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 78KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667
  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) return UCD_PROPERTY_JOIN_CONTROL | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  39. if (c == 0x200D) return UCD_PROPERTY_JOIN_CONTROL;
  40. if (c >= 0x200E && c <= 0x200F) return UCD_PROPERTY_BIDI_CONTROL;
  41. if (c >= 0x202A && c <= 0x202E) return UCD_PROPERTY_BIDI_CONTROL;
  42. if (c >= 0x2061 && c <= 0x2064) return UCD_PROPERTY_OTHER_MATH;
  43. if (c >= 0x2066 && c <= 0x2069) return UCD_PROPERTY_BIDI_CONTROL;
  44. if (c >= 0x206A && c <= 0x206F) return UCD_PROPERTY_DEPRECATED;
  45. break;
  46. case 0x0E0000:
  47. if (c == 0x0E0001) return UCD_PROPERTY_DEPRECATED;
  48. if (c >= 0x0E0020 && c <= 0x0E007F) return UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  49. break;
  50. }
  51. return 0;
  52. }
  53. static int properties_Cn(codepoint_t c)
  54. {
  55. switch (c & 0xFFFF0000)
  56. {
  57. case 0x0000:
  58. if (c == 0x2065) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  59. if (c >= 0xFFF0 && c <= 0xFFF8) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  60. if (c >= 0xFDD0 && c <= 0xFDEF) return UCD_PROPERTY_NONCHARACTER_CODE_POINT;
  61. break;
  62. case 0x0E0000:
  63. if (c == 0xE0000) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  64. if (c >= 0xE0002 && c <= 0xE001F) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  65. if (c >= 0xE0080 && c <= 0xE00FF) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  66. if (c >= 0xE01F0 && c <= 0xE0FFF) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  67. break;
  68. }
  69. if ((c & 0x0000FFFF) >= 0xFFFE) return UCD_PROPERTY_NONCHARACTER_CODE_POINT;
  70. return 0;
  71. }
  72. static int properties_Ll(codepoint_t c)
  73. {
  74. switch (c & 0xFFFFFF00)
  75. {
  76. case 0x0000:
  77. if (c >= 0x0061 && c <= 0x0066) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT;
  78. if (c >= 0x0069 && c <= 0x006A) return UCD_PROPERTY_SOFT_DOTTED;
  79. break;
  80. case 0x0100:
  81. if (c == 0x012F) return UCD_PROPERTY_SOFT_DOTTED;
  82. if (c == 0x0149) return UCD_PROPERTY_DEPRECATED;
  83. break;
  84. case 0x0200:
  85. if (c == 0x0249) return UCD_PROPERTY_SOFT_DOTTED;
  86. if (c == 0x0268) return UCD_PROPERTY_SOFT_DOTTED;
  87. if (c == 0x029D) return UCD_PROPERTY_SOFT_DOTTED;
  88. break;
  89. case 0x0300:
  90. if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  91. if (c == 0x03D5) return UCD_PROPERTY_OTHER_MATH;
  92. if (c >= 0x03F0 && c <= 0x03F1) return UCD_PROPERTY_OTHER_MATH;
  93. if (c == 0x03F3) return UCD_PROPERTY_SOFT_DOTTED;
  94. if (c >= 0x03F4 && c <= 0x03F5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  95. break;
  96. case 0x0400:
  97. if (c == 0x0456) return UCD_PROPERTY_SOFT_DOTTED;
  98. if (c == 0x0458) return UCD_PROPERTY_SOFT_DOTTED;
  99. break;
  100. case 0x1D00:
  101. if (c == 0x1D96) return UCD_PROPERTY_SOFT_DOTTED;
  102. break;
  103. case 0x1E00:
  104. if (c == 0x1E2D) return UCD_PROPERTY_SOFT_DOTTED;
  105. if (c == 0x1ECB) return UCD_PROPERTY_SOFT_DOTTED;
  106. break;
  107. case 0x2100:
  108. if (c >= 0x210A && c <= 0x2113) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  109. if (c >= 0x212F && c <= 0x2131) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  110. if (c >= 0x2133 && c <= 0x2134) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  111. if (c >= 0x213C && c <= 0x213F) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  112. if (c >= 0x2145 && c <= 0x2147) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  113. if (c >= 0x2148 && c <= 0x2149) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  114. break;
  115. case 0xFF00:
  116. if (c >= 0xFF41 && c <= 0xFF46) return UCD_PROPERTY_HEX_DIGIT;
  117. break;
  118. case 0x01D400:
  119. if (c >= 0x01D422 && c <= 0x01D423) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  120. if (c >= 0x01D400 && c <= 0x01D454) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  121. if (c >= 0x01D456 && c <= 0x01D457) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  122. if (c >= 0x01D48A && c <= 0x01D48B) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  123. if (c >= 0x01D458 && c <= 0x01D49C) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  124. if (c >= 0x01D4AE && c <= 0x01D4B9) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  125. if (c == 0x01D4BB) return UCD_PROPERTY_OTHER_MATH;
  126. if (c >= 0x01D4BE && c <= 0x01D4BF) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  127. if (c >= 0x01D4BD && c <= 0x01D4C3) return UCD_PROPERTY_OTHER_MATH;
  128. if (c >= 0x01D4F2 && c <= 0x01D4F3) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  129. if (c >= 0x01D4C5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  130. break;
  131. case 0x01D500:
  132. if (c <= 0x01D505) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  133. if (c >= 0x01D526 && c <= 0x01D527) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  134. if (c >= 0x01D51E && c <= 0x01D539) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  135. if (c >= 0x01D55A && c <= 0x01D55B) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  136. if (c >= 0x01D58E && c <= 0x01D58F) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  137. if (c >= 0x01D5C2 && c <= 0x01D5C3) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  138. if (c >= 0x01D5F6 && c <= 0x01D5F7) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  139. if (c >= 0x01D552) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  140. break;
  141. case 0x01D600:
  142. if (c >= 0x01D62A && c <= 0x01D62B) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  143. if (c >= 0x01D65E && c <= 0x01D65F) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  144. if (c >= 0x01D692 && c <= 0x01D693) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  145. if (c <= 0x01D6A5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  146. if (c >= 0x01D6C2 && c <= 0x01D6DA) return UCD_PROPERTY_OTHER_MATH;
  147. if (c >= 0x01D6DC && c <= 0x01D6FA) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  148. if (c >= 0x01D6FC) return UCD_PROPERTY_OTHER_MATH;
  149. break;
  150. case 0x01D700:
  151. if (c <= 0x01D714) return UCD_PROPERTY_OTHER_MATH;
  152. if (c >= 0x01D716 && c <= 0x01D734) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  153. if (c >= 0x01D736 && c <= 0x01D74E) return UCD_PROPERTY_OTHER_MATH;
  154. if (c >= 0x01D750 && c <= 0x01D76E) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  155. if (c >= 0x01D770 && c <= 0x01D788) return UCD_PROPERTY_OTHER_MATH;
  156. if (c >= 0x01D78A && c <= 0x01D7A8) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  157. if (c >= 0x01D7AA && c <= 0x01D7C2) return UCD_PROPERTY_OTHER_MATH;
  158. if (c >= 0x01D7C4 && c <= 0x01D7CB) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  159. break;
  160. }
  161. return 0;
  162. }
  163. static int properties_Lm(codepoint_t c)
  164. {
  165. switch (c & 0xFFFFFF00)
  166. {
  167. case 0x0200:
  168. if (c == 0x02B2) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_SOFT_DOTTED;
  169. if (c >= 0x02B0 && c <= 0x02B8) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  170. if (c >= 0x02B9 && c <= 0x02BF) return UCD_PROPERTY_DIACRITIC;
  171. if (c >= 0x02C0 && c <= 0x02C1) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  172. if (c >= 0x02C6 && c <= 0x02CF) return UCD_PROPERTY_DIACRITIC;
  173. if (c >= 0x02D0 && c <= 0x02D1) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  174. if (c >= 0x02E0 && c <= 0x02E4) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  175. if (c == 0x02EC) return UCD_PROPERTY_DIACRITIC;
  176. if (c == 0x02EE) return UCD_PROPERTY_DIACRITIC;
  177. break;
  178. case 0x0300:
  179. if (c == 0x0374) return UCD_PROPERTY_DIACRITIC;
  180. if (c == 0x037A) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  181. break;
  182. case 0x0500:
  183. if (c == 0x0559) return UCD_PROPERTY_DIACRITIC;
  184. break;
  185. case 0x0600:
  186. if (c == 0x0640) return UCD_PROPERTY_EXTENDER;
  187. if (c >= 0x06E5 && c <= 0x06E6) return UCD_PROPERTY_DIACRITIC;
  188. break;
  189. case 0x0700:
  190. if (c >= 0x07F4 && c <= 0x07F5) return UCD_PROPERTY_DIACRITIC;
  191. if (c == 0x07FA) return UCD_PROPERTY_EXTENDER;
  192. break;
  193. case 0x0900:
  194. if (c == 0x0971) return UCD_PROPERTY_DIACRITIC;
  195. break;
  196. case 0x0E00:
  197. if (c == 0x0E46) return UCD_PROPERTY_EXTENDER;
  198. if (c == 0x0EC6) return UCD_PROPERTY_EXTENDER;
  199. break;
  200. case 0x1800:
  201. if (c == 0x1843) return UCD_PROPERTY_EXTENDER;
  202. break;
  203. case 0x1A00:
  204. if (c == 0x1AA7) return UCD_PROPERTY_EXTENDER;
  205. break;
  206. case 0x1C00:
  207. if (c == 0x1C7B) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  208. if (c >= 0x1C78 && c <= 0x1C7D) return UCD_PROPERTY_DIACRITIC;
  209. break;
  210. case 0x1D00:
  211. if (c == 0x1D62) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_SOFT_DOTTED;
  212. if (c >= 0x1D2C && c <= 0x1D6A) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_DIACRITIC;
  213. if (c == 0x1D78) return UCD_PROPERTY_OTHER_LOWERCASE;
  214. if (c == 0x1DA4) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_SOFT_DOTTED;
  215. if (c == 0x1DA8) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_SOFT_DOTTED;
  216. if (c >= 0x1D9B && c <= 0x1DBF) return UCD_PROPERTY_OTHER_LOWERCASE;
  217. break;
  218. case 0x2000:
  219. if (c == 0x2071) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_SOFT_DOTTED;
  220. if (c == 0x207F) return UCD_PROPERTY_OTHER_LOWERCASE;
  221. if (c >= 0x2090 && c <= 0x209C) return UCD_PROPERTY_OTHER_LOWERCASE;
  222. break;
  223. case 0x2C00:
  224. if (c == 0x2C7C) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_SOFT_DOTTED;
  225. if (c == 0x2C7D) return UCD_PROPERTY_OTHER_LOWERCASE;
  226. break;
  227. case 0x2E00:
  228. if (c == 0x2E2F) return UCD_PROPERTY_DIACRITIC;
  229. break;
  230. case 0x3000:
  231. if (c == 0x3005) return UCD_PROPERTY_EXTENDER;
  232. if (c >= 0x3031 && c <= 0x3035) return UCD_PROPERTY_EXTENDER;
  233. if (c >= 0x309D && c <= 0x309E) return UCD_PROPERTY_EXTENDER;
  234. if (c == 0x30FC) return UCD_PROPERTY_EXTENDER | UCD_PROPERTY_DIACRITIC;
  235. if (c >= 0x30FD && c <= 0x30FE) return UCD_PROPERTY_EXTENDER;
  236. break;
  237. case 0xA000:
  238. if (c == 0xA015) return UCD_PROPERTY_EXTENDER;
  239. break;
  240. case 0xA600:
