GenerateUnicodeDateTimeFormat.cpp 95 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396
  1. /*
  2. * Copyright (c) 2021-2022, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "GeneratorUtil.h"
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/Find.h>
  10. #include <AK/Format.h>
  11. #include <AK/GenericLexer.h>
  12. #include <AK/HashFunctions.h>
  13. #include <AK/HashMap.h>
  14. #include <AK/JsonObject.h>
  15. #include <AK/JsonParser.h>
  16. #include <AK/JsonValue.h>
  17. #include <AK/LexicalPath.h>
  18. #include <AK/SourceGenerator.h>
  19. #include <AK/String.h>
  20. #include <AK/StringBuilder.h>
  21. #include <AK/Traits.h>
  22. #include <AK/Utf8View.h>
  23. #include <LibCore/ArgsParser.h>
  24. #include <LibCore/DirIterator.h>
  25. #include <LibCore/Stream.h>
  26. #include <LibTimeZone/TimeZone.h>
  27. #include <LibUnicode/DateTimeFormat.h>
  28. using StringIndexType = u16;
  29. constexpr auto s_string_index_type = "u16"sv;
  30. using CalendarPatternIndexType = u16;
  31. constexpr auto s_calendar_pattern_index_type = "u16"sv;
  32. using CalendarPatternListIndexType = u8;
  33. constexpr auto s_calendar_pattern_list_index_type = "u8"sv;
  34. using CalendarRangePatternIndexType = u16;
  35. constexpr auto s_calendar_range_pattern_index_type = "u16"sv;
  36. using CalendarRangePatternListIndexType = u8;
  37. constexpr auto s_calendar_range_pattern_list_index_type = "u8"sv;
  38. using CalendarFormatIndexType = u8;
  39. constexpr auto s_calendar_format_index_type = "u8"sv;
  40. using SymbolListIndexType = u16;
  41. constexpr auto s_symbol_list_index_type = "u16"sv;
  42. using CalendarSymbolsIndexType = u16;
  43. constexpr auto s_calendar_symbols_index_type = "u16"sv;
  44. using CalendarSymbolsListIndexType = u8;
  45. constexpr auto s_calendar_symbols_list_index_type = "u8"sv;
  46. using CalendarIndexType = u8;
  47. constexpr auto s_calendar_index_type = "u8"sv;
  48. using TimeZoneNamesIndexType = u16;
  49. constexpr auto s_time_zone_index_type = "u16"sv;
  50. using TimeZoneNamesListIndexType = u8;
  51. constexpr auto s_time_zone_list_index_type = "u8"sv;
  52. using TimeZoneFormatIndexType = u8;
  53. constexpr auto s_time_zone_format_index_type = "u8"sv;
  54. using DayPeriodIndexType = u8;
  55. constexpr auto s_day_period_index_type = "u8"sv;
  56. using DayPeriodListIndexType = u8;
  57. constexpr auto s_day_period_list_index_type = "u8"sv;
  58. using HourCycleListIndexType = u8;
  59. constexpr auto s_hour_cycle_list_index_type = "u8"sv;
  60. struct CalendarPattern : public Unicode::CalendarPattern {
  61. bool contains_only_date_fields() const
  62. {
  63. return !day_period.has_value() && !hour.has_value() && !minute.has_value() && !second.has_value() && !fractional_second_digits.has_value() && !time_zone_name.has_value();
  64. }
  65. bool contains_only_time_fields() const
  66. {
  67. return !weekday.has_value() && !era.has_value() && !year.has_value() && !month.has_value() && !day.has_value();
  68. }
  69. unsigned hash() const
  70. {
  71. auto hash = pair_int_hash(pattern_index, pattern12_index);
  72. hash = pair_int_hash(hash, skeleton_index);
  73. auto hash_field = [&](auto const& field) {
  74. if (field.has_value())
  75. hash = pair_int_hash(hash, static_cast<u8>(*field));
  76. else
  77. hash = pair_int_hash(hash, -1);
  78. };
  79. hash_field(era);
  80. hash_field(year);
  81. hash_field(month);
  82. hash_field(weekday);
  83. hash_field(day);
  84. hash_field(day_period);
  85. hash_field(hour);
  86. hash_field(minute);
  87. hash_field(second);
  88. hash_field(fractional_second_digits);
  89. hash_field(time_zone_name);
  90. return hash;
  91. }
  92. bool operator==(CalendarPattern const& other) const
  93. {
  94. return (skeleton_index == other.skeleton_index)
  95. && (pattern_index == other.pattern_index)
  96. && (pattern12_index == other.pattern12_index)
  97. && (era == other.era)
  98. && (year == other.year)
  99. && (month == other.month)
  100. && (weekday == other.weekday)
  101. && (day == other.day)
  102. && (day_period == other.day_period)
  103. && (hour == other.hour)
  104. && (minute == other.minute)
  105. && (second == other.second)
  106. && (fractional_second_digits == other.fractional_second_digits)
  107. && (time_zone_name == other.time_zone_name);
  108. }
  109. StringIndexType skeleton_index { 0 };
  110. StringIndexType pattern_index { 0 };
  111. StringIndexType pattern12_index { 0 };
  112. };
  113. template<>
  114. struct AK::Formatter<CalendarPattern> : Formatter<FormatString> {
  115. ErrorOr<void> format(FormatBuilder& builder, CalendarPattern const& pattern)
  116. {
  117. auto field_to_i8 = [](auto const& field) -> i8 {
  118. if (!field.has_value())
  119. return -1;
  120. return static_cast<i8>(*field);
  121. };
  122. return Formatter<FormatString>::format(builder,
  123. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  124. pattern.skeleton_index,
  125. pattern.pattern_index,
  126. pattern.pattern12_index,
  127. field_to_i8(pattern.era),
  128. field_to_i8(pattern.year),
  129. field_to_i8(pattern.month),
  130. field_to_i8(pattern.weekday),
  131. field_to_i8(pattern.day),
  132. field_to_i8(pattern.day_period),
  133. field_to_i8(pattern.hour),
  134. field_to_i8(pattern.minute),
  135. field_to_i8(pattern.second),
  136. field_to_i8(pattern.fractional_second_digits),
  137. field_to_i8(pattern.time_zone_name));
  138. }
  139. };
  140. template<>
  141. struct AK::Traits<CalendarPattern> : public GenericTraits<CalendarPattern> {
  142. static unsigned hash(CalendarPattern const& c) { return c.hash(); }
  143. };
  144. struct CalendarRangePattern : public CalendarPattern {
  145. unsigned hash() const
  146. {
  147. auto hash = CalendarPattern::hash();
  148. if (field.has_value())
  149. hash = pair_int_hash(hash, static_cast<u8>(*field));
  150. hash = pair_int_hash(hash, start_range);
  151. hash = pair_int_hash(hash, separator);
  152. hash = pair_int_hash(hash, end_range);
  153. return hash;
  154. }
  155. bool operator==(CalendarRangePattern const& other) const
  156. {
  157. if (!CalendarPattern::operator==(other))
  158. return false;
  159. return (field == other.field)
  160. && (start_range == other.start_range)
  161. && (separator == other.separator)
  162. && (end_range == other.end_range);
  163. }
  164. Optional<Unicode::CalendarRangePattern::Field> field {};
  165. StringIndexType start_range { 0 };
  166. StringIndexType separator { 0 };
  167. StringIndexType end_range { 0 };
  168. };
  169. template<>
  170. struct AK::Formatter<CalendarRangePattern> : Formatter<FormatString> {
  171. ErrorOr<void> format(FormatBuilder& builder, CalendarRangePattern const& pattern)
  172. {
  173. auto field_to_i8 = [](auto const& field) -> i8 {
  174. if (!field.has_value())
  175. return -1;
  176. return static_cast<i8>(*field);
  177. };
  178. return Formatter<FormatString>::format(builder,
  179. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  180. pattern.skeleton_index,
  181. field_to_i8(pattern.field),
  182. pattern.start_range,
  183. pattern.separator,
  184. pattern.end_range,
  185. field_to_i8(pattern.era),
  186. field_to_i8(pattern.year),
  187. field_to_i8(pattern.month),
  188. field_to_i8(pattern.weekday),
  189. field_to_i8(pattern.day),
  190. field_to_i8(pattern.day_period),
  191. field_to_i8(pattern.hour),
  192. field_to_i8(pattern.minute),
  193. field_to_i8(pattern.second),
  194. field_to_i8(pattern.fractional_second_digits),
  195. field_to_i8(pattern.time_zone_name));
  196. }
  197. };
  198. template<>
  199. struct AK::Traits<CalendarRangePattern> : public GenericTraits<CalendarRangePattern> {
  200. static unsigned hash(CalendarRangePattern const& c) { return c.hash(); }
  201. };
  202. struct CalendarFormat {
  203. unsigned hash() const
  204. {
  205. auto hash = pair_int_hash(full_format, long_format);
  206. hash = pair_int_hash(hash, medium_format);
  207. hash = pair_int_hash(hash, short_format);
  208. return hash;
  209. }
  210. bool operator==(CalendarFormat const& other) const
  211. {
  212. return (full_format == other.full_format)
  213. && (long_format == other.long_format)
  214. && (medium_format == other.medium_format)
  215. && (short_format == other.short_format);
  216. }
  217. CalendarPatternIndexType full_format { 0 };
  218. CalendarPatternIndexType long_format { 0 };
  219. CalendarPatternIndexType medium_format { 0 };
  220. CalendarPatternIndexType short_format { 0 };
  221. };
  222. template<>
  223. struct AK::Formatter<CalendarFormat> : Formatter<FormatString> {
  224. ErrorOr<void> format(FormatBuilder& builder, CalendarFormat const& pattern)
  225. {
  226. return Formatter<FormatString>::format(builder,
  227. "{{ {}, {}, {}, {} }}"sv,
  228. pattern.full_format,
  229. pattern.long_format,
  230. pattern.medium_format,
  231. pattern.short_format);
  232. }
  233. };
  234. template<>
  235. struct AK::Traits<CalendarFormat> : public GenericTraits<CalendarFormat> {
  236. static unsigned hash(CalendarFormat const& c) { return c.hash(); }
  237. };
  238. using SymbolList = Vector<StringIndexType>;
  239. struct CalendarSymbols {
  240. unsigned hash() const
  241. {
  242. auto hash = pair_int_hash(narrow_symbols, short_symbols);
  243. hash = pair_int_hash(hash, long_symbols);
  244. return hash;
  245. }
  246. bool operator==(CalendarSymbols const& other) const
  247. {
  248. return (narrow_symbols == other.narrow_symbols)
  249. && (short_symbols == other.short_symbols)
  250. && (long_symbols == other.long_symbols);
  251. }
  252. SymbolListIndexType narrow_symbols { 0 };
  253. SymbolListIndexType short_symbols { 0 };
  254. SymbolListIndexType long_symbols { 0 };
  255. };
  256. template<>
  257. struct AK::Formatter<CalendarSymbols> : Formatter<FormatString> {
  258. ErrorOr<void> format(FormatBuilder& builder, CalendarSymbols const& symbols)
  259. {
  260. return Formatter<FormatString>::format(builder,
  261. "{{ {}, {}, {} }}"sv,
  262. symbols.narrow_symbols,
  263. symbols.short_symbols,
  264. symbols.long_symbols);
  265. }
  266. };
  267. template<>
  268. struct AK::Traits<CalendarSymbols> : public GenericTraits<CalendarSymbols> {
  269. static unsigned hash(CalendarSymbols const& c) { return c.hash(); }
  270. };
  271. using CalendarPatternList = Vector<CalendarPatternIndexType>;
  272. using CalendarRangePatternList = Vector<CalendarRangePatternIndexType>;
  273. using CalendarSymbolsList = Vector<CalendarSymbolsIndexType>;
  274. struct Calendar {
  275. unsigned hash() const
  276. {
  277. auto hash = int_hash(date_formats);
  278. hash = pair_int_hash(hash, time_formats);
  279. hash = pair_int_hash(hash, date_time_formats);
  280. hash = pair_int_hash(hash, available_formats);
