, // GREEK CAPITAL LETTER ETA WITH PSILI AND PERISPOMENI
  536. { 0x1F2F, 0x1F27 }, // GREEK CAPITAL LETTER ETA WITH DASIA AND PERISPOMENI
  537. { 0x1F38, 0x1F30 }, // GREEK CAPITAL LETTER IOTA WITH PSILI
  538. { 0x1F39, 0x1F31 }, // GREEK CAPITAL LETTER IOTA WITH DASIA
  539. { 0x1F3A, 0x1F32 }, // GREEK CAPITAL LETTER IOTA WITH PSILI AND VARIA
  540. { 0x1F3B, 0x1F33 }, // GREEK CAPITAL LETTER IOTA WITH DASIA AND VARIA
  541. { 0x1F3C, 0x1F34 }, // GREEK CAPITAL LETTER IOTA WITH PSILI AND OXIA
  542. { 0x1F3D, 0x1F35 }, // GREEK CAPITAL LETTER IOTA WITH DASIA AND OXIA
  543. { 0x1F3E, 0x1F36 }, // GREEK CAPITAL LETTER IOTA WITH PSILI AND PERISPOMENI
  544. { 0x1F3F, 0x1F37 }, // GREEK CAPITAL LETTER IOTA WITH DASIA AND PERISPOMENI
  545. { 0x1F48, 0x1F40 }, // GREEK CAPITAL LETTER OMICRON WITH PSILI
  546. { 0x1F49, 0x1F41 }, // GREEK CAPITAL LETTER OMICRON WITH DASIA
  547. { 0x1F4A, 0x1F42 }, // GREEK CAPITAL LETTER OMICRON WITH PSILI AND VARIA
  548. { 0x1F4B, 0x1F43 }, // GREEK CAPITAL LETTER OMICRON WITH DASIA AND VARIA
  549. { 0x1F4C, 0x1F44 }, // GREEK CAPITAL LETTER OMICRON WITH PSILI AND OXIA
  550. { 0x1F4D, 0x1F45 }, // GREEK CAPITAL LETTER OMICRON WITH DASIA AND OXIA
  551. { 0x1F59, 0x1F51 }, // GREEK CAPITAL LETTER UPSILON WITH DASIA
  552. { 0x1F5B, 0x1F53 }, // GREEK CAPITAL LETTER UPSILON WITH DASIA AND VARIA
  553. { 0x1F5D, 0x1F55 }, // GREEK CAPITAL LETTER UPSILON WITH DASIA AND OXIA
  554. { 0x1F5F, 0x1F57 }, // GREEK CAPITAL LETTER UPSILON WITH DASIA AND PERISPOMENI
  555. { 0x1F68, 0x1F60 }, // GREEK CAPITAL LETTER OMEGA WITH PSILI
  556. { 0x1F69, 0x1F61 }, // GREEK CAPITAL LETTER OMEGA WITH DASIA
  557. { 0x1F6A, 0x1F62 }, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND VARIA
  558. { 0x1F6B, 0x1F63 }, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND VARIA
  559. { 0x1F6C, 0x1F64 }, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND OXIA
  560. { 0x1F6D, 0x1F65 }, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND OXIA
  561. { 0x1F6E, 0x1F66 }, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PERISPOMENI
  562. { 0x1F6F, 0x1F67 }, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PERISPOMENI
  563. { 0x1F88, 0x1F80 }, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PROSGEGRAMMENI
  564. { 0x1F89, 0x1F81 }, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PROSGEGRAMMENI
  565. { 0x1F8A, 0x1F82 }, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND VARIA AND PROSGEGRAMMENI
  566. { 0x1F8B, 0x1F83 }, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND VARIA AND PROSGEGRAMMENI
  567. { 0x1F8C, 0x1F84 }, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND OXIA AND PROSGEGRAMMENI
  568. { 0x1F8D, 0x1F85 }, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND OXIA AND PROSGEGRAMMENI
  569. { 0x1F8E, 0x1F86 }, // GREEK CAPITAL LETTER ALPHA WITH PSILI AND PERISPOMENI AND PROSGEGRAMMENI
