GenerateUnicodeDateTimeFormat.cpp 89 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279
  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/Format.h>
  10. #include <AK/GenericLexer.h>
  11. #include <AK/HashFunctions.h>
  12. #include <AK/HashMap.h>
  13. #include <AK/JsonObject.h>
  14. #include <AK/JsonParser.h>
  15. #include <AK/JsonValue.h>
  16. #include <AK/LexicalPath.h>
  17. #include <AK/SourceGenerator.h>
  18. #include <AK/String.h>
  19. #include <AK/StringBuilder.h>
  20. #include <AK/Traits.h>
  21. #include <AK/Utf8View.h>
  22. #include <LibCore/ArgsParser.h>
  23. #include <LibCore/DirIterator.h>
  24. #include <LibCore/Stream.h>
  25. #include <LibTimeZone/TimeZone.h>
  26. #include <LibUnicode/DateTimeFormat.h>
  27. using StringIndexType = u16;
  28. constexpr auto s_string_index_type = "u16"sv;
  29. using CalendarPatternIndexType = u16;
  30. constexpr auto s_calendar_pattern_index_type = "u16"sv;
  31. using CalendarPatternListIndexType = u8;
  32. constexpr auto s_calendar_pattern_list_index_type = "u8"sv;
  33. using CalendarRangePatternIndexType = u16;
  34. constexpr auto s_calendar_range_pattern_index_type = "u16"sv;
  35. using CalendarRangePatternListIndexType = u8;
  36. constexpr auto s_calendar_range_pattern_list_index_type = "u8"sv;
  37. using CalendarFormatIndexType = u8;
  38. constexpr auto s_calendar_format_index_type = "u8"sv;
  39. using SymbolListIndexType = u16;
  40. constexpr auto s_symbol_list_index_type = "u16"sv;
  41. using CalendarSymbolsIndexType = u16;
  42. constexpr auto s_calendar_symbols_index_type = "u16"sv;
  43. using CalendarSymbolsListIndexType = u8;
  44. constexpr auto s_calendar_symbols_list_index_type = "u8"sv;
  45. using CalendarIndexType = u8;
  46. constexpr auto s_calendar_index_type = "u8"sv;
  47. using TimeZoneNamesIndexType = u16;
  48. constexpr auto s_time_zone_index_type = "u16"sv;
  49. using TimeZoneNamesListIndexType = u8;
  50. constexpr auto s_time_zone_list_index_type = "u8"sv;
  51. using TimeZoneFormatIndexType = u8;
  52. constexpr auto s_time_zone_format_index_type = "u8"sv;
  53. using DayPeriodIndexType = u8;
  54. constexpr auto s_day_period_index_type = "u8"sv;
  55. using DayPeriodListIndexType = u8;
  56. constexpr auto s_day_period_list_index_type = "u8"sv;
  57. using HourCycleListIndexType = u8;
  58. constexpr auto s_hour_cycle_list_index_type = "u8"sv;
  59. struct CalendarPattern : public Unicode::CalendarPattern {
  60. bool contains_only_date_fields() const
  61. {
  62. return !day_period.has_value() && !hour.has_value() && !minute.has_value() && !second.has_value() && !fractional_second_digits.has_value() && !time_zone_name.has_value();
  63. }
  64. bool contains_only_time_fields() const
  65. {
  66. return !weekday.has_value() && !era.has_value() && !year.has_value() && !month.has_value() && !day.has_value();
  67. }
  68. unsigned hash() const
  69. {
  70. auto hash = pair_int_hash(pattern_index, pattern12_index);
  71. hash = pair_int_hash(hash, skeleton_index);
  72. auto hash_field = [&](auto const& field) {
  73. if (field.has_value())
  74. hash = pair_int_hash(hash, static_cast<u8>(*field));
  75. else
  76. hash = pair_int_hash(hash, -1);
  77. };
  78. hash_field(era);
  79. hash_field(year);
  80. hash_field(month);
  81. hash_field(weekday);
  82. hash_field(day);
  83. hash_field(day_period);
  84. hash_field(hour);
  85. hash_field(minute);
  86. hash_field(second);
  87. hash_field(fractional_second_digits);
  88. hash_field(time_zone_name);
  89. return hash;
  90. }
  91. bool operator==(CalendarPattern const& other) const
  92. {
  93. return (skeleton_index == other.skeleton_index)
  94. && (pattern_index == other.pattern_index)
  95. && (pattern12_index == other.pattern12_index)
  96. && (era == other.era)
  97. && (year == other.year)
  98. && (month == other.month)
  99. && (weekday == other.weekday)
  100. && (day == other.day)
  101. && (day_period == other.day_period)
  102. && (hour == other.hour)
  103. && (minute == other.minute)
  104. && (second == other.second)
  105. && (fractional_second_digits == other.fractional_second_digits)
  106. && (time_zone_name == other.time_zone_name);
  107. }
  108. StringIndexType skeleton_index { 0 };
  109. StringIndexType pattern_index { 0 };
  110. StringIndexType pattern12_index { 0 };
  111. };
  112. template<>
  113. struct AK::Formatter<CalendarPattern> : Formatter<FormatString> {
  114. ErrorOr<void> format(FormatBuilder& builder, CalendarPattern const& pattern)
  115. {
  116. auto field_to_i8 = [](auto const& field) -> i8 {
  117. if (!field.has_value())
  118. return -1;
  119. return static_cast<i8>(*field);
  120. };
  121. return Formatter<FormatString>::format(builder,
  122. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}",
  123. pattern.skeleton_index,
  124. pattern.pattern_index,
  125. pattern.pattern12_index,
  126. field_to_i8(pattern.era),
  127. field_to_i8(pattern.year),
  128. field_to_i8(pattern.month),
  129. field_to_i8(pattern.weekday),
  130. field_to_i8(pattern.day),
  131. field_to_i8(pattern.day_period),
  132. field_to_i8(pattern.hour),
  133. field_to_i8(pattern.minute),
  134. field_to_i8(pattern.second),
  135. field_to_i8(pattern.fractional_second_digits),
  136. field_to_i8(pattern.time_zone_name));
  137. }
  138. };
  139. template<>
  140. struct AK::Traits<CalendarPattern> : public GenericTraits<CalendarPattern> {
  141. static unsigned hash(CalendarPattern const& c) { return c.hash(); }
  142. };
  143. struct CalendarRangePattern : public CalendarPattern {
  144. unsigned hash() const
  145. {
  146. auto hash = CalendarPattern::hash();
  147. if (field.has_value())
  148. hash = pair_int_hash(hash, static_cast<u8>(*field));
  149. hash = pair_int_hash(hash, start_range);
  150. hash = pair_int_hash(hash, separator);
  151. hash = pair_int_hash(hash, end_range);
  152. return hash;
  153. }
  154. bool operator==(CalendarRangePattern const& other) const
  155. {
  156. if (!CalendarPattern::operator==(other))
  157. return false;
  158. return (field == other.field)
  159. && (start_range == other.start_range)
  160. && (separator == other.separator)
  161. && (end_range == other.end_range);
  162. }
  163. Optional<Unicode::CalendarRangePattern::Field> field {};
  164. StringIndexType start_range { 0 };
  165. StringIndexType separator { 0 };
  166. StringIndexType end_range { 0 };
  167. };
  168. template<>
  169. struct AK::Formatter<CalendarRangePattern> : Formatter<FormatString> {
  170. ErrorOr<void> format(FormatBuilder& builder, CalendarRangePattern const& pattern)
  171. {
  172. auto field_to_i8 = [](auto const& field) -> i8 {
  173. if (!field.has_value())
  174. return -1;
  175. return static_cast<i8>(*field);
  176. };
  177. return Formatter<FormatString>::format(builder,
  178. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}",
  179. pattern.skeleton_index,
  180. field_to_i8(pattern.field),
  181. pattern.start_range,
  182. pattern.separator,
  183. pattern.end_range,
  184. field_to_i8(pattern.era),
  185. field_to_i8(pattern.year),
  186. field_to_i8(pattern.month),
  187. field_to_i8(pattern.weekday),
  188. field_to_i8(pattern.day),
  189. field_to_i8(pattern.day_period),
  190. field_to_i8(pattern.hour),
  191. field_to_i8(pattern.minute),
  192. field_to_i8(pattern.second),
  193. field_to_i8(pattern.fractional_second_digits),
  194. field_to_i8(pattern.time_zone_name));
  195. }
  196. };
  197. template<>
  198. struct AK::Traits<CalendarRangePattern> : public GenericTraits<CalendarRangePattern> {
