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 106KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066
  1. /* PropList APIs.
  2. *
  3. * Copyright (C) 2017-2018 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 ucd_property properties_Cc(codepoint_t c)
  22. {
  23. if (c >= 0x0009 && c <= 0x000D) return UCD_PROPERTY_WHITE_SPACE | UCD_PROPERTY_PATTERN_WHITE_SPACE;
  24. if (c == 0x0085) return UCD_PROPERTY_WHITE_SPACE | UCD_PROPERTY_PATTERN_WHITE_SPACE;
  25. return 0;
  26. }
  27. static ucd_property 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 >= 0x0600 && c <= 0x0605) return UCD_PROPERTY_PREPENDED_CONCATENATION_MARK;
  36. if (c == 0x061C) return UCD_PROPERTY_BIDI_CONTROL;
  37. if (c == 0x06DD) return UCD_PROPERTY_PREPENDED_CONCATENATION_MARK;
  38. break;
  39. case 0x0700:
  40. if (c == 0x070F) return UCD_PROPERTY_PREPENDED_CONCATENATION_MARK;
  41. break;
  42. case 0x0800:
  43. if (c == 0x08E2) return UCD_PROPERTY_PREPENDED_CONCATENATION_MARK;
  44. break;
  45. case 0x2000:
  46. if (c == 0x200C) return UCD_PROPERTY_JOIN_CONTROL | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  47. if (c == 0x200D) return UCD_PROPERTY_JOIN_CONTROL;
  48. if (c >= 0x200E && c <= 0x200F) return UCD_PROPERTY_BIDI_CONTROL | UCD_PROPERTY_PATTERN_WHITE_SPACE;
  49. if (c >= 0x202A && c <= 0x202E) return UCD_PROPERTY_BIDI_CONTROL;
  50. if (c >= 0x2061 && c <= 0x2064) return UCD_PROPERTY_OTHER_MATH;
  51. if (c >= 0x2066 && c <= 0x2069) return UCD_PROPERTY_BIDI_CONTROL;
  52. if (c >= 0x206A && c <= 0x206F) return UCD_PROPERTY_DEPRECATED;
  53. break;
  54. case 0x011000:
  55. if (c == 0x0110BD) return UCD_PROPERTY_PREPENDED_CONCATENATION_MARK;
  56. if (c == 0x0110CD) return UCD_PROPERTY_PREPENDED_CONCATENATION_MARK;
  57. break;
  58. case 0x0E0000:
  59. if (c == 0x0E0001) return UCD_PROPERTY_DEPRECATED;
  60. if (c >= 0x0E0020 && c <= 0x0E007F) return UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  61. break;
  62. }
  63. return 0;
  64. }
  65. static ucd_property properties_Cn(codepoint_t c)
  66. {
  67. switch (c & 0xFFFFFF00)
  68. {
  69. case 0x2000:
  70. if (c == 0x2065) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  71. break;
  72. case 0x2400:
  73. if (c >= 0x2427 && c <= 0x243F) return UCD_PROPERTY_PATTERN_SYNTAX;
  74. if (c >= 0x244B && c <= 0x245F) return UCD_PROPERTY_PATTERN_SYNTAX;
  75. break;
  76. case 0x2B00:
  77. if (c >= 0x2B74 && c <= 0x2B75) return UCD_PROPERTY_PATTERN_SYNTAX;
  78. if (c >= 0x2B96 && c <= 0x2B97) return UCD_PROPERTY_PATTERN_SYNTAX;
  79. if (c >= 0x2BBA && c <= 0x2BBC) return UCD_PROPERTY_PATTERN_SYNTAX;
  80. if (c == 0x2BC9) return UCD_PROPERTY_PATTERN_SYNTAX;
  81. if (c >= 0x2BD3 && c <= 0x2BEB) return UCD_PROPERTY_PATTERN_SYNTAX;
  82. if (c >= 0x2BF0 && c <= 0x2BFF) return UCD_PROPERTY_PATTERN_SYNTAX;
  83. break;
  84. case 0x2E00:
  85. if (c >= 0x2E45 && c <= 0x2E7F) return UCD_PROPERTY_PATTERN_SYNTAX;
  86. break;
  87. case 0xFD00:
  88. if (c >= 0xFDD0 && c <= 0xFDEF) return UCD_PROPERTY_NONCHARACTER_CODE_POINT;
  89. break;
  90. case 0xFF00:
  91. if (c >= 0xFFF0 && c <= 0xFFF8) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  92. break;
  93. case 0x0E0000:
  94. if (c == 0xE0000) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  95. if (c >= 0xE0002 && c <= 0xE001F) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  96. if (c >= 0xE0080 && c <= 0xE00FF) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  97. break;
  98. case 0x0E0100:
  99. if (c >= 0xE01F0) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  100. break;
  101. case 0x0E0200:
  102. case 0x0E0300:
  103. case 0x0E0400:
  104. case 0x0E0500:
  105. case 0x0E0600:
  106. case 0x0E0700:
  107. case 0x0E0800:
  108. case 0x0E0900:
  109. case 0x0E0A00:
  110. case 0x0E0B00:
  111. case 0x0E0C00:
  112. case 0x0E0D00:
  113. case 0x0E0E00:
  114. case 0x0E0F00:
  115. return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  116. }
  117. if ((c & 0x0000FFFF) >= 0xFFFE) return UCD_PROPERTY_NONCHARACTER_CODE_POINT;
  118. return 0;
  119. }
  120. static ucd_property properties_Ll(codepoint_t c)
  121. {
  122. switch (c & 0xFFFFFF00)
  123. {
  124. case 0x0000:
  125. if (c >= 0x0061 && c <= 0x0066) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT;
  126. if (c >= 0x0069 && c <= 0x006A) return UCD_PROPERTY_SOFT_DOTTED;
  127. break;
  128. case 0x0100:
  129. if (c == 0x012F) return UCD_PROPERTY_SOFT_DOTTED;
  130. if (c == 0x0149) return UCD_PROPERTY_DEPRECATED;
  131. break;
  132. case 0x0200:
  133. if (c == 0x0249) return UCD_PROPERTY_SOFT_DOTTED;
  134. if (c == 0x0268) return UCD_PROPERTY_SOFT_DOTTED;
  135. if (c == 0x029D) return UCD_PROPERTY_SOFT_DOTTED;
  136. break;
  137. case 0x0300:
  138. if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  139. if (c == 0x03D5) return UCD_PROPERTY_OTHER_MATH;
  140. if (c >= 0x03F0 && c <= 0x03F1) return UCD_PROPERTY_OTHER_MATH;
  141. if (c == 0x03F3) return UCD_PROPERTY_SOFT_DOTTED;
  142. if (c >= 0x03F4 && c <= 0x03F5) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  143. break;
  144. case 0x0400:
  145. if (c == 0x0456) return UCD_PROPERTY_SOFT_DOTTED;
  146. if (c == 0x0458) return UCD_PROPERTY_SOFT_DOTTED;
  147. break;
  148. case 0x1D00:
  149. if (c == 0x1D96) return UCD_PROPERTY_SOFT_DOTTED;
  150. break;
  151. case 0x1E00:
  152. if (c == 0x1E2D) return UCD_PROPERTY_SOFT_DOTTED;
  153. if (c == 0x1ECB) return UCD_PROPERTY_SOFT_DOTTED;
  154. break;
  155. case 0x2100:
  156. if (c >= 0x210A && c <= 0x2113) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  157. if (c >= 0x212F && c <= 0x2131) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  158. if (c >= 0x2133 && c <= 0x2134) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  159. if (c == 0x2139) return UCD_PROPERTY_EMOJI;
  160. if (c >= 0x213C && c <= 0x213F) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  161. if (c >= 0x2145 && c <= 0x2147) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  162. if (c >= 0x2148 && c <= 0x2149) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  163. break;
  164. case 0xFF00:
  165. if (c >= 0xFF41 && c <= 0xFF46) return UCD_PROPERTY_HEX_DIGIT;
  166. break;
  167. case 0x01D400:
  168. if (c >= 0x01D422 && c <= 0x01D423) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  169. if (c >= 0x01D400 && c <= 0x01D454) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  170. if (c >= 0x01D456 && c <= 0x01D457) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  171. if (c >= 0x01D48A && c <= 0x01D48B) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  172. if (c >= 0x01D458 && c <= 0x01D49C) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  173. if (c >= 0x01D4AE && c <= 0x01D4B9) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  174. if (c == 0x01D4BB) return UCD_PROPERTY_OTHER_MATH;
  175. if (c >= 0x01D4BE && c <= 0x01D4BF) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  176. if (c >= 0x01D4BD && c <= 0x01D4C3) return UCD_PROPERTY_OTHER_MATH;
  177. if (c >= 0x01D4F2 && c <= 0x01D4F3) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  178. if (c >= 0x01D4C5) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  179. break;
  180. case 0x01D500:
  181. if (c <= 0x01D505) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  182. if (c >= 0x01D526 && c <= 0x01D527) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  183. if (c >= 0x01D51E && c <= 0x01D539) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  184. if (c >= 0x01D55A && c <= 0x01D55B) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  185. if (c >= 0x01D58E && c <= 0x01D58F) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  186. if (c >= 0x01D5C2 && c <= 0x01D5C3) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  187. if (c >= 0x01D5F6 && c <= 0x01D5F7) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  188. if (c >= 0x01D552) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  189. break;
  190. case 0x01D600:
  191. if (c >= 0x01D62A && c <= 0x01D62B) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  192. if (c >= 0x01D65E && c <= 0x01D65F) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  193. if (c >= 0x01D692 && c <= 0x01D693) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_SOFT_DOTTED;
  194. if (c <= 0x01D6A5) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  195. if (c >= 0x01D6C2 && c <= 0x01D6DA) return UCD_PROPERTY_OTHER_MATH;
  196. if (c >= 0x01D6DC && c <= 0x01D6FA) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  197. if (c >= 0x01D6FC) return UCD_PROPERTY_OTHER_MATH;
  198. break;
  199. case 0x01D700:
  200. if (c <= 0x01D714) return UCD_PROPERTY_OTHER_MATH;
  201. if (c >= 0x01D716 && c <= 0x01D734) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  202. if (c >= 0x01D736 && c <= 0x01D74E) return UCD_PROPERTY_OTHER_MATH;
  203. if (c >= 0x01D750 && c <= 0x01D76E) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  204. if (c >= 0x01D770 && c <= 0x01D788) return UCD_PROPERTY_OTHER_MATH;
  205. if (c >= 0x01D78A && c <= 0x01D7A8) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  206. if (c >= 0x01D7AA && c <= 0x01D7C2) return UCD_PROPERTY_OTHER_MATH;
  207. if (c >= 0x01D7C4 && c <= 0x01D7CB) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  208. break;
  209. }
  210. return 0;
  211. }
  212. static ucd_property properties_Lm(codepoint_t c)
  213. {
  214. switch (c & 0xFFFFFF00)
  215. {
  216. case 0x0200:
  217. if (c == 0x02B2) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_SOFT_DOTTED;
  218. if (c >= 0x02B0 && c <= 0x02B8) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  219. if (c >= 0x02B9 && c <= 0x02BF) return UCD_PROPERTY_DIACRITIC;
  220. if (c >= 0x02C0 && c <= 0x02C1) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  221. if (c >= 0x02C6 && c <= 0x02CF) return UCD_PROPERTY_DIACRITIC;
  222. if (c >= 0x02D0 && c <= 0x02D1) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  223. if (c >= 0x02E0 && c <= 0x02E4) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  224. if (c == 0x02EC) return UCD_PROPERTY_DIACRITIC;
  225. if (c == 0x02EE) return UCD_PROPERTY_DIACRITIC;
  226. break;
  227. case 0x0300:
  228. if (c == 0x0374) return UCD_PROPERTY_DIACRITIC;
  229. if (c == 0x037A) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  230. break;
  231. case 0x0500:
  232. if (c == 0x0559) return UCD_PROPERTY_DIACRITIC;
  233. break;
  234. case 0x0600:
  235. if (c == 0x0640) return UCD_PROPERTY_EXTENDER;
  236. if (c >= 0x06E5 && c <= 0x06E6) return UCD_PROPERTY_DIACRITIC;
  237. break;
  238. case 0x0700:
  239. if (c >= 0x07F4 && c <= 0x07F5) return UCD_PROPERTY_DIACRITIC;
  240. if (c == 0x07FA) return UCD_PROPERTY_EXTENDER;
  241. break;
  242. case 0x0900:
  243. if (c == 0x0971) return UCD_PROPERTY_DIACRITIC;
  244. break;
  245. case 0x0E00:
  246. if (c == 0x0E46) return UCD_PROPERTY_EXTENDER;
  247. if (c == 0x0EC6) return UCD_PROPERTY_EXTENDER;
  248. break;
  249. case 0x1800:
  250. if (c == 0x1843) return UCD_PROPERTY_EXTENDER;
  251. break;
  252. case 0x1A00:
  253. if (c == 0x1AA7) return UCD_PROPERTY_EXTENDER;
  254. break;
  255. case 0x1C00:
  256. if (c == 0x1C7B) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  257. if (c >= 0x1C78 && c <= 0x1C7D) return UCD_PROPERTY_DIACRITIC;
  258. break;
  259. case 0x1D00:
  260. if (c == 0x1D62) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_SOFT_DOTTED;
  261. if (c >= 0x1D2C && c <= 0x1D6A) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_DIACRITIC;
  262. if (c == 0x1D78) return UCD_PROPERTY_OTHER_LOWERCASE;
  263. if (c == 0x1DA4) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_SOFT_DOTTED;
  264. if (c == 0x1DA8) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_SOFT_DOTTED;
  265. if (c >= 0x1D9B && c <= 0x1DBF) return UCD_PROPERTY_OTHER_LOWERCASE;
  266. break;
  267. case 0x2000:
  268. if (c == 0x2071) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_SOFT_DOTTED;
  269. if (c == 0x207F) return UCD_PROPERTY_OTHER_LOWERCASE;
  270. if (c >= 0x2090 && c <= 0x209C) return UCD_PROPERTY_OTHER_LOWERCASE;
  271. break;
  272. case 0x2C00:
  273. if (c == 0x2C7C) return UCD_PROPERTY_OTHER_LOWERCASE | UCD_PROPERTY_SOFT_DOTTED;
  274. if (c == 0x2C7D) return UCD_PROPERTY_OTHER_LOWERCASE;
  275. break;
  276. case 0x2E00:
  277. if (c == 0x2E2F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_PATTERN_SYNTAX;
  278. break;
  279. case 0x3000:
  280. if (c == 0x3005) return UCD_PROPERTY_EXTENDER;
  281. if (c >= 0x3031 && c <= 0x3035) return UCD_PROPERTY_EXTENDER;
  282. if (c >= 0x309D && c <= 0x309E) return UCD_PROPERTY_EXTENDER;
  283. if (c == 0x30FC) return UCD_PROPERTY_EXTENDER | UCD_PROPERTY_DIACRITIC;
  284. if (c >= 0x30FD && c <= 0x30FE) return UCD_PROPERTY_EXTENDER;
  285. break;
  286. case 0xA000:
  287. if (c == 0xA015) return UCD_PROPERTY_EXTENDER;
  288. break;
  289. case 0xA600:
  290. if (c == 0xA60C) return UCD_PROPERTY_EXTENDER;
  291. if (c == 0xA67F) return UCD_PROPERTY_DIACRITIC;
  292. if (c >= 0xA69C && c <= 0xA69D) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  293. break;
  294. case 0xA700:
  295. if (c >= 0xA717 && c <= 0xA71F) return UCD_PROPERTY_DIACRITIC;