  570. { 0x1F8F, 0x1F87 }, // GREEK CAPITAL LETTER ALPHA WITH DASIA AND PERISPOMENI AND PROSGEGRAMMENI
  571. { 0x1F98, 0x1F90 }, // GREEK CAPITAL LETTER ETA WITH PSILI AND PROSGEGRAMMENI
  572. { 0x1F99, 0x1F91 }, // GREEK CAPITAL LETTER ETA WITH DASIA AND PROSGEGRAMMENI
  573. { 0x1F9A, 0x1F92 }, // GREEK CAPITAL LETTER ETA WITH PSILI AND VARIA AND PROSGEGRAMMENI
  574. { 0x1F9B, 0x1F93 }, // GREEK CAPITAL LETTER ETA WITH DASIA AND VARIA AND PROSGEGRAMMENI
  575. { 0x1F9C, 0x1F94 }, // GREEK CAPITAL LETTER ETA WITH PSILI AND OXIA AND PROSGEGRAMMENI
  576. { 0x1F9D, 0x1F95 }, // GREEK CAPITAL LETTER ETA WITH DASIA AND OXIA AND PROSGEGRAMMENI
  577. { 0x1F9E, 0x1F96 }, // GREEK CAPITAL LETTER ETA WITH PSILI AND PERISPOMENI AND PROSGEGRAMMENI
  578. { 0x1F9F, 0x1F97 }, // GREEK CAPITAL LETTER ETA WITH DASIA AND PERISPOMENI AND PROSGEGRAMMENI
  579. { 0x1FA8, 0x1FA0 }, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PROSGEGRAMMENI
  580. { 0x1FA9, 0x1FA1 }, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PROSGEGRAMMENI
  581. { 0x1FAA, 0x1FA2 }, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND VARIA AND PROSGEGRAMMENI
  582. { 0x1FAB, 0x1FA3 }, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND VARIA AND PROSGEGRAMMENI
  583. { 0x1FAC, 0x1FA4 }, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND OXIA AND PROSGEGRAMMENI
  584. { 0x1FAD, 0x1FA5 }, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND OXIA AND PROSGEGRAMMENI
  585. { 0x1FAE, 0x1FA6 }, // GREEK CAPITAL LETTER OMEGA WITH PSILI AND PERISPOMENI AND PROSGEGRAMMENI
  586. { 0x1FAF, 0x1FA7 }, // GREEK CAPITAL LETTER OMEGA WITH DASIA AND PERISPOMENI AND PROSGEGRAMMENI
  587. { 0x1FB8, 0x1FB0 }, // GREEK CAPITAL LETTER ALPHA WITH VRACHY
  588. { 0x1FB9, 0x1FB1 }, // GREEK CAPITAL LETTER ALPHA WITH MACRON
  589. { 0x1FBA, 0x1F70 }, // GREEK CAPITAL LETTER ALPHA WITH VARIA
  590. { 0x1FBB, 0x1F71 }, // GREEK CAPITAL LETTER ALPHA WITH OXIA
  591. { 0x1FBC, 0x1FB3 }, // GREEK CAPITAL LETTER ALPHA WITH PROSGEGRAMMENI
  592. { 0x1FC8, 0x1F72 }, // GREEK CAPITAL LETTER EPSILON WITH VARIA
  593. { 0x1FC9, 0x1F73 }, // GREEK CAPITAL LETTER EPSILON WITH OXIA
  594. { 0x1FCA, 0x1F74 }, // GREEK CAPITAL LETTER ETA WITH VARIA
  595. { 0x1FCB, 0x1F75 }, // GREEK CAPITAL LETTER ETA WITH OXIA
  596. { 0x1FCC, 0x1FC3 }, // GREEK CAPITAL LETTER ETA WITH PROSGEGRAMMENI
  597. { 0x1FD8, 0x1FD0 }, // GREEK CAPITAL LETTER IOTA WITH VRACHY
  598. { 0x1FD9, 0x1FD1 }, // GREEK CAPITAL LETTER IOTA WITH MACRON
  599. { 0x1FDA, 0x1F76 }, // GREEK CAPITAL LETTER IOTA WITH VARIA
  600. { 0x1FDB, 0x1F77 }, // GREEK CAPITAL LETTER IOTA WITH OXIA
  601. { 0x1FE8, 0x1FE0 }, // GREEK CAPITAL LETTER UPSILON WITH VRACHY
  602. { 0x1FE9, 0x1FE1 }, // GREEK CAPITAL LETTER UPSILON WITH MACRON
  603. { 0x1FEA, 0x1F7A }, // GREEK CAPITAL LETTER UPSILON WITH VARIA
  604. { 0x1FEB, 0x1F7B }, // GREEK CAPITAL LETTER UPSILON WITH OXIA
  605. { 0x1FEC, 0x1FE5 }, // GREEK CAPITAL LETTER RHO WITH DASIA