  199. static unsigned hash(CalendarRangePattern const& c) { return c.hash(); }
  200. };
  201. struct CalendarFormat {
  202. unsigned hash() const
  203. {
  204. auto hash = pair_int_hash(full_format, long_format);
  205. hash = pair_int_hash(hash, medium_format);
  206. hash = pair_int_hash(hash, short_format);
  207. return hash;
  208. }
  209. bool operator==(CalendarFormat const& other) const
  210. {
  211. return (full_format == other.full_format)
  212. && (long_format == other.long_format)
  213. && (medium_format == other.medium_format)
  214. && (short_format == other.short_format);
  215. }
  216. CalendarPatternIndexType full_format { 0 };
  217. CalendarPatternIndexType long_format { 0 };
  218. CalendarPatternIndexType medium_format { 0 };
  219. CalendarPatternIndexType short_format { 0 };
  220. };
  221. template<>
  222. struct AK::Formatter<CalendarFormat> : Formatter<FormatString> {
  223. ErrorOr<void> format(FormatBuilder& builder, CalendarFormat const& pattern)
  224. {
  225. return Formatter<FormatString>::format(builder,
  226. "{{ {}, {}, {}, {} }}",
  227. pattern.full_format,
  228. pattern.long_format,
  229. pattern.medium_format,
  230. pattern.short_format);
  231. }
  232. };
  233. template<>
  234. struct AK::Traits<CalendarFormat> : public GenericTraits<CalendarFormat> {
  235. static unsigned hash(CalendarFormat const& c) { return c.hash(); }
  236. };
  237. using SymbolList = Vector<StringIndexType>;
  238. struct CalendarSymbols {
  239. unsigned hash() const
  240. {
  241. auto hash = pair_int_hash(narrow_symbols, short_symbols);
  242. hash = pair_int_hash(hash, long_symbols);
  243. return hash;
  244. }
  245. bool operator==(CalendarSymbols const& other) const
  246. {
  247. return (narrow_symbols == other.narrow_symbols)
  248. && (short_symbols == other.short_symbols)
  249. && (long_symbols == other.long_symbols);
  250. }
  251. SymbolListIndexType narrow_symbols { 0 };
  252. SymbolListIndexType short_symbols { 0 };
  253. SymbolListIndexType long_symbols { 0 };
  254. };
  255. template<>
  256. struct AK::Formatter<CalendarSymbols> : Formatter<FormatString> {
  257. ErrorOr<void> format(FormatBuilder& builder, CalendarSymbols const& symbols)
  258. {
  259. return Formatter<FormatString>::format(builder,
  260. "{{ {}, {}, {} }}",
  261. symbols.narrow_symbols,
  262. symbols.short_symbols,
  263. symbols.long_symbols);
  264. }
  265. };
  266. template<>
  267. struct AK::Traits<CalendarSymbols> : public GenericTraits<CalendarSymbols> {
  268. static unsigned hash(CalendarSymbols const& c) { return c.hash(); }
  269. };
  270. using CalendarPatternList = Vector<CalendarPatternIndexType>;
  271. using CalendarRangePatternList = Vector<CalendarRangePatternIndexType>;
  272. using CalendarSymbolsList = Vector<CalendarSymbolsIndexType>;
  273. struct Calendar {
  274. unsigned hash() const
  275. {
  276. auto hash = int_hash(date_formats);
  277. hash = pair_int_hash(hash, time_formats);
  278. hash = pair_int_hash(hash, date_time_formats);
  279. hash = pair_int_hash(hash, available_formats);
  280. hash = pair_int_hash(hash, default_range_format);
  281. hash = pair_int_hash(hash, range_formats);
  282. hash = pair_int_hash(hash, range12_formats);
  283. hash = pair_int_hash(hash, symbols);
  284. return hash;
  285. }
  286. bool operator==(Calendar const& other) const
  287. {
  288. return (date_formats == other.date_formats)
  289. && (time_formats == other.time_formats)
  290. && (date_time_formats == other.date_time_formats)
  291. && (available_formats == other.available_formats)
  292. && (default_range_format == other.default_range_format)
  293. && (range_formats == other.range_formats)
  294. && (range12_formats == other.range12_formats)
  295. && (symbols == other.symbols);
  296. }
  297. CalendarFormatIndexType date_formats { 0 };
  298. CalendarFormatIndexType time_formats { 0 };
  299. CalendarFormatIndexType date_time_formats { 0 };
  300. CalendarPatternListIndexType available_formats { 0 };
  301. CalendarRangePatternIndexType default_range_format { 0 };
  302. CalendarRangePatternListIndexType range_formats { 0 };
  303. CalendarRangePatternListIndexType range12_formats { 0 };
  304. CalendarSymbolsListIndexType symbols { 0 };
  305. };
  306. template<>
  307. struct AK::Formatter<Calendar> : Formatter<FormatString> {
  308. ErrorOr<void> format(FormatBuilder& builder, Calendar const& calendar)
  309. {
  310. return Formatter<FormatString>::format(builder,
  311. "{{ {}, {}, {}, {}, {}, {}, {}, {} }}",
  312. calendar.date_formats,
  313. calendar.time_formats,
  314. calendar.date_time_formats,
  315. calendar.available_formats,
  316. calendar.default_range_format,
  317. calendar.range_formats,
  318. calendar.range12_formats,
  319. calendar.symbols);
  320. }
  321. };
  322. template<>
  323. struct AK::Traits<Calendar> : public GenericTraits<Calendar> {
  324. static unsigned hash(Calendar const& c) { return c.hash(); }
  325. };
  326. struct TimeZoneNames {
  327. unsigned hash() const
  328. {
  329. auto hash = pair_int_hash(short_standard_name, long_standard_name);
  330. hash = pair_int_hash(hash, short_daylight_name);
  331. hash = pair_int_hash(hash, long_daylight_name);
  332. hash = pair_int_hash(hash, short_generic_name);
  333. hash = pair_int_hash(hash, long_generic_name);
  334. return hash;
  335. }
  336. bool operator==(TimeZoneNames const& other) const
  337. {
  338. return (short_standard_name == other.short_standard_name)
  339. && (long_standard_name == other.long_standard_name)
  340. && (short_daylight_name == other.short_daylight_name)
  341. && (long_daylight_name == other.long_daylight_name)
  342. && (short_generic_name == other.short_generic_name)
  343. && (long_generic_name == other.long_generic_name);
  344. }
  345. StringIndexType short_standard_name { 0 };
  346. StringIndexType long_standard_name { 0 };
  347. StringIndexType short_daylight_name { 0 };
  348. StringIndexType long_daylight_name { 0 };
  349. StringIndexType short_generic_name { 0 };
  350. StringIndexType long_generic_name { 0 };
  351. };
  352. template<>
  353. struct AK::Formatter<TimeZoneNames> : Formatter<FormatString> {
  354. ErrorOr<void> format(FormatBuilder& builder, TimeZoneNames const& time_zone)
  355. {
  356. return Formatter<FormatString>::format(builder,
  357. "{{ {}, {}, {}, {}, {}, {} }}",
  358. time_zone.short_standard_name,
  359. time_zone.long_standard_name,
  360. time_zone.short_daylight_name,
  361. time_zone.long_daylight_name,
  362. time_zone.short_generic_name,
  363. time_zone.long_generic_name);
  364. }
  365. };
  366. template<>
  367. struct AK::Traits<TimeZoneNames> : public GenericTraits<TimeZoneNames> {
  368. static unsigned hash(TimeZoneNames const& t) { return t.hash(); }
  369. };
  370. struct TimeZoneFormat {
  371. unsigned hash() const
  372. {
  373. auto hash = int_hash(symbol_ahead_sign);
  374. hash = pair_int_hash(hash, symbol_ahead_separator);
  375. hash = pair_int_hash(hash, symbol_behind_sign);
  376. hash = pair_int_hash(hash, symbol_behind_separator);
  377. hash = pair_int_hash(hash, gmt_format);
  378. hash = pair_int_hash(hash, gmt_zero_format);
  379. return hash;
  380. }
  381. bool operator==(TimeZoneFormat const& other) const
  382. {
  383. return (symbol_ahead_sign == other.symbol_ahead_sign)
  384. && (symbol_ahead_separator == other.symbol_ahead_separator)
  385. && (symbol_behind_sign == other.symbol_behind_sign)