  296. if (c == 0xA770) return UCD_PROPERTY_OTHER_LOWERCASE;
  297. if (c == 0xA788) return UCD_PROPERTY_DIACRITIC;
  298. if (c >= 0xA7F8 && c <= 0xA7F9) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  299. break;
  300. case 0xA900:
  301. if (c == 0xA9CF) return UCD_PROPERTY_EXTENDER;
  302. if (c == 0xA9E6) return UCD_PROPERTY_EXTENDER;
  303. break;
  304. case 0xAA00:
  305. if (c == 0xAA70) return UCD_PROPERTY_EXTENDER;
  306. if (c == 0xAADD) return UCD_PROPERTY_EXTENDER;
  307. if (c >= 0xAAF3 && c <= 0xAAF4) return UCD_PROPERTY_EXTENDER;
  308. break;
  309. case 0xAB00:
  310. if (c >= 0xAB5C && c <= 0xAB5F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_LOWERCASE;
  311. break;
  312. case 0xFF00:
  313. if (c == 0xFF70) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  314. if (c >= 0xFF9E && c <= 0xFF9F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  315. break;
  316. case 0x016B00:
  317. if (c >= 0x016B42 && c <= 0x016B43) return UCD_PROPERTY_EXTENDER;
  318. break;
  319. case 0x016F00:
  320. if (c >= 0x016F93 && c <= 0x016F9F) return UCD_PROPERTY_DIACRITIC;
  321. if (c >= 0x016FE0 && c <= 0x016FE1) return UCD_PROPERTY_EXTENDER;
  322. break;
  323. }
  324. return 0;
  325. }
  326. static ucd_property properties_Lo(codepoint_t c)
  327. {
  328. switch (c & 0xFFFFFF00)
  329. {
  330. case 0x0000:
  331. if (c == 0x00AA) return UCD_PROPERTY_OTHER_LOWERCASE;
  332. if (c == 0x00BA) return UCD_PROPERTY_OTHER_LOWERCASE;
  333. break;
  334. case 0x0600:
  335. if (c == 0x0673) return UCD_PROPERTY_DEPRECATED;
  336. break;
  337. case 0x0E00:
  338. if (c >= 0x0E40 && c <= 0x0E44) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  339. if (c >= 0x0EC0 && c <= 0x0EC4) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  340. break;
  341. case 0x1100:
  342. if (c >= 0x115F && c <= 0x1160) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  343. break;
  344. case 0x1700:
  345. if (c >= 0x17A3 && c <= 0x17A4) return UCD_PROPERTY_DEPRECATED;
  346. break;
  347. case 0x1900:
  348. if (c >= 0x19B5 && c <= 0x19B7) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  349. if (c == 0x19BA) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  350. break;
  351. case 0x2100:
  352. if (c >= 0x2135 && c <= 0x2138) return UCD_PROPERTY_OTHER_MATH;
  353. break;
  354. case 0x3000:
  355. if (c == 0x3006) return UCD_PROPERTY_IDEOGRAPHIC;
  356. break;
  357. case 0x3100:
  358. if (c == 0x3164) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  359. break;
  360. case 0xAA00:
  361. if (c >= 0xAAB5 && c <= 0xAAB6) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  362. if (c == 0xAAB9) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  363. if (c >= 0xAABB && c <= 0xAABC) return UCD_PROPERTY_LOGICAL_ORDER_EXCEPTION;
  364. if (c == 0xAAC0) return UCD_PROPERTY_DIACRITIC;
  365. if (c == 0xAAC2) return UCD_PROPERTY_DIACRITIC;
  366. break;
  367. case 0xFA00:
  368. if (c >= 0xFA0E && c <= 0xFA0F) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  369. if (c == 0xFA11) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  370. if (c >= 0xFA13 && c <= 0xFA14) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  371. if (c == 0xFA1F) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  372. if (c == 0xFA21) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  373. if (c >= 0xFA23 && c <= 0xFA24) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  374. if (c >= 0xFA27 && c <= 0xFA29) return UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  375. break;
  376. case 0xFF00:
  377. if (c == 0xFFA0) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  378. break;
  379. case 0x10D00:
  380. if (c == 0x10D22) return UCD_PROPERTY_DIACRITIC;
  381. if (c == 0x10D23) return UCD_PROPERTY_DIACRITIC;
  382. break;
  383. case 0x18700:
  384. if (c >= 0x187ED && c <= 0x187F1) return UCD_PROPERTY_IDEOGRAPHIC;
  385. break;
  386. case 0x11300:
  387. if (c == 0x1135D) return UCD_PROPERTY_EXTENDER;
  388. break;
  389. case 0x1EE00:
  390. return UCD_PROPERTY_OTHER_MATH;
  391. }
  392. return 0;
  393. }
  394. static ucd_property properties_Lo_ideographic(codepoint_t c)
  395. {
  396. switch (c & 0xFFFF0000)
  397. {
  398. case 0x000000:
  399. if (c >= 0x3400 && c <= 0x4DB5) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  400. if (c >= 0x4E00 && c <= 0x9FEF) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  401. if (c >= 0xF900 && c <= 0xFA6D) return UCD_PROPERTY_IDEOGRAPHIC;
  402. if (c >= 0xFA70 && c <= 0xFAD9) return UCD_PROPERTY_IDEOGRAPHIC;
  403. break;
  404. case 0x010000:
  405. if (c >= 0x017000 && c <= 0x0187EC) return UCD_PROPERTY_IDEOGRAPHIC;
  406. if (c >= 0x018800 && c <= 0x018AF2) return UCD_PROPERTY_IDEOGRAPHIC;
  407. if (c >= 0x01B170 && c <= 0x01B2FB) return UCD_PROPERTY_IDEOGRAPHIC;
  408. break;
  409. case 0x020000:
  410. if (c >= 0x020000 && c <= 0x02A6D6) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  411. if (c >= 0x02A700 && c <= 0x02B734) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  412. if (c >= 0x02B740 && c <= 0x02B81D) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  413. if (c >= 0x02B820 && c <= 0x02CEA1) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  414. if (c >= 0x02CEB0 && c <= 0x02EBE0) return UCD_PROPERTY_IDEOGRAPHIC | UCD_PROPERTY_UNIFIED_IDEOGRAPH;
  415. if (c >= 0x02F800 && c <= 0x02FA1D) return UCD_PROPERTY_IDEOGRAPHIC;
  416. break;
  417. }
  418. return 0;
  419. }
  420. static ucd_property properties_Lu(codepoint_t c)
  421. {
  422. switch (c & 0xFFFFFF00)
  423. {
  424. case 0x0000:
  425. if (c >= 0x0041 && c <= 0x0046) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT;
  426. break;
  427. case 0x0300:
  428. if (c >= 0x03D0 && c <= 0x03D2) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  429. if (c >= 0x03F4 && c <= 0x03F5) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  430. break;
  431. case 0xFF00:
  432. if (c >= 0xFF21 && c <= 0xFF26) return UCD_PROPERTY_HEX_DIGIT;
  433. break;
  434. case 0x2100:
  435. if (c == 0x2102) return UCD_PROPERTY_OTHER_MATH;
  436. if (c == 0x2107) return UCD_PROPERTY_OTHER_MATH;
  437. if (c >= 0x210A && c <= 0x2113) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  438. if (c == 0x2115) return UCD_PROPERTY_OTHER_MATH;
  439. if (c >= 0x2119 && c <= 0x211D) return UCD_PROPERTY_OTHER_MATH;
  440. if (c == 0x2124) return UCD_PROPERTY_OTHER_MATH;
  441. if (c == 0x2128) return UCD_PROPERTY_OTHER_MATH;
  442. if (c >= 0x212C && c <= 0x212D) return UCD_PROPERTY_OTHER_MATH;
  443. if (c >= 0x212F && c <= 0x2131) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  444. if (c >= 0x2133 && c <= 0x2134) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  445. if (c >= 0x213C && c <= 0x213F) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  446. if (c >= 0x2145 && c <= 0x2149) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  447. break;
  448. case 0x01D400:
  449. if (c >= 0x01D400 && c <= 0x01D454) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  450. if (c >= 0x01D456 && c <= 0x01D49C) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  451. if (c >= 0x01D49E && c <= 0x01D49F) return UCD_PROPERTY_OTHER_MATH;
  452. if (c == 0x01D4A2) return UCD_PROPERTY_OTHER_MATH;
  453. if (c >= 0x01D4A5 && c <= 0x01D4A6) return UCD_PROPERTY_OTHER_MATH;
  454. if (c >= 0x01D4A9 && c <= 0x01D4AC) return UCD_PROPERTY_OTHER_MATH;
  455. if (c >= 0x01D4AE && c <= 0x01D4B9) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  456. if (c >= 0x01D4C5) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  457. break;
  458. case 0x01D500:
  459. if (c <= 0x01D505) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  460. if (c >= 0x01D507 && c <= 0x01D50A) return UCD_PROPERTY_OTHER_MATH;
  461. if (c >= 0x01D50D && c <= 0x01D514) return UCD_PROPERTY_OTHER_MATH;
  462. if (c >= 0x01D516 && c <= 0x01D51C) return UCD_PROPERTY_OTHER_MATH;
  463. if (c >= 0x01D51E && c <= 0x01D539) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  464. if (c >= 0x01D53B && c <= 0x01D53E) return UCD_PROPERTY_OTHER_MATH;
  465. if (c >= 0x01D540 && c <= 0x01D544) return UCD_PROPERTY_OTHER_MATH;
  466. if (c == 0x01D546) return UCD_PROPERTY_OTHER_MATH;
  467. if (c >= 0x01D54A && c <= 0x01D550) return UCD_PROPERTY_OTHER_MATH;
  468. if (c >= 0x01D552) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  469. break;
  470. case 0x01D600:
  471. if (c <= 0x01D6A5) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  472. if (c >= 0x01D6A8 && c <= 0x01D6C0) return UCD_PROPERTY_OTHER_MATH;
  473. if (c >= 0x01D6DC && c <= 0x01D6FA) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  474. break;
  475. case 0x01D700:
  476. if (c >= 0x01D716 && c <= 0x01D734) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  477. if (c >= 0x01D750 && c <= 0x01D76E) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  478. if (c >= 0x01D78A && c <= 0x01D7A8) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  479. if (c >= 0x01D7C4 && c <= 0x01D7CB) return UCD_PROPERTY_OTHER_MATH; /* Ll|Lu */
  480. break;
  481. }
  482. return 0;
  483. }
  484. static ucd_property properties_Mc(codepoint_t c)
  485. {
  486. switch (c & 0xFFFFFF00)
  487. {
  488. case 0x0900:
  489. if (c == 0x09BE) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  490. if (c == 0x09D7) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  491. break;
  492. case 0x0B00:
  493. if (c == 0x0B3E) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  494. if (c == 0x0B57) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  495. if (c == 0x0BBE) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  496. if (c == 0x0BD7) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  497. break;
  498. case 0x0C00:
  499. if (c == 0x0CC2) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  500. if (c >= 0x0CD5 && c <= 0x0CD6) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  501. break;
  502. case 0x0D00:
  503. if (c == 0x0D3E) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  504. if (c == 0x0D57) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  505. if (c == 0x0DCF) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  506. if (c == 0x0DDF) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  507. break;
  508. case 0x0F00:
  509. if (c >= 0x0F3E && c <= 0x0F3F) return UCD_PROPERTY_DIACRITIC;
  510. break;
  511. case 0x1000:
  512. if (c >= 0x102B && c <= 0x102C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  513. if (c == 0x1031) return UCD_PROPERTY_OTHER_ALPHABETIC;
  514. if (c == 0x1038) return UCD_PROPERTY_OTHER_ALPHABETIC;
  515. if (c >= 0x103B && c <= 0x103C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  516. if (c >= 0x1056 && c <= 0x1057) return UCD_PROPERTY_OTHER_ALPHABETIC;
  517. if (c == 0x1062) return UCD_PROPERTY_OTHER_ALPHABETIC;
  518. if (c >= 0x1067 && c <= 0x1068) return UCD_PROPERTY_OTHER_ALPHABETIC;
  519. if (c >= 0x1083 && c <= 0x1084) return UCD_PROPERTY_OTHER_ALPHABETIC;
  520. if (c >= 0x1087 && c <= 0x108C) return UCD_PROPERTY_DIACRITIC;
  521. if (c == 0x108F) return UCD_PROPERTY_DIACRITIC;
  522. if (c >= 0x109A && c <= 0x109B) return UCD_PROPERTY_DIACRITIC;
  523. if (c == 0x109C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  524. return 0;
  525. case 0x1B00:
  526. if (c == 0x1B04) return UCD_PROPERTY_OTHER_ALPHABETIC;
  527. if (c == 0x1B35) return UCD_PROPERTY_OTHER_ALPHABETIC;
  528. if (c == 0x1B3B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  529. if (c >= 0x1B3D && c <= 0x1B41) return UCD_PROPERTY_OTHER_ALPHABETIC;
  530. if (c == 0x1B43) return UCD_PROPERTY_OTHER_ALPHABETIC;
  531. if (c == 0x1B44) return UCD_PROPERTY_DIACRITIC;
  532. if (c == 0x1B82) return UCD_PROPERTY_OTHER_ALPHABETIC;
  533. if (c == 0x1BA1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  534. if (c >= 0x1BA6 && c <= 0x1BA7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  535. if (c == 0x1BAA) return UCD_PROPERTY_DIACRITIC;
  536. if (c == 0x1BE7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  537. if (c >= 0x1BEA && c <= 0x1BEC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  538. if (c == 0x1BEE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  539. return 0;
  540. case 0x1C00:
  541. if (c == 0x1CE1) return UCD_PROPERTY_DIACRITIC;
  542. if (c == 0x1CF7) return UCD_PROPERTY_DIACRITIC;
  543. break;
  544. case 0x3000:
  545. if (c >= 0x302E && c <= 0x302F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  546. break;
  547. case 0xA900:
  548. if (c == 0xA953) return UCD_PROPERTY_DIACRITIC;
  549. if (c == 0xA9C0) return UCD_PROPERTY_DIACRITIC;
  550. break;
  551. case 0xAA00:
  552. if (c == 0xAA7B) return UCD_PROPERTY_DIACRITIC;
  553. if (c == 0xAA7D) return UCD_PROPERTY_DIACRITIC;
  554. break;
  555. case 0xAB00:
  556. if (c == 0xABEC) return UCD_PROPERTY_DIACRITIC;
  557. break;
  558. case 0x011100:
  559. if (c == 0x0111C0) return UCD_PROPERTY_DIACRITIC;
  560. break;
  561. case 0x011200:
  562. if (c == 0x011235) return UCD_PROPERTY_DIACRITIC;
  563. break;
  564. case 0x011300:
  565. if (c == 0x01133E) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  566. if (c == 0x01134D) return UCD_PROPERTY_DIACRITIC;
