eSpeak NG is an open source speech synthesizer that supports more than hundred languages and accents.
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

proplist.c 69KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522
  1. /* PropList APIs.
  2. *
  3. * Copyright (C) 2017 Reece H. Dunn
  4. *
  5. * This file is part of ucd-tools.
  6. *
  7. * ucd-tools is free software: you can redistribute it and/or modify
  8. * it under the terms of the GNU General Public License as published by
  9. * the Free Software Foundation, either version 3 of the License, or
  10. * (at your option) any later version.
  11. *
  12. * ucd-tools is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  15. * GNU General Public License for more details.
  16. *
  17. * You should have received a copy of the GNU General Public License
  18. * along with ucd-tools. If not, see <http://www.gnu.org/licenses/>.
  19. */
  20. #include "ucd/ucd.h"
  21. static int properties_Cc(codepoint_t c)
  22. {
  23. if (c >= 0x0009 && c <= 0x000D) return UCD_PROPERTY_WHITE_SPACE;
  24. if (c == 0x0085) return UCD_PROPERTY_WHITE_SPACE;
  25. return 0;
  26. }
  27. static int properties_Cf(codepoint_t c)
  28. {
  29. switch (c & 0xFFFFFF00)
  30. {
  31. case 0x0000:
  32. if (c == 0x00AD) return UCD_PROPERTY_HYPHEN;
  33. break;
  34. case 0x0600:
  35. if (c == 0x061C) return UCD_PROPERTY_BIDI_CONTROL;
  36. break;
  37. case 0x2000:
  38. if (c >= 0x200C && c <= 0x200D) return UCD_PROPERTY_JOIN_CONTROL;
  39. if (c >= 0x200E && c <= 0x200F) return UCD_PROPERTY_BIDI_CONTROL;
  40. if (c >= 0x202A && c <= 0x202E) return UCD_PROPERTY_BIDI_CONTROL;
  41. if (c >= 0x2061 && c <= 0x2064) return UCD_PROPERTY_OTHER_MATH;
  42. if (c >= 0x2066 && c <= 0x2069) return UCD_PROPERTY_BIDI_CONTROL;
  43. break;
  44. }
  45. return 0;
  46. }
  47. static int properties_Ll(codepoint_t c)
  48. {
  49. switch (c & 0xFFFFFF00)
  50. {
  51. case 0x0000:
  52. if (c >= 0x0061 && c <= 0x0066) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT;
  53. break;
  54. case 0x0300:
  55. if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  56. if (c == 0x03D5) return UCD_PROPERTY_OTHER_MATH;
  57. if (c >= 0x03F0 && c <= 0x03F1) return UCD_PROPERTY_OTHER_MATH;
  58. if (c >= 0x03F4 && c <= 0x03F5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  59. break;
  60. case 0xFF00:
  61. if (c >= 0xFF41 && c <= 0xFF46) return UCD_PROPERTY_HEX_DIGIT;
  62. break;
  63. case 0x2100:
  64. if (c >= 0x210A && c <= 0x2113) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  65. if (c >= 0x212F && c <= 0x2131) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  66. if (c >= 0x2133 && c <= 0x2134) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  67. if (c >= 0x213C && c <= 0x213F) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  68. if (c >= 0x2145 && c <= 0x2149) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  69. break;
  70. case 0x01D400:
  71. if (c >= 0x01D400 && c <= 0x01D454) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  72. if (c >= 0x01D456 && c <= 0x01D49C) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  73. if (c >= 0x01D4AE && c <= 0x01D4B9) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  74. if (c == 0x01D4BB) return UCD_PROPERTY_OTHER_MATH;
  75. if (c >= 0x01D4BD && c <= 0x01D4C3) return UCD_PROPERTY_OTHER_MATH;
  76. if (c >= 0x01D4C5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  77. break;
  78. case 0x01D500:
  79. if (c <= 0x01D505) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  80. if (c >= 0x01D51E && c <= 0x01D539) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  81. if (c >= 0x01D552) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  82. break;
  83. case 0x01D600:
  84. if (c <= 0x01D6A5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  85. if (c >= 0x01D6C2 && c <= 0x01D6DA) return UCD_PROPERTY_OTHER_MATH;
  86. if (c >= 0x01D6DC && c <= 0x01D6FA) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  87. if (c >= 0x01D6FC) return UCD_PROPERTY_OTHER_MATH;
  88. break;
  89. case 0x01D700:
  90. if (c <= 0x01D714) return UCD_PROPERTY_OTHER_MATH;
  91. if (c >= 0x01D716 && c <= 0x01D734) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  92. if (c >= 0x01D736 && c <= 0x01D74E) return UCD_PROPERTY_OTHER_MATH;
  93. if (c >= 0x01D750 && c <= 0x01D76E) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  94. if (c >= 0x01D770 && c <= 0x01D788) return UCD_PROPERTY_OTHER_MATH;
  95. if (c >= 0x01D78A && c <= 0x01D7A8) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  96. if (c >= 0x01D7AA && c <= 0x01D7C2) return UCD_PROPERTY_OTHER_MATH;
  97. if (c >= 0x01D7C4 && c <= 0x01D7CB) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  98. break;
  99. }
  100. return 0;
  101. }
  102. static int properties_Lm(codepoint_t c)
  103. {
  104. switch (c & 0xFFFFFF00)
  105. {
  106. case 0x0200:
  107. if (c >= 0x02B0 && c <= 0x02C1) return UCD_PROPERTY_DIACRITIC;
  108. if (c >= 0x02C6 && c <= 0x02CF) return UCD_PROPERTY_DIACRITIC;
  109. if (c >= 0x02D0 && c <= 0x02D1) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  110. if (c >= 0x02E0 && c <= 0x02E4) return UCD_PROPERTY_DIACRITIC;
  111. if (c == 0x02EC) return UCD_PROPERTY_DIACRITIC;
  112. if (c == 0x02EE) return UCD_PROPERTY_DIACRITIC;
  113. break;
  114. case 0x0300:
  115. if (c == 0x0374) return UCD_PROPERTY_DIACRITIC;
  116. if (c == 0x037A) return UCD_PROPERTY_DIACRITIC;
  117. break;
  118. case 0x0500:
  119. if (c == 0x0559) return UCD_PROPERTY_DIACRITIC;
  120. break;
  121. case 0x0600:
  122. if (c == 0x0640) return UCD_PROPERTY_EXTENDER;
  123. if (c >= 0x06E5 && c <= 0x06E6) return UCD_PROPERTY_DIACRITIC;
  124. break;
  125. case 0x0700:
  126. if (c >= 0x07F4 && c <= 0x07F5) return UCD_PROPERTY_DIACRITIC;
  127. if (c == 0x07FA) return UCD_PROPERTY_EXTENDER;
  128. break;
  129. case 0x0900:
  130. if (c == 0x0971) return UCD_PROPERTY_DIACRITIC;
  131. break;
  132. case 0x0E00:
  133. if (c == 0x0E46) return UCD_PROPERTY_EXTENDER;
  134. if (c == 0x0EC6) return UCD_PROPERTY_EXTENDER;
  135. break;
  136. case 0x1800:
  137. if (c == 0x1843) return UCD_PROPERTY_EXTENDER;
  138. break;
  139. case 0x1A00:
  140. if (c == 0x1AA7) return UCD_PROPERTY_EXTENDER;
  141. break;
  142. case 0x1C00:
  143. if (c == 0x1C7B) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  144. if (c >= 0x1C78 && c <= 0x1C7D) return UCD_PROPERTY_DIACRITIC;
  145. break;
  146. case 0x1D00:
  147. if (c >= 0x1D2C && c <= 0x1D6A) return UCD_PROPERTY_DIACRITIC;
  148. break;
  149. case 0x2E00:
  150. if (c == 0x2E2F) return UCD_PROPERTY_DIACRITIC;
  151. break;
  152. case 0x3000:
  153. if (c == 0x3005) return UCD_PROPERTY_EXTENDER;
  154. if (c >= 0x3031 && c <= 0x3035) return UCD_PROPERTY_EXTENDER;
  155. if (c >= 0x309D && c <= 0x309E) return UCD_PROPERTY_EXTENDER;
  156. if (c == 0x30FC) return UCD_PROPERTY_EXTENDER | UCD_PROPERTY_DIACRITIC;
  157. if (c >= 0x30FD && c <= 0x30FE) return UCD_PROPERTY_EXTENDER;
  158. break;
  159. case 0xA000:
  160. if (c == 0xA015) return UCD_PROPERTY_EXTENDER;
  161. break;
  162. case 0xA600:
  163. if (c == 0xA60C) return UCD_PROPERTY_EXTENDER;
  164. if (c == 0xA67F) return UCD_PROPERTY_DIACRITIC;
  165. if (c >= 0xA69C && c <= 0xA69D) return UCD_PROPERTY_DIACRITIC;
  166. break;
  167. case 0xA700:
  168. if (c >= 0xA717 && c <= 0xA71F) return UCD_PROPERTY_DIACRITIC;
  169. if (c == 0xA788) return UCD_PROPERTY_DIACRITIC;
  170. if (c >= 0xA7F8 && c <= 0xA7F9) return UCD_PROPERTY_DIACRITIC;
  171. break;
  172. case 0xA900:
  173. if (c == 0xA9CF) return UCD_PROPERTY_EXTENDER;
  174. if (c == 0xA9E6) return UCD_PROPERTY_EXTENDER;
  175. break;
  176. case 0xAA00:
  177. if (c == 0xAA70) return UCD_PROPERTY_EXTENDER;
  178. if (c == 0xAADD) return UCD_PROPERTY_EXTENDER;
  179. if (c >= 0xAAF3 && c <= 0xAAF4) return UCD_PROPERTY_EXTENDER;
  180. break;
  181. case 0xAB00:
  182. if (c >= 0xAB5C && c <= 0xAB5F) return UCD_PROPERTY_DIACRITIC;
  183. break;
  184. case 0xFF00:
  185. if (c == 0xFF70) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  186. if (c >= 0xFF9E && c <= 0xFF9F) return UCD_PROPERTY_DIACRITIC;
  187. break;
  188. case 0x016B00:
  189. if (c >= 0x016B42 && c <= 0x016B43) return UCD_PROPERTY_EXTENDER;
  190. break;
  191. case 0x016F00:
  192. if (c >= 0x016F93 && c <= 0x016F9F) return UCD_PROPERTY_DIACRITIC;
  193. if (c == 0x016FE0) return UCD_PROPERTY_EXTENDER;
  194. break;
  195. }
  196. return 0;
  197. }
  198. static int properties_Lo(codepoint_t c)
  199. {
  200. switch (c & 0xFFFFFF00)
  201. {
  202. case 0x2100:
  203. if (c >= 0x2135 && c <= 0x2138) return UCD_PROPERTY_OTHER_MATH;
  204. break;
  205. case 0x3000:
  206. if (c == 0x3006) return UCD_PROPERTY_IDEOGRAPHIC;
  207. break;
  208. case 0xAA00:
  209. if (c == 0xAAC0) return UCD_PROPERTY_DIACRITIC;
  210. if (c == 0xAAC2) return UCD_PROPERTY_DIACRITIC;
  211. break;
  212. case 0x11300:
  213. if (c == 0x01135D) return UCD_PROPERTY_EXTENDER;
  214. break;
  215. case 0x1EE00:
  216. return UCD_PROPERTY_OTHER_MATH;
  217. }
  218. return 0;
  219. }
  220. static int properties_Lo_ideographic(codepoint_t c)
  221. {
  222. switch (c & 0xFFFF0000)
  223. {
  224. case 0x000000:
  225. if (c >= 0x3400 && c <= 0x4DB5) return UCD_PROPERTY_IDEOGRAPHIC;
  226. if (c >= 0x4E00 && c <= 0x9FD5) return UCD_PROPERTY_IDEOGRAPHIC;
  227. if (c >= 0xF900 && c <= 0xFA6D) return UCD_PROPERTY_IDEOGRAPHIC;
  228. if (c >= 0xFA70 && c <= 0xFAD9) return UCD_PROPERTY_IDEOGRAPHIC;