  386. && (symbol_behind_separator == other.symbol_behind_separator)
  387. && (gmt_format == other.gmt_format)
  388. && (gmt_zero_format == other.gmt_zero_format);
  389. }
  390. StringIndexType symbol_ahead_sign { 0 };
  391. StringIndexType symbol_ahead_separator { 0 };
  392. StringIndexType symbol_behind_sign { 0 };
  393. StringIndexType symbol_behind_separator { 0 };
  394. StringIndexType gmt_format { 0 };
  395. StringIndexType gmt_zero_format { 0 };
  396. };
  397. template<>
  398. struct AK::Formatter<TimeZoneFormat> : Formatter<FormatString> {
  399. ErrorOr<void> format(FormatBuilder& builder, TimeZoneFormat const& time_zone_format)
  400. {
  401. return Formatter<FormatString>::format(builder, "{{ {}, {}, {}, {}, {}, {} }}",
  402. time_zone_format.symbol_ahead_sign,
  403. time_zone_format.symbol_ahead_separator,
  404. time_zone_format.symbol_behind_sign,
  405. time_zone_format.symbol_behind_separator,
  406. time_zone_format.gmt_format,
  407. time_zone_format.gmt_zero_format);
  408. }
  409. };
  410. template<>
  411. struct AK::Traits<TimeZoneFormat> : public GenericTraits<TimeZoneFormat> {
  412. static unsigned hash(TimeZoneFormat const& t) { return t.hash(); }
  413. };
  414. struct DayPeriod {
  415. unsigned hash() const
  416. {
  417. auto hash = int_hash(static_cast<u8>(day_period));
  418. hash = pair_int_hash(hash, begin);
  419. hash = pair_int_hash(hash, end);
  420. return hash;
  421. }
  422. bool operator==(DayPeriod const& other) const
  423. {
  424. return (day_period == other.day_period)
  425. && (begin == other.begin)
  426. && (end == other.end);
  427. }
  428. Unicode::DayPeriod day_period {};
  429. u8 begin { 0 };
  430. u8 end { 0 };
  431. };
  432. template<>
  433. struct AK::Formatter<DayPeriod> : Formatter<FormatString> {
  434. ErrorOr<void> format(FormatBuilder& builder, DayPeriod const& day_period)
  435. {
  436. return Formatter<FormatString>::format(builder,
  437. "{{ {}, {}, {} }}",
  438. static_cast<u8>(day_period.day_period),
  439. day_period.begin,
  440. day_period.end);
  441. }
  442. };
  443. template<>
  444. struct AK::Traits<DayPeriod> : public GenericTraits<DayPeriod> {
  445. static unsigned hash(DayPeriod const& d) { return d.hash(); }
  446. };
  447. using TimeZoneNamesList = Vector<TimeZoneNamesIndexType>;
  448. using DayPeriodList = Vector<DayPeriodIndexType>;
  449. using HourCycleList = Vector<Unicode::HourCycle>;
  450. template<>
  451. struct AK::Formatter<Unicode::HourCycle> : Formatter<FormatString> {
  452. ErrorOr<void> format(FormatBuilder& builder, Unicode::HourCycle hour_cycle)
  453. {
  454. return Formatter<FormatString>::format(builder, "{}", to_underlying(hour_cycle));
  455. }
  456. };
  457. struct Locale {
  458. HashMap<String, CalendarIndexType> calendars;
  459. TimeZoneNamesListIndexType time_zones { 0 };
  460. TimeZoneFormatIndexType time_zone_formats { 0 };
  461. DayPeriodListIndexType day_periods { 0 };
  462. };
  463. struct UnicodeLocaleData {
  464. UniqueStringStorage<StringIndexType> unique_strings;
  465. UniqueStorage<CalendarPattern, CalendarPatternIndexType> unique_patterns;
  466. UniqueStorage<CalendarPatternList, CalendarPatternListIndexType> unique_pattern_lists;
  467. UniqueStorage<CalendarRangePattern, CalendarRangePatternIndexType> unique_range_patterns;
  468. UniqueStorage<CalendarRangePatternList, CalendarRangePatternListIndexType> unique_range_pattern_lists;
  469. UniqueStorage<CalendarFormat, CalendarFormatIndexType> unique_formats;
  470. UniqueStorage<SymbolList, SymbolListIndexType> unique_symbol_lists;
  471. UniqueStorage<CalendarSymbols, CalendarSymbolsIndexType> unique_calendar_symbols;
  472. UniqueStorage<CalendarSymbolsList, CalendarSymbolsListIndexType> unique_calendar_symbols_lists;
  473. UniqueStorage<Calendar, CalendarIndexType> unique_calendars;
  474. UniqueStorage<TimeZoneNames, TimeZoneNamesIndexType> unique_time_zones;
  475. UniqueStorage<TimeZoneNamesList, TimeZoneNamesListIndexType> unique_time_zone_lists;
  476. UniqueStorage<TimeZoneFormat, TimeZoneFormatIndexType> unique_time_zone_formats;
  477. UniqueStorage<DayPeriod, DayPeriodIndexType> unique_day_periods;
  478. UniqueStorage<DayPeriodList, DayPeriodListIndexType> unique_day_period_lists;
  479. UniqueStorage<HourCycleList, HourCycleListIndexType> unique_hour_cycle_lists;
  480. HashMap<String, Locale> locales;
  481. HashMap<String, HourCycleListIndexType> hour_cycles;
  482. Vector<String> hour_cycle_regions;
  483. HashMap<String, Vector<TimeZone::TimeZone>> meta_zones;
  484. Vector<String> time_zones { "UTC"sv };
  485. Vector<String> calendars;
  486. };
  487. static Optional<Unicode::DayPeriod> day_period_from_string(StringView day_period)
  488. {
  489. if (day_period == "am"sv)
  490. return Unicode::DayPeriod::AM;
  491. if (day_period == "pm"sv)
  492. return Unicode::DayPeriod::PM;
  493. if (day_period == "morning1"sv)
  494. return Unicode::DayPeriod::Morning1;
  495. if (day_period == "morning2"sv)
  496. return Unicode::DayPeriod::Morning2;
  497. if (day_period == "afternoon1"sv)
  498. return Unicode::DayPeriod::Afternoon1;
  499. if (day_period == "afternoon2"sv)
  500. return Unicode::DayPeriod::Afternoon2;
  501. if (day_period == "evening1"sv)
  502. return Unicode::DayPeriod::Evening1;
  503. if (day_period == "evening2"sv)
  504. return Unicode::DayPeriod::Evening2;
  505. if (day_period == "night1"sv)
  506. return Unicode::DayPeriod::Night1;
  507. if (day_period == "night2"sv)
  508. return Unicode::DayPeriod::Night2;
  509. return {};
  510. }
  511. static ErrorOr<void> parse_hour_cycles(String core_path, UnicodeLocaleData& locale_data)
  512. {
  513. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  514. LexicalPath time_data_path(move(core_path));
  515. time_data_path = time_data_path.append("supplemental"sv);
  516. time_data_path = time_data_path.append("timeData.json"sv);
  517. auto time_data = TRY(read_json_file(time_data_path.string()));
  518. auto const& supplemental_object = time_data.as_object().get("supplemental"sv);
  519. auto const& time_data_object = supplemental_object.as_object().get("timeData"sv);
  520. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Unicode::HourCycle> {
  521. if (hour_cycle == "h"sv)
  522. return Unicode::HourCycle::H12;
  523. if (hour_cycle == "H"sv)
  524. return Unicode::HourCycle::H23;
  525. if (hour_cycle == "K"sv)
  526. return Unicode::HourCycle::H11;
  527. if (hour_cycle == "k"sv)
  528. return Unicode::HourCycle::H24;
  529. return {};
  530. };
  531. time_data_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  532. auto allowed_hour_cycles_string = value.as_object().get("_allowed"sv).as_string();
  533. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  534. Vector<Unicode::HourCycle> hour_cycles;
  535. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  536. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  537. hour_cycles.append(*hour_cycle);
  538. }
  539. auto hour_cycles_index = locale_data.unique_hour_cycle_lists.ensure(move(hour_cycles));
  540. locale_data.hour_cycles.set(key, hour_cycles_index);
  541. if (!locale_data.hour_cycle_regions.contains_slow(key))
  542. locale_data.hour_cycle_regions.append(key);
  543. });
  544. return {};
  545. }
  546. static ErrorOr<void> parse_meta_zones(String core_path, UnicodeLocaleData& locale_data)
  547. {
  548. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  549. LexicalPath meta_zone_path(move(core_path));
  550. meta_zone_path = meta_zone_path.append("supplemental"sv);