  567. if (c == 0x011357) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  568. break;
  569. case 0x011400:
  570. if (c == 0x0114B0) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  571. if (c == 0x0114BD) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  572. break;
  573. case 0x011500:
  574. if (c == 0x0115AF) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  575. break;
  576. case 0x011600:
  577. if (c == 0x0116B6) return UCD_PROPERTY_DIACRITIC;
  578. break;
  579. case 0x01D100:
  580. if (c == 0x01D165) return UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  581. if (c == 0x01D166) return 0;
  582. if (c == 0x01D16D) return UCD_PROPERTY_DIACRITIC;
  583. if (c >= 0x01D16E && c <= 0x01D172) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_GRAPHEME_EXTEND;
  584. break;
  585. }
  586. return UCD_PROPERTY_OTHER_ALPHABETIC;
  587. }
  588. static ucd_property properties_Mn(codepoint_t c)
  589. {
  590. switch (c & 0xFFFFFF00)
  591. {
  592. case 0x0300:
  593. if (c >= 0x0300 && c <= 0x0344) return UCD_PROPERTY_DIACRITIC;
  594. if (c == 0x0345) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_LOWERCASE;
  595. if (c >= 0x0346 && c <= 0x034E) return UCD_PROPERTY_DIACRITIC;
  596. if (c == 0x034F) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  597. if (c >= 0x0350 && c <= 0x0357) return UCD_PROPERTY_DIACRITIC;
  598. if (c >= 0x035D && c <= 0x0362) return UCD_PROPERTY_DIACRITIC;
  599. break;
  600. case 0x0400:
  601. if (c >= 0x0483 && c <= 0x0487) return UCD_PROPERTY_DIACRITIC;
  602. break;
  603. case 0x0500:
  604. if (c >= 0x0591 && c <= 0x05A1) return UCD_PROPERTY_DIACRITIC;
  605. if (c >= 0x05A3 && c <= 0x05AF) return UCD_PROPERTY_DIACRITIC;
  606. if (c >= 0x05B0 && c <= 0x05BD) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  607. if (c == 0x05BF) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  608. if (c >= 0x05C1 && c <= 0x05C2) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  609. if (c == 0x05C4) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  610. if (c == 0x05C5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  611. if (c == 0x05C7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  612. break;
  613. case 0x0600:
  614. if (c >= 0x0610 && c <= 0x061A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  615. if (c >= 0x064B && c <= 0x0652) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  616. if (c >= 0x0653 && c <= 0x0656) return UCD_PROPERTY_OTHER_ALPHABETIC;
  617. if (c == 0x0657) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  618. if (c == 0x0658) return UCD_PROPERTY_DIACRITIC;
  619. if (c >= 0x0659 && c <= 0x065F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  620. if (c == 0x0670) return UCD_PROPERTY_OTHER_ALPHABETIC;
  621. if (c >= 0x06D6 && c <= 0x06DC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  622. if (c >= 0x06DF && c <= 0x06E0) return UCD_PROPERTY_DIACRITIC;
  623. if (c >= 0x06E1 && c <= 0x06E4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  624. if (c >= 0x06E7 && c <= 0x06E8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  625. if (c >= 0x06EA && c <= 0x06EC) return UCD_PROPERTY_DIACRITIC;
  626. if (c == 0x06ED) return UCD_PROPERTY_OTHER_ALPHABETIC;
  627. break;
  628. case 0x0700:
  629. if (c == 0x0711) return UCD_PROPERTY_OTHER_ALPHABETIC;
  630. if (c >= 0x0730 && c <= 0x073F) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  631. if (c >= 0x0740 && c <= 0x074A) return UCD_PROPERTY_DIACRITIC;
  632. if (c >= 0x07A6 && c <= 0x07B0) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  633. if (c >= 0x07EB && c <= 0x07F3) return UCD_PROPERTY_DIACRITIC;
  634. break;
  635. case 0x0800:
  636. if (c >= 0x0816 && c <= 0x0817) return UCD_PROPERTY_OTHER_ALPHABETIC;
  637. if (c >= 0x0818 && c <= 0x0819) return UCD_PROPERTY_DIACRITIC;
  638. if (c >= 0x081B && c <= 0x0823) return UCD_PROPERTY_OTHER_ALPHABETIC;
  639. if (c >= 0x0825 && c <= 0x0827) return UCD_PROPERTY_OTHER_ALPHABETIC;
  640. if (c >= 0x0829 && c <= 0x082C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  641. if (c >= 0x08D4 && c <= 0x08DF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  642. if (c >= 0x08E3 && c <= 0x08E9) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  643. if (c >= 0x08EA && c <= 0x08EF) return UCD_PROPERTY_DIACRITIC;
  644. if (c >= 0x08F0 && c <= 0x08FE) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  645. if (c == 0x08FF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  646. break;
  647. case 0x0900:
  648. if (c <= 0x0902) return UCD_PROPERTY_OTHER_ALPHABETIC;
  649. if (c == 0x093A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  650. if (c == 0x093C) return UCD_PROPERTY_DIACRITIC;
  651. if (c >= 0x0941 && c <= 0x0948) return UCD_PROPERTY_OTHER_ALPHABETIC;
  652. if (c == 0x094D) return UCD_PROPERTY_DIACRITIC;
  653. if (c >= 0x0951 && c <= 0x0954) return UCD_PROPERTY_DIACRITIC;
  654. if (c >= 0x0955 && c <= 0x0957) return UCD_PROPERTY_OTHER_ALPHABETIC;
  655. if (c >= 0x0962 && c <= 0x0963) return UCD_PROPERTY_OTHER_ALPHABETIC;
  656. if (c == 0x0981) return UCD_PROPERTY_OTHER_ALPHABETIC;
  657. if (c == 0x09BC) return UCD_PROPERTY_DIACRITIC;
  658. if (c >= 0x09C1 && c <= 0x09C4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  659. if (c == 0x09CD) return UCD_PROPERTY_DIACRITIC;
  660. if (c >= 0x09E2 && c <= 0x09E3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  661. break;
  662. case 0x0A00:
  663. if (c >= 0x0A01 && c <= 0x0A02) return UCD_PROPERTY_OTHER_ALPHABETIC;
  664. if (c == 0x0A3C) return UCD_PROPERTY_DIACRITIC;
  665. if (c >= 0x0A41 && c <= 0x0A42) return UCD_PROPERTY_OTHER_ALPHABETIC;
  666. if (c >= 0x0A47 && c <= 0x0A48) return UCD_PROPERTY_OTHER_ALPHABETIC;
  667. if (c >= 0x0A4B && c <= 0x0A4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  668. if (c == 0x0A4D) return UCD_PROPERTY_DIACRITIC;
  669. if (c == 0x0A51) return UCD_PROPERTY_OTHER_ALPHABETIC;
  670. if (c >= 0x0A70 && c <= 0x0A71) return UCD_PROPERTY_OTHER_ALPHABETIC;
  671. if (c == 0x0A75) return UCD_PROPERTY_OTHER_ALPHABETIC;
  672. if (c >= 0x0A81 && c <= 0x0A82) return UCD_PROPERTY_OTHER_ALPHABETIC;
  673. if (c == 0x0ABC) return UCD_PROPERTY_DIACRITIC;
  674. if (c >= 0x0AC1 && c <= 0x0AC5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  675. if (c >= 0x0AC7 && c <= 0x0AC8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  676. if (c == 0x0ACD) return UCD_PROPERTY_DIACRITIC;
  677. if (c >= 0x0AE2 && c <= 0x0AE3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  678. if (c >= 0x0AFA && c <= 0x0AFC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  679. if (c >= 0x0AFD && c <= 0x0AFF) return UCD_PROPERTY_DIACRITIC;
  680. break;
  681. case 0x0B00:
  682. if (c == 0x0B01) return UCD_PROPERTY_OTHER_ALPHABETIC;
  683. if (c == 0x0B3C) return UCD_PROPERTY_DIACRITIC;
  684. if (c == 0x0B3F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  685. if (c >= 0x0B41 && c <= 0x0B44) return UCD_PROPERTY_OTHER_ALPHABETIC;
  686. if (c == 0x0B4D) return UCD_PROPERTY_DIACRITIC;
  687. if (c == 0x0B56) return UCD_PROPERTY_OTHER_ALPHABETIC;
  688. if (c >= 0x0B62 && c <= 0x0B63) return UCD_PROPERTY_OTHER_ALPHABETIC;
  689. if (c == 0x0B82) return UCD_PROPERTY_OTHER_ALPHABETIC;
  690. if (c == 0x0BC0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  691. if (c == 0x0BCD) return UCD_PROPERTY_DIACRITIC;
  692. break;
  693. case 0x0C00:
  694. if (c == 0x0C00) return UCD_PROPERTY_OTHER_ALPHABETIC;
  695. if (c >= 0x0C3E && c <= 0x0C40) return UCD_PROPERTY_OTHER_ALPHABETIC;
  696. if (c >= 0x0C46 && c <= 0x0C48) return UCD_PROPERTY_OTHER_ALPHABETIC;
  697. if (c >= 0x0C4A && c <= 0x0C4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  698. if (c == 0x0C4D) return UCD_PROPERTY_DIACRITIC;
  699. if (c >= 0x0C55 && c <= 0x0C56) return UCD_PROPERTY_OTHER_ALPHABETIC;
  700. if (c >= 0x0C62 && c <= 0x0C63) return UCD_PROPERTY_OTHER_ALPHABETIC;
  701. if (c == 0x0C81) return UCD_PROPERTY_OTHER_ALPHABETIC;
  702. if (c == 0x0CBC) return UCD_PROPERTY_DIACRITIC;
  703. if (c == 0x0CBF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  704. if (c == 0x0CC6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  705. if (c == 0x0CCC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  706. if (c == 0x0CCD) return UCD_PROPERTY_DIACRITIC;
  707. if (c >= 0x0CE2 && c <= 0x0CE3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  708. break;
  709. case 0x0D00:
  710. if (c >= 0x0D00 && c <= 0x0D01) return UCD_PROPERTY_OTHER_ALPHABETIC;
  711. if (c >= 0x0D3B && c <= 0x0D3C) return UCD_PROPERTY_DIACRITIC;
  712. if (c >= 0x0D41 && c <= 0x0D44) return UCD_PROPERTY_OTHER_ALPHABETIC;
  713. if (c == 0x0D4D) return UCD_PROPERTY_DIACRITIC;
  714. if (c >= 0x0D62 && c <= 0x0D63) return UCD_PROPERTY_OTHER_ALPHABETIC;
  715. if (c == 0x0DCA) return UCD_PROPERTY_DIACRITIC;
  716. if (c >= 0x0DD2 && c <= 0x0DD4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  717. if (c == 0x0DD6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  718. break;
  719. case 0x0E00:
  720. if (c == 0x0E31) return UCD_PROPERTY_OTHER_ALPHABETIC;
  721. if (c >= 0x0E34 && c <= 0x0E3A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  722. if (c >= 0x0E47 && c <= 0x0E4C) return UCD_PROPERTY_DIACRITIC;
  723. if (c == 0x0E4D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  724. if (c == 0x0E4E) return UCD_PROPERTY_DIACRITIC;
  725. if (c == 0x0EB1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  726. if (c >= 0x0EB4 && c <= 0x0EB9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  727. if (c >= 0x0EBB && c <= 0x0EBC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  728. if (c >= 0x0EC8 && c <= 0x0ECC) return UCD_PROPERTY_DIACRITIC;
  729. if (c == 0x0ECD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  730. break;
  731. case 0x0F00:
  732. if (c >= 0x0F18 && c <= 0x0F19) return UCD_PROPERTY_DIACRITIC;
  733. if (c == 0x0F35) return UCD_PROPERTY_DIACRITIC;
  734. if (c == 0x0F37) return UCD_PROPERTY_DIACRITIC;
  735. if (c == 0x0F39) return UCD_PROPERTY_DIACRITIC;
  736. if (c == 0x0F77) return UCD_PROPERTY_DEPRECATED | UCD_PROPERTY_OTHER_ALPHABETIC;
  737. if (c == 0x0F79) return UCD_PROPERTY_DEPRECATED | UCD_PROPERTY_OTHER_ALPHABETIC;
  738. if (c >= 0x0F71 && c <= 0x0F7E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  739. if (c >= 0x0F80 && c <= 0x0F81) return UCD_PROPERTY_OTHER_ALPHABETIC;
  740. if (c >= 0x0F82 && c <= 0x0F84) return UCD_PROPERTY_DIACRITIC;
  741. if (c >= 0x0F86 && c <= 0x0F87) return UCD_PROPERTY_DIACRITIC;
  742. if (c >= 0x0F8D && c <= 0x0F97) return UCD_PROPERTY_OTHER_ALPHABETIC;
  743. if (c >= 0x0F99 && c <= 0x0FBC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  744. if (c == 0x0FC6) return UCD_PROPERTY_DIACRITIC;
  745. break;
  746. case 0x1000:
  747. if (c >= 0x102D && c <= 0x1030) return UCD_PROPERTY_OTHER_ALPHABETIC;
  748. if (c >= 0x1032 && c <= 0x1036) return UCD_PROPERTY_OTHER_ALPHABETIC;
  749. if (c == 0x1037) return UCD_PROPERTY_DIACRITIC;
  750. if (c >= 0x1039 && c <= 0x103A) return UCD_PROPERTY_DIACRITIC;
  751. if (c >= 0x103D && c <= 0x103E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  752. if (c >= 0x1058 && c <= 0x1059) return UCD_PROPERTY_OTHER_ALPHABETIC;
  753. if (c >= 0x105E && c <= 0x1060) return UCD_PROPERTY_OTHER_ALPHABETIC;
  754. if (c >= 0x1071 && c <= 0x1074) return UCD_PROPERTY_OTHER_ALPHABETIC;
  755. if (c == 0x1082) return UCD_PROPERTY_OTHER_ALPHABETIC;
  756. if (c >= 0x1085 && c <= 0x1086) return UCD_PROPERTY_OTHER_ALPHABETIC;
  757. if (c == 0x108D) return UCD_PROPERTY_DIACRITIC;
  758. if (c == 0x109D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  759. break;
  760. case 0x1300:
  761. if (c == 0x135F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  762. break;
  763. case 0x1700:
  764. if (c >= 0x1712 && c <= 0x1713) return UCD_PROPERTY_OTHER_ALPHABETIC;
  765. if (c >= 0x1732 && c <= 0x1733) return UCD_PROPERTY_OTHER_ALPHABETIC;
  766. if (c >= 0x1752 && c <= 0x1753) return UCD_PROPERTY_OTHER_ALPHABETIC;
  767. if (c >= 0x1772 && c <= 0x1773) return UCD_PROPERTY_OTHER_ALPHABETIC;
  768. if (c >= 0x17B4 && c <= 0x17B5) return UCD_PROPERTY_OTHER_DEFAULT_IGNORABLE_CODE_POINT;
  769. if (c >= 0x17B7 && c <= 0x17BD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  770. if (c == 0x17C6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  771. if (c >= 0x17C9 && c <= 0x17D3) return UCD_PROPERTY_DIACRITIC;
  772. if (c == 0x17DD) return UCD_PROPERTY_DIACRITIC;
  773. break;
  774. case 0x1800:
  775. if (c >= 0x180B && c <= 0x180D) return UCD_PROPERTY_VARIATION_SELECTOR;
  776. if (c >= 0x1885 && c <= 0x1886) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_ID_START;
  777. if (c == 0x18A9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  778. break;
  779. case 0x1900:
  780. if (c >= 0x1920 && c <= 0x1922) return UCD_PROPERTY_OTHER_ALPHABETIC;
  781. if (c >= 0x1927 && c <= 0x1928) return UCD_PROPERTY_OTHER_ALPHABETIC;
  782. if (c == 0x1932) return UCD_PROPERTY_OTHER_ALPHABETIC;
  783. if (c >= 0x1939 && c <= 0x193B) return UCD_PROPERTY_DIACRITIC;
  784. break;
  785. case 0x1A00:
  786. if (c >= 0x1A17 && c <= 0x1A18) return UCD_PROPERTY_OTHER_ALPHABETIC;