  281. hash = pair_int_hash(hash, default_range_format);
  282. hash = pair_int_hash(hash, range_formats);
  283. hash = pair_int_hash(hash, range12_formats);
  284. hash = pair_int_hash(hash, symbols);
  285. return hash;
  286. }
  287. bool operator==(Calendar const& other) const
  288. {
  289. return (date_formats == other.date_formats)
  290. && (time_formats == other.time_formats)
  291. && (date_time_formats == other.date_time_formats)
  292. && (available_formats == other.available_formats)
  293. && (default_range_format == other.default_range_format)
  294. && (range_formats == other.range_formats)
  295. && (range12_formats == other.range12_formats)
  296. && (symbols == other.symbols);
  297. }
  298. CalendarFormatIndexType date_formats { 0 };
  299. CalendarFormatIndexType time_formats { 0 };
  300. CalendarFormatIndexType date_time_formats { 0 };
  301. CalendarPatternListIndexType available_formats { 0 };
  302. CalendarRangePatternIndexType default_range_format { 0 };
  303. CalendarRangePatternListIndexType range_formats { 0 };
  304. CalendarRangePatternListIndexType range12_formats { 0 };
  305. CalendarSymbolsListIndexType symbols { 0 };
  306. };
  307. template<>
  308. struct AK::Formatter<Calendar> : Formatter<FormatString> {
  309. ErrorOr<void> format(FormatBuilder& builder, Calendar const& calendar)
  310. {
  311. return Formatter<FormatString>::format(builder,
  312. "{{ {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  313. calendar.date_formats,
  314. calendar.time_formats,
  315. calendar.date_time_formats,
  316. calendar.available_formats,
  317. calendar.default_range_format,
  318. calendar.range_formats,
  319. calendar.range12_formats,
  320. calendar.symbols);
  321. }
  322. };
  323. template<>
  324. struct AK::Traits<Calendar> : public GenericTraits<Calendar> {
  325. static unsigned hash(Calendar const& c) { return c.hash(); }
  326. };
  327. struct TimeZoneNames {
  328. unsigned hash() const
  329. {
  330. auto hash = pair_int_hash(short_standard_name, long_standard_name);
  331. hash = pair_int_hash(hash, short_daylight_name);
  332. hash = pair_int_hash(hash, long_daylight_name);
  333. hash = pair_int_hash(hash, short_generic_name);
  334. hash = pair_int_hash(hash, long_generic_name);
  335. return hash;
  336. }
  337. bool operator==(TimeZoneNames const& other) const
  338. {
  339. return (short_standard_name == other.short_standard_name)
  340. && (long_standard_name == other.long_standard_name)
  341. && (short_daylight_name == other.short_daylight_name)
  342. && (long_daylight_name == other.long_daylight_name)
  343. && (short_generic_name == other.short_generic_name)
  344. && (long_generic_name == other.long_generic_name);
  345. }
  346. StringIndexType short_standard_name { 0 };
  347. StringIndexType long_standard_name { 0 };
  348. StringIndexType short_daylight_name { 0 };
  349. StringIndexType long_daylight_name { 0 };
  350. StringIndexType short_generic_name { 0 };
  351. StringIndexType long_generic_name { 0 };
  352. };
  353. template<>
  354. struct AK::Formatter<TimeZoneNames> : Formatter<FormatString> {
  355. ErrorOr<void> format(FormatBuilder& builder, TimeZoneNames const& time_zone)
  356. {
  357. return Formatter<FormatString>::format(builder,
  358. "{{ {}, {}, {}, {}, {}, {} }}"sv,
  359. time_zone.short_standard_name,
  360. time_zone.long_standard_name,
  361. time_zone.short_daylight_name,
  362. time_zone.long_daylight_name,
  363. time_zone.short_generic_name,
  364. time_zone.long_generic_name);
  365. }
  366. };
  367. template<>
  368. struct AK::Traits<TimeZoneNames> : public GenericTraits<TimeZoneNames> {
  369. static unsigned hash(TimeZoneNames const& t) { return t.hash(); }
  370. };
  371. struct TimeZoneFormat {
  372. unsigned hash() const
  373. {
  374. auto hash = int_hash(symbol_ahead_sign);
  375. hash = pair_int_hash(hash, symbol_ahead_separator);
  376. hash = pair_int_hash(hash, symbol_behind_sign);
  377. hash = pair_int_hash(hash, symbol_behind_separator);
  378. hash = pair_int_hash(hash, gmt_format);
  379. hash = pair_int_hash(hash, gmt_zero_format);
  380. return hash;
  381. }
  382. bool operator==(TimeZoneFormat const& other) const
  383. {
  384. return (symbol_ahead_sign == other.symbol_ahead_sign)
  385. && (symbol_ahead_separator == other.symbol_ahead_separator)
  386. && (symbol_behind_sign == other.symbol_behind_sign)
  387. && (symbol_behind_separator == other.symbol_behind_separator)
  388. && (gmt_format == other.gmt_format)
  389. && (gmt_zero_format == other.gmt_zero_format);
  390. }
  391. StringIndexType symbol_ahead_sign { 0 };
  392. StringIndexType symbol_ahead_separator { 0 };
  393. StringIndexType symbol_behind_sign { 0 };
  394. StringIndexType symbol_behind_separator { 0 };
  395. StringIndexType gmt_format { 0 };
  396. StringIndexType gmt_zero_format { 0 };
  397. };
  398. template<>
  399. struct AK::Formatter<TimeZoneFormat> : Formatter<FormatString> {
  400. ErrorOr<void> format(FormatBuilder& builder, TimeZoneFormat const& time_zone_format)
  401. {
  402. return Formatter<FormatString>::format(builder, "{{ {}, {}, {}, {}, {}, {} }}"sv,
  403. time_zone_format.symbol_ahead_sign,
  404. time_zone_format.symbol_ahead_separator,
  405. time_zone_format.symbol_behind_sign,
  406. time_zone_format.symbol_behind_separator,
  407. time_zone_format.gmt_format,
  408. time_zone_format.gmt_zero_format);
  409. }
  410. };
  411. template<>
  412. struct AK::Traits<TimeZoneFormat> : public GenericTraits<TimeZoneFormat> {
  413. static unsigned hash(TimeZoneFormat const& t) { return t.hash(); }
  414. };
  415. struct DayPeriod {
  416. unsigned hash() const
  417. {
  418. auto hash = int_hash(static_cast<u8>(day_period));
  419. hash = pair_int_hash(hash, begin);
  420. hash = pair_int_hash(hash, end);
  421. return hash;
  422. }
  423. bool operator==(DayPeriod const& other) const
  424. {
  425. return (day_period == other.day_period)
  426. && (begin == other.begin)
  427. && (end == other.end);
  428. }
  429. Unicode::DayPeriod day_period {};
  430. u8 begin { 0 };
  431. u8 end { 0 };
  432. };
  433. template<>
  434. struct AK::Formatter<DayPeriod> : Formatter<FormatString> {
  435. ErrorOr<void> format(FormatBuilder& builder, DayPeriod const& day_period)
  436. {
  437. return Formatter<FormatString>::format(builder,
  438. "{{ {}, {}, {} }}"sv,
  439. static_cast<u8>(day_period.day_period),
  440. day_period.begin,
  441. day_period.end);
  442. }
  443. };
  444. template<>
  445. struct AK::Traits<DayPeriod> : public GenericTraits<DayPeriod> {
  446. static unsigned hash(DayPeriod const& d) { return d.hash(); }
  447. };
  448. using TimeZoneNamesList = Vector<TimeZoneNamesIndexType>;
  449. using DayPeriodList = Vector<DayPeriodIndexType>;
  450. using HourCycleList = Vector<Unicode::HourCycle>;
  451. template<>
  452. struct AK::Formatter<Unicode::HourCycle> : Formatter<FormatString> {
  453. ErrorOr<void> format(FormatBuilder& builder, Unicode::HourCycle hour_cycle)
  454. {
  455. return builder.put_u64(to_underlying(hour_cycle));
  456. }
  457. };
  458. struct Locale {
  459. HashMap<String, CalendarIndexType> calendars;
  460. TimeZoneNamesListIndexType time_zones { 0 };
  461. TimeZoneFormatIndexType time_zone_formats { 0 };
  462. DayPeriodListIndexType day_periods { 0 };
  463. };
  464. struct UnicodeLocaleData {
  465. UniqueStringStorage<StringIndexType> unique_strings;
  466. UniqueStorage<CalendarPattern, CalendarPatternIndexType> unique_patterns;
  467. UniqueStorage<CalendarPatternList, CalendarPatternListIndexType> unique_pattern_lists;
  468. UniqueStorage<CalendarRangePattern, CalendarRangePatternIndexType> unique_range_patterns;
  469. UniqueStorage<CalendarRangePatternList, CalendarRangePatternListIndexType> unique_range_pattern_lists;
  470. UniqueStorage<CalendarFormat, CalendarFormatIndexType> unique_formats;
  471. UniqueStorage<SymbolList, SymbolListIndexType> unique_symbol_lists;
  472. UniqueStorage<CalendarSymbols, CalendarSymbolsIndexType> unique_calendar_symbols;
  473. UniqueStorage<CalendarSymbolsList, CalendarSymbolsListIndexType> unique_calendar_symbols_lists;
  474. UniqueStorage<Calendar, CalendarIndexType> unique_calendars;
  475. UniqueStorage<TimeZoneNames, TimeZoneNamesIndexType> unique_time_zones;
  476. UniqueStorage<TimeZoneNamesList, TimeZoneNamesListIndexType> unique_time_zone_lists;
  477. UniqueStorage<TimeZoneFormat, TimeZoneFormatIndexType> unique_time_zone_formats;
  478. UniqueStorage<DayPeriod, DayPeriodIndexType> unique_day_periods;
  479. UniqueStorage<DayPeriodList, DayPeriodListIndexType> unique_day_period_lists;
  480. UniqueStorage<HourCycleList, HourCycleListIndexType> unique_hour_cycle_lists;
  481. HashMap<String, Locale> locales;
  482. HashMap<String, HourCycleListIndexType> hour_cycles;
  483. Vector<String> hour_cycle_regions;
  484. HashMap<String, u8> minimum_days;
  485. Vector<String> minimum_days_regions;
  486. HashMap<String, Unicode::Weekday> first_day;
  487. Vector<String> first_day_regions;
  488. HashMap<String, Unicode::Weekday> weekend_start;
  489. Vector<String> weekend_start_regions;
  490. HashMap<String, Unicode::Weekday> weekend_end;
  491. Vector<String> weekend_end_regions;
  492. HashMap<String, Vector<TimeZone::TimeZone>> meta_zones;
  493. Vector<String> time_zones { "UTC"sv };
  494. Vector<String> calendars;
  495. };
  496. static Optional<Unicode::DayPeriod> day_period_from_string(StringView day_period)
  497. {
  498. if (day_period == "am"sv)
  499. return Unicode::DayPeriod::AM;
  500. if (day_period == "pm"sv)
  501. return Unicode::DayPeriod::PM;
  502. if (day_period == "morning1"sv)
  503. return Unicode::DayPeriod::Morning1;
  504. if (day_period == "morning2"sv)
  505. return Unicode::DayPeriod::Morning2;
  506. if (day_period == "afternoon1"sv)
  507. return Unicode::DayPeriod::Afternoon1;
  508. if (day_period == "afternoon2"sv)
  509. return Unicode::DayPeriod::Afternoon2;
  510. if (day_period == "evening1"sv)
  511. return Unicode::DayPeriod::Evening1;
  512. if (day_period == "evening2"sv)
  513. return Unicode::DayPeriod::Evening2;
  514. if (day_period == "night1"sv)
  515. return Unicode::DayPeriod::Night1;
  516. if (day_period == "night2"sv)
  517. return Unicode::DayPeriod::Night2;
  518. return {};
  519. }
  520. static ErrorOr<void> parse_hour_cycles(String core_path, UnicodeLocaleData& locale_data)
  521. {
  522. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  523. LexicalPath time_data_path(move(core_path));
  524. time_data_path = time_data_path.append("supplemental"sv);
  525. time_data_path = time_data_path.append("timeData.json"sv);
  526. auto time_data = TRY(read_json_file(time_data_path.string()));
  527. auto const& supplemental_object = time_data.as_object().get("supplemental"sv);