  551. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  552. auto meta_zone = TRY(read_json_file(meta_zone_path.string()));
  553. auto const& supplemental_object = meta_zone.as_object().get("supplemental"sv);
  554. auto const& meta_zone_object = supplemental_object.as_object().get("metaZones"sv);
  555. auto const& meta_zone_array = meta_zone_object.as_object().get("metazones"sv);
  556. meta_zone_array.as_array().for_each([&](JsonValue const& value) {
  557. auto const& mapping = value.as_object().get("mapZone"sv);
  558. auto const& meta_zone = mapping.as_object().get("_other"sv);
  559. auto const& golden_zone = mapping.as_object().get("_type"sv);
  560. if (auto time_zone = TimeZone::time_zone_from_string(golden_zone.as_string()); time_zone.has_value()) {
  561. auto& golden_zones = locale_data.meta_zones.ensure(meta_zone.as_string());
  562. golden_zones.append(*time_zone);
  563. }
  564. });
  565. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  566. if (auto time_zone = TimeZone::time_zone_from_string("UTC"sv); time_zone.has_value())
  567. locale_data.meta_zones.set("UTC"sv, { *time_zone });
  568. return {};
  569. }
  570. static constexpr auto is_char(char ch)
  571. {
  572. return [ch](auto c) { return c == ch; };
  573. }
  574. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  575. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  576. // extra spaces around the day period. Some example expected removals:
  577. //
  578. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  579. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  580. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  581. static String remove_period_from_pattern(String pattern)
  582. {
  583. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  584. auto index = pattern.find(remove);
  585. if (!index.has_value())
  586. continue;
  587. constexpr u32 space = ' ';
  588. constexpr u32 open = '{';
  589. constexpr u32 close = '}';
  590. Utf8View utf8_pattern { pattern };
  591. Optional<u32> before_removal;
  592. Optional<u32> after_removal;
  593. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  594. before_removal = *it;
  595. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  596. after_removal = *it;
  597. if ((before_removal == space) && (after_removal != open)) {
  598. pattern = String::formatted("{}{}",
  599. pattern.substring_view(0, *index - 1),
  600. pattern.substring_view(*index + remove.length()));
  601. } else if ((after_removal == space) && (before_removal != close)) {
  602. pattern = String::formatted("{}{}",
  603. pattern.substring_view(0, *index),
  604. pattern.substring_view(*index + remove.length() + 1));
  605. } else {
  606. pattern = String::formatted("{}{}",
  607. pattern.substring_view(0, *index),
  608. pattern.substring_view(*index + remove.length()));
  609. }
  610. }
  611. return pattern;
  612. }
  613. static Optional<CalendarPattern> parse_date_time_pattern_raw(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  614. {
  615. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  616. using Unicode::CalendarPatternStyle;
  617. CalendarPattern format {};
  618. if (!skeleton.is_empty())
  619. format.skeleton_index = locale_data.unique_strings.ensure(move(skeleton));
  620. GenericLexer lexer { pattern };
  621. StringBuilder builder;
  622. bool hour12 { false };
  623. while (!lexer.is_eof()) {
  624. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  625. // processing (this just avoids conflicts with the patterns below).
  626. if (lexer.next_is(is_quote)) {
  627. builder.append(lexer.consume_quoted_string());
  628. continue;
  629. }
  630. auto starting_char = lexer.peek();
  631. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  632. // Era
  633. if (all_of(segment, is_char('G'))) {
  634. builder.append("{era}");
  635. if (segment.length() <= 3)
  636. format.era = CalendarPatternStyle::Short;
  637. else if (segment.length() == 4)
  638. format.era = CalendarPatternStyle::Long;
  639. else
  640. format.era = CalendarPatternStyle::Narrow;
  641. }
  642. // Year
  643. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  644. builder.append("{year}");
  645. if (segment.length() == 2)
  646. format.year = CalendarPatternStyle::TwoDigit;
  647. else
  648. format.year = CalendarPatternStyle::Numeric;
  649. }
  650. // Quarter
  651. else if (all_of(segment, is_any_of("qQ"sv))) {
  652. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  653. return {};
  654. }
  655. // Month
  656. else if (all_of(segment, is_any_of("ML"sv))) {
  657. builder.append("{month}");
  658. if (segment.length() == 1)
  659. format.month = CalendarPatternStyle::Numeric;
  660. else if (segment.length() == 2)
  661. format.month = CalendarPatternStyle::TwoDigit;
  662. else if (segment.length() == 3)
  663. format.month = CalendarPatternStyle::Short;
  664. else if (segment.length() == 4)
  665. format.month = CalendarPatternStyle::Long;
  666. else if (segment.length() == 5)
  667. format.month = CalendarPatternStyle::Narrow;
  668. } else if (all_of(segment, is_char('l'))) {
  669. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  670. return {};
  671. }
  672. // Week
  673. else if (all_of(segment, is_any_of("wW"sv))) {
  674. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  675. return {};
  676. }
  677. // Day
  678. else if (all_of(segment, is_char('d'))) {
  679. builder.append("{day}");
  680. if (segment.length() == 1)
  681. format.day = CalendarPatternStyle::Numeric;
  682. else
  683. format.day = CalendarPatternStyle::TwoDigit;
  684. } else if (all_of(segment, is_any_of("DFg"sv))) {
  685. builder.append("{day}");
  686. format.day = CalendarPatternStyle::Numeric;
  687. }
  688. // Weekday
  689. else if (all_of(segment, is_char('E'))) {
  690. builder.append("{weekday}");
  691. if (segment.length() == 4)
  692. format.weekday = CalendarPatternStyle::Long;
  693. else if (segment.length() == 5)
  694. format.weekday = CalendarPatternStyle::Narrow;
  695. else
  696. format.weekday = CalendarPatternStyle::Short;
  697. } else if (all_of(segment, is_any_of("ec"sv))) {
  698. builder.append("{weekday}");
  699. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  700. // pattern styles are not supported by Intl.DateTimeFormat.
  701. if (segment.length() <= 2)
  702. return {};
  703. if (segment.length() == 4)
  704. format.weekday = CalendarPatternStyle::Long;
  705. else if (segment.length() == 5)
  706. format.weekday = CalendarPatternStyle::Narrow;
  707. else
  708. format.weekday = CalendarPatternStyle::Short;
  709. }
  710. // Period
  711. else if (all_of(segment, is_any_of("ab"sv))) {
  712. builder.append("{ampm}");
  713. hour12 = true;
  714. } else if (all_of(segment, is_char('B'))) {
  715. builder.append("{dayPeriod}");
  716. hour12 = true;
  717. if (segment.length() == 4)
  718. format.day_period = CalendarPatternStyle::Long;
  719. else if (segment.length() == 5)
  720. format.day_period = CalendarPatternStyle::Narrow;
  721. else
  722. format.day_period = CalendarPatternStyle::Short;
  723. }
  724. // Hour
  725. else if (all_of(segment, is_any_of("hHKk"sv))) {
  726. builder.append("{hour}");
  727. if ((segment[0] == 'h') || (segment[0] == 'K'))
  728. hour12 = true;
  729. if (segment.length() == 1)
  730. format.hour = CalendarPatternStyle::Numeric;
  731. else
  732. format.hour = CalendarPatternStyle::TwoDigit;
  733. } else if (all_of(segment, is_any_of("jJC"sv))) {
  734. // TR-35 indicates these should not be used.