  787. if (c == 0x1A1B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  788. if (c == 0x1A56) return UCD_PROPERTY_OTHER_ALPHABETIC;
  789. if (c >= 0x1A58 && c <= 0x1A5E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  790. if (c == 0x1A62) return UCD_PROPERTY_OTHER_ALPHABETIC;
  791. if (c >= 0x1A65 && c <= 0x1A6C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  792. if (c >= 0x1A73 && c <= 0x1A74) return UCD_PROPERTY_OTHER_ALPHABETIC;
  793. if (c >= 0x1A75 && c <= 0x1A7C) return UCD_PROPERTY_DIACRITIC;
  794. if (c == 0x1A7F) return UCD_PROPERTY_DIACRITIC;
  795. if (c >= 0x1AB0 && c <= 0x1ABD) return UCD_PROPERTY_DIACRITIC;
  796. break;
  797. case 0x1B00:
  798. if (c >= 0x1B00 && c <= 0x1B03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  799. if (c == 0x1B34) return UCD_PROPERTY_DIACRITIC;
  800. if (c >= 0x1B36 && c <= 0x1B3A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  801. if (c == 0x1B3C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  802. if (c == 0x1B42) return UCD_PROPERTY_OTHER_ALPHABETIC;
  803. if (c >= 0x1B6B && c <= 0x1B73) return UCD_PROPERTY_DIACRITIC;
  804. if (c >= 0x1B80 && c <= 0x1B81) return UCD_PROPERTY_OTHER_ALPHABETIC;
  805. if (c >= 0x1BA2 && c <= 0x1BA5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  806. if (c >= 0x1BA8 && c <= 0x1BA9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  807. if (c == 0x1BAB) return UCD_PROPERTY_DIACRITIC;
  808. if (c >= 0x1BAC && c <= 0x1BAD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  809. if (c >= 0x1BE8 && c <= 0x1BE9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  810. if (c == 0x1BED) return UCD_PROPERTY_OTHER_ALPHABETIC;
  811. if (c >= 0x1BEF && c <= 0x1BF1) return UCD_PROPERTY_OTHER_ALPHABETIC;
  812. break;
  813. case 0x1C00:
  814. if (c >= 0x1C2C && c <= 0x1C33) return UCD_PROPERTY_OTHER_ALPHABETIC;
  815. if (c == 0x1C36) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  816. if (c == 0x1C37) return UCD_PROPERTY_DIACRITIC;
  817. if (c >= 0x1CD0 && c <= 0x1CD2) return UCD_PROPERTY_DIACRITIC;
  818. if (c >= 0x1CD4 && c <= 0x1CE0) return UCD_PROPERTY_DIACRITIC;
  819. if (c >= 0x1CE2 && c <= 0x1CE8) return UCD_PROPERTY_DIACRITIC;
  820. if (c == 0x1CED) return UCD_PROPERTY_DIACRITIC;
  821. if (c == 0x1CF4) return UCD_PROPERTY_DIACRITIC;
  822. if (c >= 0x1CF8 && c <= 0x1CF9) return UCD_PROPERTY_DIACRITIC;
  823. break;
  824. case 0x1D00:
  825. if (c >= 0x1DC4 && c <= 0x1DCF) return UCD_PROPERTY_DIACRITIC;
  826. if (c >= 0x1DE7 && c <= 0x1DF4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  827. if (c >= 0x1DF5 && c <= 0x1DF9) return UCD_PROPERTY_DIACRITIC;
  828. if (c >= 0x1DFD && c <= 0x1DFF) return UCD_PROPERTY_DIACRITIC;
  829. break;
  830. case 0x2000:
  831. if (c >= 0x20D0 && c <= 0x20DC) return UCD_PROPERTY_OTHER_MATH;
  832. if (c == 0x20E1) return UCD_PROPERTY_OTHER_MATH;
  833. if (c >= 0x20E5 && c <= 0x20E6) return UCD_PROPERTY_OTHER_MATH;
  834. if (c >= 0x20EB && c <= 0x20EF) return UCD_PROPERTY_OTHER_MATH;
  835. break;
  836. case 0x2C00:
  837. if (c >= 0x2CEF && c <= 0x2CF1) return UCD_PROPERTY_DIACRITIC;
  838. break;
  839. case 0x2D00:
  840. if (c >= 0x2DE0 && c <= 0x2DFF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  841. break;
  842. case 0x3000:
  843. if (c >= 0x302A && c <= 0x302D) return UCD_PROPERTY_DIACRITIC;
  844. if (c >= 0x3099 && c <= 0x309A) return UCD_PROPERTY_DIACRITIC;
  845. break;
  846. case 0xA600:
  847. if (c == 0xA66F) return UCD_PROPERTY_DIACRITIC;
  848. if (c >= 0xA674 && c <= 0xA67B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  849. if (c >= 0xA67C && c <= 0xA67D) return UCD_PROPERTY_DIACRITIC;
  850. if (c >= 0xA69E && c <= 0xA69F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  851. if (c >= 0xA6F0 && c <= 0xA6F1) return UCD_PROPERTY_DIACRITIC;
  852. break;
  853. case 0xA800:
  854. if (c >= 0xA825 && c <= 0xA826) return UCD_PROPERTY_OTHER_ALPHABETIC;
  855. if (c == 0xA8C4) return UCD_PROPERTY_DIACRITIC;
  856. if (c == 0xA8C5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  857. if (c >= 0xA8E0 && c <= 0xA8F1) return UCD_PROPERTY_DIACRITIC;
  858. break;
  859. case 0xA900:
  860. if (c >= 0xA926 && c <= 0xA92A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  861. if (c >= 0xA92B && c <= 0xA92D) return UCD_PROPERTY_DIACRITIC;
  862. if (c >= 0xA947 && c <= 0xA951) return UCD_PROPERTY_OTHER_ALPHABETIC;
  863. if (c >= 0xA980 && c <= 0xA982) return UCD_PROPERTY_OTHER_ALPHABETIC;
  864. if (c == 0xA9B3) return UCD_PROPERTY_DIACRITIC;
  865. if (c >= 0xA9B6 && c <= 0xA9B9) return UCD_PROPERTY_OTHER_ALPHABETIC;
  866. if (c == 0xA9BC) return UCD_PROPERTY_OTHER_ALPHABETIC;
  867. if (c == 0xA9E5) return UCD_PROPERTY_DIACRITIC;
  868. break;
  869. case 0xAA00:
  870. if (c >= 0xAA29 && c <= 0xAA2E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  871. if (c >= 0xAA31 && c <= 0xAA32) return UCD_PROPERTY_OTHER_ALPHABETIC;
  872. if (c >= 0xAA35 && c <= 0xAA36) return UCD_PROPERTY_OTHER_ALPHABETIC;
  873. if (c == 0xAA43) return UCD_PROPERTY_OTHER_ALPHABETIC;
  874. if (c == 0xAA4C) return UCD_PROPERTY_OTHER_ALPHABETIC;
  875. if (c == 0xAA7C) return UCD_PROPERTY_DIACRITIC;
  876. if (c == 0xAAB0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  877. if (c >= 0xAAB2 && c <= 0xAAB4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  878. if (c >= 0xAAB7 && c <= 0xAAB8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  879. if (c == 0xAABE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  880. if (c == 0xAABF) return UCD_PROPERTY_DIACRITIC;
  881. if (c == 0xAAC1) return UCD_PROPERTY_DIACRITIC;
  882. if (c >= 0xAAEC && c <= 0xAAED) return UCD_PROPERTY_OTHER_ALPHABETIC;
  883. if (c == 0xAAF6) return UCD_PROPERTY_DIACRITIC;
  884. break;
  885. case 0xAB00:
  886. if (c == 0xABE5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  887. if (c == 0xABE8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  888. if (c == 0xABED) return UCD_PROPERTY_DIACRITIC;
  889. break;
  890. case 0xFB00:
  891. if (c == 0xFB1E) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  892. break;
  893. case 0xFE00:
  894. if (c >= 0xFE00 && c <= 0xFE0F) return UCD_PROPERTY_VARIATION_SELECTOR;
  895. if (c >= 0xFE20 && c <= 0xFE2F) return UCD_PROPERTY_DIACRITIC;
  896. break;
  897. case 0x010200:
  898. if (c == 0x0102E0) return UCD_PROPERTY_DIACRITIC;
  899. break;
  900. case 0x010300:
  901. if (c >= 0x010376 && c <= 0x01037A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  902. break;
  903. case 0x010A00:
  904. if (c >= 0x010A01 && c <= 0x010A03) return UCD_PROPERTY_OTHER_ALPHABETIC;
  905. if (c >= 0x010A05 && c <= 0x010A06) return UCD_PROPERTY_OTHER_ALPHABETIC;
  906. if (c >= 0x010A0C && c <= 0x010A0F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  907. if (c >= 0x010AE5 && c <= 0x010AE6) return UCD_PROPERTY_DIACRITIC;
  908. break;
  909. case 0x010D00:
  910. if (c >= 0x010D24 && c <= 0x010D27) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ALPHABETIC;
  911. break;
  912. case 0x010F00:
  913. if (c >= 0x010F46 && c <= 0x010F50) return UCD_PROPERTY_DIACRITIC;
  914. break;
  915. case 0x011000:
  916. if (c == 0x011001) return UCD_PROPERTY_OTHER_ALPHABETIC;
  917. if (c >= 0x011038 && c <= 0x011045) return UCD_PROPERTY_OTHER_ALPHABETIC;
  918. if (c >= 0x0110B3 && c <= 0x0110B6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  919. if (c >= 0x0110B9 && c <= 0x0110BA) return UCD_PROPERTY_DIACRITIC;
  920. break;
  921. case 0x011100:
  922. if (c >= 0x011100 && c <= 0x011102) return UCD_PROPERTY_OTHER_ALPHABETIC;
  923. if (c >= 0x011127 && c <= 0x01112B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  924. if (c >= 0x01112D && c <= 0x011132) return UCD_PROPERTY_OTHER_ALPHABETIC;
  925. if (c >= 0x011133 && c <= 0x011134) return UCD_PROPERTY_DIACRITIC;
  926. if (c == 0x011173) return UCD_PROPERTY_DIACRITIC;
  927. if (c >= 0x011180 && c <= 0x011181) return UCD_PROPERTY_OTHER_ALPHABETIC;
  928. if (c >= 0x0111B6 && c <= 0x0111BE) return UCD_PROPERTY_OTHER_ALPHABETIC;
  929. if (c >= 0x0111CA && c <= 0x0111CC) return UCD_PROPERTY_DIACRITIC;
  930. break;
  931. case 0x011200:
  932. if (c >= 0x01122F && c <= 0x011231) return UCD_PROPERTY_OTHER_ALPHABETIC;
  933. if (c == 0x011234) return UCD_PROPERTY_OTHER_ALPHABETIC;
  934. if (c == 0x011236) return UCD_PROPERTY_DIACRITIC;
  935. if (c == 0x011237) return UCD_PROPERTY_OTHER_ALPHABETIC;
  936. if (c == 0x01123E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  937. if (c == 0x0112DF) return UCD_PROPERTY_OTHER_ALPHABETIC;
  938. if (c >= 0x0112E3 && c <= 0x0112E8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  939. if (c >= 0x0112E9 && c <= 0x0112EA) return UCD_PROPERTY_DIACRITIC;
  940. break;
  941. case 0x011300:
  942. if (c >= 0x011300 && c <= 0x011301) return UCD_PROPERTY_OTHER_ALPHABETIC;
  943. if (c == 0x01133C) return UCD_PROPERTY_DIACRITIC;
  944. if (c == 0x011340) return UCD_PROPERTY_OTHER_ALPHABETIC;
  945. if (c >= 0x011366 && c <= 0x01136C) return UCD_PROPERTY_DIACRITIC;
  946. if (c >= 0x011370 && c <= 0x011374) return UCD_PROPERTY_DIACRITIC;
  947. break;
  948. case 0x011400:
  949. if (c >= 0x011438 && c <= 0x01143F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  950. if (c == 0x011442) return UCD_PROPERTY_DIACRITIC;
  951. if (c >= 0x011443 && c <= 0x011444) return UCD_PROPERTY_OTHER_ALPHABETIC;
  952. if (c == 0x011446) return UCD_PROPERTY_DIACRITIC;
  953. if (c >= 0x0114B3 && c <= 0x0114B8) return UCD_PROPERTY_OTHER_ALPHABETIC;
  954. if (c == 0x0114BA) return UCD_PROPERTY_OTHER_ALPHABETIC;
  955. if (c >= 0x0114BF && c <= 0x0114C0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  956. if (c >= 0x0114C2 && c <= 0x0114C3) return UCD_PROPERTY_DIACRITIC;
  957. break;
  958. case 0x011500:
  959. if (c >= 0x0115B2 && c <= 0x0115B5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  960. if (c >= 0x0115BC && c <= 0x0115BD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  961. if (c >= 0x0115BF && c <= 0x0115C0) return UCD_PROPERTY_DIACRITIC;
  962. if (c >= 0x0115DC && c <= 0x0115DD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  963. break;
  964. case 0x011600:
  965. if (c >= 0x011633 && c <= 0x01163A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  966. if (c == 0x01163D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  967. if (c == 0x01163F) return UCD_PROPERTY_DIACRITIC;
  968. if (c == 0x011640) return UCD_PROPERTY_OTHER_ALPHABETIC;
  969. if (c == 0x0116AB) return UCD_PROPERTY_OTHER_ALPHABETIC;
  970. if (c == 0x0116AD) return UCD_PROPERTY_OTHER_ALPHABETIC;
  971. if (c >= 0x0116B0 && c <= 0x0116B5) return UCD_PROPERTY_OTHER_ALPHABETIC;
  972. if (c == 0x0116B7) return UCD_PROPERTY_DIACRITIC;
  973. break;
  974. case 0x011700:
  975. if (c >= 0x01171D && c <= 0x01171F) return UCD_PROPERTY_OTHER_ALPHABETIC;
  976. if (c >= 0x011722 && c <= 0x011725) return UCD_PROPERTY_OTHER_ALPHABETIC;
  977. if (c >= 0x011727 && c <= 0x01172A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  978. if (c == 0x01172B) return UCD_PROPERTY_DIACRITIC;
  979. break;
  980. case 0x011800:
  981. if (c >= 0x01182F && c <= 0x011838) return UCD_PROPERTY_OTHER_ALPHABETIC;
  982. if (c >= 0x011839 && c <= 0x01183A) return UCD_PROPERTY_DIACRITIC;
  983. break;
  984. case 0x011A00:
  985. if (c >= 0x011A01 && c <= 0x011A0A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  986. if (c == 0x011A34) return UCD_PROPERTY_DIACRITIC;
  987. if (c >= 0x011A35 && c <= 0x011A3E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  988. if (c == 0x011A47) return UCD_PROPERTY_DIACRITIC;
  989. if (c >= 0x011A51 && c <= 0x011A5B) return UCD_PROPERTY_OTHER_ALPHABETIC;
  990. if (c >= 0x011A8A && c <= 0x011A96) return UCD_PROPERTY_OTHER_ALPHABETIC;
  991. if (c == 0x011A98) return UCD_PROPERTY_EXTENDER;
  992. if (c == 0x011A99) return UCD_PROPERTY_DIACRITIC;
  993. break;
  994. case 0x011C00:
  995. if (c >= 0x011C30 && c <= 0x011C36) return UCD_PROPERTY_OTHER_ALPHABETIC;
  996. if (c >= 0x011C38 && c <= 0x011C3D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  997. if (c == 0x011C3F) return UCD_PROPERTY_DIACRITIC;
  998. if (c >= 0x011C92 && c <= 0x011CA7) return UCD_PROPERTY_OTHER_ALPHABETIC;
  999. if (c >= 0x011CAA && c <= 0x011CB0) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1000. if (c >= 0x011CB2 && c <= 0x011CB3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1001. if (c >= 0x011CB5 && c <= 0x011CB6) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1002. break;
  1003. case 0x011D00:
  1004. if (c >= 0x011D31 && c <= 0x011D36) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1005. if (c == 0x011D3A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1006. if (c >= 0x011D3C && c <= 0x011D3D) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1007. if (c >= 0x011D3F && c <= 0x011D41) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1008. if (c == 0x011D42) return UCD_PROPERTY_DIACRITIC;
  1009. if (c == 0x011D43) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1010. if (c >= 0x011D44 && c <= 0x011D45) return UCD_PROPERTY_DIACRITIC;
  1011. if (c == 0x011D47) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1012. if (c == 0x011D90) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1013. if (c == 0x011D91) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1014. if (c == 0x011D95) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1015. if (c == 0x011D97) return UCD_PROPERTY_DIACRITIC;