  229. break;
  230. case 0x010000:
  231. if (c >= 0x017000 && c <= 0x0187EC) return UCD_PROPERTY_IDEOGRAPHIC;
  232. if (c >= 0x018800 && c <= 0x018AF2) return UCD_PROPERTY_IDEOGRAPHIC;
  233. break;
  234. case 0x020000:
  235. if (c >= 0x020000 && c <= 0x02A6D6) return UCD_PROPERTY_IDEOGRAPHIC;
  236. if (c >= 0x02A700 && c <= 0x02B734) return UCD_PROPERTY_IDEOGRAPHIC;
  237. if (c >= 0x02B740 && c <= 0x02B81D) return UCD_PROPERTY_IDEOGRAPHIC;
  238. if (c >= 0x02B820 && c <= 0x02CEA1) return UCD_PROPERTY_IDEOGRAPHIC;
  239. if (c >= 0x02F800 && c <= 0x02FA1D) return UCD_PROPERTY_IDEOGRAPHIC;
  240. break;
  241. }
  242. return 0;
  243. }
  244. static int properties_Lu(codepoint_t c)
  245. {
  246. switch (c & 0xFFFFFF00)
  247. {
  248. case 0x0000:
  249. if (c >= 0x0041 && c <= 0x0046) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT;
  250. break;
  251. case 0x0300:
  252. if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  253. if (c >= 0x03F4 && c <= 0x03F5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  254. break;
  255. case 0xFF00:
  256. if (c >= 0xFF21 && c <= 0xFF26) return UCD_PROPERTY_HEX_DIGIT;
  257. break;
  258. case 0x2100:
  259. if (c == 0x2102) return UCD_PROPERTY_OTHER_MATH;
  260. if (c == 0x2107) return UCD_PROPERTY_OTHER_MATH;
  261. if (c >= 0x210A && c <= 0x2113) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  262. if (c == 0x2115) return UCD_PROPERTY_OTHER_MATH;
  263. if (c >= 0x2119 && c <= 0x211D) return UCD_PROPERTY_OTHER_MATH;
  264. if (c == 0x2124) return UCD_PROPERTY_OTHER_MATH;
  265. if (c == 0x2128) return UCD_PROPERTY_OTHER_MATH;
  266. if (c >= 0x212C && c <= 0x212D) return UCD_PROPERTY_OTHER_MATH;
  267. if (c >= 0x212F && c <= 0x2131) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  268. if (c >= 0x2133 && c <= 0x2134) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  269. if (c >= 0x213C && c <= 0x213F) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  270. if (c >= 0x2145 && c <= 0x2149) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  271. break;
  272. case 0x01D400:
  273. if (c >= 0x01D400 && c <= 0x01D454) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  274. if (c >= 0x01D456 && c <= 0x01D49C) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  275. if (c >= 0x01D49E && c <= 0x01D49F) return UCD_PROPERTY_OTHER_MATH;
  276. if (c == 0x01D4A2) return UCD_PROPERTY_OTHER_MATH;
  277. if (c >= 0x01D4A5 && c <= 0x01D4A6) return UCD_PROPERTY_OTHER_MATH;
  278. if (c >= 0x01D4A9 && c <= 0x01D4AC) return UCD_PROPERTY_OTHER_MATH;
  279. if (c >= 0x01D4AE && c <= 0x01D4B9) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  280. if (c >= 0x01D4C5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  281. break;
  282. case 0x01D500:
  283. if (c <= 0x01D505) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  284. if (c >= 0x01D507 && c <= 0x01D50A) return UCD_PROPERTY_OTHER_MATH;
  285. if (c >= 0x01D50D && c <= 0x01D514) return UCD_PROPERTY_OTHER_MATH;
  286. if (c >= 0x01D516 && c <= 0x01D51C) return UCD_PROPERTY_OTHER_MATH;
  287. if (c >= 0x01D51E && c <= 0x01D539) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  288. if (c >= 0x01D53B && c <= 0x01D53E) return UCD_PROPERTY_OTHER_MATH;
  289. if (c >= 0x01D540 && c <= 0x01D544) return UCD_PROPERTY_OTHER_MATH;
  290. if (c == 0x01D546) return UCD_PROPERTY_OTHER_MATH;
  291. if (c >= 0x01D54A && c <= 0x01D550) return UCD_PROPERTY_OTHER_MATH;
  292. if (c >= 0x01D552) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  293. break;
  294. case 0x01D600:
  295. if (c <= 0x01D6A5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  296. if (c >= 0x01D6A8 && c <= 0x01D6C0) return UCD_PROPERTY_OTHER_MATH;
  297. if (c >= 0x01D6DC && c <= 0x01D6FA) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  298. break;
  299. case 0x01D700:
  300. if (c >= 0x01D716 && c <= 0x01D734) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  301. if (c >= 0x01D750 && c <= 0x01D76E) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  302. if (c >= 0x01D78A && c <= 0x01D7A8) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  303. if (c >= 0x01D7C4 && c <= 0x01D7CB) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  304. break;
  305. }
  306. return 0;
  307. }
  308. static int properties_Mc(codepoint_t c)
  309. {
  310. switch (c & 0xFFFFFF00)
  311. {
  312. case 0x0900:
  313. if (c == 0x0903) return UCD_PROPERTY_OTHER_ALPHABETIC;
  314. if (c == 0x093B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  315. if (c >= 0x093E && c <= 0x0940) return UCD_PROPERTY_OTHER_ALPHABETIC;
  316. if (c >= 0x0949 && c <= 0x094C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  317. if (c >= 0x094E && c <= 0x094F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  318. if (c >= 0x0982 && c <= 0x0983) return UCD_PROPERTY_OTHER_ALPHABETIC;
  319. if (c >= 0x09BE && c <= 0x09C0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  320. if (c >= 0x09C7 && c <= 0x09C8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  321. if (c >= 0x09CB && c <= 0x09CC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  322. if (c == 0x09D7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  323. break;
  324. case 0x0A00:
  325. if (c == 0x0A03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  326. if (c >= 0x0A3E && c <= 0x0A40) return UCD_PROPERTY_OTHER_ALPHABETIC;
  327. if (c == 0x0A83) return UCD_PROPERTY_OTHER_ALPHABETIC;
  328. if (c >= 0x0ABE && c <= 0x0AC0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  329. if (c == 0x0AC9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  330. if (c >= 0x0ACB && c <= 0x0ACC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  331. break;
  332. case 0x0B00:
  333. if (c >= 0x0B02 && c <= 0x0B03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  334. if (c == 0x0B3E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  335. if (c == 0x0B40) return UCD_PROPERTY_OTHER_ALPHABETIC;
  336. if (c >= 0x0B47 && c <= 0x0B48) return UCD_PROPERTY_OTHER_ALPHABETIC;
  337. if (c >= 0x0B4B && c <= 0x0B4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  338. if (c == 0x0B57) return UCD_PROPERTY_OTHER_ALPHABETIC;
  339. if (c >= 0x0BBE && c <= 0x0BBF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  340. if (c >= 0x0BC1 && c <= 0x0BC2) return UCD_PROPERTY_OTHER_ALPHABETIC;
  341. if (c >= 0x0BC6 && c <= 0x0BC8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  342. if (c >= 0x0BCA && c <= 0x0BCC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  343. if (c == 0x0BD7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  344. break;
  345. case 0x0C00:
  346. if (c >= 0x0C01 && c <= 0x0C03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  347. if (c >= 0x0C41 && c <= 0x0C44) return UCD_PROPERTY_OTHER_ALPHABETIC;
  348. if (c >= 0x0C82 && c <= 0x0C83) return UCD_PROPERTY_OTHER_ALPHABETIC;
  349. if (c == 0x0CBE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  350. if (c >= 0x0CC0 && c <= 0x0CC4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  351. if (c >= 0x0CC7 && c <= 0x0CC8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  352. if (c >= 0x0CCA && c <= 0x0CCB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  353. if (c >= 0x0CD5 && c <= 0x0CD6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  354. break;
  355. case 0x0D00:
  356. if (c >= 0x0D02 && c <= 0x0D03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  357. if (c >= 0x0D3E && c <= 0x0D40) return UCD_PROPERTY_OTHER_ALPHABETIC;
  358. if (c >= 0x0D46 && c <= 0x0D48) return UCD_PROPERTY_OTHER_ALPHABETIC;
  359. if (c >= 0x0D4A && c <= 0x0D4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  360. if (c == 0x0D57) return UCD_PROPERTY_OTHER_ALPHABETIC;
  361. if (c >= 0x0D82 && c <= 0x0D83) return UCD_PROPERTY_OTHER_ALPHABETIC;
  362. if (c >= 0x0DCF && c <= 0x0DD1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  363. if (c >= 0x0DD8 && c <= 0x0DDF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  364. if (c >= 0x0DF2 && c <= 0x0DF3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  365. break;
  366. case 0x0F00:
  367. if (c >= 0x0F3E && c <= 0x0F3F) return UCD_PROPERTY_DIACRITIC;
  368. if (c == 0x0F7F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  369. break;
  370. case 0x1000:
  371. if (c >= 0x102B && c <= 0x102C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  372. if (c == 0x1031) return UCD_PROPERTY_OTHER_ALPHABETIC;
  373. if (c == 0x1038) return UCD_PROPERTY_OTHER_ALPHABETIC;
  374. if (c >= 0x103B && c <= 0x103C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  375. if (c >= 0x1056 && c <= 0x1057) return UCD_PROPERTY_OTHER_ALPHABETIC;
  376. if (c == 0x1062) return UCD_PROPERTY_OTHER_ALPHABETIC;
  377. if (c >= 0x1067 && c <= 0x1068) return UCD_PROPERTY_OTHER_ALPHABETIC;
  378. if (c >= 0x1083 && c <= 0x1084) return UCD_PROPERTY_OTHER_ALPHABETIC;
  379. if (c >= 0x1087 && c <= 0x108C) return UCD_PROPERTY_DIACRITIC;
  380. if (c == 0x108F) return UCD_PROPERTY_DIACRITIC;
  381. if (c >= 0x109A && c <= 0x109B) return UCD_PROPERTY_DIACRITIC;
  382. if (c == 0x109C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  383. break;
  384. case 0x1700:
  385. if (c == 0x17B6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  386. if (c >= 0x17BE && c <= 0x17C5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  387. if (c >= 0x17C7 && c <= 0x17C8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  388. break;