  606. { 0x1FF8, 0x1F78 }, // GREEK CAPITAL LETTER OMICRON WITH VARIA
  607. { 0x1FF9, 0x1F79 }, // GREEK CAPITAL LETTER OMICRON WITH OXIA
  608. { 0x1FFA, 0x1F7C }, // GREEK CAPITAL LETTER OMEGA WITH VARIA
  609. { 0x1FFB, 0x1F7D }, // GREEK CAPITAL LETTER OMEGA WITH OXIA
  610. { 0x1FFC, 0x1FF3 }, // GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI
  611. { 0x2126, 0x03C9 }, // OHM SIGN
  612. { 0x212A, 0x006B }, // KELVIN SIGN
  613. { 0x212B, 0x00E5 }, // ANGSTROM SIGN
  614. { 0x2132, 0x214E }, // TURNED CAPITAL F
  615. { 0x2160, 0x2170 }, // ROMAN NUMERAL ONE
  616. { 0x2161, 0x2171 }, // ROMAN NUMERAL TWO
  617. { 0x2162, 0x2172 }, // ROMAN NUMERAL THREE
  618. { 0x2163, 0x2173 }, // ROMAN NUMERAL FOUR
  619. { 0x2164, 0x2174 }, // ROMAN NUMERAL FIVE
  620. { 0x2165, 0x2175 }, // ROMAN NUMERAL SIX
  621. { 0x2166, 0x2176 }, // ROMAN NUMERAL SEVEN
  622. { 0x2167, 0x2177 }, // ROMAN NUMERAL EIGHT
  623. { 0x2168, 0x2178 }, // ROMAN NUMERAL NINE
  624. { 0x2169, 0x2179 }, // ROMAN NUMERAL TEN
  625. { 0x216A, 0x217A }, // ROMAN NUMERAL ELEVEN
  626. { 0x216B, 0x217B }, // ROMAN NUMERAL TWELVE
  627. { 0x216C, 0x217C }, // ROMAN NUMERAL FIFTY
  628. { 0x216D, 0x217D }, // ROMAN NUMERAL ONE HUNDRED
  629. { 0x216E, 0x217E }, // ROMAN NUMERAL FIVE HUNDRED
  630. { 0x216F, 0x217F }, // ROMAN NUMERAL ONE THOUSAND
  631. { 0x2183, 0x2184 }, // ROMAN NUMERAL REVERSED ONE HUNDRED
  632. { 0x24B6, 0x24D0 }, // CIRCLED LATIN CAPITAL LETTER A
  633. { 0x24B7, 0x24D1 }, // CIRCLED LATIN CAPITAL LETTER B
  634. { 0x24B8, 0x24D2 }, // CIRCLED LATIN CAPITAL LETTER C
  635. { 0x24B9, 0x24D3 }, // CIRCLED LATIN CAPITAL LETTER D
  636. { 0x24BA, 0x24D4 }, // CIRCLED LATIN CAPITAL LETTER E
  637. { 0x24BB, 0x24D5 }, // CIRCLED LATIN CAPITAL LETTER F
  638. { 0x24BC, 0x24D6 }, // CIRCLED LATIN CAPITAL LETTER G
  639. { 0x24BD, 0x24D7 }, // CIRCLED LATIN CAPITAL LETTER H
  640. { 0x24BE, 0x24D8 }, // CIRCLED LATIN CAPITAL LETTER I
  641. { 0x24BF, 0x24D9 }, // CIRCLED LATIN CAPITAL LETTER J
  642. { 0x24C0, 0x24DA }, // CIRCLED LATIN CAPITAL LETTER K
  643. { 0x24C1, 0x24DB }, // CIRCLED LATIN CAPITAL LETTER L
  644. { 0x24C2, 0x24DC }, // CIRCLED LATIN CAPITAL LETTER M
  645. { 0x24C3, 0x24DD }, // CIRCLED LATIN CAPITAL LETTER N
  646. { 0x24C4, 0x24DE }, // CIRCLED LATIN CAPITAL LETTER O
  647. { 0x24C5, 0x24DF }, // CIRCLED LATIN CAPITAL LETTER P
  648. { 0x24C6, 0x24E0 }, // CIRCLED LATIN CAPITAL LETTER Q
  649. { 0x24C7, 0x24E1 }, // CIRCLED LATIN CAPITAL LETTER R
  650. { 0x24C8, 0x24E2 }, // CIRCLED LATIN CAPITAL LETTER S
  651. { 0x24C9, 0x24E3 }, // CIRCLED LATIN CAPITAL LETTER T
  652. { 0x24CA, 0x24E4 }, // CIRCLED LATIN CAPITAL LETTER U
  653. { 0x24CB, 0x24E5 }, // CIRCLED LATIN CAPITAL LETTER V
  654. { 0x24CC, 0x24E6 }, // CIRCLED LATIN CAPITAL LETTER W
  655. { 0x24CD, 0x24E7 }, // CIRCLED LATIN CAPITAL LETTER X
  656. { 0x24CE, 0x24E8 }, // CIRCLED LATIN CAPITAL LETTER Y