  1016. break;
  1017. case 0x011E00:
  1018. if (c == 0x011EF3) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1019. if (c == 0x011EF4) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1020. break;
  1021. case 0x016A00:
  1022. if (c >= 0x016AF0 && c <= 0x016AF4) return UCD_PROPERTY_DIACRITIC;
  1023. break;
  1024. case 0x016B00:
  1025. if (c >= 0x016B30 && c <= 0x016B36) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1026. break;
  1027. case 0x016F00:
  1028. if (c >= 0x016F8F && c <= 0x016F92) return UCD_PROPERTY_DIACRITIC;
  1029. break;
  1030. case 0x01BC00:
  1031. if (c == 0x01BC9E) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1032. break;
  1033. case 0x01D100:
  1034. if (c >= 0x01D167 && c <= 0x01D169) return UCD_PROPERTY_DIACRITIC;
  1035. if (c >= 0x01D17B && c <= 0x01D182) return UCD_PROPERTY_DIACRITIC;
  1036. if (c >= 0x01D185 && c <= 0x01D18B) return UCD_PROPERTY_DIACRITIC;
  1037. if (c >= 0x01D1AA && c <= 0x01D1AD) return UCD_PROPERTY_DIACRITIC;
  1038. break;
  1039. case 0x01E000:
  1040. if (c >= 0x01E000 && c <= 0x01E006) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1041. if (c >= 0x01E008 && c <= 0x01E018) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1042. if (c >= 0x01E01B && c <= 0x01E021) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1043. if (c >= 0x01E023 && c <= 0x01E024) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1044. if (c >= 0x01E026 && c <= 0x01E02A) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1045. break;
  1046. case 0x01E800:
  1047. if (c >= 0x01E8D0 && c <= 0x01E8D6) return UCD_PROPERTY_DIACRITIC;
  1048. break;
  1049. case 0x01E900:
  1050. if (c >= 0x01E944 && c <= 0x01E946) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER;
  1051. if (c == 0x01E947) return UCD_PROPERTY_OTHER_ALPHABETIC;
  1052. if (c >= 0x01E948 && c <= 0x01E94A) return UCD_PROPERTY_DIACRITIC;
  1053. break;
  1054. case 0x0E0100:
  1055. if (c >= 0x0E0100 && c <= 0x0E01EF) return UCD_PROPERTY_VARIATION_SELECTOR;
  1056. break;
  1057. }
  1058. return 0;
  1059. }
  1060. static ucd_property properties_Nd(codepoint_t c)
  1061. {
  1062. switch (c & 0xFFFFFF00)
  1063. {
  1064. case 0x0000:
  1065. if (c >= 0x0030 && c <= 0x0039) return UCD_PROPERTY_HEX_DIGIT | UCD_PROPERTY_ASCII_HEX_DIGIT | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_COMPONENT;
  1066. break;
  1067. case 0xFF00:
  1068. if (c >= 0xFF10 && c <= 0xFF19) return UCD_PROPERTY_HEX_DIGIT;
  1069. break;
  1070. case 0x01D700:
  1071. if (c >= 0x01D7CE && c <= 0x01D7FF) return UCD_PROPERTY_OTHER_MATH;
  1072. break;
  1073. }
  1074. return 0;
  1075. }
  1076. static ucd_property properties_Nl(codepoint_t c)
  1077. {
  1078. switch (c & 0xFFFFFF00)
  1079. {
  1080. case 0x2100:
  1081. if (c >= 0x2160 && c <= 0x216F) return UCD_PROPERTY_OTHER_UPPERCASE;
  1082. if (c >= 0x2170 && c <= 0x217F) return UCD_PROPERTY_OTHER_LOWERCASE;
  1083. break;
  1084. case 0x3000:
  1085. if (c == 0x3007) return UCD_PROPERTY_IDEOGRAPHIC;
  1086. if (c >= 0x3021 && c <= 0x3029) return UCD_PROPERTY_IDEOGRAPHIC;
  1087. if (c >= 0x3038 && c <= 0x303A) return UCD_PROPERTY_IDEOGRAPHIC;
  1088. break;
  1089. }
  1090. return 0;
  1091. }
  1092. static ucd_property properties_No(codepoint_t c)
  1093. {
  1094. switch (c & 0xFFFFFF00)
  1095. {
  1096. case 0x1300:
  1097. if (c >= 0x1369 && c <= 0x1371) return UCD_PROPERTY_OTHER_ID_CONTINUE;
  1098. break;
  1099. case 0x1900:
  1100. if (c == 0x19DA) return UCD_PROPERTY_OTHER_ID_CONTINUE;
  1101. break;
  1102. }
  1103. return 0;
  1104. }
  1105. static ucd_property properties_Pc(codepoint_t c)
  1106. {
  1107. switch (c & 0xFFFFFF00)
  1108. {
  1109. case 0x2000:
  1110. if (c == 0x2040) return UCD_PROPERTY_OTHER_MATH;
  1111. break;
  1112. }
  1113. return 0;
  1114. }
  1115. static ucd_property properties_Pd(codepoint_t c)
  1116. {
  1117. switch (c & 0xFFFFFF00)
  1118. {
  1119. case 0x0000:
  1120. return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN | UCD_PROPERTY_PATTERN_SYNTAX;
  1121. case 0x0500:
  1122. if (c == 0x058A) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1123. break;
  1124. case 0x1800:
  1125. if (c == 0x1806) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1126. break;
  1127. case 0x2000:
  1128. if (c >= 0x2010 && c <= 0x2011) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN | UCD_PROPERTY_PATTERN_SYNTAX;
  1129. return UCD_PROPERTY_DASH | UCD_PROPERTY_PATTERN_SYNTAX;
  1130. case 0x2E00:
  1131. if (c == 0x2E17) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN | UCD_PROPERTY_PATTERN_SYNTAX;
  1132. return UCD_PROPERTY_DASH | UCD_PROPERTY_PATTERN_SYNTAX;
  1133. case 0x3000:
  1134. if (c == 0x301C) return UCD_PROPERTY_DASH | UCD_PROPERTY_PATTERN_SYNTAX;
  1135. if (c == 0x3030) return UCD_PROPERTY_DASH | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1136. break;
  1137. case 0xFE00:
  1138. if (c == 0xFE63) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN | UCD_PROPERTY_OTHER_MATH;
  1139. break;
  1140. case 0xFF00:
  1141. if (c == 0xFF0D) return UCD_PROPERTY_DASH | UCD_PROPERTY_HYPHEN;
  1142. break;
  1143. }
  1144. return UCD_PROPERTY_DASH;
  1145. }
  1146. static ucd_property properties_Pe(codepoint_t c)
  1147. {
  1148. switch (c & 0xFFFFFF00)
  1149. {
  1150. case 0x0000:
  1151. return UCD_PROPERTY_PATTERN_SYNTAX;
  1152. case 0x2000:
  1153. if (c == 0x2046) return UCD_PROPERTY_PATTERN_SYNTAX;
  1154. if (c == 0x207E) return UCD_PROPERTY_OTHER_MATH;
  1155. if (c == 0x208E) return UCD_PROPERTY_OTHER_MATH;
  1156. break;
  1157. case 0x2300:
  1158. if (c == 0x2309) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1159. if (c == 0x230B) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1160. if (c == 0x232A) return UCD_PROPERTY_DEPRECATED | UCD_PROPERTY_PATTERN_SYNTAX;
  1161. break;
  1162. case 0x2700:
  1163. if (c == 0x27C6) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1164. if (c >= 0x27E6 && c <= 0x27EF) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX; /* Pe|Ps */
  1165. return UCD_PROPERTY_PATTERN_SYNTAX;
  1166. case 0x2900:
  1167. return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX; /* Pe|Ps */
  1168. case 0x2E00:
  1169. return UCD_PROPERTY_PATTERN_SYNTAX;
  1170. case 0x3000:
  1171. if (c == 0x300D) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_QUOTATION_MARK;
  1172. if (c == 0x300F) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_QUOTATION_MARK;
  1173. if (c >= 0x301E && c <= 0x301F) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_QUOTATION_MARK;
  1174. return UCD_PROPERTY_PATTERN_SYNTAX;
  1175. case 0xFD00:
  1176. if (c == 0xFD3E) return UCD_PROPERTY_PATTERN_SYNTAX;
  1177. break;
  1178. case 0xFE00:
  1179. if (c == 0xFE42) return UCD_PROPERTY_QUOTATION_MARK;
  1180. if (c == 0xFE44) return UCD_PROPERTY_QUOTATION_MARK;
  1181. break;
  1182. case 0xFF00:
  1183. if (c == 0xFF63) return UCD_PROPERTY_QUOTATION_MARK;
  1184. break;
  1185. }
  1186. return 0;
  1187. }
  1188. static ucd_property properties_Pf(codepoint_t c)
  1189. {
  1190. switch (c & 0xFFFFFF00)
  1191. {
  1192. case 0x0000:
  1193. case 0x2000:
  1194. return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
  1195. case 0x2E00:
  1196. return UCD_PROPERTY_PATTERN_SYNTAX;
  1197. }
  1198. return 0;
  1199. }
  1200. static ucd_property properties_Pi(codepoint_t c)
  1201. {
  1202. switch (c & 0xFFFFFF00)
  1203. {
  1204. case 0x0000:
  1205. case 0x2000:
  1206. return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
  1207. case 0x2E00:
  1208. return UCD_PROPERTY_PATTERN_SYNTAX;
  1209. }
  1210. return 0;
  1211. }
  1212. static ucd_property properties_Po(codepoint_t c)
  1213. {
  1214. switch (c & 0xFFFFFF00)
  1215. {
  1216. case 0x0000:
  1217. if (c == 0x0021) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX;
  1218. if (c == 0x0022) return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
  1219. if (c == 0x0023) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI_COMPONENT;
  1220. if (c == 0x0027) return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
  1221. if (c == 0x002A) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI_COMPONENT;
  1222. if (c == 0x002C) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX;
  1223. if (c == 0x002E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX;
  1224. if (c >= 0x003A && c <= 0x003B) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX;
  1225. if (c == 0x003F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX;
  1226. if (c == 0x00B7) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_EXTENDER | UCD_PROPERTY_OTHER_ID_CONTINUE;
  1227. return UCD_PROPERTY_PATTERN_SYNTAX;
  1228. case 0x0300:
  1229. if (c == 0x037E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1230. if (c == 0x0387) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_OTHER_ID_CONTINUE;
  1231. break;
  1232. case 0x0500:
  1233. if (c == 0x0589) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1234. if (c == 0x05C3) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1235. break;
  1236. case 0x0600:
  1237. if (c == 0x060C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1238. if (c == 0x061B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1239. if (c == 0x061E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1240. if (c == 0x061F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1241. if (c == 0x06D4) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1242. break;
  1243. case 0x0700:
  1244. if (c >= 0x0700 && c <= 0x0702) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1245. if (c >= 0x0703 && c <= 0x070A) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1246. if (c == 0x070C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1247. if (c == 0x07F8) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1248. if (c == 0x07F9) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1249. break;
  1250. case 0x0800:
  1251. if (c == 0x0837) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1252. if (c == 0x0839) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1253. if (c == 0x083D) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1254. if (c == 0x083E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1255. if (c >= 0x0830 && c <= 0x083E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1256. if (c == 0x085E) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1257. break;
  1258. case 0x0900:
  1259. if (c >= 0x0964 && c <= 0x0965) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1260. break;
  1261. case 0x0E00:
  1262. if (c >= 0x0E5A && c <= 0x0E5B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1263. break;
  1264. case 0x0F00:
  1265. if (c == 0x0F08) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1266. if (c >= 0x0F0D && c <= 0x0F12) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1267. break;
  1268. case 0x1000:
  1269. if (c >= 0x104A && c <= 0x104B) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1270. break;
  1271. case 0x1300:
  1272. if (c == 0x1361) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1273. if (c == 0x1362) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1274. if (c >= 0x1363 && c <= 0x1366) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1275. if (c >= 0x1367 && c <= 0x1368) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1276. break;
  1277. case 0x1600:
  1278. if (c == 0x166D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1279. if (c == 0x166E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1280. if (c >= 0x16EB && c <= 0x16ED) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1281. break;
  1282. case 0x1700:
  1283. if (c >= 0x1735 && c <= 0x1736) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1284. if (c >= 0x17D4 && c <= 0x17D6) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1285. if (c == 0x17DA) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1286. break;
  1287. case 0x1800:
  1288. if (c == 0x1803) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1289. if (c >= 0x1802 && c <= 0x1805) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1290. if (c == 0x1808) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1291. if (c == 0x1809) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1292. if (c == 0x180A) return UCD_PROPERTY_EXTENDER;
  1293. break;
  1294. case 0x1900:
  1295. if (c >= 0x1944 && c <= 0x1945) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1296. break;
  1297. case 0x1A00:
  1298. if (c >= 0x1AA8 && c <= 0x1AAB) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1299. break;
  1300. case 0x1B00:
  1301. if (c >= 0x1B5A && c <= 0x1B5B) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1302. if (c == 0x1B5D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1303. if (c >= 0x1B5E && c <= 0x1B5F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1304. break;
  1305. case 0x1C00:
  1306. if (c >= 0x1C3B && c <= 0x1C3C) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1307. if (c >= 0x1C3D && c <= 0x1C3F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1308. if (c >= 0x1C7E && c <= 0x1C7F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1309. if (c == 0x1CD3) return UCD_PROPERTY_DIACRITIC;