  389. case 0x1900:
  390. if (c >= 0x1923 && c <= 0x1926) return UCD_PROPERTY_OTHER_ALPHABETIC;
  391. if (c >= 0x1929 && c <= 0x192B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  392. if (c >= 0x1930 && c <= 0x1931) return UCD_PROPERTY_OTHER_ALPHABETIC;
  393. if (c >= 0x1933 && c <= 0x1938) return UCD_PROPERTY_OTHER_ALPHABETIC;
  394. break;
  395. case 0x1A00:
  396. if (c >= 0x1A19 && c <= 0x1A1A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  397. if (c == 0x1A55) return UCD_PROPERTY_OTHER_ALPHABETIC;
  398. if (c == 0x1A57) return UCD_PROPERTY_OTHER_ALPHABETIC;
  399. if (c == 0x1A61) return UCD_PROPERTY_OTHER_ALPHABETIC;
  400. if (c >= 0x1A63 && c <= 0x1A64) return UCD_PROPERTY_OTHER_ALPHABETIC;
  401. if (c >= 0x1A6D && c <= 0x1A72) return UCD_PROPERTY_OTHER_ALPHABETIC;
  402. break;
  403. case 0x1B00:
  404. if (c == 0x1B04) return UCD_PROPERTY_OTHER_ALPHABETIC;
  405. if (c == 0x1B35) return UCD_PROPERTY_OTHER_ALPHABETIC;
  406. if (c == 0x1B3B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  407. if (c >= 0x1B3D && c <= 0x1B41) return UCD_PROPERTY_OTHER_ALPHABETIC;
  408. if (c == 0x1B43) return UCD_PROPERTY_OTHER_ALPHABETIC;
  409. if (c == 0x1B44) return UCD_PROPERTY_DIACRITIC;
  410. if (c == 0x1B82) return UCD_PROPERTY_OTHER_ALPHABETIC;
  411. if (c == 0x1BA1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  412. if (c >= 0x1BA6 && c <= 0x1BA7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  413. if (c == 0x1BAA) return UCD_PROPERTY_DIACRITIC;
  414. if (c == 0x1BE7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  415. if (c >= 0x1BEA && c <= 0x1BEC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  416. if (c == 0x1BEE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  417. break;
  418. case 0x1C00:
  419. if (c >= 0x1C24 && c <= 0x1C2B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  420. if (c >= 0x1C34 && c <= 0x1C35) return UCD_PROPERTY_OTHER_ALPHABETIC;
  421. if (c == 0x1CE1) return UCD_PROPERTY_DIACRITIC;
  422. if (c >= 0x1CF2 && c <= 0x1CF3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  423. break;
  424. case 0x3000:
  425. if (c >= 0x302E && c <= 0x302F) return UCD_PROPERTY_DIACRITIC;
  426. break;
  427. case 0xA800:
  428. if (c >= 0xA823 && c <= 0xA824) return UCD_PROPERTY_OTHER_ALPHABETIC;
  429. if (c == 0xA827) return UCD_PROPERTY_OTHER_ALPHABETIC;
  430. if (c >= 0xA880 && c <= 0xA881) return UCD_PROPERTY_OTHER_ALPHABETIC;
  431. if (c >= 0xA8B4 && c <= 0xA8C3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  432. break;
  433. case 0xA900:
  434. if (c == 0xA952) return UCD_PROPERTY_OTHER_ALPHABETIC;
  435. if (c == 0xA953) return UCD_PROPERTY_DIACRITIC;
  436. if (c == 0xA983) return UCD_PROPERTY_OTHER_ALPHABETIC;
  437. if (c >= 0xA9B4 && c <= 0xA9B5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  438. if (c >= 0xA9BA && c <= 0xA9BB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  439. if (c >= 0xA9BD && c <= 0xA9BF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  440. if (c == 0xA9C0) return UCD_PROPERTY_DIACRITIC;
  441. break;
  442. case 0xAA00:
  443. if (c >= 0xAA2F && c <= 0xAA30) return UCD_PROPERTY_OTHER_ALPHABETIC;
  444. if (c >= 0xAA33 && c <= 0xAA34) return UCD_PROPERTY_OTHER_ALPHABETIC;
  445. if (c == 0xAA4D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  446. if (c == 0xAA7B) return UCD_PROPERTY_DIACRITIC;
  447. if (c == 0xAA7D) return UCD_PROPERTY_DIACRITIC;
  448. if (c == 0xAAEB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  449. if (c >= 0xAAEE && c <= 0xAAEF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  450. if (c == 0xAAF5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  451. break;
  452. case 0xAB00:
  453. if (c >= 0xABE3 && c <= 0xABE4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  454. if (c >= 0xABE6 && c <= 0xABE7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  455. if (c >= 0xABE9 && c <= 0xABEA) return UCD_PROPERTY_OTHER_ALPHABETIC;
  456. if (c == 0xABEC) return UCD_PROPERTY_DIACRITIC;
  457. break;
  458. case 0x011000:
  459. if (c == 0x011000) return UCD_PROPERTY_OTHER_ALPHABETIC;
  460. if (c == 0x011002) return UCD_PROPERTY_OTHER_ALPHABETIC;
  461. if (c == 0x011082) return UCD_PROPERTY_OTHER_ALPHABETIC;
  462. if (c >= 0x0110B0 && c <= 0x0110B2) return UCD_PROPERTY_OTHER_ALPHABETIC;
  463. if (c >= 0x0110B7 && c <= 0x0110B8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  464. break;
  465. case 0x011100:
  466. if (c == 0x01112C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  467. if (c == 0x011182) return UCD_PROPERTY_OTHER_ALPHABETIC;
  468. if (c >= 0x0111B3 && c <= 0x0111B5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  469. if (c == 0x0111BF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  470. if (c == 0x0111C0) return UCD_PROPERTY_DIACRITIC;
  471. break;
  472. case 0x011200:
  473. if (c >= 0x01122C && c <= 0x01122E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  474. if (c >= 0x011232 && c <= 0x011233) return UCD_PROPERTY_OTHER_ALPHABETIC;
  475. if (c == 0x011235) return UCD_PROPERTY_DIACRITIC;
  476. if (c >= 0x0112E0 && c <= 0x0112E2) return UCD_PROPERTY_OTHER_ALPHABETIC;
  477. break;
  478. case 0x011300:
  479. if (c >= 0x011302 && c <= 0x011303) return UCD_PROPERTY_OTHER_ALPHABETIC;
  480. if (c >= 0x01133E && c <= 0x01133F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  481. if (c >= 0x011341 && c <= 0x011344) return UCD_PROPERTY_OTHER_ALPHABETIC;
  482. if (c >= 0x011347 && c <= 0x011348) return UCD_PROPERTY_OTHER_ALPHABETIC;
  483. if (c >= 0x01134B && c <= 0x01134C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  484. if (c == 0x01134D) return UCD_PROPERTY_DIACRITIC;
  485. if (c == 0x011357) return UCD_PROPERTY_OTHER_ALPHABETIC;
  486. if (c >= 0x011362 && c <= 0x011363) return UCD_PROPERTY_OTHER_ALPHABETIC;
  487. break;
  488. case 0x011400:
  489. if (c >= 0x011435 && c <= 0x011437) return UCD_PROPERTY_OTHER_ALPHABETIC;
  490. if (c >= 0x011440 && c <= 0x011441) return UCD_PROPERTY_OTHER_ALPHABETIC;
  491. if (c == 0x011445) return UCD_PROPERTY_OTHER_ALPHABETIC;
  492. if (c >= 0x0114B0 && c <= 0x0114B2) return UCD_PROPERTY_OTHER_ALPHABETIC;
  493. if (c == 0x0114B9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  494. if (c >= 0x0114BB && c <= 0x0114BE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  495. if (c == 0x0114C1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  496. break;
  497. case 0x011500:
  498. if (c >= 0x0115AF && c <= 0x0115B1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  499. if (c >= 0x0115B8 && c <= 0x0115BB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  500. if (c == 0x0115BE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  501. break;
  502. case 0x011600:
  503. if (c >= 0x011630 && c <= 0x011632) return UCD_PROPERTY_OTHER_ALPHABETIC;
  504. if (c >= 0x01163B && c <= 0x01163C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  505. if (c == 0x01163E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  506. if (c == 0x0116AC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  507. if (c >= 0x0116AE && c <= 0x0116AF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  508. if (c == 0x0116B6) return UCD_PROPERTY_DIACRITIC;
  509. break;
  510. case 0x011700:
  511. if (c >= 0x011720 && c <= 0x011721) return UCD_PROPERTY_OTHER_ALPHABETIC;
  512. if (c == 0x011726) return UCD_PROPERTY_OTHER_ALPHABETIC;
  513. break;
  514. case 0x011C00:
  515. if (c == 0x011C2F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  516. if (c == 0x011C3E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  517. if (c == 0x011CA9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  518. if (c == 0x011CB1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  519. if (c == 0x011CB4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  520. break;
  521. case 0x01D100:
  522. if (c >= 0x01D16D && c <= 0x01D172) return UCD_PROPERTY_DIACRITIC;
  523. break;
  524. case 0x016F00:
  525. if (c >= 0x016F51 && c <= 0x016F7E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  526. break;
  527. }
  528. return 0;
  529. }
  530. static int properties_Mn(codepoint_t c)
  531. {
  532. switch (c & 0xFFFFFF00)
  533. {
  534. case 0x0300:
  535. if (c >= 0x0300 && c <= 0x0344) return UCD_PROPERTY_DIACRITIC;
  536. if (c == 0x0345) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  537. if (c >= 0x0346 && c <= 0x034E) return UCD_PROPERTY_DIACRITIC;
  538. if (c >= 0x0350 && c <= 0x0357) return UCD_PROPERTY_DIACRITIC;
  539. if (c >= 0x035D && c <= 0x0362) return UCD_PROPERTY_DIACRITIC;
  540. break;
  541. case 0x0400:
  542. if (c >= 0x0483 && c <= 0x0487) return UCD_PROPERTY_DIACRITIC;
  543. break;
  544. case 0x0500:
  545. if (c >= 0x0591 && c <= 0x05A1) return UCD_PROPERTY_DIACRITIC;
  546. if (c >= 0x05A3 && c <= 0x05AF) return UCD_PROPERTY_DIACRITIC;
  547. if (c >= 0x05B0 && c <= 0x05BD) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  548. if (c == 0x05BF) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  549. if (c >= 0x05C1 && c <= 0x05C2) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  550. if (c == 0x05C4) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  551. if (c == 0x05C5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  552. if (c == 0x05C7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  553. break;
  554. case 0x0600:
  555. if (c >= 0x0610 && c <= 0x061A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  556. if (c >= 0x064B && c <= 0x0652) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  557. if (c >= 0x0653 && c <= 0x0656) return UCD_PROPERTY_OTHER_ALPHABETIC;
  558. if (c == 0x0657) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  559. if (c == 0x0658) return UCD_PROPERTY_DIACRITIC;
  560. if (c >= 0x0659 && c <= 0x065F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  561. if (c == 0x0670) return UCD_PROPERTY_OTHER_ALPHABETIC;
  562. if (c >= 0x06D6 && c <= 0x06DC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  563. if (c >= 0x06DF && c <= 0x06E0) return UCD_PROPERTY_DIACRITIC;
  564. if (c >= 0x06E1 && c <= 0x06E4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  565. if (c >= 0x06E7 && c <= 0x06E8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  566. if (c >= 0x06EA && c <= 0x06EC) return UCD_PROPERTY_DIACRITIC;
  567. if (c == 0x06ED) return UCD_PROPERTY_OTHER_ALPHABETIC;
  568. break;
  569. case 0x0700:
  570. if (c == 0x0711) return UCD_PROPERTY_OTHER_ALPHABETIC;
  571. if (c >= 0x0730 && c <= 0x073F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  572. if (c >= 0x0740 && c <= 0x074A) return UCD_PROPERTY_DIACRITIC;
  573. if (c >= 0x07A6 && c <= 0x07B0) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  574. if (c >= 0x07EB && c <= 0x07F3) return UCD_PROPERTY_DIACRITIC;
  575. break;
  576. case 0x0800:
  577. if (c >= 0x0816 && c <= 0x0817) return UCD_PROPERTY_OTHER_ALPHABETIC;
  578. if (c >= 0x0818 && c <= 0x0819) return UCD_PROPERTY_DIACRITIC;
  579. if (c >= 0x081B && c <= 0x0823) return UCD_PROPERTY_OTHER_ALPHABETIC;
  580. if (c >= 0x0825 && c <= 0x0827) return UCD_PROPERTY_OTHER_ALPHABETIC;
  581. if (c >= 0x0829 && c <= 0x082C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  582. if (c >= 0x08D4 && c <= 0x08DF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  583. if (c >= 0x08E3 && c <= 0x08E9) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  584. if (c >= 0x08EA && c <= 0x08EF) return UCD_PROPERTY_DIACRITIC;
  585. if (c >= 0x08F0 && c <= 0x08FE) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  586. if (c == 0x08FF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  587. break;
  588. case 0x0900:
  589. if (c <= 0x0902) return UCD_PROPERTY_OTHER_ALPHABETIC;
  590. if (c == 0x093A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  591. if (c == 0x093C) return UCD_PROPERTY_DIACRITIC;
  592. if (c >= 0x0941 && c <= 0x0948) return UCD_PROPERTY_OTHER_ALPHABETIC;
  593. if (c == 0x094D) return UCD_PROPERTY_DIACRITIC;
  594. if (c >= 0x0951 && c <= 0x0954) return UCD_PROPERTY_DIACRITIC;
  595. if (c >= 0x0955 && c <= 0x0957) return UCD_PROPERTY_OTHER_ALPHABETIC;
  596. if (c >= 0x0962 && c <= 0x0963) return UCD_PROPERTY_OTHER_ALPHABETIC;
  597. if (c == 0x0981) return UCD_PROPERTY_OTHER_ALPHABETIC;
  598. if (c == 0x09BC) return UCD_PROPERTY_DIACRITIC;
  599. if (c >= 0x09C1 && c <= 0x09C4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  600. if (c == 0x09CD) return UCD_PROPERTY_DIACRITIC;
  601. if (c >= 0x09E2 && c <= 0x09E3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  602. break;
  603. case 0x0A00:
  604. if (c >= 0x0A01 && c <= 0x0A02) return UCD_PROPERTY_OTHER_ALPHABETIC;
  605. if (c == 0x0A3C) return UCD_PROPERTY_DIACRITIC;
  606. if (c >= 0x0A41 && c <= 0x0A42) return UCD_PROPERTY_OTHER_ALPHABETIC;
  607. if (c >= 0x0A47 && c <= 0x0A48) return UCD_PROPERTY_OTHER_ALPHABETIC;
  608. if (c >= 0x0A4B && c <= 0x0A4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  609. if (c == 0x0A4D) return UCD_PROPERTY_DIACRITIC;
  610. if (c == 0x0A51) return UCD_PROPERTY_OTHER_ALPHABETIC;
  611. if (c >= 0x0A70 && c <= 0x0A71) return UCD_PROPERTY_OTHER_ALPHABETIC;
  612. if (c == 0x0A75) return UCD_PROPERTY_OTHER_ALPHABETIC;
  613. if (c >= 0x0A81 && c <= 0x0A82) return UCD_PROPERTY_OTHER_ALPHABETIC;
  614. if (c == 0x0ABC) return UCD_PROPERTY_DIACRITIC;
  615. if (c >= 0x0AC1 && c <= 0x0AC5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  616. if (c >= 0x0AC7 && c <= 0x0AC8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  617. if (c == 0x0ACD) return UCD_PROPERTY_DIACRITIC;
  618. if (c >= 0x0AE2 && c <= 0x0AE3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  619. break;
  620. case 0x0B00:
  621. if (c == 0x0B01) return UCD_PROPERTY_OTHER_ALPHABETIC;
  622. if (c == 0x0B3C) return UCD_PROPERTY_DIACRITIC;
  623. if (c == 0x0B3F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  624. if (c >= 0x0B41 && c <= 0x0B44) return UCD_PROPERTY_OTHER_ALPHABETIC;
  625. if (c == 0x0B4D) return UCD_PROPERTY_DIACRITIC;
  626. if (c == 0x0B56) return UCD_PROPERTY_OTHER_ALPHABETIC;
  627. if (c >= 0x0B62 && c <= 0x0B63) return UCD_PROPERTY_OTHER_ALPHABETIC;
  628. if (c == 0x0B82) return UCD_PROPERTY_OTHER_ALPHABETIC;
  629. if (c == 0x0BC0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  630. if (c == 0x0BCD) return UCD_PROPERTY_DIACRITIC;
  631. break;
  632. case 0x0C00:
  633. if (c == 0x0C00) return UCD_PROPERTY_OTHER_ALPHABETIC;
  634. if (c >= 0x0C3E && c <= 0x0C40) return UCD_PROPERTY_OTHER_ALPHABETIC;
  635. if (c >= 0x0C46 && c <= 0x0C48) return UCD_PROPERTY_OTHER_ALPHABETIC;
  636. if (c >= 0x0C4A && c <= 0x0C4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  637. if (c == 0x0C4D) return UCD_PROPERTY_DIACRITIC;
  638. if (c >= 0x0C55 && c <= 0x0C56) return UCD_PROPERTY_OTHER_ALPHABETIC;
  639. if (c >= 0x0C62 && c <= 0x0C63) return UCD_PROPERTY_OTHER_ALPHABETIC;
  640. if (c == 0x0C81) return UCD_PROPERTY_OTHER_ALPHABETIC;
  641. if (c == 0x0CBC) return UCD_PROPERTY_DIACRITIC;
  642. if (c == 0x0CBF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  643. if (c == 0x0CC6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  644. if (c == 0x0CCC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  645. if (c == 0x0CCD) return UCD_PROPERTY_DIACRITIC;
  646. if (c >= 0x0CE2 && c <= 0x0CE3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  647. break;
  648. case 0x0D00:
  649. if (c == 0x0D01) return UCD_PROPERTY_OTHER_ALPHABETIC;
  650. if (c >= 0x0D41 && c <= 0x0D44) return UCD_PROPERTY_OTHER_ALPHABETIC;
  651. if (c == 0x0D4D) return UCD_PROPERTY_DIACRITIC;
  652. if (c >= 0x0D62 && c <= 0x0D63) return UCD_PROPERTY_OTHER_ALPHABETIC;
  653. if (c == 0x0DCA) return UCD_PROPERTY_DIACRITIC;
  654. if (c >= 0x0DD2 && c <= 0x0DD4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  655. if (c == 0x0DD6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  656. break;
  657. case 0x0E00:
  658. if (c == 0x0E31) return UCD_PROPERTY_OTHER_ALPHABETIC;
  659. if (c >= 0x0E34 && c <= 0x0E3A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  660. if (c >= 0x0E47 && c <= 0x0E4C) return UCD_PROPERTY_DIACRITIC;
  661. if (c == 0x0E4D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  662. if (c == 0x0E4E) return UCD_PROPERTY_DIACRITIC;
  663. if (c == 0x0EB1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  664. if (c >= 0x0EB4 && c <= 0x0EB9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  665. if (c >= 0x0EBB && c <= 0x0EBC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  666. if (c >= 0x0EC8 && c <= 0x0ECC) return UCD_PROPERTY_DIACRITIC;
  667. if (c == 0x0ECD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  668. break;
  669. case 0x0F00:
  670. if (c >= 0x0F18 && c <= 0x0F19) return UCD_PROPERTY_DIACRITIC;
  671. if (c == 0x0F35) return UCD_PROPERTY_DIACRITIC;
  672. if (c == 0x0F37) return UCD_PROPERTY_DIACRITIC;
  673. if (c == 0x0F39) return UCD_PROPERTY_DIACRITIC;
  674. if (c >= 0x0F71 && c <= 0x0F7E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  675. if (c >= 0x0F80 && c <= 0x0F81) return UCD_PROPERTY_OTHER_ALPHABETIC;
  676. if (c >= 0x0F82 && c <= 0x0F84) return UCD_PROPERTY_DIACRITIC;
  677. if (c >= 0x0F86 && c <= 0x0F87) return UCD_PROPERTY_DIACRITIC;
  678. if (c >= 0x0F8D && c <= 0x0F97) return UCD_PROPERTY_OTHER_ALPHABETIC;
  679. if (c >= 0x0F99 && c <= 0x0FBC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  680. if (c == 0x0FC6) return UCD_PROPERTY_DIACRITIC;
  681. break;
  682. case 0x1000:
  683. if (c >= 0x102D && c <= 0x1030) return UCD_PROPERTY_OTHER_ALPHABETIC;
  684. if (c >= 0x1032 && c <= 0x1036) return UCD_PROPERTY_OTHER_ALPHABETIC;
  685. if (c == 0x1037) return UCD_PROPERTY_DIACRITIC;
  686. if (c >= 0x1039 && c <= 0x103A) return UCD_PROPERTY_DIACRITIC;
  687. if (c >= 0x103D && c <= 0x103E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  688. if (c >= 0x1058 && c <= 0x1059) return UCD_PROPERTY_OTHER_ALPHABETIC;
  689. if (c >= 0x105E && c <= 0x1060) return UCD_PROPERTY_OTHER_ALPHABETIC;
  690. if (c >= 0x1071 && c <= 0x1074) return UCD_PROPERTY_OTHER_ALPHABETIC;
  691. if (c == 0x1082) return UCD_PROPERTY_OTHER_ALPHABETIC;
  692. if (c >= 0x1085 && c <= 0x1086) return UCD_PROPERTY_OTHER_ALPHABETIC;
  693. if (c == 0x108D) return UCD_PROPERTY_DIACRITIC;
  694. if (c == 0x109D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  695. break;
  696. case 0x1300:
  697. if (c == 0x135F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  698. break;
  699. case 0x1700:
  700. if (c >= 0x1712 && c <= 0x1713) return UCD_PROPERTY_OTHER_ALPHABETIC;
  701. if (c >= 0x1732 && c <= 0x1733) return UCD_PROPERTY_OTHER_ALPHABETIC;
  702. if (c >= 0x1752 && c <= 0x1753) return UCD_PROPERTY_OTHER_ALPHABETIC;
  703. if (c >= 0x1772 && c <= 0x1773) return UCD_PROPERTY_OTHER_ALPHABETIC;
  704. if (c >= 0x17B7 && c <= 0x17BD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  705. if (c == 0x17C6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  706. if (c >= 0x17C9 && c <= 0x17D3) return UCD_PROPERTY_DIACRITIC;