  528. auto const& time_data_object = supplemental_object.as_object().get("timeData"sv);
  529. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Unicode::HourCycle> {
  530. if (hour_cycle == "h"sv)
  531. return Unicode::HourCycle::H12;
  532. if (hour_cycle == "H"sv)
  533. return Unicode::HourCycle::H23;
  534. if (hour_cycle == "K"sv)
  535. return Unicode::HourCycle::H11;
  536. if (hour_cycle == "k"sv)
  537. return Unicode::HourCycle::H24;
  538. return {};
  539. };
  540. time_data_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  541. auto allowed_hour_cycles_string = value.as_object().get("_allowed"sv).as_string();
  542. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  543. Vector<Unicode::HourCycle> hour_cycles;
  544. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  545. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  546. hour_cycles.append(*hour_cycle);
  547. }
  548. auto hour_cycles_index = locale_data.unique_hour_cycle_lists.ensure(move(hour_cycles));
  549. locale_data.hour_cycles.set(key, hour_cycles_index);
  550. if (!locale_data.hour_cycle_regions.contains_slow(key))
  551. locale_data.hour_cycle_regions.append(key);
  552. });
  553. return {};
  554. }
  555. static ErrorOr<void> parse_week_data(String core_path, UnicodeLocaleData& locale_data)
  556. {
  557. // https://unicode.org/reports/tr35/tr35-dates.html#Week_Data
  558. LexicalPath week_data_path(move(core_path));
  559. week_data_path = week_data_path.append("supplemental"sv);
  560. week_data_path = week_data_path.append("weekData.json"sv);
  561. auto week_data = TRY(read_json_file(week_data_path.string()));
  562. auto const& supplemental_object = week_data.as_object().get("supplemental"sv);
  563. auto const& week_data_object = supplemental_object.as_object().get("weekData"sv);
  564. auto parse_weekday = [](StringView day) -> Unicode::Weekday {
  565. if (day == "sun"sv)
  566. return Unicode::Weekday::Sunday;
  567. if (day == "mon"sv)
  568. return Unicode::Weekday::Monday;
  569. if (day == "tue"sv)
  570. return Unicode::Weekday::Tuesday;
  571. if (day == "wed"sv)
  572. return Unicode::Weekday::Wednesday;
  573. if (day == "thu"sv)
  574. return Unicode::Weekday::Thursday;
  575. if (day == "fri"sv)
  576. return Unicode::Weekday::Friday;
  577. if (day == "sat"sv)
  578. return Unicode::Weekday::Saturday;
  579. VERIFY_NOT_REACHED();
  580. };
  581. auto parse_regional_weekdays = [&](auto const& region, auto const& weekday, auto& weekdays_map, auto& weekday_regions) {
  582. if (region.ends_with("alt-variant"sv))
  583. return;
  584. weekdays_map.set(region, parse_weekday(weekday));
  585. if (!weekday_regions.contains_slow(region))
  586. weekday_regions.append(region);
  587. };
  588. auto const& minimum_days_object = week_data_object.as_object().get("minDays"sv);
  589. auto const& first_day_object = week_data_object.as_object().get("firstDay"sv);
  590. auto const& weekend_start_object = week_data_object.as_object().get("weekendStart"sv);
  591. auto const& weekend_end_object = week_data_object.as_object().get("weekendEnd"sv);
  592. minimum_days_object.as_object().for_each_member([&](auto const& region, auto const& value) {
  593. auto minimum_days = value.as_string().template to_uint<u8>();
  594. locale_data.minimum_days.set(region, *minimum_days);
  595. if (!locale_data.minimum_days_regions.contains_slow(region))
  596. locale_data.minimum_days_regions.append(region);
  597. });
  598. first_day_object.as_object().for_each_member([&](auto const& region, auto const& value) {
  599. parse_regional_weekdays(region, value.as_string(), locale_data.first_day, locale_data.first_day_regions);
  600. });
  601. weekend_start_object.as_object().for_each_member([&](auto const& region, auto const& value) {
  602. parse_regional_weekdays(region, value.as_string(), locale_data.weekend_start, locale_data.weekend_start_regions);
  603. });
  604. weekend_end_object.as_object().for_each_member([&](auto const& region, auto const& value) {
  605. parse_regional_weekdays(region, value.as_string(), locale_data.weekend_end, locale_data.weekend_end_regions);
  606. });
  607. return {};
  608. }
  609. static ErrorOr<void> parse_meta_zones(String core_path, UnicodeLocaleData& locale_data)
  610. {
  611. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  612. LexicalPath meta_zone_path(move(core_path));
  613. meta_zone_path = meta_zone_path.append("supplemental"sv);
  614. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  615. auto meta_zone = TRY(read_json_file(meta_zone_path.string()));
  616. auto const& supplemental_object = meta_zone.as_object().get("supplemental"sv);
  617. auto const& meta_zone_object = supplemental_object.as_object().get("metaZones"sv);
  618. auto const& meta_zone_array = meta_zone_object.as_object().get("metazones"sv);
  619. meta_zone_array.as_array().for_each([&](JsonValue const& value) {
  620. auto const& mapping = value.as_object().get("mapZone"sv);
  621. auto const& meta_zone = mapping.as_object().get("_other"sv);
  622. auto const& golden_zone = mapping.as_object().get("_type"sv);
  623. if (auto time_zone = TimeZone::time_zone_from_string(golden_zone.as_string()); time_zone.has_value()) {
  624. auto& golden_zones = locale_data.meta_zones.ensure(meta_zone.as_string());
  625. golden_zones.append(*time_zone);
  626. }
  627. });
  628. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  629. if (auto time_zone = TimeZone::time_zone_from_string("UTC"sv); time_zone.has_value())
  630. locale_data.meta_zones.set("UTC"sv, { *time_zone });
  631. return {};
  632. }
  633. static constexpr auto is_char(char ch)
  634. {
  635. return [ch](auto c) { return c == ch; };
  636. }
  637. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  638. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  639. // extra spaces around the day period. Some example expected removals:
  640. //
  641. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  642. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  643. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  644. static String remove_period_from_pattern(String pattern)
  645. {
  646. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  647. auto index = pattern.find(remove);
  648. if (!index.has_value())
  649. continue;
  650. constexpr u32 space = ' ';
  651. constexpr u32 open = '{';
  652. constexpr u32 close = '}';
  653. Utf8View utf8_pattern { pattern };
  654. Optional<u32> before_removal;
  655. Optional<u32> after_removal;
  656. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  657. before_removal = *it;
  658. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  659. after_removal = *it;
  660. if ((before_removal == space) && (after_removal != open)) {
  661. pattern = String::formatted("{}{}",
  662. pattern.substring_view(0, *index - 1),
  663. pattern.substring_view(*index + remove.length()));
  664. } else if ((after_removal == space) && (before_removal != close)) {
  665. pattern = String::formatted("{}{}",
  666. pattern.substring_view(0, *index),
  667. pattern.substring_view(*index + remove.length() + 1));
  668. } else {
  669. pattern = String::formatted("{}{}",
  670. pattern.substring_view(0, *index),
  671. pattern.substring_view(*index + remove.length()));
  672. }
  673. }
  674. return pattern;
  675. }
  676. static Optional<CalendarPattern> parse_date_time_pattern_raw(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  677. {
  678. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  679. using Unicode::CalendarPatternStyle;
  680. CalendarPattern format {};
  681. if (!skeleton.is_empty())
  682. format.skeleton_index = locale_data.unique_strings.ensure(move(skeleton));
  683. GenericLexer lexer { pattern };
  684. StringBuilder builder;
  685. bool hour12 { false };
  686. while (!lexer.is_eof()) {
  687. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  688. // processing (this just avoids conflicts with the patterns below).
  689. if (lexer.next_is(is_quote)) {
  690. builder.append(lexer.consume_quoted_string());
  691. continue;
  692. }
  693. auto starting_char = lexer.peek();
  694. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  695. // Era
  696. if (all_of(segment, is_char('G'))) {
  697. builder.append("{era}"sv);
  698. if (segment.length() <= 3)
  699. format.era = CalendarPatternStyle::Short;
  700. else if (segment.length() == 4)
  701. format.era = CalendarPatternStyle::Long;
  702. else
  703. format.era = CalendarPatternStyle::Narrow;
  704. }
  705. // Year
  706. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  707. builder.append("{year}"sv);
  708. if (segment.length() == 2)
  709. format.year = CalendarPatternStyle::TwoDigit;
  710. else
  711. format.year = CalendarPatternStyle::Numeric;
  712. }
  713. // Quarter
  714. else if (all_of(segment, is_any_of("qQ"sv))) {
  715. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  716. return {};
  717. }
  718. // Month
  719. else if (all_of(segment, is_any_of("ML"sv))) {
  720. builder.append("{month}"sv);
  721. if (segment.length() == 1)
  722. format.month = CalendarPatternStyle::Numeric;
  723. else if (segment.length() == 2)
  724. format.month = CalendarPatternStyle::TwoDigit;
  725. else if (segment.length() == 3)
  726. format.month = CalendarPatternStyle::Short;
  727. else if (segment.length() == 4)
  728. format.month = CalendarPatternStyle::Long;
  729. else if (segment.length() == 5)
  730. format.month = CalendarPatternStyle::Narrow;
  731. } else if (all_of(segment, is_char('l'))) {
  732. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  733. return {};
  734. }
  735. // Week
  736. else if (all_of(segment, is_any_of("wW"sv))) {
  737. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  738. return {};
  739. }
  740. // Day
  741. else if (all_of(segment, is_char('d'))) {
  742. builder.append("{day}"sv);
  743. if (segment.length() == 1)
  744. format.day = CalendarPatternStyle::Numeric;
  745. else
  746. format.day = CalendarPatternStyle::TwoDigit;
  747. } else if (all_of(segment, is_any_of("DFg"sv))) {
  748. builder.append("{day}"sv);
  749. format.day = CalendarPatternStyle::Numeric;
  750. }
  751. // Weekday
  752. else if (all_of(segment, is_char('E'))) {
  753. builder.append("{weekday}"sv);
  754. if (segment.length() == 4)
  755. format.weekday = CalendarPatternStyle::Long;
  756. else if (segment.length() == 5)
  757. format.weekday = CalendarPatternStyle::Narrow;
  758. else
  759. format.weekday = CalendarPatternStyle::Short;
  760. } else if (all_of(segment, is_any_of("ec"sv))) {
  761. builder.append("{weekday}"sv);
  762. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  763. // pattern styles are not supported by Intl.DateTimeFormat.