  657. { 0x24CF, 0x24E9 }, // CIRCLED LATIN CAPITAL LETTER Z
  658. { 0x2C00, 0x2C30 }, // GLAGOLITIC CAPITAL LETTER AZU
  659. { 0x2C01, 0x2C31 }, // GLAGOLITIC CAPITAL LETTER BUKY
  660. { 0x2C02, 0x2C32 }, // GLAGOLITIC CAPITAL LETTER VEDE
  661. { 0x2C03, 0x2C33 }, // GLAGOLITIC CAPITAL LETTER GLAGOLI
  662. { 0x2C04, 0x2C34 }, // GLAGOLITIC CAPITAL LETTER DOBRO
  663. { 0x2C05, 0x2C35 }, // GLAGOLITIC CAPITAL LETTER YESTU
  664. { 0x2C06, 0x2C36 }, // GLAGOLITIC CAPITAL LETTER ZHIVETE
  665. { 0x2C07, 0x2C37 }, // GLAGOLITIC CAPITAL LETTER DZELO
  666. { 0x2C08, 0x2C38 }, // GLAGOLITIC CAPITAL LETTER ZEMLJA
  667. { 0x2C09, 0x2C39 }, // GLAGOLITIC CAPITAL LETTER IZHE
  668. { 0x2C0A, 0x2C3A }, // GLAGOLITIC CAPITAL LETTER INITIAL IZHE
  669. { 0x2C0B, 0x2C3B }, // GLAGOLITIC CAPITAL LETTER I
  670. { 0x2C0C, 0x2C3C }, // GLAGOLITIC CAPITAL LETTER DJERVI
  671. { 0x2C0D, 0x2C3D }, // GLAGOLITIC CAPITAL LETTER KAKO
  672. { 0x2C0E, 0x2C3E }, // GLAGOLITIC CAPITAL LETTER LJUDIJE
  673. { 0x2C0F, 0x2C3F }, // GLAGOLITIC CAPITAL LETTER MYSLITE
  674. { 0x2C10, 0x2C40 }, // GLAGOLITIC CAPITAL LETTER NASHI
  675. { 0x2C11, 0x2C41 }, // GLAGOLITIC CAPITAL LETTER ONU
  676. { 0x2C12, 0x2C42 }, // GLAGOLITIC CAPITAL LETTER POKOJI
  677. { 0x2C13, 0x2C43 }, // GLAGOLITIC CAPITAL LETTER RITSI
  678. { 0x2C14, 0x2C44 }, // GLAGOLITIC CAPITAL LETTER SLOVO
  679. { 0x2C15, 0x2C45 }, // GLAGOLITIC CAPITAL LETTER TVRIDO
  680. { 0x2C16, 0x2C46 }, // GLAGOLITIC CAPITAL LETTER UKU
  681. { 0x2C17, 0x2C47 }, // GLAGOLITIC CAPITAL LETTER FRITU
  682. { 0x2C18, 0x2C48 }, // GLAGOLITIC CAPITAL LETTER HERU
  683. { 0x2C19, 0x2C49 }, // GLAGOLITIC CAPITAL LETTER OTU
  684. { 0x2C1A, 0x2C4A }, // GLAGOLITIC CAPITAL LETTER PE
  685. { 0x2C1B, 0x2C4B }, // GLAGOLITIC CAPITAL LETTER SHTA
  686. { 0x2C1C, 0x2C4C }, // GLAGOLITIC CAPITAL LETTER TSI
  687. { 0x2C1D, 0x2C4D }, // GLAGOLITIC CAPITAL LETTER CHRIVI
  688. { 0x2C1E, 0x2C4E }, // GLAGOLITIC CAPITAL LETTER SHA
  689. { 0x2C1F, 0x2C4F }, // GLAGOLITIC CAPITAL LETTER YERU
  690. { 0x2C20, 0x2C50 }, // GLAGOLITIC CAPITAL LETTER YERI
  691. { 0x2C21, 0x2C51 }, // GLAGOLITIC CAPITAL LETTER YATI
  692. { 0x2C22, 0x2C52 }, // GLAGOLITIC CAPITAL LETTER SPIDERY HA
  693. { 0x2C23, 0x2C53 }, // GLAGOLITIC CAPITAL LETTER YU
  694. { 0x2C24, 0x2C54 }, // GLAGOLITIC CAPITAL LETTER SMALL YUS
  695. { 0x2C25, 0x2C55 }, // GLAGOLITIC CAPITAL LETTER SMALL YUS WITH TAIL
  696. { 0x2C26, 0x2C56 }, // GLAGOLITIC CAPITAL LETTER YO
  697. { 0x2C27, 0x2C57 }, // GLAGOLITIC CAPITAL LETTER IOTATED SMALL YUS
  698. { 0x2C28, 0x2C58 }, // GLAGOLITIC CAPITAL LETTER BIG YUS
  699. { 0x2C29, 0x2C59 }, // GLAGOLITIC CAPITAL LETTER IOTATED BIG YUS
  700. { 0x2C2A, 0x2C5A }, // GLAGOLITIC CAPITAL LETTER FITA
  701. { 0x2C2B, 0x2C5B }, // GLAGOLITIC CAPITAL LETTER IZHITSA
  702. { 0x2C2C, 0x2C5C }, // GLAGOLITIC CAPITAL LETTER SHTAPIC