  707. if (c == 0x17DD) return UCD_PROPERTY_DIACRITIC;
  708. break;
  709. case 0x1800:
  710. if (c >= 0x1885 && c <= 0x1886) return UCD_PROPERTY_OTHER_ALPHABETIC;
  711. if (c == 0x18A9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  712. break;
  713. case 0x1900:
  714. if (c >= 0x1920 && c <= 0x1922) return UCD_PROPERTY_OTHER_ALPHABETIC;
  715. if (c >= 0x1927 && c <= 0x1928) return UCD_PROPERTY_OTHER_ALPHABETIC;
  716. if (c == 0x1932) return UCD_PROPERTY_OTHER_ALPHABETIC;
  717. if (c >= 0x1939 && c <= 0x193B) return UCD_PROPERTY_DIACRITIC;
  718. break;
  719. case 0x1A00:
  720. if (c >= 0x1A17 && c <= 0x1A18) return UCD_PROPERTY_OTHER_ALPHABETIC;
  721. if (c == 0x1A1B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  722. if (c == 0x1A56) return UCD_PROPERTY_OTHER_ALPHABETIC;
  723. if (c >= 0x1A58 && c <= 0x1A5E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  724. if (c == 0x1A62) return UCD_PROPERTY_OTHER_ALPHABETIC;
  725. if (c >= 0x1A65 && c <= 0x1A6C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  726. if (c >= 0x1A73 && c <= 0x1A74) return UCD_PROPERTY_OTHER_ALPHABETIC;
  727. if (c >= 0x1A75 && c <= 0x1A7C) return UCD_PROPERTY_DIACRITIC;
  728. if (c == 0x1A7F) return UCD_PROPERTY_DIACRITIC;
  729. if (c >= 0x1AB0 && c <= 0x1ABD) return UCD_PROPERTY_DIACRITIC;
  730. break;
  731. case 0x1B00:
  732. if (c >= 0x1B00 && c <= 0x1B03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  733. if (c == 0x1B34) return UCD_PROPERTY_DIACRITIC;
  734. if (c >= 0x1B36 && c <= 0x1B3A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  735. if (c == 0x1B3C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  736. if (c == 0x1B42) return UCD_PROPERTY_OTHER_ALPHABETIC;
  737. if (c >= 0x1B6B && c <= 0x1B73) return UCD_PROPERTY_DIACRITIC;
  738. if (c >= 0x1B80 && c <= 0x1B81) return UCD_PROPERTY_OTHER_ALPHABETIC;
  739. if (c >= 0x1BA2 && c <= 0x1BA5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  740. if (c >= 0x1BA8 && c <= 0x1BA9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  741. if (c == 0x1BAB) return UCD_PROPERTY_DIACRITIC;
  742. if (c >= 0x1BAC && c <= 0x1BAD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  743. if (c >= 0x1BE8 && c <= 0x1BE9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  744. if (c == 0x1BED) return UCD_PROPERTY_OTHER_ALPHABETIC;
  745. if (c >= 0x1BEF && c <= 0x1BF1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  746. break;
  747. case 0x1C00:
  748. if (c >= 0x1C2C && c <= 0x1C33) return UCD_PROPERTY_OTHER_ALPHABETIC;
  749. if (c == 0x1C36) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  750. if (c == 0x1C37) return UCD_PROPERTY_DIACRITIC;
  751. if (c >= 0x1CD0 && c <= 0x1CD2) return UCD_PROPERTY_DIACRITIC;
  752. if (c >= 0x1CD4 && c <= 0x1CE0) return UCD_PROPERTY_DIACRITIC;
  753. if (c >= 0x1CE2 && c <= 0x1CE8) return UCD_PROPERTY_DIACRITIC;
  754. if (c == 0x1CED) return UCD_PROPERTY_DIACRITIC;
  755. if (c == 0x1CF4) return UCD_PROPERTY_DIACRITIC;
  756. if (c >= 0x1CF8 && c <= 0x1CF9) return UCD_PROPERTY_DIACRITIC;
  757. break;
  758. case 0x1D00:
  759. if (c >= 0x1DC4 && c <= 0x1DCF) return UCD_PROPERTY_DIACRITIC;
  760. if (c >= 0x1DE7 && c <= 0x1DF4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  761. if (c == 0x1DF5) return UCD_PROPERTY_DIACRITIC;
  762. if (c >= 0x1DFD && c <= 0x1DFF) return UCD_PROPERTY_DIACRITIC;
  763. break;
  764. case 0x2000:
  765. if (c >= 0x20D0 && c <= 0x20DC) return UCD_PROPERTY_OTHER_MATH;
  766. if (c == 0x20E1) return UCD_PROPERTY_OTHER_MATH;
  767. if (c >= 0x20E5 && c <= 0x20E6) return UCD_PROPERTY_OTHER_MATH;
  768. if (c >= 0x20EB && c <= 0x20EF) return UCD_PROPERTY_OTHER_MATH;
  769. break;
  770. case 0x2C00:
  771. if (c >= 0x2CEF && c <= 0x2CF1) return UCD_PROPERTY_DIACRITIC;
  772. break;
  773. case 0x2D00:
  774. if (c >= 0x2DE0 && c <= 0x2DFF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  775. break;
  776. case 0x3000:
  777. if (c >= 0x302A && c <= 0x302D) return UCD_PROPERTY_DIACRITIC;
  778. if (c >= 0x3099 && c <= 0x309A) return UCD_PROPERTY_DIACRITIC;
  779. break;
  780. case 0xA600:
  781. if (c == 0xA66F) return UCD_PROPERTY_DIACRITIC;
  782. if (c >= 0xA674 && c <= 0xA67B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  783. if (c >= 0xA67C && c <= 0xA67D) return UCD_PROPERTY_DIACRITIC;
  784. if (c >= 0xA69E && c <= 0xA69F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  785. if (c >= 0xA6F0 && c <= 0xA6F1) return UCD_PROPERTY_DIACRITIC;
  786. break;
  787. case 0xA800:
  788. if (c >= 0xA825 && c <= 0xA826) return UCD_PROPERTY_OTHER_ALPHABETIC;
  789. if (c == 0xA8C4) return UCD_PROPERTY_DIACRITIC;
  790. if (c == 0xA8C5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  791. if (c >= 0xA8E0 && c <= 0xA8F1) return UCD_PROPERTY_DIACRITIC;
  792. break;
  793. case 0xA900:
  794. if (c >= 0xA926 && c <= 0xA92A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  795. if (c >= 0xA92B && c <= 0xA92D) return UCD_PROPERTY_DIACRITIC;
  796. if (c >= 0xA947 && c <= 0xA951) return UCD_PROPERTY_OTHER_ALPHABETIC;
  797. if (c >= 0xA980 && c <= 0xA982) return UCD_PROPERTY_OTHER_ALPHABETIC;
  798. if (c == 0xA9B3) return UCD_PROPERTY_DIACRITIC;
  799. if (c >= 0xA9B6 && c <= 0xA9B9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  800. if (c == 0xA9BC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  801. if (c == 0xA9E5) return UCD_PROPERTY_DIACRITIC;
  802. break;
  803. case 0xAA00:
  804. if (c >= 0xAA29 && c <= 0xAA2E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  805. if (c >= 0xAA31 && c <= 0xAA32) return UCD_PROPERTY_OTHER_ALPHABETIC;
  806. if (c >= 0xAA35 && c <= 0xAA36) return UCD_PROPERTY_OTHER_ALPHABETIC;
  807. if (c == 0xAA43) return UCD_PROPERTY_OTHER_ALPHABETIC;
  808. if (c == 0xAA4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  809. if (c == 0xAA7C) return UCD_PROPERTY_DIACRITIC;
  810. if (c == 0xAAB0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  811. if (c >= 0xAAB2 && c <= 0xAAB4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  812. if (c >= 0xAAB7 && c <= 0xAAB8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  813. if (c == 0xAABE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  814. if (c == 0xAABF) return UCD_PROPERTY_DIACRITIC;
  815. if (c == 0xAAC1) return UCD_PROPERTY_DIACRITIC;
  816. if (c >= 0xAAEC && c <= 0xAAED) return UCD_PROPERTY_OTHER_ALPHABETIC;
  817. if (c == 0xAAF6) return UCD_PROPERTY_DIACRITIC;
  818. break;
  819. case 0xAB00:
  820. if (c == 0xABE5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  821. if (c == 0xABE8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  822. if (c == 0xABED) return UCD_PROPERTY_DIACRITIC;
  823. break;
  824. case 0xFB00:
  825. if (c == 0xFB1E) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  826. break;
  827. case 0xFE00:
  828. if (c >= 0xFE20 && c <= 0xFE2F) return UCD_PROPERTY_DIACRITIC;
  829. break;
  830. case 0x010200:
  831. if (c == 0x0102E0) return UCD_PROPERTY_DIACRITIC;
  832. break;
  833. case 0x010300:
  834. if (c >= 0x010376 && c <= 0x01037A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  835. break;
  836. case 0x010A00:
  837. if (c >= 0x010A01 && c <= 0x010A03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  838. if (c >= 0x010A05 && c <= 0x010A06) return UCD_PROPERTY_OTHER_ALPHABETIC;
  839. if (c >= 0x010A0C && c <= 0x010A0F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  840. if (c >= 0x010AE5 && c <= 0x010AE6) return UCD_PROPERTY_DIACRITIC;
  841. break;
  842. case 0x011000:
  843. if (c == 0x011001) return UCD_PROPERTY_OTHER_ALPHABETIC;
  844. if (c >= 0x011038 && c <= 0x011045) return UCD_PROPERTY_OTHER_ALPHABETIC;
  845. if (c >= 0x0110B3 && c <= 0x0110B6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  846. if (c >= 0x0110B9 && c <= 0x0110BA) return UCD_PROPERTY_DIACRITIC;
  847. break;
  848. case 0x011100:
  849. if (c >= 0x011100 && c <= 0x011102) return UCD_PROPERTY_OTHER_ALPHABETIC;
  850. if (c >= 0x011127 && c <= 0x01112B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  851. if (c >= 0x01112D && c <= 0x011132) return UCD_PROPERTY_OTHER_ALPHABETIC;
  852. if (c >= 0x011133 && c <= 0x011134) return UCD_PROPERTY_DIACRITIC;
  853. if (c == 0x011173) return UCD_PROPERTY_DIACRITIC;
  854. if (c >= 0x011180 && c <= 0x011181) return UCD_PROPERTY_OTHER_ALPHABETIC;
  855. if (c >= 0x0111B6 && c <= 0x0111BE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  856. if (c >= 0x0111CA && c <= 0x0111CC) return UCD_PROPERTY_DIACRITIC;
  857. break;
  858. case 0x011200:
  859. if (c >= 0x01122F && c <= 0x011231) return UCD_PROPERTY_OTHER_ALPHABETIC;
  860. if (c == 0x011234) return UCD_PROPERTY_OTHER_ALPHABETIC;
  861. if (c == 0x011236) return UCD_PROPERTY_DIACRITIC;
  862. if (c == 0x011237) return UCD_PROPERTY_OTHER_ALPHABETIC;
  863. if (c == 0x01123E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  864. if (c == 0x0112DF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  865. if (c >= 0x0112E3 && c <= 0x0112E8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  866. if (c >= 0x0112E9 && c <= 0x0112EA) return UCD_PROPERTY_DIACRITIC;
  867. break;
  868. case 0x011300:
  869. if (c >= 0x011300 && c <= 0x011301) return UCD_PROPERTY_OTHER_ALPHABETIC;
  870. if (c == 0x01133C) return UCD_PROPERTY_DIACRITIC;
  871. if (c == 0x011340) return UCD_PROPERTY_OTHER_ALPHABETIC;