  764. if (segment.length() <= 2)
  765. return {};
  766. if (segment.length() == 4)
  767. format.weekday = CalendarPatternStyle::Long;
  768. else if (segment.length() == 5)
  769. format.weekday = CalendarPatternStyle::Narrow;
  770. else
  771. format.weekday = CalendarPatternStyle::Short;
  772. }
  773. // Period
  774. else if (all_of(segment, is_any_of("ab"sv))) {
  775. builder.append("{ampm}"sv);
  776. hour12 = true;
  777. } else if (all_of(segment, is_char('B'))) {
  778. builder.append("{dayPeriod}"sv);
  779. hour12 = true;
  780. if (segment.length() == 4)
  781. format.day_period = CalendarPatternStyle::Long;
  782. else if (segment.length() == 5)
  783. format.day_period = CalendarPatternStyle::Narrow;
  784. else
  785. format.day_period = CalendarPatternStyle::Short;
  786. }
  787. // Hour
  788. else if (all_of(segment, is_any_of("hHKk"sv))) {
  789. builder.append("{hour}"sv);
  790. if ((segment[0] == 'h') || (segment[0] == 'K'))
  791. hour12 = true;
  792. if (segment.length() == 1)
  793. format.hour = CalendarPatternStyle::Numeric;
  794. else
  795. format.hour = CalendarPatternStyle::TwoDigit;
  796. } else if (all_of(segment, is_any_of("jJC"sv))) {
  797. // TR-35 indicates these should not be used.
  798. return {};
  799. }
  800. // Minute
  801. else if (all_of(segment, is_char('m'))) {
  802. builder.append("{minute}"sv);
  803. if (segment.length() == 1)
  804. format.minute = CalendarPatternStyle::Numeric;
  805. else
  806. format.minute = CalendarPatternStyle::TwoDigit;
  807. }
  808. // Second
  809. else if (all_of(segment, is_char('s'))) {
  810. builder.append("{second}"sv);
  811. if (segment.length() == 1)
  812. format.second = CalendarPatternStyle::Numeric;
  813. else
  814. format.second = CalendarPatternStyle::TwoDigit;
  815. } else if (all_of(segment, is_char('S'))) {
  816. builder.append("{fractionalSecondDigits}"sv);
  817. VERIFY(segment.length() <= 3);
  818. format.fractional_second_digits = static_cast<u8>(segment.length());
  819. } else if (all_of(segment, is_char('A'))) {
  820. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  821. return {};
  822. }
  823. // Zone
  824. else if (all_of(segment, is_any_of("zV"sv))) {
  825. builder.append("{timeZoneName}"sv);
  826. if (segment.length() < 4)
  827. format.time_zone_name = CalendarPatternStyle::Short;
  828. else
  829. format.time_zone_name = CalendarPatternStyle::Long;
  830. } else if (all_of(segment, is_any_of("ZOXx"sv))) {
  831. builder.append("{timeZoneName}"sv);
  832. if (segment.length() < 4)
  833. format.time_zone_name = CalendarPatternStyle::ShortOffset;
  834. else
  835. format.time_zone_name = CalendarPatternStyle::LongOffset;
  836. } else if (all_of(segment, is_char('v'))) {
  837. builder.append("{timeZoneName}"sv);
  838. if (segment.length() < 4)
  839. format.time_zone_name = CalendarPatternStyle::ShortGeneric;
  840. else
  841. format.time_zone_name = CalendarPatternStyle::LongGeneric;
  842. }
  843. // Non-patterns
  844. else {
  845. builder.append(segment);
  846. }
  847. }
  848. pattern = builder.build();
  849. if (hour12) {
  850. format.pattern = remove_period_from_pattern(pattern);
  851. format.pattern12 = move(pattern);
  852. } else {
  853. format.pattern = move(pattern);
  854. }
  855. return format;
  856. }
  857. static Optional<CalendarPatternIndexType> parse_date_time_pattern(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  858. {
  859. auto format = parse_date_time_pattern_raw(move(pattern), move(skeleton), locale_data);
  860. if (!format.has_value())
  861. return {};
  862. format->pattern_index = locale_data.unique_strings.ensure(move(format->pattern));
  863. if (format->pattern12.has_value())
  864. format->pattern12_index = locale_data.unique_strings.ensure(format->pattern12.release_value());
  865. return locale_data.unique_patterns.ensure(format.release_value());
  866. }
  867. template<typename... Chars>
  868. static constexpr bool char_is_one_of(char ch, Chars&&... chars)
  869. {
  870. return ((ch == chars) || ...);
  871. }
  872. static void parse_interval_patterns(Calendar& calendar, JsonObject const& interval_formats_object, UnicodeLocaleData& locale_data)
  873. {
  874. // https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats
  875. CalendarRangePatternList range_formats {};
  876. CalendarRangePatternList range12_formats {};
  877. auto name_of_field = [&](char field) {
  878. if (char_is_one_of(field, 'G'))
  879. return Unicode::CalendarRangePattern::Field::Era;
  880. if (char_is_one_of(field, 'y', 'Y', 'u', 'U', 'r'))
  881. return Unicode::CalendarRangePattern::Field::Year;
  882. if (char_is_one_of(field, 'M', 'L'))
  883. return Unicode::CalendarRangePattern::Field::Month;
  884. if (char_is_one_of(field, 'd', 'D', 'F', 'g'))
  885. return Unicode::CalendarRangePattern::Field::Day;
  886. if (char_is_one_of(field, 'a', 'b'))
  887. return Unicode::CalendarRangePattern::Field::AmPm;
  888. if (char_is_one_of(field, 'B'))
  889. return Unicode::CalendarRangePattern::Field::DayPeriod;
  890. if (char_is_one_of(field, 'h', 'H', 'K', 'k'))
  891. return Unicode::CalendarRangePattern::Field::Hour;
  892. if (char_is_one_of(field, 'm'))
  893. return Unicode::CalendarRangePattern::Field::Minute;
  894. if (char_is_one_of(field, 's'))
  895. return Unicode::CalendarRangePattern::Field::Second;
  896. if (char_is_one_of(field, 'S'))
  897. return Unicode::CalendarRangePattern::Field::FractionalSecondDigits;
  898. VERIFY_NOT_REACHED();
  899. };
  900. auto split_default_range_pattern = [&](auto skeleton, auto const& pattern) {
  901. auto start_range_end = pattern.find('}').value() + 1;
  902. auto end_range_begin = pattern.find_last('{').value();
  903. auto start_range = pattern.substring_view(0, start_range_end);
  904. auto separator = pattern.substring_view(start_range_end, end_range_begin - start_range_end);
  905. auto end_range = pattern.substring_view(end_range_begin);
  906. CalendarRangePattern format {};
  907. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  908. format.start_range = locale_data.unique_strings.ensure(start_range);
  909. format.separator = locale_data.unique_strings.ensure(separator);
  910. format.end_range = locale_data.unique_strings.ensure(end_range);
  911. return format;
  912. };
  913. auto split_range_pattern = [&](auto skeleton, auto field, auto const& pattern, auto const& parsed_fields) {
  914. HashMap<StringView, size_t> partitions;
  915. StringView last_partition;
  916. auto begin_index = pattern.find('{');
  917. size_t end_index = 0;
  918. while (begin_index.has_value()) {
  919. end_index = pattern.find('}', *begin_index).value();
  920. auto partition = pattern.substring_view(*begin_index, end_index - *begin_index);
  921. if (partitions.contains(partition))
  922. break;
  923. partitions.set(partition, *begin_index);
  924. last_partition = partition;
  925. begin_index = pattern.find('{', end_index + 1);
  926. }
  927. VERIFY(begin_index.has_value() && !last_partition.is_empty());
  928. auto start_range_end = partitions.get(last_partition).value() + last_partition.length() + 1;
  929. auto start_range = pattern.substring_view(0, start_range_end);
  930. auto separator = pattern.substring_view(start_range_end, *begin_index - start_range_end);
  931. auto end_range = pattern.substring_view(*begin_index);
  932. CalendarRangePattern format {};
  933. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  934. format.field = field;
  935. format.start_range = locale_data.unique_strings.ensure(start_range);
  936. format.separator = locale_data.unique_strings.ensure(separator);
  937. format.end_range = locale_data.unique_strings.ensure(end_range);
  938. format.for_each_calendar_field_zipped_with(parsed_fields, [](auto& format_field, auto const& parsed_field, auto) {
  939. format_field = parsed_field;
  940. });
  941. return format;
  942. };
  943. interval_formats_object.for_each_member([&](auto const& skeleton, auto const& value) {
  944. if (skeleton == "intervalFormatFallback"sv) {
  945. auto range_format = split_default_range_pattern(skeleton, value.as_string());
  946. calendar.default_range_format = locale_data.unique_range_patterns.ensure(move(range_format));
  947. return;
  948. }
  949. value.as_object().for_each_member([&](auto const& field, auto const& pattern) {
  950. if (field.ends_with("alt-variant"sv))
  951. return;
  952. VERIFY(field.length() == 1);
  953. auto name = name_of_field(field[0]);
  954. auto format = parse_date_time_pattern_raw(pattern.as_string(), skeleton, locale_data).release_value();
  955. auto range_format = split_range_pattern(skeleton, name, format.pattern, format);
  956. range_formats.append(locale_data.unique_range_patterns.ensure(move(range_format)));
  957. if (format.pattern12.has_value()) {
  958. auto range12_pattern = split_range_pattern(skeleton, name, *format.pattern12, format);
  959. range12_formats.append(locale_data.unique_range_patterns.ensure(move(range12_pattern)));
  960. } else {
  961. range12_formats.append(range_formats.last());
  962. }
  963. });
  964. });
  965. calendar.range_formats = locale_data.unique_range_pattern_lists.ensure(move(range_formats));
  966. calendar.range12_formats = locale_data.unique_range_pattern_lists.ensure(move(range12_formats));
  967. }
  968. static void generate_missing_patterns(Calendar& calendar, CalendarPatternList& formats, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, UnicodeLocaleData& locale_data)
  969. {
  970. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  971. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  972. auto pattern = locale_data.unique_strings.get(format);
  973. auto time_pattern = locale_data.unique_strings.get(time_format);
  974. auto date_pattern = locale_data.unique_strings.get(date_format);
  975. auto new_pattern = pattern.replace("{0}"sv, time_pattern, ReplaceMode::FirstOnly).replace("{1}"sv, date_pattern, ReplaceMode::FirstOnly);
  976. return locale_data.unique_strings.ensure(move(new_pattern));
  977. };
  978. auto inject_fractional_second_digits = [&](auto format) {
  979. auto pattern = locale_data.unique_strings.get(format);
  980. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv, ReplaceMode::FirstOnly);
  981. return locale_data.unique_strings.ensure(move(new_pattern));
  982. };
  983. auto append_if_unique = [&](auto format) {
  984. auto format_index = locale_data.unique_patterns.ensure(move(format));
  985. if (!formats.contains_slow(format_index))
  986. formats.append(format_index);
  987. };
  988. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  989. for (auto const& format : date_formats)
  990. append_if_unique(format);
  991. for (auto const& format : time_formats) {
  992. append_if_unique(format);
  993. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  994. auto new_format = format;
  995. new_format.fractional_second_digits = 2;
  996. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  997. if (new_format.pattern12_index != 0)
  998. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  999. time_formats_with_fractional_second_digits.append(new_format);
  1000. append_if_unique(move(new_format));
  1001. }
  1002. }
  1003. time_formats.extend(move(time_formats_with_fractional_second_digits));
  1004. for (auto const& date_format : date_formats) {
  1005. auto const& date_time_formats = locale_data.unique_formats.get(calendar.date_time_formats);
  1006. CalendarPatternIndexType date_time_format_index = 0;
  1007. if (date_format.month == Unicode::CalendarPatternStyle::Long) {
  1008. if (date_format.weekday.has_value())
  1009. date_time_format_index = date_time_formats.full_format;
  1010. else
  1011. date_time_format_index = date_time_formats.long_format;
  1012. } else if (date_format.month == Unicode::CalendarPatternStyle::Short) {
  1013. date_time_format_index = date_time_formats.medium_format;
  1014. } else {
  1015. date_time_format_index = date_time_formats.short_format;
  1016. }
  1017. for (auto const& time_format : time_formats) {
  1018. auto format = locale_data.unique_patterns.get(date_time_format_index);