  703. { 0x2C2D, 0x2C5D }, // GLAGOLITIC CAPITAL LETTER TROKUTASTI A
  704. { 0x2C2E, 0x2C5E }, // GLAGOLITIC CAPITAL LETTER LATINATE MYSLITE
  705. { 0x2C60, 0x2C61 }, // LATIN CAPITAL LETTER L WITH DOUBLE BAR
  706. { 0x2C62, 0x026B }, // LATIN CAPITAL LETTER L WITH MIDDLE TILDE
  707. { 0x2C63, 0x1D7D }, // LATIN CAPITAL LETTER P WITH STROKE
  708. { 0x2C64, 0x027D }, // LATIN CAPITAL LETTER R WITH TAIL
  709. { 0x2C67, 0x2C68 }, // LATIN CAPITAL LETTER H WITH DESCENDER
  710. { 0x2C69, 0x2C6A }, // LATIN CAPITAL LETTER K WITH DESCENDER
  711. { 0x2C6B, 0x2C6C }, // LATIN CAPITAL LETTER Z WITH DESCENDER
  712. { 0x2C6D, 0x0251 }, // LATIN CAPITAL LETTER ALPHA
  713. { 0x2C6E, 0x0271 }, // LATIN CAPITAL LETTER M WITH HOOK
  714. { 0x2C6F, 0x0250 }, // LATIN CAPITAL LETTER TURNED A
  715. { 0x2C70, 0x0252 }, // LATIN CAPITAL LETTER TURNED ALPHA
  716. { 0x2C72, 0x2C73 }, // LATIN CAPITAL LETTER W WITH HOOK
  717. { 0x2C75, 0x2C76 }, // LATIN CAPITAL LETTER HALF H
  718. { 0x2C7E, 0x023F }, // LATIN CAPITAL LETTER S WITH SWASH TAIL
  719. { 0x2C7F, 0x0240 }, // LATIN CAPITAL LETTER Z WITH SWASH TAIL
  720. { 0x2C80, 0x2C81 }, // COPTIC CAPITAL LETTER ALFA
  721. { 0x2C82, 0x2C83 }, // COPTIC CAPITAL LETTER VIDA
  722. { 0x2C84, 0x2C85 }, // COPTIC CAPITAL LETTER GAMMA
  723. { 0x2C86, 0x2C87 }, // COPTIC CAPITAL LETTER DALDA
  724. { 0x2C88, 0x2C89 }, // COPTIC CAPITAL LETTER EIE
  725. { 0x2C8A, 0x2C8B }, // COPTIC CAPITAL LETTER SOU
  726. { 0x2C8C, 0x2C8D }, // COPTIC CAPITAL LETTER ZATA
  727. { 0x2C8E, 0x2C8F }, // COPTIC CAPITAL LETTER HATE
  728. { 0x2C90, 0x2C91 }, // COPTIC CAPITAL LETTER THETHE
  729. { 0x2C92, 0x2C93 }, // COPTIC CAPITAL LETTER IAUDA
  730. { 0x2C94, 0x2C95 }, // COPTIC CAPITAL LETTER KAPA
  731. { 0x2C96, 0x2C97 }, // COPTIC CAPITAL LETTER LAULA
  732. { 0x2C98, 0x2C99 }, // COPTIC CAPITAL LETTER MI
  733. { 0x2C9A, 0x2C9B }, // COPTIC CAPITAL LETTER NI
  734. { 0x2C9C, 0x2C9D }, // COPTIC CAPITAL LETTER KSI
  735. { 0x2C9E, 0x2C9F }, // COPTIC CAPITAL LETTER O
  736. { 0x2CA0, 0x2CA1 }, // COPTIC CAPITAL LETTER PI
  737. { 0x2CA2, 0x2CA3 }, // COPTIC CAPITAL LETTER RO
  738. { 0x2CA4, 0x2CA5 }, // COPTIC CAPITAL LETTER SIMA
  739. { 0x2CA6, 0x2CA7 }, // COPTIC CAPITAL LETTER TAU
  740. { 0x2CA8, 0x2CA9 }, // COPTIC CAPITAL LETTER UA
  741. { 0x2CAA, 0x2CAB }, // COPTIC CAPITAL LETTER FI
  742. { 0x2CAC, 0x2CAD }, // COPTIC CAPITAL LETTER KHI
  743. { 0x2CAE, 0x2CAF }, // COPTIC CAPITAL LETTER PSI
  744. { 0x2CB0, 0x2CB1 }, // COPTIC CAPITAL LETTER OOU
  745. { 0x2CB2, 0x2CB3 }, // COPTIC CAPITAL LETTER DIALECT-P ALEF
  746. { 0x2CB4, 0x2CB5 }, // COPTIC CAPITAL LETTER OLD COPTIC AIN
  747. { 0x2CB6, 0x2CB7 }, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC EIE
  748. { 0x2CB8, 0x2CB9 }, // COPTIC CAPITAL LETTER DIALECT-P KAPA
  749. { 0x2CBA, 0x2CBB }, // COPTIC CAPITAL LETTER DIALECT-P NI
  750. { 0x2CBC, 0x2CBD }, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC NI