  241. if (c == 0xA60C) return UCD_PROPERTY_EXTENDER;
  242. if (c == 0xA67F) return UCD_PROPERTY_DIACRITIC;
  243. if (c >= 0xA69C && c <= 0xA69D) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  244. break;
  245. case 0xA700:
  246. if (c >= 0xA717 && c <= 0xA71F) return UCD_PROPERTY_DIACRITIC;
  247. if (c == 0xA770) return UCD_PROPERTY_OTHER_LOWERCASE;
  248. if (c == 0xA788) return UCD_PROPERTY_DIACRITIC;
  249. if (c >= 0xA7F8 && c <= 0xA7F9) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  250. break;
  251. case 0xA900:
  252. if (c == 0xA9CF) return UCD_PROPERTY_EXTENDER;
  253. if (c == 0xA9E6) return UCD_PROPERTY_EXTENDER;
  254. break;
  255. case 0xAA00:
  256. if (c == 0xAA70) return UCD_PROPERTY_EXTENDER;
  257. if (c == 0xAADD) return UCD_PROPERTY_EXTENDER;
  258. if (c >= 0xAAF3 && c <= 0xAAF4) return UCD_PROPERTY_EXTENDER;
  259. break;
  260. case 0xAB00:
  261. if (c >= 0xAB5C && c <= 0xAB5F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  262. break;
  263. case 0xFF00:
  264. if (c == 0xFF70) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  265. if (c >= 0xFF9E && c <= 0xFF9F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  266. break;
  267. case 0x016B00:
  268. if (c >= 0x016B42 && c <= 0x016B43) return UCD_PROPERTY_EXTENDER;
  269. break;
  270. case 0x016F00:
  271. if (c >= 0x016F93 && c <= 0x016F9F) return UCD_PROPERTY_DIACRITIC;
  272. if (c == 0x016FE0) return UCD_PROPERTY_EXTENDER;
  273. break;
  274. }
  275. return 0;
  276. }
  277. static int properties_Lo(codepoint_t c)
  278. {
  279. switch (c & 0xFFFFFF00)
  280. {
  281. case 0x0000:
  282. if (c == 0x00AA) return UCD_PROPERTY_OTHER_LOWERCASE;
  283. if (c == 0x00BA) return UCD_PROPERTY_OTHER_LOWERCASE;
  284. break;
  285. case 0x0600:
  286. if (c == 0x0673) return UCD_PROPERTY_DEPRECATED;
  287. break;
  288. case 0x0E00:
  289. if (c >= 0x0E40 && c <= 0x0E44) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  290. if (c >= 0x0EC0 && c <= 0x0EC4) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  291. break;
  292. case 0x1100:
  293. if (c >= 0x115F && c <= 0x1160) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  294. break;
  295. case 0x1700:
  296. if (c >= 0x17A3 && c <= 0x17A4) return UCD_PROPERTY_DEPRECATED;
  297. break;
  298. case 0x1900:
  299. if (c >= 0x19B5 && c <= 0x19B7) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  300. if (c == 0x19BA) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  301. break;
  302. case 0x2100:
  303. if (c >= 0x2135 && c <= 0x2138) return UCD_PROPERTY_OTHER_MATH;
  304. break;
  305. case 0x3000:
  306. if (c == 0x3006) return UCD_PROPERTY_IDEOGRAPHIC;
  307. break;
  308. case 0x3100:
  309. if (c == 0x3164) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  310. break;
  311. case 0xAA00:
  312. if (c >= 0xAAB5 && c <= 0xAAB6) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  313. if (c == 0xAAB9) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  314. if (c >= 0xAABB && c <= 0xAABC) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  315. if (c == 0xAAC0) return UCD_PROPERTY_DIACRITIC;
  316. if (c == 0xAAC2) return UCD_PROPERTY_DIACRITIC;
  317. break;
  318. case 0xFA00:
  319. if (c >= 0xFA0E && c <= 0xFA0F) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  320. if (c == 0xFA11) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  321. if (c >= 0xFA13 && c <= 0xFA14) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  322. if (c == 0xFA1F) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  323. if (c == 0xFA21) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  324. if (c >= 0xFA23 && c <= 0xFA24) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  325. if (c >= 0xFA27 && c <= 0xFA29) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  326. break;
  327. case 0xFF00:
  328. if (c == 0xFFA0) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  329. break;
  330. case 0x11300:
  331. if (c == 0x01135D) return UCD_PROPERTY_EXTENDER;
  332. break;
  333. case 0x1EE00:
  334. return UCD_PROPERTY_OTHER_MATH;
  335. }
  336. return 0;
  337. }
  338. static int properties_Lo_ideographic(codepoint_t c)
  339. {
  340. switch (c & 0xFFFF0000)
  341. {
  342. case 0x000000:
  343. if (c >= 0x3400 && c <= 0x4DB5) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  344. if (c >= 0x4E00 && c <= 0x9FD5) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  345. if (c >= 0xF900 && c <= 0xFA6D) return UCD_PROPERTY_IDEOGRAPHIC;
  346. if (c >= 0xFA70 && c <= 0xFAD9) return UCD_PROPERTY_IDEOGRAPHIC;
  347. break;
  348. case 0x010000:
  349. if (c >= 0x017000 && c <= 0x0187EC) return UCD_PROPERTY_IDEOGRAPHIC;
  350. if (c >= 0x018800 && c <= 0x018AF2) return UCD_PROPERTY_IDEOGRAPHIC;
  351. break;
  352. case 0x020000:
  353. if (c >= 0x020000 && c <= 0x02A6D6) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  354. if (c >= 0x02A700 && c <= 0x02B734) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  355. if (c >= 0x02B740 && c <= 0x02B81D) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  356. if (c >= 0x02B820 && c <= 0x02CEA1) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  357. if (c >= 0x02F800 && c <= 0x02FA1D) return UCD_PROPERTY_IDEOGRAPHIC;
  358. break;
  359. }
  360. return 0;
  361. }
  362. static int properties_Lu(codepoint_t c)
  363. {
  364. switch (c & 0xFFFFFF00)
  365. {
  366. case 0x0000:
  367. if (c >= 0x0041 && c <= 0x0046) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT;
  368. break;
  369. case 0x0300:
  370. if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  371. if (c >= 0x03F4 && c <= 0x03F5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  372. break;
  373. case 0xFF00:
  374. if (c >= 0xFF21 && c <= 0xFF26) return UCD_PROPERTY_HEX_DIGIT;
  375. break;
  376. case 0x2100:
  377. if (c == 0x2102) return UCD_PROPERTY_OTHER_MATH;
  378. if (c == 0x2107) return UCD_PROPERTY_OTHER_MATH;
  379. if (c >= 0x210A && c <= 0x2113) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  380. if (c == 0x2115) return UCD_PROPERTY_OTHER_MATH;
  381. if (c >= 0x2119 && c <= 0x211D) return UCD_PROPERTY_OTHER_MATH;
  382. if (c == 0x2124) return UCD_PROPERTY_OTHER_MATH;
  383. if (c == 0x2128) return UCD_PROPERTY_OTHER_MATH;
  384. if (c >= 0x212C && c <= 0x212D) return UCD_PROPERTY_OTHER_MATH;
  385. if (c >= 0x212F && c <= 0x2131) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  386. if (c >= 0x2133 && c <= 0x2134) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  387. if (c >= 0x213C && c <= 0x213F) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  388. if (c >= 0x2145 && c <= 0x2149) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  389. break;
  390. case 0x01D400:
  391. if (c >= 0x01D400 && c <= 0x01D454) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  392. if (c >= 0x01D456 && c <= 0x01D49C) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  393. if (c >= 0x01D49E && c <= 0x01D49F) return UCD_PROPERTY_OTHER_MATH;
  394. if (c == 0x01D4A2) return UCD_PROPERTY_OTHER_MATH;
  395. if (c >= 0x01D4A5 && c <= 0x01D4A6) return UCD_PROPERTY_OTHER_MATH;
  396. if (c >= 0x01D4A9 && c <= 0x01D4AC) return UCD_PROPERTY_OTHER_MATH;
  397. if (c >= 0x01D4AE && c <= 0x01D4B9) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  398. if (c >= 0x01D4C5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  399. break;
  400. case 0x01D500:
  401. if (c <= 0x01D505) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  402. if (c >= 0x01D507 && c <= 0x01D50A) return UCD_PROPERTY_OTHER_MATH;
  403. if (c >= 0x01D50D && c <= 0x01D514) return UCD_PROPERTY_OTHER_MATH;
  404. if (c >= 0x01D516 && c <= 0x01D51C) return UCD_PROPERTY_OTHER_MATH;
  405. if (c >= 0x01D51E && c <= 0x01D539) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  406. if (c >= 0x01D53B && c <= 0x01D53E) return UCD_PROPERTY_OTHER_MATH;
  407. if (c >= 0x01D540 && c <= 0x01D544) return UCD_PROPERTY_OTHER_MATH;
  408. if (c == 0x01D546) return UCD_PROPERTY_OTHER_MATH;
  409. if (c >= 0x01D54A && c <= 0x01D550) return UCD_PROPERTY_OTHER_MATH;
  410. if (c >= 0x01D552) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  411. break;
  412. case 0x01D600:
  413. if (c <= 0x01D6A5) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  414. if (c >= 0x01D6A8 && c <= 0x01D6C0) return UCD_PROPERTY_OTHER_MATH;
  415. if (c >= 0x01D6DC && c <= 0x01D6FA) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  416. break;
  417. case 0x01D700:
  418. if (c >= 0x01D716 && c <= 0x01D734) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  419. if (c >= 0x01D750 && c <= 0x01D76E) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  420. if (c >= 0x01D78A && c <= 0x01D7A8) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  421. if (c >= 0x01D7C4 && c <= 0x01D7CB) return UCD_PROPERTY_OTHER_MATH; // Ll|Lu
  422. break;
  423. }
  424. return 0;
  425. }
  426. static int properties_Mc(codepoint_t c)
  427. {
  428. switch (c & 0xFFFFFF00)
  429. {
  430. case 0x0900:
  431. if (c == 0x0903) return UCD_PROPERTY_OTHER_ALPHABETIC;
  432. if (c == 0x093B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  433. if (c >= 0x093E && c <= 0x0940) return UCD_PROPERTY_OTHER_ALPHABETIC;
  434. if (c >= 0x0949 && c <= 0x094C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  435. if (c >= 0x094E && c <= 0x094F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  436. if (c >= 0x0982 && c <= 0x0983) return UCD_PROPERTY_OTHER_ALPHABETIC;
  437. if (c == 0x09BE) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  438. if (c >= 0x09BF && c <= 0x09C0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  439. if (c >= 0x09C7 && c <= 0x09C8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  440. if (c >= 0x09CB && c <= 0x09CC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  441. if (c == 0x09D7) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  442. break;
  443. case 0x0A00:
  444. if (c == 0x0A03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  445. if (c >= 0x0A3E && c <= 0x0A40) return UCD_PROPERTY_OTHER_ALPHABETIC;
  446. if (c == 0x0A83) return UCD_PROPERTY_OTHER_ALPHABETIC;
  447. if (c >= 0x0ABE && c <= 0x0AC0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  448. if (c == 0x0AC9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  449. if (c >= 0x0ACB && c <= 0x0ACC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  450. break;
  451. case 0x0B00:
  452. if (c >= 0x0B02 && c <= 0x0B03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  453. if (c == 0x0B3E) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  454. if (c == 0x0B40) return UCD_PROPERTY_OTHER_ALPHABETIC;
  455. if (c >= 0x0B47 && c <= 0x0B48) return UCD_PROPERTY_OTHER_ALPHABETIC;
  456. if (c >= 0x0B4B && c <= 0x0B4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  457. if (c == 0x0B57) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  458. if (c == 0x0BBE) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  459. if (c == 0x0BBF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  460. if (c >= 0x0BC1 && c <= 0x0BC2) return UCD_PROPERTY_OTHER_ALPHABETIC;
  461. if (c >= 0x0BC6 && c <= 0x0BC8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  462. if (c >= 0x0BCA && c <= 0x0BCC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  463. if (c == 0x0BD7) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  464. break;
  465. case 0x0C00:
  466. if (c >= 0x0C01 && c <= 0x0C03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  467. if (c >= 0x0C41 && c <= 0x0C44) return UCD_PROPERTY_OTHER_ALPHABETIC;
  468. if (c >= 0x0C82 && c <= 0x0C83) return UCD_PROPERTY_OTHER_ALPHABETIC;
  469. if (c == 0x0CBE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  470. if (c >= 0x0CC0 && c <= 0x0CC1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  471. if (c == 0x0CC2) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  472. if (c >= 0x0CC3 && c <= 0x0CC4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  473. if (c >= 0x0CC7 && c <= 0x0CC8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  474. if (c >= 0x0CCA && c <= 0x0CCB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  475. if (c >= 0x0CD5 && c <= 0x0CD6) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  476. break;
  477. case 0x0D00:
  478. if (c >= 0x0D02 && c <= 0x0D03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  479. if (c == 0x0D3E) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  480. if (c >= 0x0D3F && c <= 0x0D40) return UCD_PROPERTY_OTHER_ALPHABETIC;
  481. if (c >= 0x0D46 && c <= 0x0D48) return UCD_PROPERTY_OTHER_ALPHABETIC;
  482. if (c >= 0x0D4A && c <= 0x0D4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  483. if (c == 0x0D57) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  484. if (c >= 0x0D82 && c <= 0x0D83) return UCD_PROPERTY_OTHER_ALPHABETIC;
  485. if (c == 0x0DCF) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  486. if (c >= 0x0DD0 && c <= 0x0DD1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  487. if (c >= 0x0DD8 && c <= 0x0DDE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  488. if (c == 0x0DDF) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  489. if (c >= 0x0DF2 && c <= 0x0DF3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  490. break;
  491. case 0x0F00:
  492. if (c >= 0x0F3E && c <= 0x0F3F) return UCD_PROPERTY_DIACRITIC;
  493. if (c == 0x0F7F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  494. break;
  495. case 0x1000:
  496. if (c >= 0x102B && c <= 0x102C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  497. if (c == 0x1031) return UCD_PROPERTY_OTHER_ALPHABETIC;
  498. if (c == 0x1038) return UCD_PROPERTY_OTHER_ALPHABETIC;
  499. if (c >= 0x103B && c <= 0x103C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  500. if (c >= 0x1056 && c <= 0x1057) return UCD_PROPERTY_OTHER_ALPHABETIC;
  501. if (c == 0x1062) return UCD_PROPERTY_OTHER_ALPHABETIC;
  502. if (c >= 0x1067 && c <= 0x1068) return UCD_PROPERTY_OTHER_ALPHABETIC;
  503. if (c >= 0x1083 && c <= 0x1084) return UCD_PROPERTY_OTHER_ALPHABETIC;
  504. if (c >= 0x1087 && c <= 0x108C) return UCD_PROPERTY_DIACRITIC;
  505. if (c == 0x108F) return UCD_PROPERTY_DIACRITIC;
  506. if (c >= 0x109A && c <= 0x109B) return UCD_PROPERTY_DIACRITIC;
  507. if (c == 0x109C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  508. break;
  509. case 0x1700:
  510. if (c == 0x17B6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  511. if (c >= 0x17BE && c <= 0x17C5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  512. if (c >= 0x17C7 && c <= 0x17C8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  513. break;
  514. case 0x1900:
  515. if (c >= 0x1923 && c <= 0x1926) return UCD_PROPERTY_OTHER_ALPHABETIC;
  516. if (c >= 0x1929 && c <= 0x192B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  517. if (c >= 0x1930 && c <= 0x1931) return UCD_PROPERTY_OTHER_ALPHABETIC;
  518. if (c >= 0x1933 && c <= 0x1938) return UCD_PROPERTY_OTHER_ALPHABETIC;
  519. break;
  520. case 0x1A00:
  521. if (c >= 0x1A19 && c <= 0x1A1A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  522. if (c == 0x1A55) return UCD_PROPERTY_OTHER_ALPHABETIC;
  523. if (c == 0x1A57) return UCD_PROPERTY_OTHER_ALPHABETIC;
  524. if (c == 0x1A61) return UCD_PROPERTY_OTHER_ALPHABETIC;
  525. if (c >= 0x1A63 && c <= 0x1A64) return UCD_PROPERTY_OTHER_ALPHABETIC;
  526. if (c >= 0x1A6D && c <= 0x1A72) return UCD_PROPERTY_OTHER_ALPHABETIC;
  527. break;
  528. case 0x1B00:
  529. if (c == 0x1B04) return UCD_PROPERTY_OTHER_ALPHABETIC;
  530. if (c == 0x1B35) return UCD_PROPERTY_OTHER_ALPHABETIC;
  531. if (c == 0x1B3B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  532. if (c >= 0x1B3D && c <= 0x1B41) return UCD_PROPERTY_OTHER_ALPHABETIC;
  533. if (c == 0x1B43) return UCD_PROPERTY_OTHER_ALPHABETIC;
  534. if (c == 0x1B44) return UCD_PROPERTY_DIACRITIC;
  535. if (c == 0x1B82) return UCD_PROPERTY_OTHER_ALPHABETIC;
  536. if (c == 0x1BA1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  537. if (c >= 0x1BA6 && c <= 0x1BA7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  538. if (c == 0x1BAA) return UCD_PROPERTY_DIACRITIC;
  539. if (c == 0x1BE7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  540. if (c >= 0x1BEA && c <= 0x1BEC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  541. if (c == 0x1BEE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  542. break;
  543. case 0x1C00:
  544. if (c >= 0x1C24 && c <= 0x1C2B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  545. if (c >= 0x1C34 && c <= 0x1C35) return UCD_PROPERTY_OTHER_ALPHABETIC;
  546. if (c == 0x1CE1) return UCD_PROPERTY_DIACRITIC;
  547. if (c >= 0x1CF2 && c <= 0x1CF3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  548. break;
  549. case 0x3000:
  550. if (c >= 0x302E && c <= 0x302F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  551. break;
  552. case 0xA800:
  553. if (c >= 0xA823 && c <= 0xA824) return UCD_PROPERTY_OTHER_ALPHABETIC;
  554. if (c == 0xA827) return UCD_PROPERTY_OTHER_ALPHABETIC;
  555. if (c >= 0xA880 && c <= 0xA881) return UCD_PROPERTY_OTHER_ALPHABETIC;
  556. if (c >= 0xA8B4 && c <= 0xA8C3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  557. break;
  558. case 0xA900:
  559. if (c == 0xA952) return UCD_PROPERTY_OTHER_ALPHABETIC;
  560. if (c == 0xA953) return UCD_PROPERTY_DIACRITIC;
  561. if (c == 0xA983) return UCD_PROPERTY_OTHER_ALPHABETIC;
  562. if (c >= 0xA9B4 && c <= 0xA9B5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  563. if (c >= 0xA9BA && c <= 0xA9BB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  564. if (c >= 0xA9BD && c <= 0xA9BF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  565. if (c == 0xA9C0) return UCD_PROPERTY_DIACRITIC;
  566. break;
  567. case 0xAA00:
  568. if (c >= 0xAA2F && c <= 0xAA30) return UCD_PROPERTY_OTHER_ALPHABETIC;
  569. if (c >= 0xAA33 && c <= 0xAA34) return UCD_PROPERTY_OTHER_ALPHABETIC;
  570. if (c == 0xAA4D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  571. if (c == 0xAA7B) return UCD_PROPERTY_DIACRITIC;
  572. if (c == 0xAA7D) return UCD_PROPERTY_DIACRITIC;
  573. if (c == 0xAAEB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  574. if (c >= 0xAAEE && c <= 0xAAEF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  575. if (c == 0xAAF5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  576. break;
  577. case 0xAB00:
  578. if (c >= 0xABE3 && c <= 0xABE4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  579. if (c >= 0xABE6 && c <= 0xABE7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  580. if (c >= 0xABE9 && c <= 0xABEA) return UCD_PROPERTY_OTHER_ALPHABETIC;
  581. if (c == 0xABEC) return UCD_PROPERTY_DIACRITIC;
  582. break;
  583. case 0x011000:
  584. if (c == 0x011000) return UCD_PROPERTY_OTHER_ALPHABETIC;
  585. if (c == 0x011002) return UCD_PROPERTY_OTHER_ALPHABETIC;
  586. if (c == 0x011082) return UCD_PROPERTY_OTHER_ALPHABETIC;
  587. if (c >= 0x0110B0 && c <= 0x0110B2) return UCD_PROPERTY_OTHER_ALPHABETIC;
  588. if (c >= 0x0110B7 && c <= 0x0110B8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  589. break;
  590. case 0x011100:
  591. if (c == 0x01112C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  592. if (c == 0x011182) return UCD_PROPERTY_OTHER_ALPHABETIC;
  593. if (c >= 0x0111B3 && c <= 0x0111B5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  594. if (c == 0x0111BF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  595. if (c == 0x0111C0) return UCD_PROPERTY_DIACRITIC;
  596. break;
  597. case 0x011200:
  598. if (c >= 0x01122C && c <= 0x01122E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  599. if (c >= 0x011232 && c <= 0x011233) return UCD_PROPERTY_OTHER_ALPHABETIC;
  600. if (c == 0x011235) return UCD_PROPERTY_DIACRITIC;
  601. if (c >= 0x0112E0 && c <= 0x0112E2) return UCD_PROPERTY_OTHER_ALPHABETIC;
  602. break;
  603. case 0x011300:
  604. if (c >= 0x011302 && c <= 0x011303) return UCD_PROPERTY_OTHER_ALPHABETIC;
  605. if (c == 0x01133E) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  606. if (c == 0x01133F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  607. if (c >= 0x011341 && c <= 0x011344) return UCD_PROPERTY_OTHER_ALPHABETIC;
  608. if (c >= 0x011347 && c <= 0x011348) return UCD_PROPERTY_OTHER_ALPHABETIC;
  609. if (c >= 0x01134B && c <= 0x01134C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  610. if (c == 0x01134D) return UCD_PROPERTY_DIACRITIC;
  611. if (c == 0x011357) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  612. if (c >= 0x011362 && c <= 0x011363) return UCD_PROPERTY_OTHER_ALPHABETIC;
  613. break;
  614. case 0x011400:
  615. if (c >= 0x011435 && c <= 0x011437) return UCD_PROPERTY_OTHER_ALPHABETIC;
  616. if (c >= 0x011440 && c <= 0x011441) return UCD_PROPERTY_OTHER_ALPHABETIC;
  617. if (c == 0x011445) return UCD_PROPERTY_OTHER_ALPHABETIC;
  618. if (c == 0x0114B0) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  619. if (c >= 0x0114B1 && c <= 0x0114B2) return UCD_PROPERTY_OTHER_ALPHABETIC;