  872. if (c >= 0x011366 && c <= 0x01136C) return UCD_PROPERTY_DIACRITIC;
  873. if (c >= 0x011370 && c <= 0x011374) return UCD_PROPERTY_DIACRITIC;
  874. break;
  875. case 0x011400:
  876. if (c >= 0x011438 && c <= 0x01143F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  877. if (c == 0x011442) return UCD_PROPERTY_DIACRITIC;
  878. if (c >= 0x011443 && c <= 0x011444) return UCD_PROPERTY_OTHER_ALPHABETIC;
  879. if (c == 0x011446) return UCD_PROPERTY_DIACRITIC;
  880. if (c >= 0x0114B3 && c <= 0x0114B8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  881. if (c == 0x0114BA) return UCD_PROPERTY_OTHER_ALPHABETIC;
  882. if (c >= 0x0114BF && c <= 0x0114C0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  883. if (c >= 0x0114C2 && c <= 0x0114C3) return UCD_PROPERTY_DIACRITIC;
  884. break;
  885. case 0x011500:
  886. if (c >= 0x0115B2 && c <= 0x0115B5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  887. if (c >= 0x0115BC && c <= 0x0115BD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  888. if (c >= 0x0115BF && c <= 0x0115C0) return UCD_PROPERTY_DIACRITIC;
  889. if (c >= 0x0115DC && c <= 0x0115DD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  890. break;
  891. case 0x011600:
  892. if (c >= 0x011633 && c <= 0x01163A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  893. if (c == 0x01163D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  894. if (c == 0x01163F) return UCD_PROPERTY_DIACRITIC;
  895. if (c == 0x011640) return UCD_PROPERTY_OTHER_ALPHABETIC;
  896. if (c == 0x0116AB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  897. if (c == 0x0116AD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  898. if (c >= 0x0116B0 && c <= 0x0116B5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  899. if (c == 0x0116B7) return UCD_PROPERTY_DIACRITIC;
  900. break;
  901. case 0x011700:
  902. if (c >= 0x01171D && c <= 0x01171F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  903. if (c >= 0x011722 && c <= 0x011725) return UCD_PROPERTY_OTHER_ALPHABETIC;
  904. if (c >= 0x011727 && c <= 0x01172A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  905. if (c == 0x01172B) return UCD_PROPERTY_DIACRITIC;
  906. break;
  907. case 0x011C00:
  908. if (c >= 0x011C30 && c <= 0x011C36) return UCD_PROPERTY_OTHER_ALPHABETIC;
  909. if (c >= 0x011C38 && c <= 0x011C3D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  910. if (c == 0x011C3F) return UCD_PROPERTY_DIACRITIC;
  911. if (c >= 0x011C92 && c <= 0x011CA7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  912. if (c >= 0x011CAA && c <= 0x011CB0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  913. if (c >= 0x011CB2 && c <= 0x011CB3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  914. if (c >= 0x011CB5 && c <= 0x011CB6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  915. break;
  916. case 0x016A00:
  917. if (c >= 0x016AF0 && c <= 0x016AF4) return UCD_PROPERTY_DIACRITIC;
  918. break;
  919. case 0x016B00:
  920. if (c >= 0x016B30 && c <= 0x016B36) return UCD_PROPERTY_OTHER_ALPHABETIC;
  921. break;
  922. case 0x016F00:
  923. if (c >= 0x016F8F && c <= 0x016F92) return UCD_PROPERTY_DIACRITIC;
  924. break;
  925. case 0x01BC00:
  926. if (c == 0x01BC9E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  927. break;
  928. case 0x01D100:
  929. if (c >= 0x01D167 && c <= 0x01D169) return UCD_PROPERTY_DIACRITIC;
  930. if (c >= 0x01D17B && c <= 0x01D182) return UCD_PROPERTY_DIACRITIC;
  931. if (c >= 0x01D185 && c <= 0x01D18B) return UCD_PROPERTY_DIACRITIC;
  932. if (c >= 0x01D1AA && c <= 0x01D1AD) return UCD_PROPERTY_DIACRITIC;
  933. break;
  934. case 0x01E000:
  935. if (c >= 0x01E000 && c <= 0x01E006) return UCD_PROPERTY_OTHER_ALPHABETIC;
  936. if (c >= 0x01E008 && c <= 0x01E018) return UCD_PROPERTY_OTHER_ALPHABETIC;
  937. if (c >= 0x01E01B && c <= 0x01E021) return UCD_PROPERTY_OTHER_ALPHABETIC;
  938. if (c >= 0x01E023 && c <= 0x01E024) return UCD_PROPERTY_OTHER_ALPHABETIC;
  939. if (c >= 0x01E026 && c <= 0x01E02A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  940. break;
  941. case 0x01E800:
  942. if (c >= 0x01E8D0 && c <= 0x01E8D6) return UCD_PROPERTY_DIACRITIC;
  943. break;
  944. case 0x01E900:
  945. if (c >= 0x01E944 && c <= 0x01E946) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  946. if (c == 0x01E947) return UCD_PROPERTY_OTHER_ALPHABETIC;
  947. if (c >= 0x01E948 && c <= 0x01E94A) return UCD_PROPERTY_DIACRITIC;
  948. break;
  949. }
  950. return 0;
  951. }
  952. static int properties_Nd(codepoint_t c)
  953. {
  954. switch (c & 0xFFFFFF00)
  955. {
  956. case 0x0000:
  957. if (c >= 0x0030 && c <= 0x0039) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT;
  958. break;
  959. case 0xFF00:
  960. if (c >= 0xFF10 && c <= 0xFF19) return UCD_PROPERTY_HEX_DIGIT;
  961. break;
  962. case 0x01D700:
  963. if (c >= 0x01D7CE && c <= 0x01D7FF) return UCD_PROPERTY_OTHER_MATH;
  964. break;
  965. }
  966. return 0;
  967. }
  968. static int properties_Nl(codepoint_t c)
  969. {
  970. switch (c & 0xFFFFFF00)
  971. {
  972. case 0x3000:
  973. if (c == 0x3007) return UCD_PROPERTY_IDEOGRAPHIC;
  974. if (c >= 0x3021 && c <= 0x3029) return UCD_PROPERTY_IDEOGRAPHIC;
  975. if (c >= 0x3038 && c <= 0x303A) return UCD_PROPERTY_IDEOGRAPHIC;
  976. break;
  977. }
  978. return 0;
  979. }
  980. static int properties_Pc(codepoint_t c)
  981. {
  982. switch (c & 0xFFFFFF00)
  983. {
  984. case 0x2000:
  985. if (c == 0x2040) return UCD_PROPERTY_OTHER_MATH;
  986. break;
  987. }
  988. return 0;
  989. }
  990. static int properties_Pd(codepoint_t c)
  991. {
  992. switch (c & 0xFFFFFF00)
  993. {
  994. case 0x0000:
  995. if (c == 0x002D) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  996. break;
  997. case 0x0500:
  998. if (c == 0x058A) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  999. break;
  1000. case 0x1800:
  1001. if (c == 0x1806) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1002. break;
  1003. case 0x2000:
  1004. if (c == 0x2010) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1005. if (c == 0x2011) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN | UCD_PROPERTY_NO_BREAK;
  1006. break;
  1007. case 0x2E00:
  1008. if (c == 0x2E17) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1009. break;
  1010. case 0xFE00:
  1011. if (c == 0xFE63) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN | UCD_PROPERTY_OTHER_MATH;
  1012. break;
  1013. case 0xFF00:
  1014. if (c == 0xFF0D) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1015. break;
  1016. }
  1017. return UCD_PROPERTY_DASH;
  1018. }
  1019. static int properties_Pe(codepoint_t c)
  1020. {
  1021. switch (c & 0xFFFFFF00)
  1022. {
  1023. case 0x2000:
  1024. if (c == 0x207E) return UCD_PROPERTY_OTHER_MATH;
  1025. if (c == 0x208E) return UCD_PROPERTY_OTHER_MATH;
  1026. break;
  1027. case 0x2300:
  1028. if (c == 0x2309) return UCD_PROPERTY_OTHER_MATH;
  1029. if (c == 0x230B) return UCD_PROPERTY_OTHER_MATH;
  1030. break;
  1031. case 0x2700:
  1032. if (c == 0x27C6) return UCD_PROPERTY_OTHER_MATH;
  1033. if (c >= 0x27E6 && c <= 0x27EF) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1034. break;
  1035. case 0x2900:
  1036. if (c >= 0x2983 && c <= 0x2998) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1037. if (c >= 0x29D8 && c <= 0x29DB) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1038. if (c == 0x29FD) return UCD_PROPERTY_OTHER_MATH;
  1039. break;
  1040. case 0x3000:
  1041. if (c == 0x300D) return UCD_PROPERTY_QUOTATION_MARK;
  1042. if (c == 0x300F) return UCD_PROPERTY_QUOTATION_MARK;
  1043. if (c >= 0x301E && c <= 0x301F) return UCD_PROPERTY_QUOTATION_MARK;
  1044. break;
  1045. case 0xFE00:
  1046. if (c == 0xFE42) return UCD_PROPERTY_QUOTATION_MARK;
  1047. if (c == 0xFE44) return UCD_PROPERTY_QUOTATION_MARK;
  1048. break;
  1049. case 0xFF00:
  1050. if (c == 0xFF63) return UCD_PROPERTY_QUOTATION_MARK;
  1051. break;
  1052. }
  1053. return 0;
  1054. }
  1055. static int properties_Pf(codepoint_t c)
  1056. {
  1057. switch (c & 0xFFFFFF00)
  1058. {
  1059. case 0x0000:
  1060. if (c == 0x00BB) return UCD_PROPERTY_QUOTATION_MARK;
  1061. break;
  1062. case 0x2000:
  1063. if (c == 0x2019) return UCD_PROPERTY_QUOTATION_MARK;
  1064. if (c == 0x201D) return UCD_PROPERTY_QUOTATION_MARK;
  1065. if (c == 0x203A) return UCD_PROPERTY_QUOTATION_MARK;
  1066. break;
  1067. }
  1068. return 0;
  1069. }
  1070. static int properties_Pi(codepoint_t c)
  1071. {
  1072. switch (c & 0xFFFFFF00)
  1073. {
  1074. case 0x0000:
  1075. if (c == 0x00AB) return UCD_PROPERTY_QUOTATION_MARK;
  1076. break;
  1077. case 0x2000:
  1078. if (c == 0x2018) return UCD_PROPERTY_QUOTATION_MARK;
  1079. if (c >= 0x201B && c <= 0x201C) return UCD_PROPERTY_QUOTATION_MARK;
  1080. if (c == 0x201F) return UCD_PROPERTY_QUOTATION_MARK;
  1081. if (c == 0x2039) return UCD_PROPERTY_QUOTATION_MARK;
  1082. break;
  1083. }
  1084. return 0;
  1085. }
  1086. static int properties_Po(codepoint_t c)
  1087. {
  1088. switch (c & 0xFFFFFF00)
  1089. {
  1090. case 0x0000:
  1091. if (c == 0x0021) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1092. if (c == 0x0022) return UCD_PROPERTY_QUOTATION_MARK;
  1093. if (c == 0x0027) return UCD_PROPERTY_QUOTATION_MARK;
  1094. if (c == 0x002C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1095. if (c == 0x002E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1096. if (c >= 0x003A && c <= 0x003B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1097. if (c == 0x003F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1098. if (c == 0x00B7) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  1099. break;