  751. { 0x2CBE, 0x2CBF }, // COPTIC CAPITAL LETTER OLD COPTIC OOU
  752. { 0x2CC0, 0x2CC1 }, // COPTIC CAPITAL LETTER SAMPI
  753. { 0x2CC2, 0x2CC3 }, // COPTIC CAPITAL LETTER CROSSED SHEI
  754. { 0x2CC4, 0x2CC5 }, // COPTIC CAPITAL LETTER OLD COPTIC SHEI
  755. { 0x2CC6, 0x2CC7 }, // COPTIC CAPITAL LETTER OLD COPTIC ESH
  756. { 0x2CC8, 0x2CC9 }, // COPTIC CAPITAL LETTER AKHMIMIC KHEI
  757. { 0x2CCA, 0x2CCB }, // COPTIC CAPITAL LETTER DIALECT-P HORI
  758. { 0x2CCC, 0x2CCD }, // COPTIC CAPITAL LETTER OLD COPTIC HORI
  759. { 0x2CCE, 0x2CCF }, // COPTIC CAPITAL LETTER OLD COPTIC HA
  760. { 0x2CD0, 0x2CD1 }, // COPTIC CAPITAL LETTER L-SHAPED HA
  761. { 0x2CD2, 0x2CD3 }, // COPTIC CAPITAL LETTER OLD COPTIC HEI
  762. { 0x2CD4, 0x2CD5 }, // COPTIC CAPITAL LETTER OLD COPTIC HAT
  763. { 0x2CD6, 0x2CD7 }, // COPTIC CAPITAL LETTER OLD COPTIC GANGIA
  764. { 0x2CD8, 0x2CD9 }, // COPTIC CAPITAL LETTER OLD COPTIC DJA
  765. { 0x2CDA, 0x2CDB }, // COPTIC CAPITAL LETTER OLD COPTIC SHIMA
  766. { 0x2CDC, 0x2CDD }, // COPTIC CAPITAL LETTER OLD NUBIAN SHIMA
  767. { 0x2CDE, 0x2CDF }, // COPTIC CAPITAL LETTER OLD NUBIAN NGI
  768. { 0x2CE0, 0x2CE1 }, // COPTIC CAPITAL LETTER OLD NUBIAN NYI
  769. { 0x2CE2, 0x2CE3 }, // COPTIC CAPITAL LETTER OLD NUBIAN WAU
  770. { 0x2CEB, 0x2CEC }, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC SHEI
  771. { 0x2CED, 0x2CEE }, // COPTIC CAPITAL LETTER CRYPTOGRAMMIC GANGIA
  772. { 0xA640, 0xA641 }, // CYRILLIC CAPITAL LETTER ZEMLYA
  773. { 0xA642, 0xA643 }, // CYRILLIC CAPITAL LETTER DZELO
  774. { 0xA644, 0xA645 }, // CYRILLIC CAPITAL LETTER REVERSED DZE
  775. { 0xA646, 0xA647 }, // CYRILLIC CAPITAL LETTER IOTA
  776. { 0xA648, 0xA649 }, // CYRILLIC CAPITAL LETTER DJERV
  777. { 0xA64A, 0xA64B }, // CYRILLIC CAPITAL LETTER MONOGRAPH UK
  778. { 0xA64C, 0xA64D }, // CYRILLIC CAPITAL LETTER BROAD OMEGA
  779. { 0xA64E, 0xA64F }, // CYRILLIC CAPITAL LETTER NEUTRAL YER
  780. { 0xA650, 0xA651 }, // CYRILLIC CAPITAL LETTER YERU WITH BACK YER
  781. { 0xA652, 0xA653 }, // CYRILLIC CAPITAL LETTER IOTIFIED YAT
  782. { 0xA654, 0xA655 }, // CYRILLIC CAPITAL LETTER REVERSED YU
  783. { 0xA656, 0xA657 }, // CYRILLIC CAPITAL LETTER IOTIFIED A
  784. { 0xA658, 0xA659 }, // CYRILLIC CAPITAL LETTER CLOSED LITTLE YUS
  785. { 0xA65A, 0xA65B }, // CYRILLIC CAPITAL LETTER BLENDED YUS
  786. { 0xA65C, 0xA65D }, // CYRILLIC CAPITAL LETTER IOTIFIED CLOSED LITTLE YUS
  787. { 0xA65E, 0xA65F }, // CYRILLIC CAPITAL LETTER YN
  788. { 0xA662, 0xA663 }, // CYRILLIC CAPITAL LETTER SOFT DE
  789. { 0xA664, 0xA665 }, // CYRILLIC CAPITAL LETTER SOFT EL
  790. { 0xA666, 0xA667 }, // CYRILLIC CAPITAL LETTER SOFT EM
  791. { 0xA668, 0xA669 }, // CYRILLIC CAPITAL LETTER MONOCULAR O
  792. { 0xA66A, 0xA66B }, // CYRILLIC CAPITAL LETTER BINOCULAR O
  793. { 0xA66C, 0xA66D }, // CYRILLIC CAPITAL LETTER DOUBLE MONOCULAR O
  794. { 0xA680, 0xA681 }, // CYRILLIC CAPITAL LETTER DWE