  620. if (c == 0x0114B9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  621. if (c >= 0x0114BB && c <= 0x0114BC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  622. if (c == 0x0114BD) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  623. if (c == 0x0114BE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  624. if (c == 0x0114C1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  625. break;
  626. case 0x011500:
  627. if (c == 0x0115AF) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  628. if (c >= 0x0115B0 && c <= 0x0115B1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  629. if (c >= 0x0115B8 && c <= 0x0115BB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  630. if (c == 0x0115BE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  631. break;
  632. case 0x011600:
  633. if (c >= 0x011630 && c <= 0x011632) return UCD_PROPERTY_OTHER_ALPHABETIC;
  634. if (c >= 0x01163B && c <= 0x01163C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  635. if (c == 0x01163E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  636. if (c == 0x0116AC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  637. if (c >= 0x0116AE && c <= 0x0116AF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  638. if (c == 0x0116B6) return UCD_PROPERTY_DIACRITIC;
  639. break;
  640. case 0x011700:
  641. if (c >= 0x011720 && c <= 0x011721) return UCD_PROPERTY_OTHER_ALPHABETIC;
  642. if (c == 0x011726) return UCD_PROPERTY_OTHER_ALPHABETIC;
  643. break;
  644. case 0x011C00:
  645. if (c == 0x011C2F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  646. if (c == 0x011C3E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  647. if (c == 0x011CA9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  648. if (c == 0x011CB1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  649. if (c == 0x011CB4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  650. break;
  651. case 0x01D100:
  652. if (c == 0x01D165) return UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  653. if (c == 0x01D16D) return UCD_PROPERTY_DIACRITIC;
  654. if (c >= 0x01D16E && c <= 0x01D172) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  655. break;
  656. case 0x016F00:
  657. if (c >= 0x016F51 && c <= 0x016F7E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  658. break;
  659. }
  660. return 0;
  661. }
  662. static int properties_Mn(codepoint_t c)
  663. {
  664. switch (c & 0xFFFFFF00)
  665. {
  666. case 0x0300:
  667. if (c >= 0x0300 && c <= 0x0344) return UCD_PROPERTY_DIACRITIC;
  668. if (c == 0x0345) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_LOWERCASE;
  669. if (c >= 0x0346 && c <= 0x034E) return UCD_PROPERTY_DIACRITIC;
  670. if (c == 0x034F) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  671. if (c >= 0x0350 && c <= 0x0357) return UCD_PROPERTY_DIACRITIC;
  672. if (c >= 0x035D && c <= 0x0362) return UCD_PROPERTY_DIACRITIC;
  673. break;
  674. case 0x0400:
  675. if (c >= 0x0483 && c <= 0x0487) return UCD_PROPERTY_DIACRITIC;
  676. break;
  677. case 0x0500:
  678. if (c >= 0x0591 && c <= 0x05A1) return UCD_PROPERTY_DIACRITIC;
  679. if (c >= 0x05A3 && c <= 0x05AF) return UCD_PROPERTY_DIACRITIC;
  680. if (c >= 0x05B0 && c <= 0x05BD) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  681. if (c == 0x05BF) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  682. if (c >= 0x05C1 && c <= 0x05C2) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  683. if (c == 0x05C4) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  684. if (c == 0x05C5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  685. if (c == 0x05C7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  686. break;
  687. case 0x0600:
  688. if (c >= 0x0610 && c <= 0x061A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  689. if (c >= 0x064B && c <= 0x0652) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  690. if (c >= 0x0653 && c <= 0x0656) return UCD_PROPERTY_OTHER_ALPHABETIC;
  691. if (c == 0x0657) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  692. if (c == 0x0658) return UCD_PROPERTY_DIACRITIC;
  693. if (c >= 0x0659 && c <= 0x065F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  694. if (c == 0x0670) return UCD_PROPERTY_OTHER_ALPHABETIC;
  695. if (c >= 0x06D6 && c <= 0x06DC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  696. if (c >= 0x06DF && c <= 0x06E0) return UCD_PROPERTY_DIACRITIC;
  697. if (c >= 0x06E1 && c <= 0x06E4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  698. if (c >= 0x06E7 && c <= 0x06E8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  699. if (c >= 0x06EA && c <= 0x06EC) return UCD_PROPERTY_DIACRITIC;
  700. if (c == 0x06ED) return UCD_PROPERTY_OTHER_ALPHABETIC;
  701. break;
  702. case 0x0700:
  703. if (c == 0x0711) return UCD_PROPERTY_OTHER_ALPHABETIC;
  704. if (c >= 0x0730 && c <= 0x073F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  705. if (c >= 0x0740 && c <= 0x074A) return UCD_PROPERTY_DIACRITIC;
  706. if (c >= 0x07A6 && c <= 0x07B0) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  707. if (c >= 0x07EB && c <= 0x07F3) return UCD_PROPERTY_DIACRITIC;
  708. break;
  709. case 0x0800:
  710. if (c >= 0x0816 && c <= 0x0817) return UCD_PROPERTY_OTHER_ALPHABETIC;
  711. if (c >= 0x0818 && c <= 0x0819) return UCD_PROPERTY_DIACRITIC;
  712. if (c >= 0x081B && c <= 0x0823) return UCD_PROPERTY_OTHER_ALPHABETIC;
  713. if (c >= 0x0825 && c <= 0x0827) return UCD_PROPERTY_OTHER_ALPHABETIC;
  714. if (c >= 0x0829 && c <= 0x082C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  715. if (c >= 0x08D4 && c <= 0x08DF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  716. if (c >= 0x08E3 && c <= 0x08E9) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  717. if (c >= 0x08EA && c <= 0x08EF) return UCD_PROPERTY_DIACRITIC;
  718. if (c >= 0x08F0 && c <= 0x08FE) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  719. if (c == 0x08FF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  720. break;
  721. case 0x0900:
  722. if (c <= 0x0902) return UCD_PROPERTY_OTHER_ALPHABETIC;
  723. if (c == 0x093A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  724. if (c == 0x093C) return UCD_PROPERTY_DIACRITIC;
  725. if (c >= 0x0941 && c <= 0x0948) return UCD_PROPERTY_OTHER_ALPHABETIC;
  726. if (c == 0x094D) return UCD_PROPERTY_DIACRITIC;
  727. if (c >= 0x0951 && c <= 0x0954) return UCD_PROPERTY_DIACRITIC;
  728. if (c >= 0x0955 && c <= 0x0957) return UCD_PROPERTY_OTHER_ALPHABETIC;
  729. if (c >= 0x0962 && c <= 0x0963) return UCD_PROPERTY_OTHER_ALPHABETIC;
  730. if (c == 0x0981) return UCD_PROPERTY_OTHER_ALPHABETIC;
  731. if (c == 0x09BC) return UCD_PROPERTY_DIACRITIC;
  732. if (c >= 0x09C1 && c <= 0x09C4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  733. if (c == 0x09CD) return UCD_PROPERTY_DIACRITIC;
  734. if (c >= 0x09E2 && c <= 0x09E3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  735. break;
  736. case 0x0A00:
  737. if (c >= 0x0A01 && c <= 0x0A02) return UCD_PROPERTY_OTHER_ALPHABETIC;
  738. if (c == 0x0A3C) return UCD_PROPERTY_DIACRITIC;
  739. if (c >= 0x0A41 && c <= 0x0A42) return UCD_PROPERTY_OTHER_ALPHABETIC;
  740. if (c >= 0x0A47 && c <= 0x0A48) return UCD_PROPERTY_OTHER_ALPHABETIC;
  741. if (c >= 0x0A4B && c <= 0x0A4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  742. if (c == 0x0A4D) return UCD_PROPERTY_DIACRITIC;
  743. if (c == 0x0A51) return UCD_PROPERTY_OTHER_ALPHABETIC;
  744. if (c >= 0x0A70 && c <= 0x0A71) return UCD_PROPERTY_OTHER_ALPHABETIC;
  745. if (c == 0x0A75) return UCD_PROPERTY_OTHER_ALPHABETIC;
  746. if (c >= 0x0A81 && c <= 0x0A82) return UCD_PROPERTY_OTHER_ALPHABETIC;
  747. if (c == 0x0ABC) return UCD_PROPERTY_DIACRITIC;
  748. if (c >= 0x0AC1 && c <= 0x0AC5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  749. if (c >= 0x0AC7 && c <= 0x0AC8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  750. if (c == 0x0ACD) return UCD_PROPERTY_DIACRITIC;
  751. if (c >= 0x0AE2 && c <= 0x0AE3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  752. break;
  753. case 0x0B00:
  754. if (c == 0x0B01) return UCD_PROPERTY_OTHER_ALPHABETIC;
  755. if (c == 0x0B3C) return UCD_PROPERTY_DIACRITIC;
  756. if (c == 0x0B3F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  757. if (c >= 0x0B41 && c <= 0x0B44) return UCD_PROPERTY_OTHER_ALPHABETIC;
  758. if (c == 0x0B4D) return UCD_PROPERTY_DIACRITIC;
  759. if (c == 0x0B56) return UCD_PROPERTY_OTHER_ALPHABETIC;
  760. if (c >= 0x0B62 && c <= 0x0B63) return UCD_PROPERTY_OTHER_ALPHABETIC;
  761. if (c == 0x0B82) return UCD_PROPERTY_OTHER_ALPHABETIC;
  762. if (c == 0x0BC0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  763. if (c == 0x0BCD) return UCD_PROPERTY_DIACRITIC;
  764. break;
  765. case 0x0C00:
  766. if (c == 0x0C00) return UCD_PROPERTY_OTHER_ALPHABETIC;
  767. if (c >= 0x0C3E && c <= 0x0C40) return UCD_PROPERTY_OTHER_ALPHABETIC;
  768. if (c >= 0x0C46 && c <= 0x0C48) return UCD_PROPERTY_OTHER_ALPHABETIC;
  769. if (c >= 0x0C4A && c <= 0x0C4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  770. if (c == 0x0C4D) return UCD_PROPERTY_DIACRITIC;
  771. if (c >= 0x0C55 && c <= 0x0C56) return UCD_PROPERTY_OTHER_ALPHABETIC;
  772. if (c >= 0x0C62 && c <= 0x0C63) return UCD_PROPERTY_OTHER_ALPHABETIC;
  773. if (c == 0x0C81) return UCD_PROPERTY_OTHER_ALPHABETIC;
  774. if (c == 0x0CBC) return UCD_PROPERTY_DIACRITIC;
  775. if (c == 0x0CBF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  776. if (c == 0x0CC6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  777. if (c == 0x0CCC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  778. if (c == 0x0CCD) return UCD_PROPERTY_DIACRITIC;
  779. if (c >= 0x0CE2 && c <= 0x0CE3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  780. break;
  781. case 0x0D00:
  782. if (c == 0x0D01) return UCD_PROPERTY_OTHER_ALPHABETIC;
  783. if (c >= 0x0D41 && c <= 0x0D44) return UCD_PROPERTY_OTHER_ALPHABETIC;
  784. if (c == 0x0D4D) return UCD_PROPERTY_DIACRITIC;
  785. if (c >= 0x0D62 && c <= 0x0D63) return UCD_PROPERTY_OTHER_ALPHABETIC;
  786. if (c == 0x0DCA) return UCD_PROPERTY_DIACRITIC;
  787. if (c >= 0x0DD2 && c <= 0x0DD4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  788. if (c == 0x0DD6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  789. break;
  790. case 0x0E00:
  791. if (c == 0x0E31) return UCD_PROPERTY_OTHER_ALPHABETIC;
  792. if (c >= 0x0E34 && c <= 0x0E3A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  793. if (c >= 0x0E47 && c <= 0x0E4C) return UCD_PROPERTY_DIACRITIC;