  1100. case 0x0300:
  1101. if (c == 0x037E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1102. if (c == 0x0387) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1103. break;
  1104. case 0x0500:
  1105. if (c == 0x0589) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1106. if (c == 0x05C3) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1107. break;
  1108. case 0x0600:
  1109. if (c == 0x060C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1110. if (c == 0x061B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1111. if (c == 0x061F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1112. if (c == 0x06D4) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1113. break;
  1114. case 0x0700:
  1115. if (c >= 0x0700 && c <= 0x070A) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1116. if (c == 0x070C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1117. if (c >= 0x07F8 && c <= 0x07F9) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1118. break;
  1119. case 0x0800:
  1120. if (c >= 0x0830 && c <= 0x083E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1121. if (c == 0x085E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1122. break;
  1123. case 0x0900:
  1124. if (c >= 0x0964 && c <= 0x0965) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1125. break;
  1126. case 0x0E00:
  1127. if (c >= 0x0E5A && c <= 0x0E5B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1128. break;
  1129. case 0x0F00:
  1130. if (c == 0x0F08) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1131. if (c == 0x0F0C) return UCD_PROPERTY_NO_BREAK;
  1132. if (c >= 0x0F0D && c <= 0x0F12) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1133. break;
  1134. case 0x1000:
  1135. if (c >= 0x104A && c <= 0x104B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1136. break;
  1137. case 0x1300:
  1138. if (c >= 0x1361 && c <= 0x1368) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1139. break;
  1140. case 0x1600:
  1141. if (c >= 0x166D && c <= 0x166E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1142. if (c >= 0x16EB && c <= 0x16ED) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1143. break;
  1144. case 0x1700:
  1145. if (c >= 0x1735 && c <= 0x1736) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1146. if (c >= 0x17D4 && c <= 0x17D6) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1147. if (c == 0x17DA) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1148. break;
  1149. case 0x1800:
  1150. if (c >= 0x1802 && c <= 0x1805) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1151. if (c >= 0x1808 && c <= 0x1809) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1152. if (c == 0x180A) return UCD_PROPERTY_EXTENDER;
  1153. break;
  1154. case 0x1900:
  1155. if (c >= 0x1944 && c <= 0x1945) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1156. break;
  1157. case 0x1A00:
  1158. if (c >= 0x1AA8 && c <= 0x1AAB) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1159. break;
  1160. case 0x1B00:
  1161. if (c >= 0x1B5A && c <= 0x1B5B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1162. if (c >= 0x1B5D && c <= 0x1B5F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1163. break;
  1164. case 0x1C00:
  1165. if (c >= 0x1C3B && c <= 0x1C3F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1166. if (c >= 0x1C7E && c <= 0x1C7F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1167. if (c == 0x1CD3) return UCD_PROPERTY_DIACRITIC;
  1168. break;
  1169. case 0x2000:
  1170. if (c == 0x2016) return UCD_PROPERTY_OTHER_MATH;
  1171. if (c >= 0x2032 && c <= 0x2034) return UCD_PROPERTY_OTHER_MATH;
  1172. if (c >= 0x203C && c <= 0x203D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1173. if (c >= 0x2047 && c <= 0x2049) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1174. if (c == 0x2053) return UCD_PROPERTY_DASH;
  1175. break;
  1176. case 0x2E00:
  1177. if (c == 0x2E2E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1178. if (c == 0x2E3C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1179. if (c == 0x2E41) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1180. break;
  1181. case 0x3000:
  1182. if (c >= 0x3001 && c <= 0x3002) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1183. if (c == 0x30FB) return UCD_PROPERTY_HYPHEN;
  1184. break;
  1185. case 0xA400:
  1186. if (c >= 0xA4FE && c <= 0xA4FF) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1187. break;
  1188. case 0xA600:
  1189. if (c >= 0xA60D && c <= 0xA60F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1190. if (c >= 0xA6F3 && c <= 0xA6F7) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1191. break;
  1192. case 0xA800:
  1193. if (c >= 0xA876 && c <= 0xA877) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1194. if (c >= 0xA8CE && c <= 0xA8CF) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1195. break;
  1196. case 0xA900:
  1197. if (c == 0xA92E) return UCD_PROPERTY_DIACRITIC;
  1198. if (c == 0xA92F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1199. if (c >= 0xA9C7 && c <= 0xA9C9) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1200. break;
  1201. case 0xAA00:
  1202. if (c >= 0xAA5D && c <= 0xAA5F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1203. if (c == 0xAADF) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1204. if (c >= 0xAAF0 && c <= 0xAAF1) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1205. break;
  1206. case 0xAB00:
  1207. if (c == 0xABEB) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1208. break;
  1209. case 0xFE00:
  1210. if (c >= 0xFE50 && c <= 0xFE52) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1211. if (c >= 0xFE54 && c <= 0xFE57) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1212. if (c == 0xFE61) return UCD_PROPERTY_OTHER_MATH;
  1213. if (c == 0xFE68) return UCD_PROPERTY_OTHER_MATH;
  1214. break;
  1215. case 0xFF00:
  1216. if (c == 0xFF01) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1217. if (c == 0xFF02) return UCD_PROPERTY_QUOTATION_MARK;
  1218. if (c == 0xFF07) return UCD_PROPERTY_QUOTATION_MARK;
  1219. if (c == 0xFF0C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1220. if (c == 0xFF0E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1221. if (c == 0xFF3C) return UCD_PROPERTY_OTHER_MATH;
  1222. if (c == 0xFF65) return UCD_PROPERTY_HYPHEN;
  1223. if (c >= 0xFF1A && c <= 0xFF1B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1224. if (c == 0xFF1F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1225. if (c == 0xFF61) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1226. if (c == 0xFF64) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1227. break;
  1228. case 0x10300:
  1229. if (c == 0x01039F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1230. if (c == 0x0103D0) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1231. break;
  1232. case 0x10800:
  1233. if (c == 0x010857) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1234. break;
  1235. case 0x10900:
  1236. if (c == 0x01091F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1237. break;
  1238. case 0x10A00:
  1239. if (c >= 0x010A56 && c <= 0x010A57) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1240. if (c >= 0x010AF0 && c <= 0x010AF5) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1241. break;
  1242. case 0x10B00:
  1243. if (c >= 0x010B3A && c <= 0x010B3F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1244. if (c >= 0x010B99 && c <= 0x010B9C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1245. break;
  1246. case 0x11000:
  1247. if (c >= 0x011047 && c <= 0x01104D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1248. if (c >= 0x0110BE && c <= 0x0110C1) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1249. break;
  1250. case 0x11100:
  1251. if (c >= 0x011141 && c <= 0x011143) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1252. if (c >= 0x0111C5 && c <= 0x0111C6) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1253. if (c == 0x0111CD) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1254. if (c >= 0x0111DE && c <= 0x0111DF) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1255. break;
  1256. case 0x11200:
  1257. if (c >= 0x011238 && c <= 0x01123C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1258. if (c == 0x0112A9) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1259. break;
  1260. case 0x11400:
  1261. if (c >= 0x01144B && c <= 0x01144D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1262. if (c == 0x01145B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1263. break;
  1264. case 0x11500:
  1265. if (c >= 0x0115C2 && c <= 0x0115C5) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1266. if (c >= 0x0115C6 && c <= 0x0115C8) return UCD_PROPERTY_EXTENDER;
  1267. if (c >= 0x0115C9 && c <= 0x0115D7) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1268. break;
  1269. case 0x11600:
  1270. if (c >= 0x011641 && c <= 0x011642) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1271. break;
  1272. case 0x11700:
  1273. if (c >= 0x01173C && c <= 0x01173E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1274. break;
  1275. case 0x11C00:
  1276. if (c >= 0x011C41 && c <= 0x011C43) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1277. if (c == 0x011C71) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1278. break;
  1279. case 0x12400:
  1280. if (c >= 0x012470 && c <= 0x012474) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1281. break;