  795. { 0xA682, 0xA683 }, // CYRILLIC CAPITAL LETTER DZWE
  796. { 0xA684, 0xA685 }, // CYRILLIC CAPITAL LETTER ZHWE
  797. { 0xA686, 0xA687 }, // CYRILLIC CAPITAL LETTER CCHE
  798. { 0xA688, 0xA689 }, // CYRILLIC CAPITAL LETTER DZZE
  799. { 0xA68A, 0xA68B }, // CYRILLIC CAPITAL LETTER TE WITH MIDDLE HOOK
  800. { 0xA68C, 0xA68D }, // CYRILLIC CAPITAL LETTER TWE
  801. { 0xA68E, 0xA68F }, // CYRILLIC CAPITAL LETTER TSWE
  802. { 0xA690, 0xA691 }, // CYRILLIC CAPITAL LETTER TSSE
  803. { 0xA692, 0xA693 }, // CYRILLIC CAPITAL LETTER TCHE
  804. { 0xA694, 0xA695 }, // CYRILLIC CAPITAL LETTER HWE
  805. { 0xA696, 0xA697 }, // CYRILLIC CAPITAL LETTER SHWE
  806. { 0xA722, 0xA723 }, // LATIN CAPITAL LETTER EGYPTOLOGICAL ALEF
  807. { 0xA724, 0xA725 }, // LATIN CAPITAL LETTER EGYPTOLOGICAL AIN
  808. { 0xA726, 0xA727 }, // LATIN CAPITAL LETTER HENG
  809. { 0xA728, 0xA729 }, // LATIN CAPITAL LETTER TZ
  810. { 0xA72A, 0xA72B }, // LATIN CAPITAL LETTER TRESILLO
  811. { 0xA72C, 0xA72D }, // LATIN CAPITAL LETTER CUATRILLO
  812. { 0xA72E, 0xA72F }, // LATIN CAPITAL LETTER CUATRILLO WITH COMMA
  813. { 0xA732, 0xA733 }, // LATIN CAPITAL LETTER AA
  814. { 0xA734, 0xA735 }, // LATIN CAPITAL LETTER AO
  815. { 0xA736, 0xA737 }, // LATIN CAPITAL LETTER AU
  816. { 0xA738, 0xA739 }, // LATIN CAPITAL LETTER AV
  817. { 0xA73A, 0xA73B }, // LATIN CAPITAL LETTER AV WITH HORIZONTAL BAR
  818. { 0xA73C, 0xA73D }, // LATIN CAPITAL LETTER AY
  819. { 0xA73E, 0xA73F }, // LATIN CAPITAL LETTER REVERSED C WITH DOT
  820. { 0xA740, 0xA741 }, // LATIN CAPITAL LETTER K WITH STROKE
  821. { 0xA742, 0xA743 }, // LATIN CAPITAL LETTER K WITH DIAGONAL STROKE
  822. { 0xA744, 0xA745 }, // LATIN CAPITAL LETTER K WITH STROKE AND DIAGONAL STROKE
  823. { 0xA746, 0xA747 }, // LATIN CAPITAL LETTER BROKEN L
  824. { 0xA748, 0xA749 }, // LATIN CAPITAL LETTER L WITH HIGH STROKE
  825. { 0xA74A, 0xA74B }, // LATIN CAPITAL LETTER O WITH LONG STROKE OVERLAY
  826. { 0xA74C, 0xA74D }, // LATIN CAPITAL LETTER O WITH LOOP
  827. { 0xA74E, 0xA74F }, // LATIN CAPITAL LETTER OO
  828. { 0xA750, 0xA751 }, // LATIN CAPITAL LETTER P WITH STROKE THROUGH DESCENDER
  829. { 0xA752, 0xA753 }, // LATIN CAPITAL LETTER P WITH FLOURISH
  830. { 0xA754, 0xA755 }, // LATIN CAPITAL LETTER P WITH SQUIRREL TAIL
  831. { 0xA756, 0xA757 }, // LATIN CAPITAL LETTER Q WITH STROKE THROUGH DESCENDER
  832. { 0xA758, 0xA759 }, // LATIN CAPITAL LETTER Q WITH DIAGONAL STROKE
  833. { 0xA75A, 0xA75B }, // LATIN CAPITAL LETTER R ROTUNDA
  834. { 0xA75C, 0xA75D }, // LATIN CAPITAL LETTER RUM ROTUNDA
  835. { 0xA75E, 0xA75F }, // LATIN CAPITAL LETTER V WITH DIAGONAL STROKE
  836. { 0xA760, 0xA761 }, // LATIN CAPITAL LETTER VY
  837. { 0xA762, 0xA763 }, // LATIN CAPITAL LETTER VISIGOTHIC Z
  838. { 0xA764, 0xA765 }, // LATIN CAPITAL LETTER THORN WITH STROKE
  839. { 0xA766, 0xA767 }, // LATIN CAPITAL LETTER THORN WITH STROKE THROUGH DESCENDER
  840. { 0xA768, 0xA769 }, // LATIN CAPITAL LETTER VEND