  735. return {};
  736. }
  737. // Minute
  738. else if (all_of(segment, is_char('m'))) {
  739. builder.append("{minute}");
  740. if (segment.length() == 1)
  741. format.minute = CalendarPatternStyle::Numeric;
  742. else
  743. format.minute = CalendarPatternStyle::TwoDigit;
  744. }
  745. // Second
  746. else if (all_of(segment, is_char('s'))) {
  747. builder.append("{second}");
  748. if (segment.length() == 1)
  749. format.second = CalendarPatternStyle::Numeric;
  750. else
  751. format.second = CalendarPatternStyle::TwoDigit;
  752. } else if (all_of(segment, is_char('S'))) {
  753. builder.append("{fractionalSecondDigits}");
  754. VERIFY(segment.length() <= 3);
  755. format.fractional_second_digits = static_cast<u8>(segment.length());
  756. } else if (all_of(segment, is_char('A'))) {
  757. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  758. return {};
  759. }
  760. // Zone
  761. else if (all_of(segment, is_any_of("zV"sv))) {
  762. builder.append("{timeZoneName}");
  763. if (segment.length() < 4)
  764. format.time_zone_name = CalendarPatternStyle::Short;
  765. else
  766. format.time_zone_name = CalendarPatternStyle::Long;
  767. } else if (all_of(segment, is_any_of("ZOXx"sv))) {
  768. builder.append("{timeZoneName}");
  769. if (segment.length() < 4)
  770. format.time_zone_name = CalendarPatternStyle::ShortOffset;
  771. else
  772. format.time_zone_name = CalendarPatternStyle::LongOffset;
  773. } else if (all_of(segment, is_char('v'))) {
  774. builder.append("{timeZoneName}");
  775. if (segment.length() < 4)
  776. format.time_zone_name = CalendarPatternStyle::ShortGeneric;
  777. else
  778. format.time_zone_name = CalendarPatternStyle::LongGeneric;
  779. }
  780. // Non-patterns
  781. else {
  782. builder.append(segment);
  783. }
  784. }
  785. pattern = builder.build();
  786. if (hour12) {
  787. format.pattern = remove_period_from_pattern(pattern);
  788. format.pattern12 = move(pattern);
  789. } else {
  790. format.pattern = move(pattern);
  791. }
  792. return format;
  793. }
  794. static Optional<CalendarPatternIndexType> parse_date_time_pattern(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  795. {
  796. auto format = parse_date_time_pattern_raw(move(pattern), move(skeleton), locale_data);
  797. if (!format.has_value())
  798. return {};
  799. format->pattern_index = locale_data.unique_strings.ensure(move(format->pattern));
  800. if (format->pattern12.has_value())
  801. format->pattern12_index = locale_data.unique_strings.ensure(format->pattern12.release_value());
  802. return locale_data.unique_patterns.ensure(format.release_value());
  803. }
  804. template<typename... Chars>
  805. static constexpr bool char_is_one_of(char ch, Chars&&... chars)
  806. {
  807. return ((ch == chars) || ...);
  808. }
  809. static void parse_interval_patterns(Calendar& calendar, JsonObject const& interval_formats_object, UnicodeLocaleData& locale_data)
  810. {
  811. // https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats
  812. CalendarRangePatternList range_formats {};
  813. CalendarRangePatternList range12_formats {};
  814. auto name_of_field = [&](char field) {
  815. if (char_is_one_of(field, 'G'))
  816. return Unicode::CalendarRangePattern::Field::Era;
  817. if (char_is_one_of(field, 'y', 'Y', 'u', 'U', 'r'))
  818. return Unicode::CalendarRangePattern::Field::Year;
  819. if (char_is_one_of(field, 'M', 'L'))
  820. return Unicode::CalendarRangePattern::Field::Month;
  821. if (char_is_one_of(field, 'd', 'D', 'F', 'g'))
  822. return Unicode::CalendarRangePattern::Field::Day;
  823. if (char_is_one_of(field, 'a', 'b'))
  824. return Unicode::CalendarRangePattern::Field::AmPm;
  825. if (char_is_one_of(field, 'B'))
  826. return Unicode::CalendarRangePattern::Field::DayPeriod;
  827. if (char_is_one_of(field, 'h', 'H', 'K', 'k'))
  828. return Unicode::CalendarRangePattern::Field::Hour;
  829. if (char_is_one_of(field, 'm'))
  830. return Unicode::CalendarRangePattern::Field::Minute;
  831. if (char_is_one_of(field, 's'))
  832. return Unicode::CalendarRangePattern::Field::Second;
  833. if (char_is_one_of(field, 'S'))
  834. return Unicode::CalendarRangePattern::Field::FractionalSecondDigits;
  835. VERIFY_NOT_REACHED();
  836. };
  837. auto split_default_range_pattern = [&](auto skeleton, auto const& pattern) {
  838. auto start_range_end = pattern.find('}').value() + 1;
  839. auto end_range_begin = pattern.find_last('{').value();
  840. auto start_range = pattern.substring_view(0, start_range_end);
  841. auto separator = pattern.substring_view(start_range_end, end_range_begin - start_range_end);
  842. auto end_range = pattern.substring_view(end_range_begin);
  843. CalendarRangePattern format {};
  844. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  845. format.start_range = locale_data.unique_strings.ensure(start_range);
  846. format.separator = locale_data.unique_strings.ensure(separator);
  847. format.end_range = locale_data.unique_strings.ensure(end_range);
  848. return format;
  849. };
  850. auto split_range_pattern = [&](auto skeleton, auto field, auto const& pattern, auto const& parsed_fields) {
  851. HashMap<StringView, size_t> partitions;
  852. StringView last_partition;
  853. auto begin_index = pattern.find('{');
  854. size_t end_index = 0;
  855. while (begin_index.has_value()) {
  856. end_index = pattern.find('}', *begin_index).value();
  857. auto partition = pattern.substring_view(*begin_index, end_index - *begin_index);
  858. if (partitions.contains(partition))
  859. break;
  860. partitions.set(partition, *begin_index);
  861. last_partition = partition;
  862. begin_index = pattern.find('{', end_index + 1);
  863. }
  864. VERIFY(begin_index.has_value() && !last_partition.is_empty());
  865. auto start_range_end = partitions.get(last_partition).value() + last_partition.length() + 1;
  866. auto start_range = pattern.substring_view(0, start_range_end);
  867. auto separator = pattern.substring_view(start_range_end, *begin_index - start_range_end);
  868. auto end_range = pattern.substring_view(*begin_index);
  869. CalendarRangePattern format {};
  870. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  871. format.field = field;
  872. format.start_range = locale_data.unique_strings.ensure(start_range);
  873. format.separator = locale_data.unique_strings.ensure(separator);
  874. format.end_range = locale_data.unique_strings.ensure(end_range);
  875. format.for_each_calendar_field_zipped_with(parsed_fields, [](auto& format_field, auto const& parsed_field, auto) {
  876. format_field = parsed_field;
  877. });
  878. return format;
  879. };
  880. interval_formats_object.for_each_member([&](auto const& skeleton, auto const& value) {
  881. if (skeleton == "intervalFormatFallback"sv) {
  882. auto range_format = split_default_range_pattern(skeleton, value.as_string());
  883. calendar.default_range_format = locale_data.unique_range_patterns.ensure(move(range_format));
  884. return;
  885. }
  886. value.as_object().for_each_member([&](auto const& field, auto const& pattern) {
  887. if (field.ends_with("alt-variant"sv))
  888. return;
  889. VERIFY(field.length() == 1);
  890. auto name = name_of_field(field[0]);
  891. auto format = parse_date_time_pattern_raw(pattern.as_string(), skeleton, locale_data).release_value();
  892. auto range_format = split_range_pattern(skeleton, name, format.pattern, format);
  893. range_formats.append(locale_data.unique_range_patterns.ensure(move(range_format)));
  894. if (format.pattern12.has_value()) {
  895. auto range12_pattern = split_range_pattern(skeleton, name, *format.pattern12, format);
  896. range12_formats.append(locale_data.unique_range_patterns.ensure(move(range12_pattern)));
  897. } else {
  898. range12_formats.append(range_formats.last());
  899. }
  900. });
  901. });
  902. calendar.range_formats = locale_data.unique_range_pattern_lists.ensure(move(range_formats));
  903. calendar.range12_formats = locale_data.unique_range_pattern_lists.ensure(move(range12_formats));
  904. }
  905. static void generate_missing_patterns(Calendar& calendar, CalendarPatternList& formats, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, UnicodeLocaleData& locale_data)
  906. {
  907. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  908. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  909. auto pattern = locale_data.unique_strings.get(format);
  910. auto time_pattern = locale_data.unique_strings.get(time_format);