  1282. case 0x16A00:
  1283. if (c >= 0x016A6E && c <= 0x016A6F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1284. if (c == 0x016AF5) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1285. break;
  1286. case 0x16B00:
  1287. if (c >= 0x016B37 && c <= 0x016B39) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1288. if (c == 0x016B44) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1289. break;
  1290. case 0x1BC00:
  1291. if (c == 0x01BC9F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1292. break;
  1293. case 0x1DA00:
  1294. if (c >= 0x01DA87 && c <= 0x01DA8A) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1295. break;
  1296. }
  1297. return 0;
  1298. }
  1299. static int properties_Ps(codepoint_t c)
  1300. {
  1301. switch (c & 0xFFFFFF00)
  1302. {
  1303. case 0x2000:
  1304. if (c == 0x201A) return UCD_PROPERTY_QUOTATION_MARK;
  1305. if (c == 0x201E) return UCD_PROPERTY_QUOTATION_MARK;
  1306. if (c == 0x207D) return UCD_PROPERTY_OTHER_MATH;
  1307. if (c == 0x208D) return UCD_PROPERTY_OTHER_MATH;
  1308. break;
  1309. case 0x2300:
  1310. if (c == 0x2308) return UCD_PROPERTY_OTHER_MATH;
  1311. if (c == 0x230A) return UCD_PROPERTY_OTHER_MATH;
  1312. break;
  1313. case 0x2700:
  1314. if (c == 0x27C5) return UCD_PROPERTY_OTHER_MATH;
  1315. if (c >= 0x27E6 && c <= 0x27EF) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1316. break;
  1317. case 0x2900:
  1318. if (c >= 0x2983 && c <= 0x2998) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1319. if (c >= 0x29D8 && c <= 0x29DB) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1320. if (c == 0x29FC) return UCD_PROPERTY_OTHER_MATH;
  1321. break;
  1322. case 0x2E00:
  1323. if (c == 0x2E42) return UCD_PROPERTY_QUOTATION_MARK;
  1324. break;
  1325. case 0x3000:
  1326. if (c == 0x300C) return UCD_PROPERTY_QUOTATION_MARK;
  1327. if (c == 0x300E) return UCD_PROPERTY_QUOTATION_MARK;
  1328. if (c == 0x301D) return UCD_PROPERTY_QUOTATION_MARK;
  1329. break;
  1330. case 0xFE00:
  1331. if (c == 0xFE41) return UCD_PROPERTY_QUOTATION_MARK;
  1332. if (c == 0xFE43) return UCD_PROPERTY_QUOTATION_MARK;
  1333. break;
  1334. case 0xFF00:
  1335. if (c == 0xFF62) return UCD_PROPERTY_QUOTATION_MARK;
  1336. break;
  1337. }
  1338. return 0;
  1339. }
  1340. static int properties_Sk(codepoint_t c)
  1341. {
  1342. switch (c & 0xFFFFFF00)
  1343. {
  1344. case 0x0000:
  1345. if (c == 0x005E) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_DIACRITIC;
  1346. if (c == 0x0060) return UCD_PROPERTY_DIACRITIC;
  1347. if (c == 0x00A8) return UCD_PROPERTY_DIACRITIC;
  1348. if (c == 0x00AF) return UCD_PROPERTY_DIACRITIC;
  1349. if (c == 0x00B4) return UCD_PROPERTY_DIACRITIC;
  1350. if (c == 0x00B8) return UCD_PROPERTY_DIACRITIC;
  1351. break;
  1352. case 0x0200:
  1353. if (c >= 0x02C2 && c <= 0x02C5) return UCD_PROPERTY_DIACRITIC;
  1354. if (c >= 0x02D2 && c <= 0x02DF) return UCD_PROPERTY_DIACRITIC;
  1355. if (c >= 0x02E5 && c <= 0x02EB) return UCD_PROPERTY_DIACRITIC;
  1356. if (c == 0x02ED) return UCD_PROPERTY_DIACRITIC;
  1357. if (c >= 0x02EF && c <= 0x02FF) return UCD_PROPERTY_DIACRITIC;
  1358. break;
  1359. case 0x0300:
  1360. if (c == 0x0375) return UCD_PROPERTY_DIACRITIC;
  1361. if (c >= 0x0384 && c <= 0x0385) return UCD_PROPERTY_DIACRITIC;
  1362. break;
  1363. case 0x1F00:
  1364. if (c == 0x1FBD) return UCD_PROPERTY_DIACRITIC;
  1365. if (c >= 0x1FBF && c <= 0x1FC1) return UCD_PROPERTY_DIACRITIC;
  1366. if (c >= 0x1FCD && c <= 0x1FCF) return UCD_PROPERTY_DIACRITIC;
  1367. if (c >= 0x1FDD && c <= 0x1FDF) return UCD_PROPERTY_DIACRITIC;
  1368. if (c >= 0x1FED && c <= 0x1FEF) return UCD_PROPERTY_DIACRITIC;
  1369. if (c >= 0x1FFD && c <= 0x1FFE) return UCD_PROPERTY_DIACRITIC;
  1370. break;
  1371. case 0x3000:
  1372. if (c >= 0x309B && c <= 0x309C) return UCD_PROPERTY_DIACRITIC;
  1373. break;
  1374. case 0xA700:
  1375. if (c >= 0xA720 && c <= 0xA721) return UCD_PROPERTY_DIACRITIC;
  1376. break;
  1377. case 0xAB00:
  1378. if (c == 0xAB5B) return UCD_PROPERTY_DIACRITIC;
  1379. break;
  1380. case 0xFF00:
  1381. if (c == 0xFF3E) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_MATH;
  1382. if (c == 0xFF40) return UCD_PROPERTY_DIACRITIC;
  1383. if (c == 0xFFE3) return UCD_PROPERTY_DIACRITIC;
  1384. break;
  1385. }
  1386. return 0;
  1387. }
  1388. static int properties_Sm(codepoint_t c)
  1389. {
  1390. switch (c & 0xFFFFFF00)
  1391. {
  1392. case 0x2000:
  1393. if (c == 0x207B) return UCD_PROPERTY_DASH;
  1394. if (c == 0x208B) return UCD_PROPERTY_DASH;
  1395. break;
  1396. case 0x2200:
  1397. if (c == 0x2212) return UCD_PROPERTY_DASH;
  1398. break;
  1399. }
  1400. return 0;
  1401. }
  1402. static int properties_So(codepoint_t c)
  1403. {
  1404. switch (c & 0xFFFFFF00)
  1405. {
  1406. case 0x2100:
  1407. if (c == 0x2129) return UCD_PROPERTY_OTHER_MATH;
  1408. if (c >= 0x2195 && c <= 0x2199) return UCD_PROPERTY_OTHER_MATH;
  1409. if (c >= 0x219C && c <= 0x219F) return UCD_PROPERTY_OTHER_MATH;
  1410. if (c >= 0x21A1 && c <= 0x21A2) return UCD_PROPERTY_OTHER_MATH;
  1411. if (c >= 0x21A4 && c <= 0x21A5) return UCD_PROPERTY_OTHER_MATH;
  1412. if (c == 0x21A7) return UCD_PROPERTY_OTHER_MATH;
  1413. if (c >= 0x21A9 && c <= 0x21AD) return UCD_PROPERTY_OTHER_MATH;
  1414. if (c >= 0x21B0 && c <= 0x21B1) return UCD_PROPERTY_OTHER_MATH;
  1415. if (c >= 0x21B6 && c <= 0x21B7) return UCD_PROPERTY_OTHER_MATH;
  1416. if (c >= 0x21BC && c <= 0x21CD) return UCD_PROPERTY_OTHER_MATH;
  1417. if (c >= 0x21D0 && c <= 0x21D1) return UCD_PROPERTY_OTHER_MATH;
  1418. if (c == 0x21D3) return UCD_PROPERTY_OTHER_MATH;
  1419. if (c >= 0x21D5 && c <= 0x21DB) return UCD_PROPERTY_OTHER_MATH;
  1420. if (c == 0x21DD) return UCD_PROPERTY_OTHER_MATH;
  1421. if (c >= 0x21E4 && c <= 0x21E5) return UCD_PROPERTY_OTHER_MATH;
  1422. break;
  1423. case 0x2300:
  1424. if (c >= 0x23B4 && c <= 0x23B5) return UCD_PROPERTY_OTHER_MATH;
  1425. if (c == 0x23B7) return UCD_PROPERTY_OTHER_MATH;
  1426. if (c == 0x23D0) return UCD_PROPERTY_OTHER_MATH;
  1427. if (c == 0x23E2) return UCD_PROPERTY_OTHER_MATH;
  1428. break;
  1429. case 0x2400:
  1430. if (c >= 0x24B6 && c <= 0x24E9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1431. break;
  1432. case 0x2500:
  1433. if (c >= 0x25A0 && c <= 0x25A1) return UCD_PROPERTY_OTHER_MATH;
  1434. if (c >= 0x25AE && c <= 0x25B6) return UCD_PROPERTY_OTHER_MATH;
  1435. if (c >= 0x25BC && c <= 0x25C0) return UCD_PROPERTY_OTHER_MATH;
  1436. if (c >= 0x25C6 && c <= 0x25C7) return UCD_PROPERTY_OTHER_MATH;
  1437. if (c >= 0x25CA && c <= 0x25CB) return UCD_PROPERTY_OTHER_MATH;
  1438. if (c >= 0x25CF && c <= 0x25D3) return UCD_PROPERTY_OTHER_MATH;
  1439. if (c == 0x25E2) return UCD_PROPERTY_OTHER_MATH;
  1440. if (c == 0x25E4) return UCD_PROPERTY_OTHER_MATH;
  1441. if (c >= 0x25E7 && c <= 0x25EC) return UCD_PROPERTY_OTHER_MATH;
  1442. break;
  1443. case 0x2600:
  1444. if (c >= 0x2605 && c <= 0x2606) return UCD_PROPERTY_OTHER_MATH;
  1445. if (c == 0x2640) return UCD_PROPERTY_OTHER_MATH;
  1446. if (c == 0x2642) return UCD_PROPERTY_OTHER_MATH;
  1447. if (c >= 0x2660 && c <= 0x2663) return UCD_PROPERTY_OTHER_MATH;
  1448. if (c >= 0x266D && c <= 0x266E) return UCD_PROPERTY_OTHER_MATH;
  1449. break;
  1450. case 0x01F100:
  1451. if (c >= 0x01F130 && c <= 0x01F149) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1452. if (c >= 0x01F150 && c <= 0x01F169) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1453. if (c >= 0x01F170 && c <= 0x01F189) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1454. break;
  1455. }
  1456. return 0;
  1457. }
  1458. static int properties_Zs(codepoint_t c)
  1459. {
  1460. switch (c)
  1461. {
  1462. case 0x00A0: // NO-BREAK SPACE
  1463. case 0x2007: // FIGURE SPACE
  1464. case 0x202F: // NARROW NO-BREAK SPACE
  1465. return UCD_PROPERTY_WHITE_SPACE | UCD_PROPERTY_NO_BREAK;
  1466. }
  1467. return UCD_PROPERTY_WHITE_SPACE;
  1468. }
  1469. ucd_property ucd_properties(codepoint_t c, ucd_category category)
  1470. {
  1471. switch (category)
  1472. {
  1473. case UCD_CATEGORY_Cc: return properties_Cc(c);
  1474. case UCD_CATEGORY_Cf: return properties_Cf(c);
  1475. case UCD_CATEGORY_Ll: return properties_Ll(c);
  1476. case UCD_CATEGORY_Lm: return properties_Lm(c);
  1477. case UCD_CATEGORY_Lo: return properties_Lo(c) | properties_Lo_ideographic(c);
  1478. case UCD_CATEGORY_Lu: return properties_Lu(c);
  1479. case UCD_CATEGORY_Mc: return properties_Mc(c);
  1480. case UCD_CATEGORY_Mn: return properties_Mn(c);
  1481. case UCD_CATEGORY_Nd: return properties_Nd(c);
  1482. case UCD_CATEGORY_Nl: return properties_Nl(c);
  1483. case UCD_CATEGORY_Pc: return properties_Pc(c);
  1484. case UCD_CATEGORY_Pd: return properties_Pd(c);
  1485. case UCD_CATEGORY_Pe: return properties_Pe(c);
  1486. case UCD_CATEGORY_Pf: return properties_Pf(c);
  1487. case UCD_CATEGORY_Pi: return properties_Pi(c);
  1488. case UCD_CATEGORY_Po: return properties_Po(c);
  1489. case UCD_CATEGORY_Ps: return properties_Ps(c);
  1490. case UCD_CATEGORY_Sk: return properties_Sk(c);
  1491. case UCD_CATEGORY_Sm: return properties_Sm(c);
  1492. case UCD_CATEGORY_So: return properties_So(c);
  1493. case UCD_CATEGORY_Zl: return UCD_PROPERTY_WHITE_SPACE;
  1494. case UCD_CATEGORY_Zp: return UCD_PROPERTY_WHITE_SPACE;
  1495. case UCD_CATEGORY_Zs: return properties_Zs(c);
  1496. default: return 0; // Cn Co Cs Ii Lt Me No Sc
  1497. };
  1498. }