  1019. if (time_format.pattern12_index != 0)
  1020. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  1021. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  1022. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  1023. if (date_field.has_value())
  1024. field = date_field;
  1025. });
  1026. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  1027. if (time_field.has_value())
  1028. field = time_field;
  1029. });
  1030. append_if_unique(move(format));
  1031. }
  1032. }
  1033. }
  1034. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, UnicodeLocaleData& locale_data)
  1035. {
  1036. auto create_symbol_lists = [](size_t size) {
  1037. SymbolList narrow_symbol_list;
  1038. SymbolList short_symbol_list;
  1039. SymbolList long_symbol_list;
  1040. narrow_symbol_list.resize(size);
  1041. short_symbol_list.resize(size);
  1042. long_symbol_list.resize(size);
  1043. return Array<SymbolList, 3> { {
  1044. move(narrow_symbol_list),
  1045. move(short_symbol_list),
  1046. move(long_symbol_list),
  1047. } };
  1048. };
  1049. CalendarSymbolsList symbols_list {};
  1050. auto store_symbol_lists = [&](auto symbol, auto symbol_lists) {
  1051. auto symbol_index = to_underlying(symbol);
  1052. if (symbol_index >= symbols_list.size())
  1053. symbols_list.resize(symbol_index + 1);
  1054. CalendarSymbols symbols {};
  1055. symbols.narrow_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[0]));
  1056. symbols.short_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[1]));
  1057. symbols.long_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[2]));
  1058. auto calendar_symbols_index = locale_data.unique_calendar_symbols.ensure(move(symbols));
  1059. symbols_list[symbol_index] = calendar_symbols_index;
  1060. };
  1061. auto parse_era_symbols = [&](auto const& symbols_object) {
  1062. auto const& narrow_symbols = symbols_object.get("eraNarrow"sv).as_object();
  1063. auto const& short_symbols = symbols_object.get("eraAbbr"sv).as_object();
  1064. auto const& long_symbols = symbols_object.get("eraNames"sv).as_object();
  1065. auto symbol_lists = create_symbol_lists(2);
  1066. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1067. if (auto key_index = key.to_uint(); key_index.has_value())
  1068. symbols[*key_index] = locale_data.unique_strings.ensure(move(symbol));
  1069. };
  1070. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1071. append_symbol(symbol_lists[0], key, value.as_string());
  1072. });
  1073. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1074. append_symbol(symbol_lists[1], key, value.as_string());
  1075. });
  1076. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1077. append_symbol(symbol_lists[2], key, value.as_string());
  1078. });
  1079. store_symbol_lists(Unicode::CalendarSymbol::Era, move(symbol_lists));
  1080. };
  1081. auto parse_month_symbols = [&](auto const& symbols_object) {
  1082. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1083. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1084. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1085. auto symbol_lists = create_symbol_lists(12);
  1086. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1087. auto key_index = key.to_uint().value() - 1;
  1088. symbols[key_index] = locale_data.unique_strings.ensure(move(symbol));
  1089. };
  1090. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1091. append_symbol(symbol_lists[0], key, value.as_string());
  1092. });
  1093. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1094. append_symbol(symbol_lists[1], key, value.as_string());
  1095. });
  1096. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1097. append_symbol(symbol_lists[2], key, value.as_string());
  1098. });
  1099. store_symbol_lists(Unicode::CalendarSymbol::Month, move(symbol_lists));
  1100. };
  1101. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  1102. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1103. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1104. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1105. auto symbol_lists = create_symbol_lists(7);
  1106. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1107. if (key == "sun"sv)
  1108. symbols[to_underlying(Unicode::Weekday::Sunday)] = locale_data.unique_strings.ensure(move(symbol));
  1109. else if (key == "mon"sv)
  1110. symbols[to_underlying(Unicode::Weekday::Monday)] = locale_data.unique_strings.ensure(move(symbol));
  1111. else if (key == "tue"sv)
  1112. symbols[to_underlying(Unicode::Weekday::Tuesday)] = locale_data.unique_strings.ensure(move(symbol));
  1113. else if (key == "wed"sv)
  1114. symbols[to_underlying(Unicode::Weekday::Wednesday)] = locale_data.unique_strings.ensure(move(symbol));
  1115. else if (key == "thu"sv)
  1116. symbols[to_underlying(Unicode::Weekday::Thursday)] = locale_data.unique_strings.ensure(move(symbol));
  1117. else if (key == "fri"sv)
  1118. symbols[to_underlying(Unicode::Weekday::Friday)] = locale_data.unique_strings.ensure(move(symbol));
  1119. else if (key == "sat"sv)
  1120. symbols[to_underlying(Unicode::Weekday::Saturday)] = locale_data.unique_strings.ensure(move(symbol));
  1121. };
  1122. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1123. append_symbol(symbol_lists[0], key, value.as_string());
  1124. });
  1125. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1126. append_symbol(symbol_lists[1], key, value.as_string());
  1127. });
  1128. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1129. append_symbol(symbol_lists[2], key, value.as_string());
  1130. });
  1131. store_symbol_lists(Unicode::CalendarSymbol::Weekday, move(symbol_lists));
  1132. };
  1133. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  1134. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1135. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1136. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1137. auto symbol_lists = create_symbol_lists(10);
  1138. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1139. if (auto day_period = day_period_from_string(key); day_period.has_value())
  1140. symbols[to_underlying(*day_period)] = locale_data.unique_strings.ensure(move(symbol));
  1141. };
  1142. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1143. append_symbol(symbol_lists[0], key, value.as_string());
  1144. });
  1145. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1146. append_symbol(symbol_lists[1], key, value.as_string());
  1147. });
  1148. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1149. append_symbol(symbol_lists[2], key, value.as_string());
  1150. });
  1151. store_symbol_lists(Unicode::CalendarSymbol::DayPeriod, move(symbol_lists));
  1152. };
  1153. parse_era_symbols(calendar_object.get("eras"sv).as_object());
  1154. parse_month_symbols(calendar_object.get("months"sv).as_object().get("format"sv).as_object());
  1155. parse_weekday_symbols(calendar_object.get("days"sv).as_object().get("format"sv).as_object());
  1156. parse_day_period_symbols(calendar_object.get("dayPeriods"sv).as_object().get("format"sv).as_object());
  1157. calendar.symbols = locale_data.unique_calendar_symbols_lists.ensure(move(symbols_list));
  1158. }
  1159. static ErrorOr<void> parse_calendars(String locale_calendars_path, UnicodeLocaleData& locale_data, Locale& locale)
  1160. {
  1161. LexicalPath calendars_path(move(locale_calendars_path));
  1162. if (!calendars_path.basename().starts_with("ca-"sv))
  1163. return {};
  1164. auto calendars = TRY(read_json_file(calendars_path.string()));
  1165. auto const& main_object = calendars.as_object().get("main"sv);
  1166. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  1167. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1168. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  1169. auto parse_patterns = [&](auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) {
  1170. auto parse_pattern = [&](auto name) {
  1171. auto format = patterns_object.get(name);
  1172. auto skeleton = skeletons_object.get(name);
  1173. auto format_index = parse_date_time_pattern(format.as_string(), skeleton.as_string_or(String::empty()), locale_data).value();
  1174. if (patterns)
  1175. patterns->append(locale_data.unique_patterns.get(format_index));
  1176. return format_index;
  1177. };
  1178. CalendarFormat formats {};
  1179. formats.full_format = parse_pattern("full"sv);
  1180. formats.long_format = parse_pattern("long"sv);
  1181. formats.medium_format = parse_pattern("medium"sv);
  1182. formats.short_format = parse_pattern("short"sv);
  1183. return locale_data.unique_formats.ensure(move(formats));
  1184. };
  1185. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  1186. // The generic calendar is not a supported Unicode calendar key, so skip it:
  1187. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  1188. if (calendar_name == "generic"sv)
  1189. return;
  1190. Calendar calendar {};
  1191. CalendarPatternList available_formats {};
  1192. if (!locale_data.calendars.contains_slow(calendar_name))
  1193. locale_data.calendars.append(calendar_name);
  1194. Vector<CalendarPattern> date_formats;
  1195. Vector<CalendarPattern> time_formats;
  1196. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  1197. auto const& date_skeletons_object = value.as_object().get("dateSkeletons"sv);
  1198. calendar.date_formats = parse_patterns(date_formats_object.as_object(), date_skeletons_object.as_object(), &date_formats);
  1199. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  1200. auto const& time_skeletons_object = value.as_object().get("timeSkeletons"sv);
  1201. calendar.time_formats = parse_patterns(time_formats_object.as_object(), time_skeletons_object.as_object(), &time_formats);
  1202. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  1203. calendar.date_time_formats = parse_patterns(date_time_formats_object.as_object(), JsonObject {}, nullptr);
  1204. auto const& available_formats_object = date_time_formats_object.as_object().get("availableFormats"sv);
  1205. available_formats_object.as_object().for_each_member([&](auto const& skeleton, JsonValue const& pattern) {
  1206. auto pattern_index = parse_date_time_pattern(pattern.as_string(), skeleton, locale_data);
  1207. if (!pattern_index.has_value())
  1208. return;
  1209. auto const& format = locale_data.unique_patterns.get(*pattern_index);
  1210. if (format.contains_only_date_fields())
  1211. date_formats.append(format);
  1212. else if (format.contains_only_time_fields())
  1213. time_formats.append(format);
  1214. if (!available_formats.contains_slow(*pattern_index))
  1215. available_formats.append(*pattern_index);
  1216. });
  1217. auto const& interval_formats_object = date_time_formats_object.as_object().get("intervalFormats"sv);
  1218. parse_interval_patterns(calendar, interval_formats_object.as_object(), locale_data);
  1219. generate_missing_patterns(calendar, available_formats, move(date_formats), move(time_formats), locale_data);
  1220. parse_calendar_symbols(calendar, value.as_object(), locale_data);
  1221. calendar.available_formats = locale_data.unique_pattern_lists.ensure(move(available_formats));
  1222. locale.calendars.set(calendar_name, locale_data.unique_calendars.ensure(move(calendar)));
  1223. });
  1224. return {};
  1225. }
  1226. static ErrorOr<void> parse_time_zone_names(String locale_time_zone_names_path, UnicodeLocaleData& locale_data, Locale& locale)
  1227. {
  1228. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  1229. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  1230. auto time_zone_names = TRY(read_json_file(time_zone_names_path.string()));
  1231. auto const& main_object = time_zone_names.as_object().get("main"sv);
  1232. auto const& locale_object = main_object.as_object().get(time_zone_names_path.parent().basename());
  1233. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1234. auto const& time_zone_names_object = dates_object.as_object().get("timeZoneNames"sv);
  1235. auto const& meta_zone_object = time_zone_names_object.as_object().get("metazone"sv);
  1236. auto const& hour_format_string = time_zone_names_object.as_object().get("hourFormat"sv);