  911. auto date_pattern = locale_data.unique_strings.get(date_format);
  912. auto new_pattern = pattern.replace("{0}", time_pattern, ReplaceMode::FirstOnly).replace("{1}", date_pattern, ReplaceMode::FirstOnly);
  913. return locale_data.unique_strings.ensure(move(new_pattern));
  914. };
  915. auto inject_fractional_second_digits = [&](auto format) {
  916. auto pattern = locale_data.unique_strings.get(format);
  917. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv, ReplaceMode::FirstOnly);
  918. return locale_data.unique_strings.ensure(move(new_pattern));
  919. };
  920. auto append_if_unique = [&](auto format) {
  921. auto format_index = locale_data.unique_patterns.ensure(move(format));
  922. if (!formats.contains_slow(format_index))
  923. formats.append(format_index);
  924. };
  925. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  926. for (auto const& format : date_formats)
  927. append_if_unique(format);
  928. for (auto const& format : time_formats) {
  929. append_if_unique(format);
  930. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  931. auto new_format = format;
  932. new_format.fractional_second_digits = 2;
  933. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  934. if (new_format.pattern12_index != 0)
  935. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  936. time_formats_with_fractional_second_digits.append(new_format);
  937. append_if_unique(move(new_format));
  938. }
  939. }
  940. time_formats.extend(move(time_formats_with_fractional_second_digits));
  941. for (auto const& date_format : date_formats) {
  942. auto const& date_time_formats = locale_data.unique_formats.get(calendar.date_time_formats);
  943. CalendarPatternIndexType date_time_format_index = 0;
  944. if (date_format.month == Unicode::CalendarPatternStyle::Long) {
  945. if (date_format.weekday.has_value())
  946. date_time_format_index = date_time_formats.full_format;
  947. else
  948. date_time_format_index = date_time_formats.long_format;
  949. } else if (date_format.month == Unicode::CalendarPatternStyle::Short) {
  950. date_time_format_index = date_time_formats.medium_format;
  951. } else {
  952. date_time_format_index = date_time_formats.short_format;
  953. }
  954. for (auto const& time_format : time_formats) {
  955. auto format = locale_data.unique_patterns.get(date_time_format_index);
  956. if (time_format.pattern12_index != 0)
  957. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  958. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  959. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  960. if (date_field.has_value())
  961. field = date_field;
  962. });
  963. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  964. if (time_field.has_value())
  965. field = time_field;
  966. });
  967. append_if_unique(move(format));
  968. }
  969. }
  970. }
  971. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, UnicodeLocaleData& locale_data)
  972. {
  973. auto create_symbol_lists = [](size_t size) {
  974. SymbolList narrow_symbol_list;
  975. SymbolList short_symbol_list;
  976. SymbolList long_symbol_list;
  977. narrow_symbol_list.resize(size);
  978. short_symbol_list.resize(size);
  979. long_symbol_list.resize(size);
  980. return Array<SymbolList, 3> { {
  981. move(narrow_symbol_list),
  982. move(short_symbol_list),
  983. move(long_symbol_list),
  984. } };
  985. };
  986. CalendarSymbolsList symbols_list {};
  987. auto store_symbol_lists = [&](auto symbol, auto symbol_lists) {
  988. auto symbol_index = to_underlying(symbol);
  989. if (symbol_index >= symbols_list.size())
  990. symbols_list.resize(symbol_index + 1);
  991. CalendarSymbols symbols {};
  992. symbols.narrow_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[0]));
  993. symbols.short_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[1]));
  994. symbols.long_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[2]));
  995. auto calendar_symbols_index = locale_data.unique_calendar_symbols.ensure(move(symbols));
  996. symbols_list[symbol_index] = calendar_symbols_index;
  997. };
  998. auto parse_era_symbols = [&](auto const& symbols_object) {
  999. auto const& narrow_symbols = symbols_object.get("eraNarrow"sv).as_object();
  1000. auto const& short_symbols = symbols_object.get("eraAbbr"sv).as_object();
  1001. auto const& long_symbols = symbols_object.get("eraNames"sv).as_object();
  1002. auto symbol_lists = create_symbol_lists(2);
  1003. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1004. if (auto key_index = key.to_uint(); key_index.has_value())
  1005. symbols[*key_index] = locale_data.unique_strings.ensure(move(symbol));
  1006. };
  1007. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1008. append_symbol(symbol_lists[0], key, value.as_string());
  1009. });
  1010. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1011. append_symbol(symbol_lists[1], key, value.as_string());
  1012. });
  1013. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1014. append_symbol(symbol_lists[2], key, value.as_string());
  1015. });
  1016. store_symbol_lists(Unicode::CalendarSymbol::Era, move(symbol_lists));
  1017. };
  1018. auto parse_month_symbols = [&](auto const& symbols_object) {
  1019. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1020. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1021. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1022. auto symbol_lists = create_symbol_lists(12);
  1023. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1024. auto key_index = key.to_uint().value() - 1;
  1025. symbols[key_index] = locale_data.unique_strings.ensure(move(symbol));
  1026. };
  1027. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1028. append_symbol(symbol_lists[0], key, value.as_string());
  1029. });
  1030. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1031. append_symbol(symbol_lists[1], key, value.as_string());
  1032. });
  1033. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1034. append_symbol(symbol_lists[2], key, value.as_string());
  1035. });
  1036. store_symbol_lists(Unicode::CalendarSymbol::Month, move(symbol_lists));
  1037. };
  1038. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  1039. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1040. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1041. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1042. auto symbol_lists = create_symbol_lists(7);
  1043. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1044. if (key == "sun"sv)
  1045. symbols[to_underlying(Unicode::Weekday::Sunday)] = locale_data.unique_strings.ensure(move(symbol));
  1046. else if (key == "mon"sv)
  1047. symbols[to_underlying(Unicode::Weekday::Monday)] = locale_data.unique_strings.ensure(move(symbol));
  1048. else if (key == "tue"sv)
  1049. symbols[to_underlying(Unicode::Weekday::Tuesday)] = locale_data.unique_strings.ensure(move(symbol));
  1050. else if (key == "wed"sv)
  1051. symbols[to_underlying(Unicode::Weekday::Wednesday)] = locale_data.unique_strings.ensure(move(symbol));
  1052. else if (key == "thu"sv)
  1053. symbols[to_underlying(Unicode::Weekday::Thursday)] = locale_data.unique_strings.ensure(move(symbol));
  1054. else if (key == "fri"sv)
  1055. symbols[to_underlying(Unicode::Weekday::Friday)] = locale_data.unique_strings.ensure(move(symbol));
  1056. else if (key == "sat"sv)
  1057. symbols[to_underlying(Unicode::Weekday::Saturday)] = locale_data.unique_strings.ensure(move(symbol));
  1058. };
  1059. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1060. append_symbol(symbol_lists[0], key, value.as_string());
  1061. });
  1062. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1063. append_symbol(symbol_lists[1], key, value.as_string());
  1064. });
  1065. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1066. append_symbol(symbol_lists[2], key, value.as_string());
  1067. });
  1068. store_symbol_lists(Unicode::CalendarSymbol::Weekday, move(symbol_lists));
  1069. };
  1070. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  1071. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1072. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1073. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1074. auto symbol_lists = create_symbol_lists(10);
  1075. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1076. if (auto day_period = day_period_from_string(key); day_period.has_value())
  1077. symbols[to_underlying(*day_period)] = locale_data.unique_strings.ensure(move(symbol));
  1078. };