  1310. break;
  1311. case 0x2000:
  1312. if (c == 0x2016) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1313. if (c == 0x2017) return UCD_PROPERTY_PATTERN_SYNTAX;
  1314. if (c >= 0x2020 && c <= 0x2027) return UCD_PROPERTY_PATTERN_SYNTAX;
  1315. if (c >= 0x2032 && c <= 0x2034) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_OTHER_MATH;
  1316. if (c >= 0x2030 && c <= 0x2038) return UCD_PROPERTY_PATTERN_SYNTAX;
  1317. if (c == 0x203C) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1318. if (c == 0x203D) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX;
  1319. if (c >= 0x203B && c <= 0x203E) return UCD_PROPERTY_PATTERN_SYNTAX;
  1320. if (c >= 0x2041 && c <= 0x2043) return UCD_PROPERTY_PATTERN_SYNTAX;
  1321. if (c >= 0x2047 && c <= 0x2048) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX;
  1322. if (c == 0x2049) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1323. if (c >= 0x204A && c <= 0x2051) return UCD_PROPERTY_PATTERN_SYNTAX;
  1324. if (c == 0x2053) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_DASH;
  1325. if (c >= 0x2055 && c <= 0x205E) return UCD_PROPERTY_PATTERN_SYNTAX;
  1326. break;
  1327. case 0x2E00:
  1328. if (c == 0x2E2E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX;
  1329. if (c == 0x2E3C) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX;
  1330. if (c == 0x2E41) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX;
  1331. if (c == 0x2E4C) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX;
  1332. if (c == 0x2E4E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX;
  1333. return UCD_PROPERTY_PATTERN_SYNTAX;
  1334. case 0x3000:
  1335. if (c == 0x3001) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_PATTERN_SYNTAX;
  1336. if (c == 0x3002) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL | UCD_PROPERTY_PATTERN_SYNTAX;
  1337. if (c == 0x3003) return UCD_PROPERTY_PATTERN_SYNTAX;
  1338. if (c == 0x303D) return UCD_PROPERTY_EMOJI;
  1339. if (c == 0x30FB) return UCD_PROPERTY_HYPHEN;
  1340. break;
  1341. case 0xA400:
  1342. if (c == 0xA4FE) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1343. if (c == 0xA4FF) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1344. break;
  1345. case 0xA600:
  1346. if (c == 0xA60D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1347. if (c >= 0xA60E && c <= 0xA60F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1348. if (c == 0xA6F3) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1349. if (c >= 0xA6F4 && c <= 0xA6F6) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1350. if (c == 0xA6F7) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1351. break;
  1352. case 0xA800:
  1353. if (c >= 0xA876 && c <= 0xA877) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1354. if (c >= 0xA8CE && c <= 0xA8CF) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1355. break;
  1356. case 0xA900:
  1357. if (c == 0xA92E) return UCD_PROPERTY_DIACRITIC;
  1358. if (c == 0xA92F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1359. if (c == 0xA9C7) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1360. if (c >= 0xA9C8 && c <= 0xA9C9) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1361. break;
  1362. case 0xAA00:
  1363. if (c >= 0xAA5D && c <= 0xAA5F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1364. if (c == 0xAADF) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1365. if (c >= 0xAAF0 && c <= 0xAAF1) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1366. break;
  1367. case 0xAB00:
  1368. if (c == 0xABEB) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1369. break;
  1370. case 0xFE00:
  1371. if (c >= 0xFE45 && c <= 0xFE46) return UCD_PROPERTY_PATTERN_SYNTAX;
  1372. if (c >= 0xFE50 && c <= 0xFE51) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1373. if (c == 0xFE52) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1374. if (c >= 0xFE54 && c <= 0xFE55) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1375. if (c >= 0xFE56 && c <= 0xFE57) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1376. if (c == 0xFE61) return UCD_PROPERTY_OTHER_MATH;
  1377. if (c == 0xFE68) return UCD_PROPERTY_OTHER_MATH;
  1378. break;
  1379. case 0xFF00:
  1380. if (c == 0xFF01) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1381. if (c == 0xFF02) return UCD_PROPERTY_QUOTATION_MARK;
  1382. if (c == 0xFF07) return UCD_PROPERTY_QUOTATION_MARK;
  1383. if (c == 0xFF0C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1384. if (c == 0xFF0E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1385. if (c == 0xFF3C) return UCD_PROPERTY_OTHER_MATH;
  1386. if (c == 0xFF65) return UCD_PROPERTY_HYPHEN;
  1387. if (c >= 0xFF1A && c <= 0xFF1B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1388. if (c == 0xFF1F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1389. if (c == 0xFF61) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1390. if (c == 0xFF64) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1391. break;
  1392. case 0x10300:
  1393. if (c == 0x01039F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1394. if (c == 0x0103D0) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1395. break;
  1396. case 0x10800:
  1397. if (c == 0x010857) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1398. break;
  1399. case 0x10900:
  1400. if (c == 0x01091F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1401. break;
  1402. case 0x10A00:
  1403. if (c >= 0x010A56 && c <= 0x010A57) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1404. if (c >= 0x010AF0 && c <= 0x010AF5) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1405. break;
  1406. case 0x10B00:
  1407. if (c >= 0x010B3A && c <= 0x010B3F) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1408. if (c >= 0x010B99 && c <= 0x010B9C) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1409. break;
  1410. case 0x10F00:
  1411. if (c >= 0x10F55 && c <= 0x10F59) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1412. break;
  1413. case 0x11000:
  1414. if (c >= 0x011047 && c <= 0x011048) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1415. if (c >= 0x011049 && c <= 0x01104D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1416. if (c >= 0x0110BE && c <= 0x0110C1) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1417. break;
  1418. case 0x11100:
  1419. if (c >= 0x011141 && c <= 0x011143) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1420. if (c >= 0x0111C5 && c <= 0x0111C6) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1421. if (c == 0x0111CD) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1422. if (c >= 0x0111DE && c <= 0x0111DF) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1423. break;
  1424. case 0x11200:
  1425. if (c == 0x01123A) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1426. if (c >= 0x011238 && c <= 0x01123C) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1427. if (c == 0x0112A9) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1428. break;
  1429. case 0x11400:
  1430. if (c >= 0x01144B && c <= 0x01144C) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1431. if (c == 0x01144D) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1432. if (c == 0x01145B) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1433. break;
  1434. case 0x11500:
  1435. if (c >= 0x0115C2 && c <= 0x0115C3) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1436. if (c >= 0x0115C4 && c <= 0x0115C5) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1437. if (c >= 0x0115C6 && c <= 0x0115C8) return UCD_PROPERTY_EXTENDER;
  1438. if (c >= 0x0115C9 && c <= 0x0115D7) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1439. break;
  1440. case 0x11600:
  1441. if (c >= 0x011641 && c <= 0x011642) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1442. break;
  1443. case 0x11700:
  1444. if (c >= 0x01173C && c <= 0x01173E) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1445. break;
  1446. case 0x11A00:
  1447. if (c >= 0x011A42 && c <= 0x011A43) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1448. if (c >= 0x011A9B && c <= 0x011A9C) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1449. if (c >= 0x011AA1 && c <= 0x011AA2) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1450. break;
  1451. case 0x11C00:
  1452. if (c >= 0x011C41 && c <= 0x011C42) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1453. if (c == 0x011C43) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1454. if (c == 0x011C71) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1455. break;
  1456. case 0x11E00:
  1457. if (c >= 0x11EF7 && c <= 0x11EF8) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1458. break;
  1459. case 0x12400:
  1460. if (c >= 0x012470 && c <= 0x012474) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1461. break;
  1462. case 0x16E00:
  1463. if (c == 0x016E97) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1464. if (c == 0x016E98) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1465. break;
  1466. case 0x16A00:
  1467. if (c >= 0x016A6E && c <= 0x016A6F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1468. if (c == 0x016AF5) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1469. break;
  1470. case 0x16B00:
  1471. if (c >= 0x016B37 && c <= 0x016B38) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1472. if (c == 0x016B39) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1473. if (c == 0x016B44) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1474. break;
  1475. case 0x1BC00:
  1476. if (c == 0x01BC9F) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1477. break;
  1478. case 0x1DA00:
  1479. if (c == 0x01DA88) return UCD_PROPERTY_TERMINAL_PUNCTUATION | UCD_PROPERTY_SENTENCE_TERMINAL;
  1480. if (c >= 0x01DA87 && c <= 0x01DA8A) return UCD_PROPERTY_TERMINAL_PUNCTUATION;
  1481. break;
  1482. }
  1483. return 0;
  1484. }
  1485. static ucd_property properties_Ps(codepoint_t c)
  1486. {
  1487. switch (c & 0xFFFFFF00)
  1488. {
  1489. case 0x0000:
  1490. return UCD_PROPERTY_PATTERN_SYNTAX;
  1491. case 0x2000:
  1492. if (c == 0x201A) return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
  1493. if (c == 0x201E) return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
  1494. if (c == 0x2045) return UCD_PROPERTY_PATTERN_SYNTAX;
  1495. if (c == 0x207D) return UCD_PROPERTY_OTHER_MATH;
  1496. if (c == 0x208D) return UCD_PROPERTY_OTHER_MATH;
  1497. break;
  1498. case 0x2300:
  1499. if (c == 0x2308) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1500. if (c == 0x230A) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1501. if (c == 0x2329) return UCD_PROPERTY_DEPRECATED | UCD_PROPERTY_PATTERN_SYNTAX;
  1502. break;
  1503. case 0x2700:
  1504. if (c == 0x27C5) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1505. if (c >= 0x27E6 && c <= 0x27EF) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX; /* Pe|Ps */
  1506. return UCD_PROPERTY_PATTERN_SYNTAX;
  1507. case 0x2900:
  1508. return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1509. case 0x2E00:
  1510. if (c == 0x2E42) return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
  1511. return UCD_PROPERTY_PATTERN_SYNTAX;
  1512. case 0x3000:
  1513. if (c == 0x300C) return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
  1514. if (c == 0x300E) return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
  1515. if (c == 0x301D) return UCD_PROPERTY_QUOTATION_MARK | UCD_PROPERTY_PATTERN_SYNTAX;
  1516. return UCD_PROPERTY_PATTERN_SYNTAX;
  1517. case 0xFD00:
  1518. if (c == 0xFD3F) return UCD_PROPERTY_PATTERN_SYNTAX;
  1519. break;
  1520. case 0xFE00:
  1521. if (c == 0xFE41) return UCD_PROPERTY_QUOTATION_MARK;
  1522. if (c == 0xFE43) return UCD_PROPERTY_QUOTATION_MARK;
  1523. break;
  1524. case 0xFF00:
  1525. if (c == 0xFF62) return UCD_PROPERTY_QUOTATION_MARK;
  1526. break;
  1527. }
  1528. return 0;
  1529. }
  1530. static ucd_property properties_Sc(codepoint_t c)
  1531. {
  1532. switch (c & 0xFFFFFF00)
  1533. {
  1534. case 0x0000:
  1535. return UCD_PROPERTY_PATTERN_SYNTAX;
  1536. }
  1537. return 0;
  1538. }
  1539. static ucd_property properties_Sk(codepoint_t c)
  1540. {
  1541. switch (c & 0xFFFFFF00)
  1542. {
  1543. case 0x0000:
  1544. if (c == 0x005E) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_PATTERN_SYNTAX;
  1545. if (c == 0x0060) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_PATTERN_SYNTAX;
  1546. if (c == 0x00A8) return UCD_PROPERTY_DIACRITIC;
  1547. if (c == 0x00AF) return UCD_PROPERTY_DIACRITIC;
  1548. if (c == 0x00B4) return UCD_PROPERTY_DIACRITIC;
  1549. if (c == 0x00B8) return UCD_PROPERTY_DIACRITIC;
  1550. break;
  1551. case 0x0200:
  1552. if (c >= 0x02C2 && c <= 0x02C5) return UCD_PROPERTY_DIACRITIC;
  1553. if (c >= 0x02D2 && c <= 0x02DF) return UCD_PROPERTY_DIACRITIC;
  1554. if (c >= 0x02E5 && c <= 0x02EB) return UCD_PROPERTY_DIACRITIC;
  1555. if (c == 0x02ED) return UCD_PROPERTY_DIACRITIC;
  1556. if (c >= 0x02EF && c <= 0x02FF) return UCD_PROPERTY_DIACRITIC;
  1557. break;
  1558. case 0x0300:
  1559. if (c == 0x0375) return UCD_PROPERTY_DIACRITIC;
  1560. if (c >= 0x0384 && c <= 0x0385) return UCD_PROPERTY_DIACRITIC;
  1561. break;
  1562. case 0x1F00:
  1563. if (c == 0x1FBD) return UCD_PROPERTY_DIACRITIC;
  1564. if (c >= 0x1FBF && c <= 0x1FC1) return UCD_PROPERTY_DIACRITIC;
  1565. if (c >= 0x1FCD && c <= 0x1FCF) return UCD_PROPERTY_DIACRITIC;
  1566. if (c >= 0x1FDD && c <= 0x1FDF) return UCD_PROPERTY_DIACRITIC;
  1567. if (c >= 0x1FED && c <= 0x1FEF) return UCD_PROPERTY_DIACRITIC;
  1568. if (c >= 0x1FFD && c <= 0x1FFE) return UCD_PROPERTY_DIACRITIC;
  1569. break;
  1570. case 0x3000:
  1571. if (c >= 0x309B && c <= 0x309C) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_ID_START;
  1572. break;
  1573. case 0xA700:
  1574. if (c >= 0xA720 && c <= 0xA721) return UCD_PROPERTY_DIACRITIC;
  1575. break;
  1576. case 0xAB00:
  1577. if (c == 0xAB5B) return UCD_PROPERTY_DIACRITIC;
  1578. break;
  1579. case 0xFF00:
  1580. if (c == 0xFF3E) return UCD_PROPERTY_DIACRITIC | UCD_PROPERTY_OTHER_MATH;
  1581. if (c == 0xFF40) return UCD_PROPERTY_DIACRITIC;
  1582. if (c == 0xFFE3) return UCD_PROPERTY_DIACRITIC;
  1583. break;
  1584. case 0x01F300:
  1585. return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER | UCD_PROPERTY_EMOJI_COMPONENT;
  1586. }
  1587. return 0;
  1588. }
  1589. static ucd_property properties_Sm(codepoint_t c)
  1590. {
  1591. switch (c & 0xFFFFFF00)
  1592. {
  1593. case 0x0000:
  1594. return UCD_PROPERTY_PATTERN_SYNTAX;
  1595. case 0x2000:
  1596. if (c == 0x2044) return UCD_PROPERTY_PATTERN_SYNTAX;
  1597. if (c == 0x2052) return UCD_PROPERTY_PATTERN_SYNTAX;
  1598. if (c == 0x207B) return UCD_PROPERTY_DASH;
  1599. if (c == 0x208B) return UCD_PROPERTY_DASH;
  1600. break;
  1601. case 0x2100:
  1602. if (c == 0x2118) return UCD_PROPERTY_OTHER_ID_START;
  1603. if (c == 0x2194) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1604. if (c >= 0x2190) return UCD_PROPERTY_PATTERN_SYNTAX;
  1605. break;
  1606. case 0x2200:
  1607. if (c == 0x2212) return UCD_PROPERTY_DASH | UCD_PROPERTY_PATTERN_SYNTAX;
  1608. return UCD_PROPERTY_PATTERN_SYNTAX;
  1609. case 0x2300:
  1610. return UCD_PROPERTY_PATTERN_SYNTAX;
  1611. case 0x2500:
  1612. if (c >= 0x25FB && c <= 0x25FC) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1613. if (c >= 0x25FD && c <= 0x25FE) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1614. return UCD_PROPERTY_PATTERN_SYNTAX;
  1615. case 0x2600:
  1616. case 0x2700:
  1617. return UCD_PROPERTY_PATTERN_SYNTAX;
  1618. case 0x2900:
  1619. if (c >= 0x2934 && c <= 0x2935) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1620. return UCD_PROPERTY_PATTERN_SYNTAX;
  1621. case 0x2A00:
  1622. case 0x2B00:
  1623. return UCD_PROPERTY_PATTERN_SYNTAX;
  1624. }
  1625. return 0;
  1626. }
  1627. static ucd_property properties_So_002600(codepoint_t c)
  1628. {
  1629. switch (c & 0xFFFFFFF0)
  1630. {
  1631. case 0x2600:
  1632. if (c <= 0x2604) return UCD_PROPERTY_EMOJI;
  1633. if (c <= 0x2606) return UCD_PROPERTY_OTHER_MATH;
  1634. if (c == 0x260E) return UCD_PROPERTY_EMOJI;
  1635. break;
  1636. case 0x2610:
  1637. if (c == 0x2611) return UCD_PROPERTY_EMOJI;
  1638. if (c <= 0x2613) return 0;
  1639. if (c <= 0x2615) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1640. if (c == 0x2618) return UCD_PROPERTY_EMOJI;
  1641. if (c == 0x261D) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1642. break;
  1643. case 0x2620:
  1644. if (c == 0x2620) return UCD_PROPERTY_EMOJI;
  1645. if (c >= 0x2622 && c <= 0x2623) return UCD_PROPERTY_EMOJI;
  1646. if (c == 0x2626) return UCD_PROPERTY_EMOJI;
  1647. if (c == 0x262A) return UCD_PROPERTY_EMOJI;
  1648. if (c >= 0x262E) return UCD_PROPERTY_EMOJI;
  1649. break;
  1650. case 0x2630:
  1651. if (c >= 0x2638 && c <= 0x263A) return UCD_PROPERTY_EMOJI;
  1652. break;
  1653. case 0x2640:
  1654. if (c == 0x2640) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_EMOJI;
  1655. if (c == 0x2642) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_EMOJI;
  1656. if (c >= 0x2648) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1657. break;
  1658. case 0x2650:
  1659. if (c <= 0x2653) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1660. break;
  1661. case 0x2660:
  1662. if (c == 0x2606) return UCD_PROPERTY_OTHER_MATH;
  1663. if (c == 0x2660) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_EMOJI;
  1664. if (c >= 0x2661 && c <= 0x2662) return UCD_PROPERTY_OTHER_MATH;
  1665. if (c == 0x2663) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_EMOJI;
  1666. if (c >= 0x2665 && c <= 0x2666) return UCD_PROPERTY_EMOJI;
  1667. if (c == 0x2668) return UCD_PROPERTY_EMOJI;
  1668. if (c >= 0x266D && c <= 0x266E) return UCD_PROPERTY_OTHER_MATH;
  1669. break;
  1670. case 0x2670:
  1671. if (c == 0x267B) return UCD_PROPERTY_EMOJI;
  1672. if (c == 0x267F) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1673. break;
  1674. case 0x2690:
  1675. if (c == 0x2693) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1676. if (c >= 0x2692 && c <= 0x2697) return UCD_PROPERTY_EMOJI;
  1677. if (c == 0x2699) return UCD_PROPERTY_EMOJI;
  1678. if (c >= 0x269B && c <= 0x269C) return UCD_PROPERTY_EMOJI;
  1679. break;
  1680. case 0x26A0:
  1681. if (c == 0x26A0) return UCD_PROPERTY_EMOJI;
  1682. if (c == 0x26A1) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1683. if (c >= 0x26AA && c <= 0x26AB) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1684. break;
  1685. case 0x26B0:
  1686. if (c >= 0x26B0 && c <= 0x26B1) return UCD_PROPERTY_EMOJI;
  1687. if (c >= 0x26BD && c <= 0x26BE) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1688. break;
  1689. case 0x26C0:
  1690. if (c >= 0x26C4 && c <= 0x26C5) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1691. if (c == 0x26C8) return UCD_PROPERTY_EMOJI;
  1692. if (c == 0x26CE) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1693. if (c == 0x26CF) return UCD_PROPERTY_EMOJI;
  1694. break;
  1695. case 0x26D0:
  1696. if (c == 0x26D1) return UCD_PROPERTY_EMOJI;
  1697. if (c == 0x26D3) return UCD_PROPERTY_EMOJI;
  1698. if (c == 0x26D4) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1699. break;
  1700. case 0x26E0:
  1701. if (c == 0x26E9) return UCD_PROPERTY_EMOJI;
  1702. if (c == 0x26EA) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1703. break;
  1704. case 0x26F0:
  1705. if (c <= 0x26F1) return UCD_PROPERTY_EMOJI;
  1706. if (c == 0x26F4) return UCD_PROPERTY_EMOJI;
  1707. if (c <= 0x26F5) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1708. if (c >= 0x26F7 && c <= 0x26F8) return UCD_PROPERTY_EMOJI;
  1709. if (c == 0x26F9) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1710. if (c == 0x26FA) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1711. if (c == 0x26FD) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1712. break;
  1713. }
  1714. return 0;
  1715. }
  1716. static ucd_property properties_So_002700(codepoint_t c)
  1717. {
  1718. switch (c & 0xFFFFFFF0)
  1719. {
  1720. case 0x2700:
  1721. if (c == 0x2702) return UCD_PROPERTY_EMOJI;
  1722. if (c == 0x2705) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1723. if (c >= 0x270A && c <= 0x270B) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1724. if (c >= 0x270C && c <= 0x270D) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1725. if (c >= 0x2708 && c <= 0x270D) return UCD_PROPERTY_EMOJI;
  1726. if (c == 0x270F) return UCD_PROPERTY_EMOJI;
  1727. break;
  1728. case 0x2710:
  1729. if (c == 0x2712) return UCD_PROPERTY_EMOJI;
  1730. if (c == 0x2714) return UCD_PROPERTY_EMOJI;
  1731. if (c == 0x2716) return UCD_PROPERTY_EMOJI;
  1732. if (c == 0x271D) return UCD_PROPERTY_EMOJI;
  1733. break;
  1734. case 0x2720:
  1735. if (c == 0x2721) return UCD_PROPERTY_EMOJI;
  1736. if (c == 0x2728) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1737. break;
  1738. case 0x2730:
  1739. if (c == 0x2733) return UCD_PROPERTY_EMOJI;
  1740. if (c == 0x2734) return UCD_PROPERTY_EMOJI;
  1741. break;
  1742. case 0x2740:
  1743. if (c == 0x2744) return UCD_PROPERTY_EMOJI;
  1744. if (c == 0x2747) return UCD_PROPERTY_EMOJI;
  1745. if (c == 0x274C) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1746. if (c == 0x274E) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1747. break;
  1748. case 0x2750:
  1749. if (c >= 0x2753 && c <= 0x2755) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1750. if (c == 0x2757) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1751. break;
  1752. case 0x2760:
  1753. if (c >= 0x2763 && c <= 0x2764) return UCD_PROPERTY_EMOJI;
  1754. break;
  1755. case 0x2790:
  1756. if (c >= 0x2795 && c <= 0x2797) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1757. break;
  1758. case 0x27A0:
  1759. if (c == 0x27A1) return UCD_PROPERTY_EMOJI;
  1760. break;
  1761. case 0x27B0:
  1762. if (c == 0x27B0) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1763. if (c == 0x27BF) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1764. break;
  1765. }
  1766. return 0;
  1767. }
  1768. static ucd_property properties_So(codepoint_t c)
  1769. {
  1770. switch (c & 0xFFFFFF00)
  1771. {
  1772. case 0x0000:
  1773. if (c == 0x00A9) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1774. if (c == 0x00AE) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1775. return UCD_PROPERTY_PATTERN_SYNTAX;
  1776. case 0x2100:
  1777. if (c == 0x2122) return UCD_PROPERTY_EMOJI;
  1778. if (c == 0x2129) return UCD_PROPERTY_OTHER_MATH;
  1779. if (c == 0x212E) return UCD_PROPERTY_OTHER_ID_START;
  1780. if (c == 0x21A8) return UCD_PROPERTY_PATTERN_SYNTAX;
  1781. if (c == 0x21A9) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1782. if (c == 0x21AA) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1783. if (c >= 0x2195 && c <= 0x2199) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1784. if (c >= 0x219C && c <= 0x21AD) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1785. if (c >= 0x21B0 && c <= 0x21B1) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1786. if (c >= 0x21B6 && c <= 0x21B7) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1787. if (c >= 0x21AF && c <= 0x21BB) return UCD_PROPERTY_PATTERN_SYNTAX;
  1788. if (c >= 0x21BC && c <= 0x21CD) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1789. if (c >= 0x21D0 && c <= 0x21D1) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1790. if (c == 0x21D3) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1791. if (c >= 0x21D5 && c <= 0x21DB) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1792. if (c == 0x21DD) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1793. if (c >= 0x21E4 && c <= 0x21E5) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1794. if (c >= 0x21D5 && c <= 0x21F3) return UCD_PROPERTY_PATTERN_SYNTAX;
  1795. break;
  1796. case 0x2300:
  1797. if (c <= 0x2307) return UCD_PROPERTY_PATTERN_SYNTAX;
  1798. if (c >= 0x231A && c <= 0x231B) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1799. if (c >= 0x230C && c <= 0x231F) return UCD_PROPERTY_PATTERN_SYNTAX;
  1800. if (c >= 0x2322 && c <= 0x2327) return UCD_PROPERTY_PATTERN_SYNTAX;
  1801. if (c == 0x2328) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1802. if (c >= 0x232B && c <= 0x237B) return UCD_PROPERTY_PATTERN_SYNTAX;
  1803. if (c >= 0x237D && c <= 0x239A) return UCD_PROPERTY_PATTERN_SYNTAX;
  1804. if (c >= 0x23B4 && c <= 0x23B5) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1805. if (c == 0x23B7) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1806. if (c == 0x23CF) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1807. if (c == 0x23D0) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1808. if (c >= 0x23B4 && c <= 0x23DB) return UCD_PROPERTY_PATTERN_SYNTAX;
  1809. if (c == 0x23E2) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1810. if (c >= 0x23E9 && c <= 0x23EC) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1811. if (c == 0x23F0) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1812. if (c == 0x23F3) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1813. if (c >= 0x23E9 && c <= 0x23F3) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1814. if (c >= 0x23F8 && c <= 0x23FA) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1815. if (c >= 0x23E3) return UCD_PROPERTY_PATTERN_SYNTAX;
  1816. break;
  1817. case 0x2400:
  1818. if (c >= 0x2400 && c <= 0x244A) return UCD_PROPERTY_PATTERN_SYNTAX;
  1819. if (c == 0x24C2) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE | UCD_PROPERTY_EMOJI;
  1820. if (c >= 0x24B6 && c <= 0x24CF) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE;
  1821. if (c >= 0x24D0 && c <= 0x24E9) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_LOWERCASE;
  1822. break;
  1823. case 0x2500:
  1824. if (c >= 0x25A0 && c <= 0x25A1) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1825. if (c >= 0x25AA && c <= 0x25AB) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1826. if (c >= 0x25AE && c <= 0x25B5) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1827. if (c == 0x25B6) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1828. if (c >= 0x25BC && c <= 0x25BF) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1829. if (c == 0x25C0) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1830. if (c >= 0x25C6 && c <= 0x25C7) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1831. if (c >= 0x25CA && c <= 0x25CB) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1832. if (c >= 0x25CF && c <= 0x25D3) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1833. if (c == 0x25E2) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1834. if (c == 0x25E4) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1835. if (c >= 0x25E7 && c <= 0x25EC) return UCD_PROPERTY_OTHER_MATH | UCD_PROPERTY_PATTERN_SYNTAX;