  1237. auto const& gmt_format_string = time_zone_names_object.as_object().get("gmtFormat"sv);
  1238. auto const& gmt_zero_format_string = time_zone_names_object.as_object().get("gmtZeroFormat"sv);
  1239. if (meta_zone_object.is_null())
  1240. return {};
  1241. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object, StringView key) -> Optional<StringIndexType> {
  1242. auto const& names = meta_zone_object.get(type);
  1243. if (!names.is_object())
  1244. return {};
  1245. auto const& name = names.as_object().get(key);
  1246. if (name.is_string())
  1247. return locale_data.unique_strings.ensure(name.as_string());
  1248. return {};
  1249. };
  1250. auto parse_hour_format = [&](auto const& format, auto& time_zone_formats) {
  1251. auto hour_formats = format.split_view(';');
  1252. auto hour_format_ahead_start = hour_formats[0].find('H').value();
  1253. auto separator_ahead_start = hour_formats[0].find_last('H').value() + 1;
  1254. auto separator_ahead_end = hour_formats[0].find('m').value();
  1255. auto hour_format_behind_start = hour_formats[1].find('H').value();
  1256. auto separator_behind_start = hour_formats[1].find_last('H').value() + 1;
  1257. auto separator_behind_end = hour_formats[1].find('m').value();
  1258. auto symbol_ahead_sign = hour_formats[0].substring_view(0, hour_format_ahead_start);
  1259. auto symbol_ahead_separator = hour_formats[0].substring_view(separator_ahead_start, separator_ahead_end - separator_ahead_start);
  1260. auto symbol_behind_sign = hour_formats[1].substring_view(0, hour_format_behind_start);
  1261. auto symbol_behind_separator = hour_formats[1].substring_view(separator_behind_start, separator_behind_end - separator_behind_start);
  1262. time_zone_formats.symbol_ahead_sign = locale_data.unique_strings.ensure(symbol_ahead_sign);
  1263. time_zone_formats.symbol_ahead_separator = locale_data.unique_strings.ensure(symbol_ahead_separator);
  1264. time_zone_formats.symbol_behind_sign = locale_data.unique_strings.ensure(symbol_behind_sign);
  1265. time_zone_formats.symbol_behind_separator = locale_data.unique_strings.ensure(symbol_behind_separator);
  1266. };
  1267. TimeZoneNamesList time_zones;
  1268. TimeZoneFormat time_zone_formats {};
  1269. parse_hour_format(hour_format_string.as_string(), time_zone_formats);
  1270. time_zone_formats.gmt_format = locale_data.unique_strings.ensure(gmt_format_string.as_string());
  1271. time_zone_formats.gmt_zero_format = locale_data.unique_strings.ensure(gmt_zero_format_string.as_string());
  1272. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  1273. auto golden_zones = locale_data.meta_zones.find(meta_zone);
  1274. if (golden_zones == locale_data.meta_zones.end())
  1275. return;
  1276. TimeZoneNames time_zone_names {};
  1277. if (auto name = parse_name("long"sv, meta_zone_object, "standard"sv); name.has_value())
  1278. time_zone_names.long_standard_name = name.value();
  1279. if (auto name = parse_name("short"sv, meta_zone_object, "standard"sv); name.has_value())
  1280. time_zone_names.short_standard_name = name.value();
  1281. if (auto name = parse_name("long"sv, meta_zone_object, "daylight"sv); name.has_value())
  1282. time_zone_names.long_daylight_name = name.value();
  1283. if (auto name = parse_name("short"sv, meta_zone_object, "daylight"sv); name.has_value())
  1284. time_zone_names.short_daylight_name = name.value();
  1285. if (auto name = parse_name("long"sv, meta_zone_object, "generic"sv); name.has_value())
  1286. time_zone_names.long_generic_name = name.value();
  1287. if (auto name = parse_name("short"sv, meta_zone_object, "generic"sv); name.has_value())
  1288. time_zone_names.short_generic_name = name.value();
  1289. auto time_zone_index = locale_data.unique_time_zones.ensure(move(time_zone_names));
  1290. for (auto golden_zone : golden_zones->value) {
  1291. auto time_zone = to_underlying(golden_zone);
  1292. if (time_zone >= time_zones.size())
  1293. time_zones.resize(time_zone + 1);
  1294. time_zones[time_zone] = time_zone_index;
  1295. }
  1296. };
  1297. meta_zone_object.as_object().for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  1298. parse_time_zone(meta_zone, value.as_object());
  1299. });
  1300. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  1301. auto const& zone_object = time_zone_names_object.as_object().get("zone"sv);
  1302. auto const& etc_object = zone_object.as_object().get("Etc"sv);
  1303. auto const& utc_object = etc_object.as_object().get("UTC"sv);
  1304. parse_time_zone("UTC"sv, utc_object.as_object());
  1305. locale.time_zones = locale_data.unique_time_zone_lists.ensure(move(time_zones));
  1306. locale.time_zone_formats = locale_data.unique_time_zone_formats.ensure(move(time_zone_formats));
  1307. return {};
  1308. }
  1309. static ErrorOr<void> parse_day_periods(String core_path, UnicodeLocaleData& locale_data)
  1310. {
  1311. // https://unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rule_Sets
  1312. LexicalPath day_periods_path(move(core_path));
  1313. day_periods_path = day_periods_path.append("supplemental"sv);
  1314. day_periods_path = day_periods_path.append("dayPeriods.json"sv);
  1315. auto locale_day_periods = TRY(read_json_file(day_periods_path.string()));
  1316. auto const& supplemental_object = locale_day_periods.as_object().get("supplemental"sv);
  1317. auto const& day_periods_object = supplemental_object.as_object().get("dayPeriodRuleSet"sv);
  1318. auto parse_hour = [](auto const& time) {
  1319. auto hour_end_index = time.find(':').value();
  1320. // The times are in the form "hours:minutes", but we only need the hour segment.
  1321. // TR-35 explicitly states that minutes other than :00 must not be used.
  1322. VERIFY(time.substring_view(hour_end_index) == ":00"sv);
  1323. auto hour = time.substring_view(0, hour_end_index);
  1324. return hour.template to_uint<u8>().value();
  1325. };
  1326. auto parse_day_period = [&](auto const& symbol, auto const& ranges) -> Optional<DayPeriod> {
  1327. auto day_period = day_period_from_string(symbol);
  1328. if (!day_period.has_value())
  1329. return {};
  1330. auto begin = parse_hour(ranges.get("_from"sv).as_string());
  1331. auto end = parse_hour(ranges.get("_before"sv).as_string());
  1332. return DayPeriod { *day_period, begin, end };
  1333. };
  1334. day_periods_object.as_object().for_each_member([&](auto const& language, JsonValue const& value) {
  1335. auto locale = locale_data.locales.find(language);
  1336. if (locale == locale_data.locales.end())
  1337. return;
  1338. DayPeriodList day_periods;
  1339. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& ranges) {
  1340. if (auto day_period = parse_day_period(symbol, ranges.as_object()); day_period.has_value()) {
  1341. auto day_period_index = locale_data.unique_day_periods.ensure(day_period.release_value());
  1342. day_periods.append(day_period_index);
  1343. }
  1344. });
  1345. locale->value.day_periods = locale_data.unique_day_period_lists.ensure(move(day_periods));
  1346. });
  1347. return {};
  1348. }
  1349. static ErrorOr<void> parse_all_locales(String core_path, String dates_path, UnicodeLocaleData& locale_data)
  1350. {
  1351. TRY(parse_hour_cycles(core_path, locale_data));
  1352. TRY(parse_week_data(core_path, locale_data));
  1353. TRY(parse_meta_zones(core_path, locale_data));
  1354. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  1355. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  1356. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  1357. StringBuilder builder;
  1358. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  1359. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  1360. builder.appendff("-{}", script);
  1361. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  1362. builder.appendff("-{}", region);
  1363. return builder.build();
  1364. };
  1365. while (dates_iterator.has_next()) {
  1366. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  1367. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  1368. auto language = TRY(remove_variants_from_path(dates_path));
  1369. auto& locale = locale_data.locales.ensure(language);
  1370. while (calendars_iterator.has_next()) {
  1371. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  1372. TRY(parse_calendars(move(calendars_path), locale_data, locale));
  1373. }
  1374. TRY(parse_time_zone_names(move(dates_path), locale_data, locale));
  1375. }
  1376. TRY(parse_day_periods(move(core_path), locale_data));
  1377. return {};
  1378. }
  1379. static String format_identifier(StringView owner, String identifier)
  1380. {
  1381. identifier = identifier.replace("-"sv, "_"sv, ReplaceMode::All);
  1382. identifier = identifier.replace("/"sv, "_"sv, ReplaceMode::All);
  1383. if (all_of(identifier, is_ascii_digit))
  1384. return String::formatted("{}_{}", owner[0], identifier);
  1385. if (is_ascii_lower_alpha(identifier[0]))
  1386. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  1387. return identifier;
  1388. }
  1389. static ErrorOr<void> generate_unicode_locale_header(Core::Stream::BufferedFile& file, UnicodeLocaleData& locale_data)
  1390. {
  1391. StringBuilder builder;
  1392. SourceGenerator generator { builder };
  1393. generator.append(R"~~~(
  1394. #pragma once
  1395. #include <AK/Types.h>
  1396. namespace Unicode {
  1397. )~~~");
  1398. generate_enum(generator, format_identifier, "Calendar"sv, {}, locale_data.calendars);
  1399. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, locale_data.hour_cycle_regions);
  1400. generate_enum(generator, format_identifier, "MinimumDaysRegion"sv, {}, locale_data.minimum_days_regions);
  1401. generate_enum(generator, format_identifier, "FirstDayRegion"sv, {}, locale_data.first_day_regions);
  1402. generate_enum(generator, format_identifier, "WeekendStartRegion"sv, {}, locale_data.weekend_start_regions);
  1403. generate_enum(generator, format_identifier, "WeekendEndRegion"sv, {}, locale_data.weekend_end_regions);
  1404. generator.append(R"~~~(
  1405. }
  1406. )~~~");
  1407. TRY(file.write(generator.as_string_view().bytes()));
  1408. return {};
  1409. }
  1410. static ErrorOr<void> generate_unicode_locale_implementation(Core::Stream::BufferedFile& file, UnicodeLocaleData& locale_data)
  1411. {
  1412. StringBuilder builder;
  1413. SourceGenerator generator { builder };
  1414. generator.set("string_index_type"sv, s_string_index_type);
  1415. generator.set("calendar_pattern_index_type"sv, s_calendar_pattern_index_type);
  1416. generator.set("calendar_pattern_list_index_type"sv, s_calendar_pattern_list_index_type);
  1417. generator.set("calendar_range_pattern_index_type"sv, s_calendar_range_pattern_index_type);
  1418. generator.set("calendar_range_pattern_list_index_type"sv, s_calendar_range_pattern_list_index_type);
  1419. generator.set("calendar_format_index_type"sv, s_calendar_format_index_type);
  1420. generator.set("symbol_list_index_type"sv, s_symbol_list_index_type);
  1421. generator.set("calendar_symbols_index_type"sv, s_calendar_symbols_index_type);
  1422. generator.set("calendar_symbols_list_index_type"sv, s_calendar_symbols_list_index_type);
  1423. generator.set("calendar_index_type"sv, s_calendar_index_type);
  1424. generator.set("time_zone_index_type"sv, s_time_zone_index_type);
  1425. generator.set("time_zone_list_index_type"sv, s_time_zone_list_index_type);
  1426. generator.set("day_period_index_type"sv, s_day_period_index_type);
  1427. generator.set("day_period_list_index_type"sv, s_day_period_list_index_type);
  1428. generator.append(R"~~~(
  1429. #include <AK/Array.h>
  1430. #include <AK/BinarySearch.h>
  1431. #include <AK/Optional.h>
  1432. #include <AK/StringView.h>
  1433. #include <LibTimeZone/TimeZone.h>
  1434. #include <LibUnicode/DateTimeFormat.h>
  1435. #include <LibUnicode/Locale.h>
  1436. #include <LibUnicode/UnicodeDateTimeFormat.h>
  1437. #include <LibUnicode/UnicodeLocale.h>
  1438. namespace Unicode {
  1439. )~~~");
  1440. locale_data.unique_strings.generate(generator);
  1441. generator.append(R"~~~(
  1442. template <typename SourceType, typename TargetType>
  1443. static void convert_calendar_fields(SourceType const& source, TargetType& target)