  794. if (c == 0x0E4D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  795. if (c == 0x0E4E) return UCD_PROPERTY_DIACRITIC;
  796. if (c == 0x0EB1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  797. if (c >= 0x0EB4 && c <= 0x0EB9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  798. if (c >= 0x0EBB && c <= 0x0EBC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  799. if (c >= 0x0EC8 && c <= 0x0ECC) return UCD_PROPERTY_DIACRITIC;
  800. if (c == 0x0ECD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  801. break;
  802. case 0x0F00:
  803. if (c >= 0x0F18 && c <= 0x0F19) return UCD_PROPERTY_DIACRITIC;
  804. if (c == 0x0F35) return UCD_PROPERTY_DIACRITIC;
  805. if (c == 0x0F37) return UCD_PROPERTY_DIACRITIC;
  806. if (c == 0x0F39) return UCD_PROPERTY_DIACRITIC;
  807. if (c == 0x0F77) return UCD_PROPERTY_DEPRECATED | UCD_PROPERTY_OTHER_ALPHABETIC;
  808. if (c == 0x0F79) return UCD_PROPERTY_DEPRECATED | UCD_PROPERTY_OTHER_ALPHABETIC;
  809. if (c >= 0x0F71 && c <= 0x0F7E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  810. if (c >= 0x0F80 && c <= 0x0F81) return UCD_PROPERTY_OTHER_ALPHABETIC;
  811. if (c >= 0x0F82 && c <= 0x0F84) return UCD_PROPERTY_DIACRITIC;
  812. if (c >= 0x0F86 && c <= 0x0F87) return UCD_PROPERTY_DIACRITIC;
  813. if (c >= 0x0F8D && c <= 0x0F97) return UCD_PROPERTY_OTHER_ALPHABETIC;
  814. if (c >= 0x0F99 && c <= 0x0FBC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  815. if (c == 0x0FC6) return UCD_PROPERTY_DIACRITIC;
  816. break;
  817. case 0x1000:
  818. if (c >= 0x102D && c <= 0x1030) return UCD_PROPERTY_OTHER_ALPHABETIC;
  819. if (c >= 0x1032 && c <= 0x1036) return UCD_PROPERTY_OTHER_ALPHABETIC;
  820. if (c == 0x1037) return UCD_PROPERTY_DIACRITIC;
  821. if (c >= 0x1039 && c <= 0x103A) return UCD_PROPERTY_DIACRITIC;
  822. if (c >= 0x103D && c <= 0x103E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  823. if (c >= 0x1058 && c <= 0x1059) return UCD_PROPERTY_OTHER_ALPHABETIC;
  824. if (c >= 0x105E && c <= 0x1060) return UCD_PROPERTY_OTHER_ALPHABETIC;
  825. if (c >= 0x1071 && c <= 0x1074) return UCD_PROPERTY_OTHER_ALPHABETIC;
  826. if (c == 0x1082) return UCD_PROPERTY_OTHER_ALPHABETIC;
  827. if (c >= 0x1085 && c <= 0x1086) return UCD_PROPERTY_OTHER_ALPHABETIC;
  828. if (c == 0x108D) return UCD_PROPERTY_DIACRITIC;
  829. if (c == 0x109D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  830. break;
  831. case 0x1300:
  832. if (c == 0x135F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  833. break;
  834. case 0x1700:
  835. if (c >= 0x1712 && c <= 0x1713) return UCD_PROPERTY_OTHER_ALPHABETIC;
  836. if (c >= 0x1732 && c <= 0x1733) return UCD_PROPERTY_OTHER_ALPHABETIC;
  837. if (c >= 0x1752 && c <= 0x1753) return UCD_PROPERTY_OTHER_ALPHABETIC;
  838. if (c >= 0x1772 && c <= 0x1773) return UCD_PROPERTY_OTHER_ALPHABETIC;
  839. if (c >= 0x17B4 && c <= 0x17B5) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  840. if (c >= 0x17B7 && c <= 0x17BD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  841. if (c == 0x17C6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  842. if (c >= 0x17C9 && c <= 0x17D3) return UCD_PROPERTY_DIACRITIC;
  843. if (c == 0x17DD) return UCD_PROPERTY_DIACRITIC;
  844. break;
  845. case 0x1800:
  846. if (c >= 0x1885 && c <= 0x1886) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_ID_START;
  847. if (c == 0x18A9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  848. break;
  849. case 0x1900:
  850. if (c >= 0x1920 && c <= 0x1922) return UCD_PROPERTY_OTHER_ALPHABETIC;
  851. if (c >= 0x1927 && c <= 0x1928) return UCD_PROPERTY_OTHER_ALPHABETIC;
  852. if (c == 0x1932) return UCD_PROPERTY_OTHER_ALPHABETIC;
  853. if (c >= 0x1939 && c <= 0x193B) return UCD_PROPERTY_DIACRITIC;
  854. break;
  855. case 0x1A00:
  856. if (c >= 0x1A17 && c <= 0x1A18) return UCD_PROPERTY_OTHER_ALPHABETIC;
  857. if (c == 0x1A1B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  858. if (c == 0x1A56) return UCD_PROPERTY_OTHER_ALPHABETIC;
  859. if (c >= 0x1A58 && c <= 0x1A5E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  860. if (c == 0x1A62) return UCD_PROPERTY_OTHER_ALPHABETIC;
  861. if (c >= 0x1A65 && c <= 0x1A6C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  862. if (c >= 0x1A73 && c <= 0x1A74) return UCD_PROPERTY_OTHER_ALPHABETIC;
  863. if (c >= 0x1A75 && c <= 0x1A7C) return UCD_PROPERTY_DIACRITIC;
  864. if (c == 0x1A7F) return UCD_PROPERTY_DIACRITIC;
  865. if (c >= 0x1AB0 && c <= 0x1ABD) return UCD_PROPERTY_DIACRITIC;
  866. break;
  867. case 0x1B00:
  868. if (c >= 0x1B00 && c <= 0x1B03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  869. if (c == 0x1B34) return UCD_PROPERTY_DIACRITIC;
  870. if (c >= 0x1B36 && c <= 0x1B3A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  871. if (c == 0x1B3C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  872. if (c == 0x1B42) return UCD_PROPERTY_OTHER_ALPHABETIC;
  873. if (c >= 0x1B6B && c <= 0x1B73) return UCD_PROPERTY_DIACRITIC;
  874. if (c >= 0x1B80 && c <= 0x1B81) return UCD_PROPERTY_OTHER_ALPHABETIC;
  875. if (c >= 0x1BA2 && c <= 0x1BA5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  876. if (c >= 0x1BA8 && c <= 0x1BA9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  877. if (c == 0x1BAB) return UCD_PROPERTY_DIACRITIC;
  878. if (c >= 0x1BAC && c <= 0x1BAD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  879. if (c >= 0x1BE8 && c <= 0x1BE9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  880. if (c == 0x1BED) return UCD_PROPERTY_OTHER_ALPHABETIC;
  881. if (c >= 0x1BEF && c <= 0x1BF1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  882. break;
  883. case 0x1C00:
  884. if (c >= 0x1C2C && c <= 0x1C33) return UCD_PROPERTY_OTHER_ALPHABETIC;
  885. if (c == 0x1C36) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  886. if (c == 0x1C37) return UCD_PROPERTY_DIACRITIC;
  887. if (c >= 0x1CD0 && c <= 0x1CD2) return UCD_PROPERTY_DIACRITIC;
  888. if (c >= 0x1CD4 && c <= 0x1CE0) return UCD_PROPERTY_DIACRITIC;
  889. if (c >= 0x1CE2 && c <= 0x1CE8) return UCD_PROPERTY_DIACRITIC;
  890. if (c == 0x1CED) return UCD_PROPERTY_DIACRITIC;
  891. if (c == 0x1CF4) return UCD_PROPERTY_DIACRITIC;
  892. if (c >= 0x1CF8 && c <= 0x1CF9) return UCD_PROPERTY_DIACRITIC;
  893. break;
  894. case 0x1D00:
  895. if (c >= 0x1DC4 && c <= 0x1DCF) return UCD_PROPERTY_DIACRITIC;
  896. if (c >= 0x1DE7 && c <= 0x1DF4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  897. if (c == 0x1DF5) return UCD_PROPERTY_DIACRITIC;
  898. if (c >= 0x1DFD && c <= 0x1DFF) return UCD_PROPERTY_DIACRITIC;
  899. break;
  900. case 0x2000:
  901. if (c >= 0x20D0 && c <= 0x20DC) return UCD_PROPERTY_OTHER_MATH;
  902. if (c == 0x20E1) return UCD_PROPERTY_OTHER_MATH;
  903. if (c >= 0x20E5 && c <= 0x20E6) return UCD_PROPERTY_OTHER_MATH;
  904. if (c >= 0x20EB && c <= 0x20EF) return UCD_PROPERTY_OTHER_MATH;
  905. break;
  906. case 0x2C00:
  907. if (c >= 0x2CEF && c <= 0x2CF1) return UCD_PROPERTY_DIACRITIC;
  908. break;
  909. case 0x2D00:
  910. if (c >= 0x2DE0 && c <= 0x2DFF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  911. break;
  912. case 0x3000:
  913. if (c >= 0x302A && c <= 0x302D) return UCD_PROPERTY_DIACRITIC;
  914. if (c >= 0x3099 && c <= 0x309A) return UCD_PROPERTY_DIACRITIC;
  915. break;
  916. case 0xA600:
  917. if (c == 0xA66F) return UCD_PROPERTY_DIACRITIC;
  918. if (c >= 0xA674 && c <= 0xA67B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  919. if (c >= 0xA67C && c <= 0xA67D) return UCD_PROPERTY_DIACRITIC;
  920. if (c >= 0xA69E && c <= 0xA69F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  921. if (c >= 0xA6F0 && c <= 0xA6F1) return UCD_PROPERTY_DIACRITIC;
  922. break;
  923. case 0xA800:
  924. if (c >= 0xA825 && c <= 0xA826) return UCD_PROPERTY_OTHER_ALPHABETIC;
  925. if (c == 0xA8C4) return UCD_PROPERTY_DIACRITIC;
  926. if (c == 0xA8C5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  927. if (c >= 0xA8E0 && c <= 0xA8F1) return UCD_PROPERTY_DIACRITIC;
  928. break;
  929. case 0xA900:
  930. if (c >= 0xA926 && c <= 0xA92A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  931. if (c >= 0xA92B && c <= 0xA92D) return UCD_PROPERTY_DIACRITIC;
  932. if (c >= 0xA947 && c <= 0xA951) return UCD_PROPERTY_OTHER_ALPHABETIC;
  933. if (c >= 0xA980 && c <= 0xA982) return UCD_PROPERTY_OTHER_ALPHABETIC;
  934. if (c == 0xA9B3) return UCD_PROPERTY_DIACRITIC;
  935. if (c >= 0xA9B6 && c <= 0xA9B9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  936. if (c == 0xA9BC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  937. if (c == 0xA9E5) return UCD_PROPERTY_DIACRITIC;
  938. break;
  939. case 0xAA00:
  940. if (c >= 0xAA29 && c <= 0xAA2E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  941. if (c >= 0xAA31 && c <= 0xAA32) return UCD_PROPERTY_OTHER_ALPHABETIC;
  942. if (c >= 0xAA35 && c <= 0xAA36) return UCD_PROPERTY_OTHER_ALPHABETIC;
  943. if (c == 0xAA43) return UCD_PROPERTY_OTHER_ALPHABETIC;
  944. if (c == 0xAA4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  945. if (c == 0xAA7C) return UCD_PROPERTY_DIACRITIC;
  946. if (c == 0xAAB0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  947. if (c >= 0xAAB2 && c <= 0xAAB4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  948. if (c >= 0xAAB7 && c <= 0xAAB8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  949. if (c == 0xAABE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  950. if (c == 0xAABF) return UCD_PROPERTY_DIACRITIC;
  951. if (c == 0xAAC1) return UCD_PROPERTY_DIACRITIC;
  952. if (c >= 0xAAEC && c <= 0xAAED) return UCD_PROPERTY_OTHER_ALPHABETIC;
  953. if (c == 0xAAF6) return UCD_PROPERTY_DIACRITIC;
  954. break;
  955. case 0xAB00:
  956. if (c == 0xABE5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  957. if (c == 0xABE8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  958. if (c == 0xABED) return UCD_PROPERTY_DIACRITIC;
  959. break;
  960. case 0xFB00:
  961. if (c == 0xFB1E) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  962. break;
  963. case 0xFE00:
  964. if (c >= 0xFE20 && c <= 0xFE2F) return UCD_PROPERTY_DIACRITIC;
  965. break;
  966. case 0x010200:
  967. if (c == 0x0102E0) return UCD_PROPERTY_DIACRITIC;
  968. break;
  969. case 0x010300:
  970. if (c >= 0x010376 && c <= 0x01037A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  971. break;
  972. case 0x010A00:
  973. if (c >= 0x010A01 && c <= 0x010A03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  974. if (c >= 0x010A05 && c <= 0x010A06) return UCD_PROPERTY_OTHER_ALPHABETIC;