  1079. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1080. append_symbol(symbol_lists[0], key, value.as_string());
  1081. });
  1082. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1083. append_symbol(symbol_lists[1], key, value.as_string());
  1084. });
  1085. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1086. append_symbol(symbol_lists[2], key, value.as_string());
  1087. });
  1088. store_symbol_lists(Unicode::CalendarSymbol::DayPeriod, move(symbol_lists));
  1089. };
  1090. parse_era_symbols(calendar_object.get("eras"sv).as_object());
  1091. parse_month_symbols(calendar_object.get("months"sv).as_object().get("format"sv).as_object());
  1092. parse_weekday_symbols(calendar_object.get("days"sv).as_object().get("format"sv).as_object());
  1093. parse_day_period_symbols(calendar_object.get("dayPeriods"sv).as_object().get("format"sv).as_object());
  1094. calendar.symbols = locale_data.unique_calendar_symbols_lists.ensure(move(symbols_list));
  1095. }
  1096. static ErrorOr<void> parse_calendars(String locale_calendars_path, UnicodeLocaleData& locale_data, Locale& locale)
  1097. {
  1098. LexicalPath calendars_path(move(locale_calendars_path));
  1099. if (!calendars_path.basename().starts_with("ca-"sv))
  1100. return {};
  1101. auto calendars = TRY(read_json_file(calendars_path.string()));
  1102. auto const& main_object = calendars.as_object().get("main"sv);
  1103. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  1104. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1105. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  1106. auto parse_patterns = [&](auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) {
  1107. auto parse_pattern = [&](auto name) {
  1108. auto format = patterns_object.get(name);
  1109. auto skeleton = skeletons_object.get(name);
  1110. auto format_index = parse_date_time_pattern(format.as_string(), skeleton.as_string_or(String::empty()), locale_data).value();
  1111. if (patterns)
  1112. patterns->append(locale_data.unique_patterns.get(format_index));
  1113. return format_index;
  1114. };
  1115. CalendarFormat formats {};
  1116. formats.full_format = parse_pattern("full"sv);
  1117. formats.long_format = parse_pattern("long"sv);
  1118. formats.medium_format = parse_pattern("medium"sv);
  1119. formats.short_format = parse_pattern("short"sv);
  1120. return locale_data.unique_formats.ensure(move(formats));
  1121. };
  1122. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  1123. // The generic calendar is not a supported Unicode calendar key, so skip it:
  1124. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  1125. if (calendar_name == "generic"sv)
  1126. return;
  1127. Calendar calendar {};
  1128. CalendarPatternList available_formats {};
  1129. if (!locale_data.calendars.contains_slow(calendar_name))
  1130. locale_data.calendars.append(calendar_name);
  1131. Vector<CalendarPattern> date_formats;
  1132. Vector<CalendarPattern> time_formats;
  1133. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  1134. auto const& date_skeletons_object = value.as_object().get("dateSkeletons"sv);
  1135. calendar.date_formats = parse_patterns(date_formats_object.as_object(), date_skeletons_object.as_object(), &date_formats);
  1136. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  1137. auto const& time_skeletons_object = value.as_object().get("timeSkeletons"sv);
  1138. calendar.time_formats = parse_patterns(time_formats_object.as_object(), time_skeletons_object.as_object(), &time_formats);
  1139. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  1140. calendar.date_time_formats = parse_patterns(date_time_formats_object.as_object(), JsonObject {}, nullptr);
  1141. auto const& available_formats_object = date_time_formats_object.as_object().get("availableFormats"sv);
  1142. available_formats_object.as_object().for_each_member([&](auto const& skeleton, JsonValue const& pattern) {
  1143. auto pattern_index = parse_date_time_pattern(pattern.as_string(), skeleton, locale_data);
  1144. if (!pattern_index.has_value())
  1145. return;
  1146. auto const& format = locale_data.unique_patterns.get(*pattern_index);
  1147. if (format.contains_only_date_fields())
  1148. date_formats.append(format);
  1149. else if (format.contains_only_time_fields())
  1150. time_formats.append(format);
  1151. if (!available_formats.contains_slow(*pattern_index))
  1152. available_formats.append(*pattern_index);
  1153. });
  1154. auto const& interval_formats_object = date_time_formats_object.as_object().get("intervalFormats"sv);
  1155. parse_interval_patterns(calendar, interval_formats_object.as_object(), locale_data);
  1156. generate_missing_patterns(calendar, available_formats, move(date_formats), move(time_formats), locale_data);
  1157. parse_calendar_symbols(calendar, value.as_object(), locale_data);
  1158. calendar.available_formats = locale_data.unique_pattern_lists.ensure(move(available_formats));
  1159. locale.calendars.set(calendar_name, locale_data.unique_calendars.ensure(move(calendar)));
  1160. });
  1161. return {};
  1162. }
  1163. static ErrorOr<void> parse_time_zone_names(String locale_time_zone_names_path, UnicodeLocaleData& locale_data, Locale& locale)
  1164. {
  1165. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  1166. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  1167. auto time_zone_names = TRY(read_json_file(time_zone_names_path.string()));
  1168. auto const& main_object = time_zone_names.as_object().get("main"sv);
  1169. auto const& locale_object = main_object.as_object().get(time_zone_names_path.parent().basename());
  1170. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1171. auto const& time_zone_names_object = dates_object.as_object().get("timeZoneNames"sv);
  1172. auto const& meta_zone_object = time_zone_names_object.as_object().get("metazone"sv);
  1173. auto const& hour_format_string = time_zone_names_object.as_object().get("hourFormat"sv);
  1174. auto const& gmt_format_string = time_zone_names_object.as_object().get("gmtFormat"sv);
  1175. auto const& gmt_zero_format_string = time_zone_names_object.as_object().get("gmtZeroFormat"sv);
  1176. if (meta_zone_object.is_null())
  1177. return {};
  1178. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object, StringView key) -> Optional<StringIndexType> {
  1179. auto const& names = meta_zone_object.get(type);
  1180. if (!names.is_object())
  1181. return {};
  1182. auto const& name = names.as_object().get(key);
  1183. if (name.is_string())
  1184. return locale_data.unique_strings.ensure(name.as_string());
  1185. return {};
  1186. };
  1187. auto parse_hour_format = [&](auto const& format, auto& time_zone_formats) {
  1188. auto hour_formats = format.split_view(';');
  1189. auto hour_format_ahead_start = hour_formats[0].find('H').value();
  1190. auto separator_ahead_start = hour_formats[0].find_last('H').value() + 1;
  1191. auto separator_ahead_end = hour_formats[0].find('m').value();
  1192. auto hour_format_behind_start = hour_formats[1].find('H').value();
  1193. auto separator_behind_start = hour_formats[1].find_last('H').value() + 1;
  1194. auto separator_behind_end = hour_formats[1].find('m').value();
  1195. auto symbol_ahead_sign = hour_formats[0].substring_view(0, hour_format_ahead_start);
  1196. auto symbol_ahead_separator = hour_formats[0].substring_view(separator_ahead_start, separator_ahead_end - separator_ahead_start);
  1197. auto symbol_behind_sign = hour_formats[1].substring_view(0, hour_format_behind_start);
  1198. auto symbol_behind_separator = hour_formats[1].substring_view(separator_behind_start, separator_behind_end - separator_behind_start);
  1199. time_zone_formats.symbol_ahead_sign = locale_data.unique_strings.ensure(symbol_ahead_sign);
  1200. time_zone_formats.symbol_ahead_separator = locale_data.unique_strings.ensure(symbol_ahead_separator);
  1201. time_zone_formats.symbol_behind_sign = locale_data.unique_strings.ensure(symbol_behind_sign);
  1202. time_zone_formats.symbol_behind_separator = locale_data.unique_strings.ensure(symbol_behind_separator);
  1203. };
  1204. TimeZoneNamesList time_zones;
  1205. TimeZoneFormat time_zone_formats {};
  1206. parse_hour_format(hour_format_string.as_string(), time_zone_formats);
  1207. time_zone_formats.gmt_format = locale_data.unique_strings.ensure(gmt_format_string.as_string());
  1208. time_zone_formats.gmt_zero_format = locale_data.unique_strings.ensure(gmt_zero_format_string.as_string());
  1209. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  1210. auto golden_zones = locale_data.meta_zones.find(meta_zone);