  1444. {
  1445. if (source.era != -1)
  1446. target.era = static_cast<CalendarPatternStyle>(source.era);
  1447. if (source.year != -1)
  1448. target.year = static_cast<CalendarPatternStyle>(source.year);
  1449. if (source.month != -1)
  1450. target.month = static_cast<CalendarPatternStyle>(source.month);
  1451. if (source.weekday != -1)
  1452. target.weekday = static_cast<CalendarPatternStyle>(source.weekday);
  1453. if (source.day != -1)
  1454. target.day = static_cast<CalendarPatternStyle>(source.day);
  1455. if (source.day_period != -1)
  1456. target.day_period = static_cast<CalendarPatternStyle>(source.day_period);
  1457. if (source.hour != -1)
  1458. target.hour = static_cast<CalendarPatternStyle>(source.hour);
  1459. if (source.minute != -1)
  1460. target.minute = static_cast<CalendarPatternStyle>(source.minute);
  1461. if (source.second != -1)
  1462. target.second = static_cast<CalendarPatternStyle>(source.second);
  1463. if (source.fractional_second_digits != -1)
  1464. target.fractional_second_digits = static_cast<u8>(source.fractional_second_digits);
  1465. if (source.time_zone_name != -1)
  1466. target.time_zone_name = static_cast<CalendarPatternStyle>(source.time_zone_name);
  1467. }
  1468. struct CalendarPatternImpl {
  1469. CalendarPattern to_unicode_calendar_pattern() const {
  1470. CalendarPattern calendar_pattern {};
  1471. calendar_pattern.skeleton = s_string_list[skeleton];
  1472. calendar_pattern.pattern = s_string_list[pattern];
  1473. if (pattern12 != 0)
  1474. calendar_pattern.pattern12 = s_string_list[pattern12];
  1475. convert_calendar_fields(*this, calendar_pattern);
  1476. return calendar_pattern;
  1477. }
  1478. @string_index_type@ skeleton { 0 };
  1479. @string_index_type@ pattern { 0 };
  1480. @string_index_type@ pattern12 { 0 };
  1481. i8 era { -1 };
  1482. i8 year { -1 };
  1483. i8 month { -1 };
  1484. i8 weekday { -1 };
  1485. i8 day { -1 };
  1486. i8 day_period { -1 };
  1487. i8 hour { -1 };
  1488. i8 minute { -1 };
  1489. i8 second { -1 };
  1490. i8 fractional_second_digits { -1 };
  1491. i8 time_zone_name { -1 };
  1492. };
  1493. struct CalendarRangePatternImpl {
  1494. CalendarRangePattern to_unicode_calendar_range_pattern() const {
  1495. CalendarRangePattern calendar_range_pattern {};
  1496. if (field != -1)
  1497. calendar_range_pattern.field = static_cast<CalendarRangePattern::Field>(field);
  1498. calendar_range_pattern.start_range = s_string_list[start_range];
  1499. calendar_range_pattern.separator = s_string_list[separator];
  1500. calendar_range_pattern.end_range = s_string_list[end_range];
  1501. convert_calendar_fields(*this, calendar_range_pattern);
  1502. return calendar_range_pattern;
  1503. }
  1504. @string_index_type@ skeleton { 0 };
  1505. i8 field { -1 };
  1506. @string_index_type@ start_range { 0 };
  1507. @string_index_type@ separator { 0 };
  1508. @string_index_type@ end_range { 0 };
  1509. i8 era { -1 };
  1510. i8 year { -1 };
  1511. i8 month { -1 };
  1512. i8 weekday { -1 };
  1513. i8 day { -1 };
  1514. i8 day_period { -1 };
  1515. i8 hour { -1 };
  1516. i8 minute { -1 };
  1517. i8 second { -1 };
  1518. i8 fractional_second_digits { -1 };
  1519. i8 time_zone_name { -1 };
  1520. };
  1521. )~~~");
  1522. locale_data.unique_patterns.generate(generator, "CalendarPatternImpl"sv, "s_calendar_patterns"sv, 10);
  1523. locale_data.unique_pattern_lists.generate(generator, s_calendar_pattern_index_type, "s_calendar_pattern_lists"sv);
  1524. locale_data.unique_range_patterns.generate(generator, "CalendarRangePatternImpl"sv, "s_calendar_range_patterns"sv, 10);
  1525. locale_data.unique_range_pattern_lists.generate(generator, s_calendar_range_pattern_index_type, "s_calendar_range_pattern_lists"sv);
  1526. generator.append(R"~~~(
  1527. struct CalendarFormatImpl {
  1528. CalendarFormat to_unicode_calendar_format() const {
  1529. CalendarFormat calendar_format {};
  1530. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  1531. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  1532. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  1533. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  1534. return calendar_format;
  1535. }
  1536. @calendar_pattern_index_type@ full_format { 0 };
  1537. @calendar_pattern_index_type@ long_format { 0 };
  1538. @calendar_pattern_index_type@ medium_format { 0 };
  1539. @calendar_pattern_index_type@ short_format { 0 };
  1540. };
  1541. struct CalendarSymbols {
  1542. @symbol_list_index_type@ narrow_symbols { 0 };
  1543. @symbol_list_index_type@ short_symbols { 0 };
  1544. @symbol_list_index_type@ long_symbols { 0 };
  1545. };
  1546. struct CalendarData {
  1547. @calendar_format_index_type@ date_formats { 0 };
  1548. @calendar_format_index_type@ time_formats { 0 };
  1549. @calendar_format_index_type@ date_time_formats { 0 };
  1550. @calendar_pattern_list_index_type@ available_formats { 0 };
  1551. @calendar_range_pattern_index_type@ default_range_format { 0 };
  1552. @calendar_range_pattern_list_index_type@ range_formats { 0 };
  1553. @calendar_range_pattern_list_index_type@ range12_formats { 0 };
  1554. @calendar_symbols_list_index_type@ symbols { 0 };
  1555. };
  1556. struct TimeZoneNames {
  1557. @string_index_type@ short_standard_name { 0 };
  1558. @string_index_type@ long_standard_name { 0 };
  1559. @string_index_type@ short_daylight_name { 0 };
  1560. @string_index_type@ long_daylight_name { 0 };
  1561. @string_index_type@ short_generic_name { 0 };
  1562. @string_index_type@ long_generic_name { 0 };
  1563. };
  1564. struct TimeZoneFormatImpl {
  1565. TimeZoneFormat to_time_zone_format() const {
  1566. TimeZoneFormat time_zone_format {};
  1567. time_zone_format.symbol_ahead_sign = s_string_list[symbol_ahead_sign];
  1568. time_zone_format.symbol_ahead_separator = s_string_list[symbol_ahead_separator];
  1569. time_zone_format.symbol_behind_sign = s_string_list[symbol_behind_sign];
  1570. time_zone_format.symbol_behind_separator = s_string_list[symbol_behind_separator];
  1571. time_zone_format.gmt_format = s_string_list[gmt_format];
  1572. time_zone_format.gmt_zero_format = s_string_list[gmt_zero_format];
  1573. return time_zone_format;
  1574. }
  1575. @string_index_type@ symbol_ahead_sign { 0 };
  1576. @string_index_type@ symbol_ahead_separator { 0 };
  1577. @string_index_type@ symbol_behind_sign { 0 };
  1578. @string_index_type@ symbol_behind_separator { 0 };
  1579. @string_index_type@ gmt_format { 0 };
  1580. @string_index_type@ gmt_zero_format { 0 };
  1581. };
  1582. struct DayPeriodData {
  1583. u8 day_period { 0 };
  1584. u8 begin { 0 };
  1585. u8 end { 0 };
  1586. };
  1587. )~~~");
  1588. locale_data.unique_formats.generate(generator, "CalendarFormatImpl"sv, "s_calendar_formats"sv, 10);
  1589. locale_data.unique_symbol_lists.generate(generator, s_string_index_type, "s_symbol_lists"sv);
  1590. locale_data.unique_calendar_symbols.generate(generator, "CalendarSymbols"sv, "s_calendar_symbols"sv, 10);
  1591. locale_data.unique_calendar_symbols_lists.generate(generator, s_calendar_symbols_index_type, "s_calendar_symbol_lists"sv);
  1592. locale_data.unique_calendars.generate(generator, "CalendarData"sv, "s_calendars"sv, 10);
  1593. locale_data.unique_time_zones.generate(generator, "TimeZoneNames"sv, "s_time_zones"sv, 30);
  1594. locale_data.unique_time_zone_lists.generate(generator, s_time_zone_index_type, "s_time_zone_lists"sv);
  1595. locale_data.unique_time_zone_formats.generate(generator, "TimeZoneFormatImpl"sv, "s_time_zone_formats"sv, 30);
  1596. locale_data.unique_day_periods.generate(generator, "DayPeriodData"sv, "s_day_periods"sv, 30);
  1597. locale_data.unique_day_period_lists.generate(generator, s_day_period_index_type, "s_day_period_lists"sv);
  1598. locale_data.unique_hour_cycle_lists.generate(generator, "u8"sv, "s_hour_cycle_lists"sv);
  1599. auto append_calendars = [&](String name, auto const& calendars) {
  1600. generator.set("name", name);
  1601. generator.set("size", String::number(calendars.size()));
  1602. generator.append(R"~~~(
  1603. static constexpr Array<@calendar_index_type@, @size@> @name@ { {)~~~");
  1604. bool first = true;
  1605. for (auto const& calendar_key : locale_data.calendars) {
  1606. auto calendar = calendars.find(calendar_key)->value;
  1607. generator.append(first ? " "sv : ", "sv);
  1608. generator.append(String::number(calendar));
  1609. first = false;
  1610. }
  1611. generator.append(" } };");
  1612. };
  1613. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  1614. generator.set("type", type);
  1615. generator.set("name", name);
  1616. generator.set("size", String::number(keys.size()));
  1617. generator.append(R"~~~(
  1618. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1619. bool first = true;
  1620. for (auto const& key : keys) {
  1621. auto const& value = map.find(key)->value;
  1622. auto mapping = mapping_getter(value);
  1623. generator.append(first ? " "sv : ", "sv);
  1624. generator.append(String::number(mapping));
  1625. first = false;
  1626. }
  1627. generator.append(" } };");
  1628. };
  1629. auto locales = locale_data.locales.keys();
  1630. quick_sort(locales);
  1631. generate_mapping(generator, locale_data.locales, s_calendar_index_type, "s_locale_calendars"sv, "s_calendars_{}"sv, format_identifier, [&](auto const& name, auto const& value) { append_calendars(name, value.calendars); });
  1632. append_mapping(locales, locale_data.locales, s_time_zone_index_type, "s_locale_time_zones"sv, [](auto const& locale) { return locale.time_zones; });
  1633. append_mapping(locales, locale_data.locales, s_time_zone_format_index_type, "s_locale_time_zone_formats"sv, [](auto const& locale) { return locale.time_zone_formats; });
  1634. append_mapping(locales, locale_data.locales, s_day_period_index_type, "s_locale_day_periods"sv, [](auto const& locale) { return locale.day_periods; });
  1635. append_mapping(locale_data.hour_cycle_regions, locale_data.hour_cycles, s_hour_cycle_list_index_type, "s_hour_cycles"sv, [](auto const& hour_cycles) { return hour_cycles; });
  1636. append_mapping(locale_data.minimum_days_regions, locale_data.minimum_days, "u8"sv, "s_minimum_days"sv, [](auto minimum_days) { return minimum_days; });
  1637. append_mapping(locale_data.first_day_regions, locale_data.first_day, "u8"sv, "s_first_day"sv, [](auto first_day) { return to_underlying(first_day); });
  1638. append_mapping(locale_data.weekend_start_regions, locale_data.weekend_start, "u8"sv, "s_weekend_start"sv, [](auto weekend_start) { return to_underlying(weekend_start); });
  1639. append_mapping(locale_data.weekend_end_regions, locale_data.weekend_end, "u8"sv, "s_weekend_end"sv, [](auto weekend_end) { return to_underlying(weekend_end); });
  1640. generator.append("\n");
  1641. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1642. HashValueMap<String> hashes;
  1643. hashes.ensure_capacity(values.size());
  1644. for (auto const& value : values)
  1645. hashes.set(value.hash(), format_identifier(enum_title, value));
  1646. for (auto const& alias : aliases)
  1647. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1648. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1649. };
  1650. append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, locale_data.hour_cycle_regions);
  1651. append_from_string("MinimumDaysRegion"sv, "minimum_days_region"sv, locale_data.minimum_days_regions);
  1652. append_from_string("FirstDayRegion"sv, "first_day_region"sv, locale_data.first_day_regions);
  1653. append_from_string("WeekendStartRegion"sv, "weekend_start_region"sv, locale_data.weekend_start_regions);
  1654. append_from_string("WeekendEndRegion"sv, "weekend_end_region"sv, locale_data.weekend_end_regions);
  1655. generator.append(R"~~~(
  1656. static Optional<Calendar> keyword_to_calendar(KeywordCalendar keyword)
  1657. {
  1658. switch (keyword) {)~~~");
  1659. for (auto const& calendar : locale_data.calendars) {
  1660. generator.set("name"sv, format_identifier({}, calendar));
  1661. generator.append(R"~~~(
  1662. case KeywordCalendar::@name@:
  1663. return Calendar::@name@;)~~~");
  1664. }
  1665. generator.append(R"~~~(
  1666. default:
  1667. return {};
  1668. }
  1669. }
  1670. Vector<HourCycle> get_regional_hour_cycles(StringView region)
  1671. {
  1672. auto region_value = hour_cycle_region_from_string(region);
  1673. if (!region_value.has_value())
  1674. return {};
  1675. auto region_index = to_underlying(*region_value);
  1676. auto regional_hour_cycles_index = s_hour_cycles.at(region_index);
  1677. auto const& regional_hour_cycles = s_hour_cycle_lists.at(regional_hour_cycles_index);
  1678. Vector<HourCycle> hour_cycles;
  1679. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1680. for (auto hour_cycle : regional_hour_cycles)
  1681. hour_cycles.unchecked_append(static_cast<HourCycle>(hour_cycle));
  1682. return hour_cycles;
  1683. }
  1684. )~~~");
  1685. auto append_regional_lookup = [&](StringView return_type, StringView lookup_type) {
  1686. generator.set("return_type", return_type);
  1687. generator.set("lookup_type", lookup_type);
  1688. generator.append(R"~~~(
  1689. Optional<@return_type@> get_regional_@lookup_type@(StringView region)
  1690. {
  1691. auto region_value = @lookup_type@_region_from_string(region);
  1692. if (!region_value.has_value())
  1693. return {};
  1694. auto region_index = to_underlying(*region_value);
  1695. auto @lookup_type@ = s_@lookup_type@.at(region_index);
  1696. return static_cast<@return_type@>(@lookup_type@);
  1697. }
  1698. )~~~");
  1699. };
  1700. append_regional_lookup("u8"sv, "minimum_days"sv);
  1701. append_regional_lookup("Unicode::Weekday"sv, "first_day"sv);
  1702. append_regional_lookup("Unicode::Weekday"sv, "weekend_start"sv);
  1703. append_regional_lookup("Unicode::Weekday"sv, "weekend_end"sv);
  1704. generator.append(R"~~~(
  1705. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1706. {
  1707. auto locale_value = locale_from_string(locale);
  1708. if (!locale_value.has_value())
  1709. return nullptr;
  1710. auto calendar_keyword = keyword_ca_from_string(calendar);
  1711. if (!calendar_keyword.has_value())
  1712. return {};
  1713. auto calendar_value = keyword_to_calendar(*calendar_keyword);
  1714. if (!calendar_value.has_value())
  1715. return {};
  1716. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1717. size_t calendar_index = to_underlying(*calendar_value);
  1718. auto const& calendar_indices = s_locale_calendars.at(locale_index);
  1719. calendar_index = calendar_indices[calendar_index];
  1720. return &s_calendars[calendar_index];
  1721. }
  1722. Optional<CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1723. {
  1724. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1725. auto const& formats = s_calendar_formats.at(data->date_formats);
  1726. return formats.to_unicode_calendar_format();
  1727. }
  1728. return {};
  1729. }
  1730. Optional<CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1731. {
  1732. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1733. auto const& formats = s_calendar_formats.at(data->time_formats);
  1734. return formats.to_unicode_calendar_format();
  1735. }
  1736. return {};
  1737. }
  1738. Optional<CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1739. {
  1740. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1741. auto const& formats = s_calendar_formats.at(data->date_time_formats);
  1742. return formats.to_unicode_calendar_format();
  1743. }
  1744. return {};
  1745. }
  1746. Vector<CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1747. {
  1748. Vector<CalendarPattern> result {};
  1749. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1750. auto const& available_formats = s_calendar_pattern_lists.at(data->available_formats);
  1751. result.ensure_capacity(available_formats.size());
  1752. for (auto const& format : available_formats)
  1753. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1754. }
  1755. return result;
  1756. }
  1757. Optional<CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar)
  1758. {
  1759. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1760. auto const& pattern = s_calendar_range_patterns[data->default_range_format];
  1761. return pattern.to_unicode_calendar_range_pattern();
  1762. }
  1763. return {};
  1764. }
  1765. Vector<CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton)
  1766. {
  1767. Vector<CalendarRangePattern> result {};
  1768. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1769. auto const& range_formats = s_calendar_range_pattern_lists.at(data->range_formats);
  1770. for (auto format : range_formats) {
  1771. auto const& pattern = s_calendar_range_patterns[format];
  1772. if (skeleton == s_string_list[pattern.skeleton])
  1773. result.append(pattern.to_unicode_calendar_range_pattern());
  1774. }
  1775. }
  1776. return result;
  1777. }
  1778. Vector<CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton)
  1779. {
  1780. Vector<CalendarRangePattern> result {};
  1781. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1782. auto const& range12_formats = s_calendar_range_pattern_lists.at(data->range12_formats);
  1783. for (auto format : range12_formats) {
  1784. auto const& pattern = s_calendar_range_patterns[format];
  1785. if (skeleton == s_string_list[pattern.skeleton])
  1786. result.append(pattern.to_unicode_calendar_range_pattern());
  1787. }
  1788. }
  1789. return result;
  1790. }
  1791. static Span<@string_index_type@ const> find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1792. {
  1793. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1794. auto const& symbols_list = s_calendar_symbol_lists[data->symbols];
  1795. auto symbol_index = to_underlying(symbol);
  1796. auto calendar_symbols_index = symbols_list.at(symbol_index);
  1797. auto const& symbols = s_calendar_symbols.at(calendar_symbols_index);
  1798. @symbol_list_index_type@ symbol_list_index = 0;
  1799. switch (style) {
  1800. case CalendarPatternStyle::Narrow:
  1801. symbol_list_index = symbols.narrow_symbols;
  1802. break;
  1803. case CalendarPatternStyle::Short:
  1804. symbol_list_index = symbols.short_symbols;
  1805. break;
  1806. case CalendarPatternStyle::Long:
  1807. symbol_list_index = symbols.long_symbols;
  1808. break;
  1809. default:
  1810. VERIFY_NOT_REACHED();
  1811. }
  1812. return s_symbol_lists.at(symbol_list_index);
  1813. }
  1814. return {};
  1815. }
  1816. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Era value)
  1817. {
  1818. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1819. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1820. return s_string_list[symbols.at(value_index)];
  1821. return {};
  1822. }
  1823. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Month value)
  1824. {
  1825. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1826. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1827. return s_string_list[symbols.at(value_index)];
  1828. return {};
  1829. }
  1830. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Weekday value)
  1831. {
  1832. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1833. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1834. return s_string_list[symbols.at(value_index)];
  1835. return {};
  1836. }
  1837. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, DayPeriod value)
  1838. {
  1839. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1840. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1841. return s_string_list[symbols.at(value_index)];
  1842. return {};
  1843. }
  1844. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour)
  1845. {
  1846. auto locale_value = locale_from_string(locale);
  1847. if (!locale_value.has_value())
  1848. return {};
  1849. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1850. auto day_periods_index = s_locale_day_periods[locale_index];
  1851. auto day_periods = s_day_period_lists[day_periods_index];
  1852. for (auto day_period_index : day_periods) {
  1853. auto day_period = s_day_periods[day_period_index];
  1854. auto h = hour;
  1855. if (day_period.begin > day_period.end) {
  1856. day_period.end += 24;
  1857. h += 24;
  1858. }
  1859. if ((day_period.begin <= h) && (h < day_period.end)) {
  1860. auto period = static_cast<DayPeriod>(day_period.day_period);
  1861. return get_calendar_day_period_symbol(locale, calendar, style, period);
  1862. }
  1863. }
  1864. // Fallback to fixed periods if the locale does not have flexible day periods.
  1865. // TR-35 states that the meaning of AM and PM does not change with locale.
  1866. if (hour < 12)
  1867. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::AM);
  1868. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::PM);
  1869. }
  1870. Optional<TimeZoneFormat> get_time_zone_format(StringView locale)
  1871. {
  1872. auto locale_value = locale_from_string(locale);
  1873. if (!locale_value.has_value())
  1874. return {};
  1875. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1876. auto time_zone_format_index = s_locale_time_zone_formats.at(locale_index);
  1877. auto const& time_zone_format = s_time_zone_formats.at(time_zone_format_index);
  1878. return time_zone_format.to_time_zone_format();
  1879. }
  1880. static TimeZoneNames const* find_time_zone_names(StringView locale, StringView time_zone)
  1881. {
  1882. auto locale_value = locale_from_string(locale);
  1883. if (!locale_value.has_value())
  1884. return nullptr;
  1885. auto time_zone_value = ::TimeZone::time_zone_from_string(time_zone);
  1886. if (!time_zone_value.has_value())
  1887. return nullptr;
  1888. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1889. size_t time_zone_index = to_underlying(*time_zone_value);
  1890. auto time_zone_list_index = s_locale_time_zones.at(locale_index);
  1891. auto const& time_zone_list = s_time_zone_lists.at(time_zone_list_index);
  1892. if (time_zone_list.size() <= time_zone_index)
  1893. return nullptr;
  1894. time_zone_index = time_zone_list.at(time_zone_index);
  1895. return &s_time_zones[time_zone_index];
  1896. }
  1897. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style, TimeZone::InDST in_dst)
  1898. {
  1899. if (auto const* data = find_time_zone_names(locale, time_zone); data != nullptr) {
  1900. size_t name_index = 0;
  1901. switch (style) {
  1902. case CalendarPatternStyle::Short:
  1903. name_index = (in_dst == TimeZone::InDST::No) ? data->short_standard_name : data->short_daylight_name;
  1904. break;
  1905. case CalendarPatternStyle::Long:
  1906. name_index = (in_dst == TimeZone::InDST::No) ? data->long_standard_name : data->long_daylight_name;
  1907. break;
  1908. case CalendarPatternStyle::ShortGeneric:
  1909. name_index = data->short_generic_name;
  1910. break;
  1911. case CalendarPatternStyle::LongGeneric:
  1912. name_index = data->long_generic_name;
  1913. break;
  1914. default:
  1915. VERIFY_NOT_REACHED();
  1916. }
  1917. if (name_index != 0)
  1918. return s_string_list[name_index];
  1919. }
  1920. return {};
  1921. }
  1922. }
  1923. )~~~");
  1924. TRY(file.write(generator.as_string_view().bytes()));
  1925. return {};
  1926. }
  1927. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1928. {
  1929. StringView generated_header_path;
  1930. StringView generated_implementation_path;
  1931. StringView core_path;
  1932. StringView dates_path;
  1933. Core::ArgsParser args_parser;
  1934. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1935. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1936. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1937. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1938. args_parser.parse(arguments);
  1939. auto generated_header_file = TRY(open_file(generated_header_path, Core::Stream::OpenMode::Write));
  1940. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::Stream::OpenMode::Write));
  1941. UnicodeLocaleData locale_data;
  1942. TRY(parse_all_locales(core_path, dates_path, locale_data));
  1943. TRY(generate_unicode_locale_header(*generated_header_file, locale_data));
  1944. TRY(generate_unicode_locale_implementation(*generated_implementation_file, locale_data));
  1945. return 0;
  1946. }