  975. if (c >= 0x010A0C && c <= 0x010A0F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  976. if (c >= 0x010AE5 && c <= 0x010AE6) return UCD_PROPERTY_DIACRITIC;
  977. break;
  978. case 0x011000:
  979. if (c == 0x011001) return UCD_PROPERTY_OTHER_ALPHABETIC;
  980. if (c >= 0x011038 && c <= 0x011045) return UCD_PROPERTY_OTHER_ALPHABETIC;
  981. if (c >= 0x0110B3 && c <= 0x0110B6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  982. if (c >= 0x0110B9 && c <= 0x0110BA) return UCD_PROPERTY_DIACRITIC;
  983. break;
  984. case 0x011100:
  985. if (c >= 0x011100 && c <= 0x011102) return UCD_PROPERTY_OTHER_ALPHABETIC;
  986. if (c >= 0x011127 && c <= 0x01112B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  987. if (c >= 0x01112D && c <= 0x011132) return UCD_PROPERTY_OTHER_ALPHABETIC;
  988. if (c >= 0x011133 && c <= 0x011134) return UCD_PROPERTY_DIACRITIC;
  989. if (c == 0x011173) return UCD_PROPERTY_DIACRITIC;
  990. if (c >= 0x011180 && c <= 0x011181) return UCD_PROPERTY_OTHER_ALPHABETIC;
  991. if (c >= 0x0111B6 && c <= 0x0111BE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  992. if (c >= 0x0111CA && c <= 0x0111CC) return UCD_PROPERTY_DIACRITIC;
  993. break;
  994. case 0x011200:
  995. if (c >= 0x01122F && c <= 0x011231) return UCD_PROPERTY_OTHER_ALPHABETIC;
  996. if (c == 0x011234) return UCD_PROPERTY_OTHER_ALPHABETIC;
  997. if (c == 0x011236) return UCD_PROPERTY_DIACRITIC;
  998. if (c == 0x011237) return UCD_PROPERTY_OTHER_ALPHABETIC;
  999. if (c == 0x01123E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1000. if (c == 0x0112DF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1001. if (c >= 0x0112E3 && c <= 0x0112E8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1002. if (c >= 0x0112E9 && c <= 0x0112EA) return UCD_PROPERTY_DIACRITIC;
  1003. break;
  1004. case 0x011300:
  1005. if (c >= 0x011300 && c <= 0x011301) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1006. if (c == 0x01133C) return UCD_PROPERTY_DIACRITIC;
  1007. if (c == 0x011340) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1008. if (c >= 0x011366 && c <= 0x01136C) return UCD_PROPERTY_DIACRITIC;
  1009. if (c >= 0x011370 && c <= 0x011374) return UCD_PROPERTY_DIACRITIC;
  1010. break;
  1011. case 0x011400:
  1012. if (c >= 0x011438 && c <= 0x01143F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1013. if (c == 0x011442) return UCD_PROPERTY_DIACRITIC;
  1014. if (c >= 0x011443 && c <= 0x011444) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1015. if (c == 0x011446) return UCD_PROPERTY_DIACRITIC;
  1016. if (c >= 0x0114B3 && c <= 0x0114B8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1017. if (c == 0x0114BA) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1018. if (c >= 0x0114BF && c <= 0x0114C0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1019. if (c >= 0x0114C2 && c <= 0x0114C3) return UCD_PROPERTY_DIACRITIC;
  1020. break;
  1021. case 0x011500:
  1022. if (c >= 0x0115B2 && c <= 0x0115B5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1023. if (c >= 0x0115BC && c <= 0x0115BD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1024. if (c >= 0x0115BF && c <= 0x0115C0) return UCD_PROPERTY_DIACRITIC;
  1025. if (c >= 0x0115DC && c <= 0x0115DD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1026. break;
  1027. case 0x011600:
  1028. if (c >= 0x011633 && c <= 0x01163A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1029. if (c == 0x01163D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1030. if (c == 0x01163F) return UCD_PROPERTY_DIACRITIC;
  1031. if (c == 0x011640) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1032. if (c == 0x0116AB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1033. if (c == 0x0116AD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1034. if (c >= 0x0116B0 && c <= 0x0116B5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1035. if (c == 0x0116B7) return UCD_PROPERTY_DIACRITIC;
  1036. break;
  1037. case 0x011700:
  1038. if (c >= 0x01171D && c <= 0x01171F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1039. if (c >= 0x011722 && c <= 0x011725) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1040. if (c >= 0x011727 && c <= 0x01172A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1041. if (c == 0x01172B) return UCD_PROPERTY_DIACRITIC;
  1042. break;
  1043. case 0x011C00:
  1044. if (c >= 0x011C30 && c <= 0x011C36) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1045. if (c >= 0x011C38 && c <= 0x011C3D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1046. if (c == 0x011C3F) return UCD_PROPERTY_DIACRITIC;
  1047. if (c >= 0x011C92 && c <= 0x011CA7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1048. if (c >= 0x011CAA && c <= 0x011CB0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1049. if (c >= 0x011CB2 && c <= 0x011CB3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1050. if (c >= 0x011CB5 && c <= 0x011CB6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1051. break;
  1052. case 0x016A00:
  1053. if (c >= 0x016AF0 && c <= 0x016AF4) return UCD_PROPERTY_DIACRITIC;
  1054. break;
  1055. case 0x016B00:
  1056. if (c >= 0x016B30 && c <= 0x016B36) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1057. break;
  1058. case 0x016F00:
  1059. if (c >= 0x016F8F && c <= 0x016F92) return UCD_PROPERTY_DIACRITIC;
  1060. break;
  1061. case 0x01BC00:
  1062. if (c == 0x01BC9E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1063. break;
  1064. case 0x01D100:
  1065. if (c >= 0x01D167 && c <= 0x01D169) return UCD_PROPERTY_DIACRITIC;
  1066. if (c >= 0x01D17B && c <= 0x01D182) return UCD_PROPERTY_DIACRITIC;
  1067. if (c >= 0x01D185 && c <= 0x01D18B) return UCD_PROPERTY_DIACRITIC;
  1068. if (c >= 0x01D1AA && c <= 0x01D1AD) return UCD_PROPERTY_DIACRITIC;
  1069. break;
  1070. case 0x01E000:
  1071. if (c >= 0x01E000 && c <= 0x01E006) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1072. if (c >= 0x01E008 && c <= 0x01E018) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1073. if (c >= 0x01E01B && c <= 0x01E021) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1074. if (c >= 0x01E023 && c <= 0x01E024) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1075. if (c >= 0x01E026 && c <= 0x01E02A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1076. break;
  1077. case 0x01E800:
  1078. if (c >= 0x01E8D0 && c <= 0x01E8D6) return UCD_PROPERTY_DIACRITIC;
  1079. break;
  1080. case 0x01E900:
  1081. if (c >= 0x01E944 && c <= 0x01E946) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  1082. if (c == 0x01E947) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1083. if (c >= 0x01E948 && c <= 0x01E94A) return UCD_PROPERTY_DIACRITIC;
  1084. break;
  1085. }
  1086. return 0;
  1087. }
  1088. static int properties_Nd(codepoint_t c)
  1089. {
  1090. switch (c & 0xFFFFFF00)
  1091. {
  1092. case 0x0000:
  1093. if (c >= 0x0030 && c <= 0x0039) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT;
  1094. break;
  1095. case 0xFF00:
  1096. if (c >= 0xFF10 && c <= 0xFF19) return UCD_PROPERTY_HEX_DIGIT;
  1097. break;
  1098. case 0x01D700:
  1099. if (c >= 0x01D7CE && c <= 0x01D7FF) return UCD_PROPERTY_OTHER_MATH;
  1100. break;
  1101. }
  1102. return 0;
  1103. }
  1104. static int properties_Nl(codepoint_t c)
  1105. {
  1106. switch (c & 0xFFFFFF00)
  1107. {
  1108. case 0x2100:
  1109. if (c >= 0x2160 && c <= 0x216F) return UCD_PROPERTY_OTHER_UPPERCASE;
  1110. if (c >= 0x2170 && c <= 0x217F) return UCD_PROPERTY_OTHER_LOWERCASE;
  1111. break;
  1112. case 0x3000:
  1113. if (c == 0x3007) return UCD_PROPERTY_IDEOGRAPHIC;
  1114. if (c >= 0x3021 && c <= 0x3029) return UCD_PROPERTY_IDEOGRAPHIC;
  1115. if (c >= 0x3038 && c <= 0x303A) return UCD_PROPERTY_IDEOGRAPHIC;
  1116. break;
  1117. }
  1118. return 0;
  1119. }
  1120. static int properties_Pc(codepoint_t c)
  1121. {
  1122. switch (c & 0xFFFFFF00)
  1123. {
  1124. case 0x2000:
  1125. if (c == 0x2040) return UCD_PROPERTY_OTHER_MATH;
  1126. break;
  1127. }
  1128. return 0;
  1129. }
  1130. static int properties_Pd(codepoint_t c)
  1131. {
  1132. switch (c & 0xFFFFFF00)
  1133. {
  1134. case 0x0000:
  1135. if (c == 0x002D) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1136. break;
  1137. case 0x0500:
  1138. if (c == 0x058A) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1139. break;
  1140. case 0x1800:
  1141. if (c == 0x1806) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1142. break;
  1143. case 0x2000:
  1144. if (c >= 0x2010 && c <= 0x2011) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1145. break;
  1146. case 0x2E00:
  1147. if (c == 0x2E17) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1148. break;
  1149. case 0xFE00:
  1150. if (c == 0xFE63) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN | UCD_PROPERTY_OTHER_MATH;
  1151. break;
  1152. case 0xFF00:
  1153. if (c == 0xFF0D) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1154. break;
  1155. }
  1156. return UCD_PROPERTY_DASH;
  1157. }
  1158. static int properties_Pe(codepoint_t c)
  1159. {
  1160. switch (c & 0xFFFFFF00)
  1161. {
  1162. case 0x2000:
  1163. if (c == 0x207E) return UCD_PROPERTY_OTHER_MATH;
  1164. if (c == 0x208E) return UCD_PROPERTY_OTHER_MATH;
  1165. break;
  1166. case 0x2300:
  1167. if (c == 0x2309) return UCD_PROPERTY_OTHER_MATH;
  1168. if (c == 0x230B) return UCD_PROPERTY_OTHER_MATH;
  1169. if (c == 0x232A) return UCD_PROPERTY_DEPRECATED;
  1170. break;
  1171. case 0x2700:
  1172. if (c == 0x27C6) return UCD_PROPERTY_OTHER_MATH;
  1173. if (c >= 0x27E6 && c <= 0x27EF) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1174. break;
  1175. case 0x2900:
  1176. if (c >= 0x2983 && c <= 0x2998) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1177. if (c >= 0x29D8 && c <= 0x29DB) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1178. if (c == 0x29FD) return UCD_PROPERTY_OTHER_MATH;
  1179. break;
  1180. case 0x3000:
  1181. if (c == 0x300D) return UCD_PROPERTY_QUOTATION_MARK;
  1182. if (c == 0x300F) return UCD_PROPERTY_QUOTATION_MARK;
  1183. if (c >= 0x301E && c <= 0x301F) return UCD_PROPERTY_QUOTATION_MARK;
  1184. break;
  1185. case 0xFE00:
  1186. if (c == 0xFE42) return UCD_PROPERTY_QUOTATION_MARK;
  1187. if (c == 0xFE44) return UCD_PROPERTY_QUOTATION_MARK;
  1188. break;
  1189. case 0xFF00:
  1190. if (c == 0xFF63) return UCD_PROPERTY_QUOTATION_MARK;
  1191. break;
  1192. }
  1193. return 0;
  1194. }
  1195. static int properties_Pf(codepoint_t c)
  1196. {
  1197. switch (c & 0xFFFFFF00)
  1198. {