  841. { 0xA76A, 0xA76B }, // LATIN CAPITAL LETTER ET
  842. { 0xA76C, 0xA76D }, // LATIN CAPITAL LETTER IS
  843. { 0xA76E, 0xA76F }, // LATIN CAPITAL LETTER CON
  844. { 0xA779, 0xA77A }, // LATIN CAPITAL LETTER INSULAR D
  845. { 0xA77B, 0xA77C }, // LATIN CAPITAL LETTER INSULAR F
  846. { 0xA77D, 0x1D79 }, // LATIN CAPITAL LETTER INSULAR G
  847. { 0xA77E, 0xA77F }, // LATIN CAPITAL LETTER TURNED INSULAR G
  848. { 0xA780, 0xA781 }, // LATIN CAPITAL LETTER TURNED L
  849. { 0xA782, 0xA783 }, // LATIN CAPITAL LETTER INSULAR R
  850. { 0xA784, 0xA785 }, // LATIN CAPITAL LETTER INSULAR S
  851. { 0xA786, 0xA787 }, // LATIN CAPITAL LETTER INSULAR T
  852. { 0xA78B, 0xA78C }, // LATIN CAPITAL LETTER SALTILLO
  853. { 0xFF21, 0xFF41 }, // FULLWIDTH LATIN CAPITAL LETTER A
  854. { 0xFF22, 0xFF42 }, // FULLWIDTH LATIN CAPITAL LETTER B
  855. { 0xFF23, 0xFF43 }, // FULLWIDTH LATIN CAPITAL LETTER C
  856. { 0xFF24, 0xFF44 }, // FULLWIDTH LATIN CAPITAL LETTER D
  857. { 0xFF25, 0xFF45 }, // FULLWIDTH LATIN CAPITAL LETTER E
  858. { 0xFF26, 0xFF46 }, // FULLWIDTH LATIN CAPITAL LETTER F
  859. { 0xFF27, 0xFF47 }, // FULLWIDTH LATIN CAPITAL LETTER G
  860. { 0xFF28, 0xFF48 }, // FULLWIDTH LATIN CAPITAL LETTER H
  861. { 0xFF29, 0xFF49 }, // FULLWIDTH LATIN CAPITAL LETTER I
  862. { 0xFF2A, 0xFF4A }, // FULLWIDTH LATIN CAPITAL LETTER J
  863. { 0xFF2B, 0xFF4B }, // FULLWIDTH LATIN CAPITAL LETTER K
  864. { 0xFF2C, 0xFF4C }, // FULLWIDTH LATIN CAPITAL LETTER L
  865. { 0xFF2D, 0xFF4D }, // FULLWIDTH LATIN CAPITAL LETTER M
  866. { 0xFF2E, 0xFF4E }, // FULLWIDTH LATIN CAPITAL LETTER N
  867. { 0xFF2F, 0xFF4F }, // FULLWIDTH LATIN CAPITAL LETTER O
  868. { 0xFF30, 0xFF50 }, // FULLWIDTH LATIN CAPITAL LETTER P
  869. { 0xFF31, 0xFF51 }, // FULLWIDTH LATIN CAPITAL LETTER Q
  870. { 0xFF32, 0xFF52 }, // FULLWIDTH LATIN CAPITAL LETTER R
  871. { 0xFF33, 0xFF53 }, // FULLWIDTH LATIN CAPITAL LETTER S
  872. { 0xFF34, 0xFF54 }, // FULLWIDTH LATIN CAPITAL LETTER T
  873. { 0xFF35, 0xFF55 }, // FULLWIDTH LATIN CAPITAL LETTER U
  874. { 0xFF36, 0xFF56 }, // FULLWIDTH LATIN CAPITAL LETTER V
  875. { 0xFF37, 0xFF57 }, // FULLWIDTH LATIN CAPITAL LETTER W
  876. { 0xFF38, 0xFF58 }, // FULLWIDTH LATIN CAPITAL LETTER X
  877. { 0xFF39, 0xFF59 }, // FULLWIDTH LATIN CAPITAL LETTER Y
  878. { 0xFF3A, 0xFF5A } // FULLWIDTH LATIN CAPITAL LETTER Z
  879. };
  880. static int compare_pair_capital(const void *a, const void *b) {
  881. return (int)(*(unsigned short *)a)
  882. - (int)((struct LatinCapitalSmallPair*)b)->capital;
  883. }
  884. unsigned short latin_tolower(unsigned short c) {
  885. struct LatinCapitalSmallPair *p =
  886. (struct LatinCapitalSmallPair *)bsearch(&c, SORTED_CHAR_MAP,
  887. sizeof(SORTED_CHAR_MAP) / sizeof(SORTED_CHAR_MAP[0]),
  888. sizeof(SORTED_CHAR_MAP[0]),
  889. compare_pair_capital);
  890. return p ? p->small : c;
  891. }
  892. } // namespace latinime