  1836. return UCD_PROPERTY_PATTERN_SYNTAX;
  1837. case 0x2600:
  1838. return properties_So_002600(c) | UCD_PROPERTY_PATTERN_SYNTAX;
  1839. case 0x2700:
  1840. return properties_So_002700(c) | UCD_PROPERTY_PATTERN_SYNTAX;
  1841. case 0x2800:
  1842. return UCD_PROPERTY_PATTERN_SYNTAX;
  1843. case 0x2B00:
  1844. if (c >= 0x2B05 && c <= 0x2B07) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI;
  1845. if (c >= 0x2B1B && c <= 0x2B1C) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1846. if (c == 0x2B50) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1847. if (c == 0x2B55) return UCD_PROPERTY_PATTERN_SYNTAX | UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1848. return UCD_PROPERTY_PATTERN_SYNTAX;
  1849. case 0x2E00:
  1850. if (c >= 0x2E80 && c <= 0x2E99) return UCD_PROPERTY_RADICAL;
  1851. if (c >= 0x2E9B && c <= 0x2EF3) return UCD_PROPERTY_RADICAL;
  1852. break;
  1853. case 0x2F00:
  1854. if (c <= 0x2FD5) return UCD_PROPERTY_RADICAL;
  1855. if (c >= 0x2FF0 && c <= 0x2FF1) return UCD_PROPERTY_IDS_BINARY_OPERATOR;
  1856. if (c >= 0x2FF2 && c <= 0x2FF3) return UCD_PROPERTY_IDS_TRINARY_OPERATOR;
  1857. if (c >= 0x2FF4 && c <= 0x2FFB) return UCD_PROPERTY_IDS_BINARY_OPERATOR;
  1858. break;
  1859. case 0x3000:
  1860. if (c >= 0x3012 && c <= 0x3013) return UCD_PROPERTY_PATTERN_SYNTAX;
  1861. if (c == 0x3020) return UCD_PROPERTY_PATTERN_SYNTAX;
  1862. break;
  1863. case 0x3200:
  1864. if (c == 0x3297) return UCD_PROPERTY_EMOJI;
  1865. if (c == 0x3299) return UCD_PROPERTY_EMOJI;
  1866. break;
  1867. case 0x01F000:
  1868. if (c == 0x01F004) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1869. if (c == 0x01F0CF) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1870. break;
  1871. case 0x01F100:
  1872. if (c >= 0x01F130 && c <= 0x01F149) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE;
  1873. if (c >= 0x01F150 && c <= 0x01F169) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE;
  1874. if (c >= 0x01F170 && c <= 0x01F171) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE | UCD_PROPERTY_EMOJI;
  1875. if (c >= 0x01F17E && c <= 0x01F17F) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE | UCD_PROPERTY_EMOJI;
  1876. if (c >= 0x01F170 && c <= 0x01F189) return UCD_PROPERTY_OTHER_ALPHABETIC | UCD_PROPERTY_OTHER_UPPERCASE;
  1877. if (c == 0x01F18E) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1878. if (c >= 0x01F191 && c <= 0x01F19A) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1879. if (c >= 0x01F1E6) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_REGIONAL_INDICATOR | UCD_PROPERTY_EMOJI_COMPONENT;
  1880. break;
  1881. case 0x01F200:
  1882. if (c == 0x01F201) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1883. if (c == 0x01F202) return UCD_PROPERTY_EMOJI;
  1884. if (c == 0x01F21A) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1885. if (c == 0x01F22F) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1886. if (c == 0x01F237) return UCD_PROPERTY_EMOJI;
  1887. if (c >= 0x01F232 && c <= 0x01F23A) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1888. if (c >= 0x01F250 && c <= 0x01F251) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1889. break;
  1890. case 0x01F300:
  1891. if (c <= 0x01F320) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1892. if (c == 0x01F321) return UCD_PROPERTY_EMOJI;
  1893. if (c >= 0x01F32D && c <= 0x01F335) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1894. if (c >= 0x01F337 && c <= 0x01F37C) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1895. if (c == 0x01F385) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1896. if (c >= 0x01F37E && c <= 0x01F393) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1897. if (c >= 0x01F324 && c <= 0x01F393) return UCD_PROPERTY_EMOJI;
  1898. if (c >= 0x01F396 && c <= 0x01F397) return UCD_PROPERTY_EMOJI;
  1899. if (c >= 0x01F399 && c <= 0x01F39B) return UCD_PROPERTY_EMOJI;
  1900. if (c >= 0x01F3C2 && c <= 0x01F3C4) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1901. if (c == 0x01F3C7) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1902. if (c >= 0x01F3A0 && c <= 0x01F3C9) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1903. if (c == 0x01F3CA) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1904. if (c >= 0x01F3CB && c <= 0x01F3CC) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1905. if (c >= 0x01F3CF && c <= 0x01F3D3) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1906. if (c >= 0x01F3E0 && c <= 0x01F3F0) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1907. if (c >= 0x01F39E && c <= 0x01F3F0) return UCD_PROPERTY_EMOJI;
  1908. if (c == 0x01F3F3) return UCD_PROPERTY_EMOJI;
  1909. if (c == 0x01F3F4) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1910. if (c == 0x01F3F5) return UCD_PROPERTY_EMOJI;
  1911. if (c == 0x01F3F7) return UCD_PROPERTY_EMOJI;
  1912. if (c >= 0x01F3F8 && c <= 0x01F3FA) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1913. break;
  1914. case 0x01F400:
  1915. if (c == 0x01F43F) return UCD_PROPERTY_EMOJI;
  1916. if (c == 0x01F441) return UCD_PROPERTY_EMOJI;
  1917. if (c >= 0x01F442 && c <= 0x01F443) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1918. if (c >= 0x01F446 && c <= 0x01F450) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1919. if (c >= 0x01F466 && c <= 0x01F469) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1920. if (c == 0x01F46E) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1921. if (c >= 0x01F470 && c <= 0x01F478) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1922. if (c == 0x01F47C) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1923. if (c >= 0x01F481 && c <= 0x01F483) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1924. if (c >= 0x01F485 && c <= 0x01F487) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1925. if (c == 0x01F4AA) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1926. if (c == 0x01F4FE) return 0;
  1927. if (c == 0x01F4FD) return UCD_PROPERTY_EMOJI;
  1928. return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1929. case 0x01F500:
  1930. if (c <= 0x01F53D) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1931. if (c >= 0x01F549 && c <= 0x01F54A) return UCD_PROPERTY_EMOJI;
  1932. if (c >= 0x01F54B && c <= 0x01F54E) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1933. if (c >= 0x01F550 && c <= 0x01F567) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1934. if (c >= 0x01F56F && c <= 0x01F570) return UCD_PROPERTY_EMOJI;
  1935. if (c >= 0x01F574 && c <= 0x01F575) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1936. if (c >= 0x01F573 && c <= 0x01F579) return UCD_PROPERTY_EMOJI;
  1937. if (c == 0x01F57A) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1938. if (c == 0x01F587) return UCD_PROPERTY_EMOJI;
  1939. if (c >= 0x01F58A && c <= 0x01F58D) return UCD_PROPERTY_EMOJI;
  1940. if (c == 0x01F590) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1941. if (c >= 0x01F595 && c <= 0x01F596) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1942. if (c == 0x01F5A4) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1943. if (c == 0x01F5A5) return UCD_PROPERTY_EMOJI;
  1944. if (c == 0x01F5A8) return UCD_PROPERTY_EMOJI;
  1945. if (c >= 0x01F5B1 && c <= 0x01F5B2) return UCD_PROPERTY_EMOJI;
  1946. if (c == 0x01F5BC) return UCD_PROPERTY_EMOJI;
  1947. if (c >= 0x01F5C2 && c <= 0x01F5C4) return UCD_PROPERTY_EMOJI;
  1948. if (c >= 0x01F5D1 && c <= 0x01F5D3) return UCD_PROPERTY_EMOJI;
  1949. if (c >= 0x01F5DC && c <= 0x01F5DE) return UCD_PROPERTY_EMOJI;
  1950. if (c == 0x01F5E1) return UCD_PROPERTY_EMOJI;
  1951. if (c == 0x01F5E3) return UCD_PROPERTY_EMOJI;
  1952. if (c == 0x01F5E8) return UCD_PROPERTY_EMOJI;
  1953. if (c == 0x01F5EF) return UCD_PROPERTY_EMOJI;
  1954. if (c == 0x01F5F3) return UCD_PROPERTY_EMOJI;
  1955. if (c == 0x01F5FA) return UCD_PROPERTY_EMOJI;
  1956. if (c >= 0x01F5FB) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1957. break;
  1958. case 0x01F600:
  1959. if (c >= 0x01F645 && c <= 0x01F647) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1960. if (c >= 0x01F64B && c <= 0x01F64F) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1961. if (c <= 0x01F64F) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1962. if (c == 0x01F6A3) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1963. if (c >= 0x01F6B4 && c <= 0x01F6B6) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1964. if (c == 0x01F6C0) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1965. if (c >= 0x01F680 && c <= 0x01F6C5) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1966. if (c == 0x01F6CC) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1967. if (c >= 0x01F6CB && c <= 0x01F6CF) return UCD_PROPERTY_EMOJI;
  1968. if (c >= 0x01F6D0 && c <= 0x01F6D2) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1969. if (c >= 0x01F6E0 && c <= 0x01F6E5) return UCD_PROPERTY_EMOJI;
  1970. if (c == 0x01F6E9) return UCD_PROPERTY_EMOJI;
  1971. if (c >= 0x01F6EB && c <= 0x01F6EC) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1972. if (c == 0x01F6F0) return UCD_PROPERTY_EMOJI;
  1973. if (c == 0x01F6F3) return UCD_PROPERTY_EMOJI;
  1974. if (c >= 0x01F6F4 && c <= 0x01F6F8) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1975. break;
  1976. case 0x01F900:
  1977. if (c >= 0x01F918 && c <= 0x01F91C) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1978. if (c >= 0x01F910 && c <= 0x01F91D) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1979. if (c >= 0x01F91E && c <= 0x01F91F) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1980. if (c == 0x01F926) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1981. if (c >= 0x01F920 && c <= 0x01F92F) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1982. if (c >= 0x01F930 && c <= 0x01F932) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1983. if (c == 0x01F93B) return 0;
  1984. if (c >= 0x01F93A && c <= 0x01F93C) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1985. if (c >= 0x01F933 && c <= 0x01F93E) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1986. if (c == 0x01F946) return 0;
  1987. if (c >= 0x01F940 && c <= 0x01F94B) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1988. if (c == 0x01F94C) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1989. if (c >= 0x01F950 && c <= 0x01F95E) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1990. if (c >= 0x01F95F && c <= 0x01F96B) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1991. if (c >= 0x01F980 && c <= 0x01F991) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1992. if (c >= 0x01F992 && c <= 0x01F997) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1993. if (c == 0x01F9C0) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1994. if (c >= 0x01F9D1 && c <= 0x01F9DD) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION | UCD_PROPERTY_EMOJI_MODIFIER_BASE;
  1995. if (c >= 0x01F9D0 && c <= 0x01F9E6) return UCD_PROPERTY_EMOJI | UCD_PROPERTY_EMOJI_PRESENTATION;
  1996. return 0;
  1997. }
  1998. return 0;
  1999. }
  2000. static ucd_property properties_Zs(codepoint_t c)
  2001. {
  2002. if (c == 0x0020) return UCD_PROPERTY_WHITE_SPACE | UCD_PROPERTY_PATTERN_WHITE_SPACE;
  2003. return UCD_PROPERTY_WHITE_SPACE;
  2004. }
  2005. ucd_property ucd_properties(codepoint_t c, ucd_category category)
  2006. {
  2007. switch (category)
  2008. {
  2009. case UCD_CATEGORY_Cc: return properties_Cc(c);
  2010. case UCD_CATEGORY_Cf: return properties_Cf(c);
  2011. case UCD_CATEGORY_Cn: return properties_Cn(c);
  2012. case UCD_CATEGORY_Ll: return properties_Ll(c);
  2013. case UCD_CATEGORY_Lm: return properties_Lm(c);
  2014. case UCD_CATEGORY_Lo: return properties_Lo(c) | properties_Lo_ideographic(c);
  2015. case UCD_CATEGORY_Lu: return properties_Lu(c);
  2016. case UCD_CATEGORY_Mc: return properties_Mc(c);
  2017. case UCD_CATEGORY_Mn: return properties_Mn(c);
  2018. case UCD_CATEGORY_Nd: return properties_Nd(c);
  2019. case UCD_CATEGORY_Nl: return properties_Nl(c);
  2020. case UCD_CATEGORY_No: return properties_No(c);
  2021. case UCD_CATEGORY_Pc: return properties_Pc(c);
  2022. case UCD_CATEGORY_Pd: return properties_Pd(c);
  2023. case UCD_CATEGORY_Pe: return properties_Pe(c);
  2024. case UCD_CATEGORY_Pf: return properties_Pf(c);
  2025. case UCD_CATEGORY_Pi: return properties_Pi(c);
  2026. case UCD_CATEGORY_Po: return properties_Po(c);
  2027. case UCD_CATEGORY_Ps: return properties_Ps(c);
  2028. case UCD_CATEGORY_Sc: return properties_Sc(c);
  2029. case UCD_CATEGORY_Sk: return properties_Sk(c);
  2030. case UCD_CATEGORY_Sm: return properties_Sm(c);
  2031. case UCD_CATEGORY_So: return properties_So(c);
  2032. case UCD_CATEGORY_Zl: return UCD_PROPERTY_WHITE_SPACE | UCD_PROPERTY_PATTERN_WHITE_SPACE;
  2033. case UCD_CATEGORY_Zp: return UCD_PROPERTY_WHITE_SPACE | UCD_PROPERTY_PATTERN_WHITE_SPACE;
  2034. case UCD_CATEGORY_Zs: return properties_Zs(c);
  2035. default: return 0; /* Co Cs Ii Lt Me */
  2036. };
  2037. }