  1199. case 0x0000:
  1200. if (c == 0x00BB) return UCD_PROPERTY_QUOTATION_MARK;
  1201. break;
  1202. case 0x2000:
  1203. if (c == 0x2019) return UCD_PROPERTY_QUOTATION_MARK;
  1204. if (c == 0x201D) return UCD_PROPERTY_QUOTATION_MARK;
  1205. if (c == 0x203A) return UCD_PROPERTY_QUOTATION_MARK;
  1206. break;
  1207. }
  1208. return 0;
  1209. }
  1210. static int properties_Pi(codepoint_t c)
  1211. {
  1212. switch (c & 0xFFFFFF00)
  1213. {
  1214. case 0x0000:
  1215. if (c == 0x00AB) return UCD_PROPERTY_QUOTATION_MARK;
  1216. break;
  1217. case 0x2000:
  1218. if (c == 0x2018) return UCD_PROPERTY_QUOTATION_MARK;
  1219. if (c >= 0x201B && c <= 0x201C) return UCD_PROPERTY_QUOTATION_MARK;
  1220. if (c == 0x201F) return UCD_PROPERTY_QUOTATION_MARK;
  1221. if (c == 0x2039) return UCD_PROPERTY_QUOTATION_MARK;
  1222. break;
  1223. }
  1224. return 0;
  1225. }
  1226. static int properties_Po(codepoint_t c)
  1227. {
  1228. switch (c & 0xFFFFFF00)
  1229. {
  1230. case 0x0000:
  1231. if (c == 0x0021) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1232. if (c == 0x0022) return UCD_PROPERTY_QUOTATION_MARK;
  1233. if (c == 0x0027) return UCD_PROPERTY_QUOTATION_MARK;
  1234. if (c == 0x002C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1235. if (c == 0x002E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1236. if (c >= 0x003A && c <= 0x003B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1237. if (c == 0x003F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1238. if (c == 0x00B7) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  1239. break;
  1240. case 0x0300:
  1241. if (c == 0x037E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1242. if (c == 0x0387) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1243. break;
  1244. case 0x0500:
  1245. if (c == 0x0589) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1246. if (c == 0x05C3) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1247. break;
  1248. case 0x0600:
  1249. if (c == 0x060C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1250. if (c == 0x061B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1251. if (c == 0x061F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1252. if (c == 0x06D4) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1253. break;
  1254. case 0x0700:
  1255. if (c >= 0x0700 && c <= 0x070A) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1256. if (c == 0x070C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1257. if (c >= 0x07F8 && c <= 0x07F9) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1258. break;
  1259. case 0x0800:
  1260. if (c >= 0x0830 && c <= 0x083E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1261. if (c == 0x085E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1262. break;
  1263. case 0x0900:
  1264. if (c >= 0x0964 && c <= 0x0965) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1265. break;
  1266. case 0x0E00:
  1267. if (c >= 0x0E5A && c <= 0x0E5B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1268. break;
  1269. case 0x0F00:
  1270. if (c == 0x0F08) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1271. if (c >= 0x0F0D && c <= 0x0F12) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1272. break;
  1273. case 0x1000:
  1274. if (c >= 0x104A && c <= 0x104B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1275. break;
  1276. case 0x1300:
  1277. if (c >= 0x1361 && c <= 0x1368) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1278. break;
  1279. case 0x1600:
  1280. if (c >= 0x166D && c <= 0x166E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1281. if (c >= 0x16EB && c <= 0x16ED) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1282. break;
  1283. case 0x1700:
  1284. if (c >= 0x1735 && c <= 0x1736) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1285. if (c >= 0x17D4 && c <= 0x17D6) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1286. if (c == 0x17DA) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1287. break;
  1288. case 0x1800:
  1289. if (c >= 0x1802 && c <= 0x1805) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1290. if (c >= 0x1808 && c <= 0x1809) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1291. if (c == 0x180A) return UCD_PROPERTY_EXTENDER;
  1292. break;
  1293. case 0x1900:
  1294. if (c >= 0x1944 && c <= 0x1945) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1295. break;
  1296. case 0x1A00:
  1297. if (c >= 0x1AA8 && c <= 0x1AAB) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1298. break;
  1299. case 0x1B00:
  1300. if (c >= 0x1B5A && c <= 0x1B5B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1301. if (c >= 0x1B5D && c <= 0x1B5F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1302. break;
  1303. case 0x1C00:
  1304. if (c >= 0x1C3B && c <= 0x1C3F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1305. if (c >= 0x1C7E && c <= 0x1C7F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1306. if (c == 0x1CD3) return UCD_PROPERTY_DIACRITIC;
  1307. break;
  1308. case 0x2000:
  1309. if (c == 0x2016) return UCD_PROPERTY_OTHER_MATH;
  1310. if (c >= 0x2032 && c <= 0x2034) return UCD_PROPERTY_OTHER_MATH;
  1311. if (c >= 0x203C && c <= 0x203D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1312. if (c >= 0x2047 && c <= 0x2049) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1313. if (c == 0x2053) return UCD_PROPERTY_DASH;
  1314. break;
  1315. case 0x2E00:
  1316. if (c == 0x2E2E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1317. if (c == 0x2E3C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1318. if (c == 0x2E41) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1319. break;
  1320. case 0x3000:
  1321. if (c >= 0x3001 && c <= 0x3002) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1322. if (c == 0x30FB) return UCD_PROPERTY_HYPHEN;
  1323. break;
  1324. case 0xA400:
  1325. if (c >= 0xA4FE && c <= 0xA4FF) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1326. break;
  1327. case 0xA600:
  1328. if (c >= 0xA60D && c <= 0xA60F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1329. if (c >= 0xA6F3 && c <= 0xA6F7) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1330. break;
  1331. case 0xA800:
  1332. if (c >= 0xA876 && c <= 0xA877) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1333. if (c >= 0xA8CE && c <= 0xA8CF) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1334. break;
  1335. case 0xA900:
  1336. if (c == 0xA92E) return UCD_PROPERTY_DIACRITIC;
  1337. if (c == 0xA92F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1338. if (c >= 0xA9C7 && c <= 0xA9C9) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1339. break;
  1340. case 0xAA00:
  1341. if (c >= 0xAA5D && c <= 0xAA5F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1342. if (c == 0xAADF) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1343. if (c >= 0xAAF0 && c <= 0xAAF1) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1344. break;
  1345. case 0xAB00:
  1346. if (c == 0xABEB) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1347. break;
  1348. case 0xFE00:
  1349. if (c >= 0xFE50 && c <= 0xFE52) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1350. if (c >= 0xFE54 && c <= 0xFE57) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1351. if (c == 0xFE61) return UCD_PROPERTY_OTHER_MATH;
  1352. if (c == 0xFE68) return UCD_PROPERTY_OTHER_MATH;
  1353. break;
  1354. case 0xFF00:
  1355. if (c == 0xFF01) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1356. if (c == 0xFF02) return UCD_PROPERTY_QUOTATION_MARK;
  1357. if (c == 0xFF07) return UCD_PROPERTY_QUOTATION_MARK;
  1358. if (c == 0xFF0C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1359. if (c == 0xFF0E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1360. if (c == 0xFF3C) return UCD_PROPERTY_OTHER_MATH;
  1361. if (c == 0xFF65) return UCD_PROPERTY_HYPHEN;
  1362. if (c >= 0xFF1A && c <= 0xFF1B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1363. if (c == 0xFF1F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1364. if (c == 0xFF61) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1365. if (c == 0xFF64) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1366. break;
  1367. case 0x10300:
  1368. if (c == 0x01039F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1369. if (c == 0x0103D0) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1370. break;
  1371. case 0x10800:
  1372. if (c == 0x010857) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1373. break;
  1374. case 0x10900:
  1375. if (c == 0x01091F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1376. break;
  1377. case 0x10A00:
  1378. if (c >= 0x010A56 && c <= 0x010A57) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1379. if (c >= 0x010AF0 && c <= 0x010AF5) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1380. break;
  1381. case 0x10B00:
  1382. if (c >= 0x010B3A && c <= 0x010B3F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1383. if (c >= 0x010B99 && c <= 0x010B9C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1384. break;
  1385. case 0x11000:
  1386. if (c >= 0x011047 && c <= 0x01104D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1387. if (c >= 0x0110BE && c <= 0x0110C1) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1388. break;
  1389. case 0x11100:
  1390. if (c >= 0x011141 && c <= 0x011143) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1391. if (c >= 0x0111C5 && c <= 0x0111C6) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1392. if (c == 0x0111CD) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1393. if (c >= 0x0111DE && c <= 0x0111DF) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1394. break;
  1395. case 0x11200:
  1396. if (c >= 0x011238 && c <= 0x01123C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1397. if (c == 0x0112A9) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1398. break;
  1399. case 0x11400:
  1400. if (c >= 0x01144B && c <= 0x01144D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1401. if (c == 0x01145B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1402. break;
  1403. case 0x11500:
  1404. if (c >= 0x0115C2 && c <= 0x0115C5) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1405. if (c >= 0x0115C6 && c <= 0x0115C8) return UCD_PROPERTY_EXTENDER;
  1406. if (c >= 0x0115C9 && c <= 0x0115D7) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1407. break;
  1408. case 0x11600:
  1409. if (c >= 0x011641 && c <= 0x011642) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1410. break;
  1411. case 0x11700:
  1412. if (c >= 0x01173C && c <= 0x01173E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1413. break;
  1414. case 0x11C00:
  1415. if (c >= 0x011C41 && c <= 0x011C43) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1416. if (c == 0x011C71) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1417. break;
  1418. case 0x12400:
  1419. if (c >= 0x012470 && c <= 0x012474) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1420. break;
  1421. case 0x16A00:
  1422. if (c >= 0x016A6E && c <= 0x016A6F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1423. if (c == 0x016AF5) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1424. break;
  1425. case 0x16B00:
  1426. if (c >= 0x016B37 && c <= 0x016B39) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1427. if (c == 0x016B44) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1428. break;
  1429. case 0x1BC00:
  1430. if (c == 0x01BC9F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1431. break;
  1432. case 0x1DA00:
  1433. if (c >= 0x01DA87 && c <= 0x01DA8A) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1434. break;
  1435. }
  1436. return 0;
  1437. }
  1438. static int properties_Ps(codepoint_t c)
  1439. {
  1440. switch (c & 0xFFFFFF00)
  1441. {
  1442. case 0x2000:
  1443. if (c == 0x201A) return UCD_PROPERTY_QUOTATION_MARK;
  1444. if (c == 0x201E) return UCD_PROPERTY_QUOTATION_MARK;
  1445. if (c == 0x207D) return UCD_PROPERTY_OTHER_MATH;
  1446. if (c == 0x208D) return UCD_PROPERTY_OTHER_MATH;
  1447. break;
  1448. case 0x2300:
  1449. if (c == 0x2308) return UCD_PROPERTY_OTHER_MATH;
  1450. if (c == 0x230A) return UCD_PROPERTY_OTHER_MATH;
  1451. if (c == 0x2329) return UCD_PROPERTY_DEPRECATED;
  1452. break;
  1453. case 0x2700:
  1454. if (c == 0x27C5) return UCD_PROPERTY_OTHER_MATH;
  1455. if (c >= 0x27E6 && c <= 0x27EF) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1456. break;
  1457. case 0x2900:
  1458. if (c >= 0x2983 && c <= 0x2998) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1459. if (c >= 0x29D8 && c <= 0x29DB) return UCD_PROPERTY_OTHER_MATH; // Pe|Ps
  1460. if (c == 0x29FC) return UCD_PROPERTY_OTHER_MATH;
  1461. break;
  1462. case 0x2E00:
  1463. if (c == 0x2E42) return UCD_PROPERTY_QUOTATION_MARK;
  1464. break;
  1465. case 0x3000:
  1466. if (c == 0x300C) return UCD_PROPERTY_QUOTATION_MARK;
  1467. if (c == 0x300E) return UCD_PROPERTY_QUOTATION_MARK;
  1468. if (c == 0x301D) return UCD_PROPERTY_QUOTATION_MARK;
  1469. break;
  1470. case 0xFE00:
  1471. if (c == 0xFE41) return UCD_PROPERTY_QUOTATION_MARK;
  1472. if (c == 0xFE43) return UCD_PROPERTY_QUOTATION_MARK;
  1473. break;
  1474. case 0xFF00:
  1475. if (c == 0xFF62) return UCD_PROPERTY_QUOTATION_MARK;
  1476. break;
  1477. }
  1478. return 0;
  1479. }
  1480. static int properties_Sk(codepoint_t c)
  1481. {
  1482. switch (c & 0xFFFFFF00)
  1483. {
  1484. case 0x0000:
  1485. if (c == 0x005E) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_DIACRITIC;
  1486. if (c == 0x0060) return UCD_PROPERTY_DIACRITIC;
  1487. if (c == 0x00A8) return UCD_PROPERTY_DIACRITIC;
  1488. if (c == 0x00AF) return UCD_PROPERTY_DIACRITIC;
  1489. if (c == 0x00B4) return UCD_PROPERTY_DIACRITIC;
  1490. if (c == 0x00B8) return UCD_PROPERTY_DIACRITIC;
  1491. break;
  1492. case 0x0200:
  1493. if (c >= 0x02C2 && c <= 0x02C5) return UCD_PROPERTY_DIACRITIC;
  1494. if (c >= 0x02D2 && c <= 0x02DF) return UCD_PROPERTY_DIACRITIC;
  1495. if (c >= 0x02E5 && c <= 0x02EB) return UCD_PROPERTY_DIACRITIC;
  1496. if (c == 0x02ED) return UCD_PROPERTY_DIACRITIC;
  1497. if (c >= 0x02EF && c <= 0x02FF) return UCD_PROPERTY_DIACRITIC;
  1498. break;
  1499. case 0x0300:
  1500. if (c == 0x0375) return UCD_PROPERTY_DIACRITIC;
  1501. if (c >= 0x0384 && c <= 0x0385) return UCD_PROPERTY_DIACRITIC;
  1502. break;
  1503. case 0x1F00:
  1504. if (c == 0x1FBD) return UCD_PROPERTY_DIACRITIC;
  1505. if (c >= 0x1FBF && c <= 0x1FC1) return UCD_PROPERTY_DIACRITIC;
  1506. if (c >= 0x1FCD && c <= 0x1FCF) return UCD_PROPERTY_DIACRITIC;
  1507. if (c >= 0x1FDD && c <= 0x1FDF) return UCD_PROPERTY_DIACRITIC;
  1508. if (c >= 0x1FED && c <= 0x1FEF) return UCD_PROPERTY_DIACRITIC;
  1509. if (c >= 0x1FFD && c <= 0x1FFE) return UCD_PROPERTY_DIACRITIC;
  1510. break;
  1511. case 0x3000:
  1512. if (c >= 0x309B && c <= 0x309C) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ID_START;
  1513. break;
  1514. case 0xA700:
  1515. if (c >= 0xA720 && c <= 0xA721) return UCD_PROPERTY_DIACRITIC;
  1516. break;
  1517. case 0xAB00:
  1518. if (c == 0xAB5B) return UCD_PROPERTY_DIACRITIC;
  1519. break;
  1520. case 0xFF00:
  1521. if (c == 0xFF3E) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_MATH;
  1522. if (c == 0xFF40) return UCD_PROPERTY_DIACRITIC;
  1523. if (c == 0xFFE3) return UCD_PROPERTY_DIACRITIC;
  1524. break;
  1525. }
  1526. return 0;
  1527. }
  1528. static int properties_Sm(codepoint_t c)
  1529. {
  1530. switch (c & 0xFFFFFF00)
  1531. {
  1532. case 0x2000:
  1533. if (c == 0x207B) return UCD_PROPERTY_DASH;
  1534. if (c == 0x208B) return UCD_PROPERTY_DASH;
  1535. break;
  1536. case 0x2100:
  1537. if (c == 0x2118) return UCD_PROPERTY_OTHER_ID_START;
  1538. break;
  1539. case 0x2200:
  1540. if (c == 0x2212) return UCD_PROPERTY_DASH;
  1541. break;
  1542. }
  1543. return 0;
  1544. }
  1545. static int properties_So(codepoint_t c)
  1546. {
  1547. switch (c & 0xFFFFFF00)
  1548. {
  1549. case 0x2100:
  1550. if (c == 0x2129) return UCD_PROPERTY_OTHER_MATH;
  1551. if (c == 0x212E) return UCD_PROPERTY_OTHER_ID_START;
  1552. if (c >= 0x2195 && c <= 0x2199) return UCD_PROPERTY_OTHER_MATH;
  1553. if (c >= 0x219C && c <= 0x219F) return UCD_PROPERTY_OTHER_MATH;
  1554. if (c >= 0x21A1 && c <= 0x21A2) return UCD_PROPERTY_OTHER_MATH;
  1555. if (c >= 0x21A4 && c <= 0x21A5) return UCD_PROPERTY_OTHER_MATH;
  1556. if (c == 0x21A7) return UCD_PROPERTY_OTHER_MATH;
  1557. if (c >= 0x21A9 && c <= 0x21AD) return UCD_PROPERTY_OTHER_MATH;
  1558. if (c >= 0x21B0 && c <= 0x21B1) return UCD_PROPERTY_OTHER_MATH;
  1559. if (c >= 0x21B6 && c <= 0x21B7) return UCD_PROPERTY_OTHER_MATH;
  1560. if (c >= 0x21BC && c <= 0x21CD) return UCD_PROPERTY_OTHER_MATH;
  1561. if (c >= 0x21D0 && c <= 0x21D1) return UCD_PROPERTY_OTHER_MATH;
  1562. if (c == 0x21D3) return UCD_PROPERTY_OTHER_MATH;
  1563. if (c >= 0x21D5 && c <= 0x21DB) return UCD_PROPERTY_OTHER_MATH;
  1564. if (c == 0x21DD) return UCD_PROPERTY_OTHER_MATH;
  1565. if (c >= 0x21E4 && c <= 0x21E5) return UCD_PROPERTY_OTHER_MATH;
  1566. break;
  1567. case 0x2300:
  1568. if (c >= 0x23B4 && c <= 0x23B5) return UCD_PROPERTY_OTHER_MATH;
  1569. if (c == 0x23B7) return UCD_PROPERTY_OTHER_MATH;
  1570. if (c == 0x23D0) return UCD_PROPERTY_OTHER_MATH;
  1571. if (c == 0x23E2) return UCD_PROPERTY_OTHER_MATH;
  1572. break;
  1573. case 0x2400:
  1574. if (c >= 0x24B6 && c <= 0x24CF) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE;
  1575. if (c >= 0x24D0 && c <= 0x24E9) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_LOWERCASE;
  1576. break;
  1577. case 0x2500:
  1578. if (c >= 0x25A0 && c <= 0x25A1) return UCD_PROPERTY_OTHER_MATH;
  1579. if (c >= 0x25AE && c <= 0x25B6) return UCD_PROPERTY_OTHER_MATH;
  1580. if (c >= 0x25BC && c <= 0x25C0) return UCD_PROPERTY_OTHER_MATH;
  1581. if (c >= 0x25C6 && c <= 0x25C7) return UCD_PROPERTY_OTHER_MATH;
  1582. if (c >= 0x25CA && c <= 0x25CB) return UCD_PROPERTY_OTHER_MATH;
  1583. if (c >= 0x25CF && c <= 0x25D3) return UCD_PROPERTY_OTHER_MATH;
  1584. if (c == 0x25E2) return UCD_PROPERTY_OTHER_MATH;
  1585. if (c == 0x25E4) return UCD_PROPERTY_OTHER_MATH;
  1586. if (c >= 0x25E7 && c <= 0x25EC) return UCD_PROPERTY_OTHER_MATH;
  1587. break;
  1588. case 0x2600:
  1589. if (c >= 0x2605 && c <= 0x2606) return UCD_PROPERTY_OTHER_MATH;
  1590. if (c == 0x2640) return UCD_PROPERTY_OTHER_MATH;
  1591. if (c == 0x2642) return UCD_PROPERTY_OTHER_MATH;
  1592. if (c >= 0x2660 && c <= 0x2663) return UCD_PROPERTY_OTHER_MATH;
  1593. if (c >= 0x266D && c <= 0x266E) return UCD_PROPERTY_OTHER_MATH;
  1594. break;
  1595. case 0x2E00:
  1596. if (c >= 0x2E80 && c <= 0x2E99) return UCD_PROPERTY_RADICAL;
  1597. if (c >= 0x2E9B && c <= 0x2EF3) return UCD_PROPERTY_RADICAL;
  1598. break;
  1599. case 0x2F00:
  1600. if (c >= 0x2F00 && c <= 0x2FD5) return UCD_PROPERTY_RADICAL;
  1601. if (c >= 0x2FF0 && c <= 0x2FF1) return UCD_PROPERTY_IDS_BINARY_OPERATOR;
  1602. if (c >= 0x2FF2 && c <= 0x2FF3) return UCD_PROPERTY_IDS_TRINARY_OPERATOR;
  1603. if (c >= 0x2FF4 && c <= 0x2FFB) return UCD_PROPERTY_IDS_BINARY_OPERATOR;
  1604. break;
  1605. case 0x01F100:
  1606. if (c >= 0x01F130 && c <= 0x01F149) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE;
  1607. if (c >= 0x01F150 && c <= 0x01F169) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE;
  1608. if (c >= 0x01F170 && c <= 0x01F189) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE;
  1609. break;
  1610. }
  1611. return 0;
  1612. }
  1613. ucd_property ucd_properties(codepoint_t c, ucd_category category)
  1614. {
  1615. switch (category)
  1616. {
  1617. case UCD_CATEGORY_Cc: return properties_Cc(c);
  1618. case UCD_CATEGORY_Cf: return properties_Cf(c);
  1619. case UCD_CATEGORY_Cn: return properties_Cn(c);
  1620. case UCD_CATEGORY_Ll: return properties_Ll(c);
  1621. case UCD_CATEGORY_Lm: return properties_Lm(c);
  1622. case UCD_CATEGORY_Lo: return properties_Lo(c) | properties_Lo_ideographic(c);
  1623. case UCD_CATEGORY_Lu: return properties_Lu(c);
  1624. case UCD_CATEGORY_Mc: return properties_Mc(c);
  1625. case UCD_CATEGORY_Mn: return properties_Mn(c);
  1626. case UCD_CATEGORY_Nd: return properties_Nd(c);
  1627. case UCD_CATEGORY_Nl: return properties_Nl(c);
  1628. case UCD_CATEGORY_Pc: return properties_Pc(c);
  1629. case UCD_CATEGORY_Pd: return properties_Pd(c);
  1630. case UCD_CATEGORY_Pe: return properties_Pe(c);
  1631. case UCD_CATEGORY_Pf: return properties_Pf(c);
  1632. case UCD_CATEGORY_Pi: return properties_Pi(c);
  1633. case UCD_CATEGORY_Po: return properties_Po(c);
  1634. case UCD_CATEGORY_Ps: return properties_Ps(c);
  1635. case UCD_CATEGORY_Sk: return properties_Sk(c);
  1636. case UCD_CATEGORY_Sm: return properties_Sm(c);
  1637. case UCD_CATEGORY_So: return properties_So(c);
  1638. case UCD_CATEGORY_Zl: return UCD_PROPERTY_WHITE_SPACE;
  1639. case UCD_CATEGORY_Zp: return UCD_PROPERTY_WHITE_SPACE;
  1640. case UCD_CATEGORY_Zs: return UCD_PROPERTY_WHITE_SPACE;
  1641. default: return 0; // Co Cs Ii Lt Me No Sc
  1642. };
  1643. }