  1211. if (golden_zones == locale_data.meta_zones.end())
  1212. return;
  1213. TimeZoneNames time_zone_names {};
  1214. if (auto name = parse_name("long"sv, meta_zone_object, "standard"sv); name.has_value())
  1215. time_zone_names.long_standard_name = name.value();
  1216. if (auto name = parse_name("short"sv, meta_zone_object, "standard"sv); name.has_value())
  1217. time_zone_names.short_standard_name = name.value();
  1218. if (auto name = parse_name("long"sv, meta_zone_object, "daylight"sv); name.has_value())
  1219. time_zone_names.long_daylight_name = name.value();
  1220. if (auto name = parse_name("short"sv, meta_zone_object, "daylight"sv); name.has_value())
  1221. time_zone_names.short_daylight_name = name.value();
  1222. if (auto name = parse_name("long"sv, meta_zone_object, "generic"sv); name.has_value())
  1223. time_zone_names.long_generic_name = name.value();
  1224. if (auto name = parse_name("short"sv, meta_zone_object, "generic"sv); name.has_value())
  1225. time_zone_names.short_generic_name = name.value();
  1226. auto time_zone_index = locale_data.unique_time_zones.ensure(move(time_zone_names));
  1227. for (auto golden_zone : golden_zones->value) {
  1228. auto time_zone = to_underlying(golden_zone);
  1229. if (time_zone >= time_zones.size())
  1230. time_zones.resize(time_zone + 1);
  1231. time_zones[time_zone] = time_zone_index;
  1232. }
  1233. };
  1234. meta_zone_object.as_object().for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  1235. parse_time_zone(meta_zone, value.as_object());
  1236. });
  1237. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  1238. auto const& zone_object = time_zone_names_object.as_object().get("zone"sv);
  1239. auto const& etc_object = zone_object.as_object().get("Etc"sv);
  1240. auto const& utc_object = etc_object.as_object().get("UTC"sv);
  1241. parse_time_zone("UTC"sv, utc_object.as_object());
  1242. locale.time_zones = locale_data.unique_time_zone_lists.ensure(move(time_zones));
  1243. locale.time_zone_formats = locale_data.unique_time_zone_formats.ensure(move(time_zone_formats));
  1244. return {};
  1245. }
  1246. static ErrorOr<void> parse_day_periods(String core_path, UnicodeLocaleData& locale_data)
  1247. {
  1248. // https://unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rule_Sets
  1249. LexicalPath day_periods_path(move(core_path));
  1250. day_periods_path = day_periods_path.append("supplemental"sv);
  1251. day_periods_path = day_periods_path.append("dayPeriods.json"sv);
  1252. auto locale_day_periods = TRY(read_json_file(day_periods_path.string()));
  1253. auto const& supplemental_object = locale_day_periods.as_object().get("supplemental"sv);
  1254. auto const& day_periods_object = supplemental_object.as_object().get("dayPeriodRuleSet"sv);
  1255. auto parse_hour = [](auto const& time) {
  1256. auto hour_end_index = time.find(':').value();
  1257. // The times are in the form "hours:minutes", but we only need the hour segment.
  1258. // TR-35 explicitly states that minutes other than :00 must not be used.
  1259. VERIFY(time.substring_view(hour_end_index) == ":00"sv);
  1260. auto hour = time.substring_view(0, hour_end_index);
  1261. return hour.template to_uint<u8>().value();
  1262. };
  1263. auto parse_day_period = [&](auto const& symbol, auto const& ranges) -> Optional<DayPeriod> {
  1264. auto day_period = day_period_from_string(symbol);
  1265. if (!day_period.has_value())
  1266. return {};
  1267. auto begin = parse_hour(ranges.get("_from"sv).as_string());
  1268. auto end = parse_hour(ranges.get("_before"sv).as_string());
  1269. return DayPeriod { *day_period, begin, end };
  1270. };
  1271. day_periods_object.as_object().for_each_member([&](auto const& language, JsonValue const& value) {
  1272. auto locale = locale_data.locales.find(language);
  1273. if (locale == locale_data.locales.end())
  1274. return;
  1275. DayPeriodList day_periods;
  1276. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& ranges) {
  1277. if (auto day_period = parse_day_period(symbol, ranges.as_object()); day_period.has_value()) {
  1278. auto day_period_index = locale_data.unique_day_periods.ensure(day_period.release_value());
  1279. day_periods.append(day_period_index);
  1280. }
  1281. });
  1282. locale->value.day_periods = locale_data.unique_day_period_lists.ensure(move(day_periods));
  1283. });
  1284. return {};
  1285. }
  1286. static ErrorOr<void> parse_all_locales(String core_path, String dates_path, UnicodeLocaleData& locale_data)
  1287. {
  1288. TRY(parse_hour_cycles(core_path, locale_data));
  1289. TRY(parse_meta_zones(core_path, locale_data));
  1290. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  1291. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  1292. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  1293. StringBuilder builder;
  1294. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  1295. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  1296. builder.appendff("-{}", script);
  1297. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  1298. builder.appendff("-{}", region);
  1299. return builder.build();
  1300. };
  1301. while (dates_iterator.has_next()) {
  1302. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  1303. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  1304. auto language = TRY(remove_variants_from_path(dates_path));
  1305. auto& locale = locale_data.locales.ensure(language);
  1306. while (calendars_iterator.has_next()) {
  1307. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  1308. TRY(parse_calendars(move(calendars_path), locale_data, locale));
  1309. }
  1310. TRY(parse_time_zone_names(move(dates_path), locale_data, locale));
  1311. }
  1312. TRY(parse_day_periods(move(core_path), locale_data));
  1313. return {};
  1314. }
  1315. static String format_identifier(StringView owner, String identifier)
  1316. {
  1317. identifier = identifier.replace("-"sv, "_"sv, ReplaceMode::All);
  1318. identifier = identifier.replace("/"sv, "_"sv, ReplaceMode::All);
  1319. if (all_of(identifier, is_ascii_digit))
  1320. return String::formatted("{}_{}", owner[0], identifier);
  1321. if (is_ascii_lower_alpha(identifier[0]))
  1322. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  1323. return identifier;
  1324. }
  1325. static ErrorOr<void> generate_unicode_locale_header(Core::Stream::BufferedFile& file, UnicodeLocaleData& locale_data)
  1326. {
  1327. StringBuilder builder;
  1328. SourceGenerator generator { builder };
  1329. generator.append(R"~~~(
  1330. #pragma once
  1331. #include <AK/Types.h>
  1332. namespace Unicode {
  1333. )~~~");
  1334. generate_enum(generator, format_identifier, "Calendar"sv, {}, locale_data.calendars);
  1335. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, locale_data.hour_cycle_regions);
  1336. generator.append(R"~~~(
  1337. }
  1338. )~~~");
  1339. TRY(file.write(generator.as_string_view().bytes()));
  1340. return {};
  1341. }
  1342. static ErrorOr<void> generate_unicode_locale_implementation(Core::Stream::BufferedFile& file, UnicodeLocaleData& locale_data)
  1343. {
  1344. StringBuilder builder;
  1345. SourceGenerator generator { builder };
  1346. generator.set("string_index_type"sv, s_string_index_type);
  1347. generator.set("calendar_pattern_index_type"sv, s_calendar_pattern_index_type);
  1348. generator.set("calendar_pattern_list_index_type"sv, s_calendar_pattern_list_index_type);
  1349. generator.set("calendar_range_pattern_index_type"sv, s_calendar_range_pattern_index_type);
  1350. generator.set("calendar_range_pattern_list_index_type"sv, s_calendar_range_pattern_list_index_type);
  1351. generator.set("calendar_format_index_type"sv, s_calendar_format_index_type);
  1352. generator.set("symbol_list_index_type"sv, s_symbol_list_index_type);
  1353. generator.set("calendar_symbols_index_type"sv, s_calendar_symbols_index_type);
  1354. generator.set("calendar_symbols_list_index_type"sv, s_calendar_symbols_list_index_type);
  1355. generator.set("calendar_index_type"sv, s_calendar_index_type);
  1356. generator.set("time_zone_index_type"sv, s_time_zone_index_type);
  1357. generator.set("time_zone_list_index_type"sv, s_time_zone_list_index_type);
  1358. generator.set("day_period_index_type"sv, s_day_period_index_type);
  1359. generator.set("day_period_list_index_type"sv, s_day_period_list_index_type);
  1360. generator.append(R"~~~(
  1361. #include <AK/Array.h>
  1362. #include <AK/BinarySearch.h>
  1363. #include <AK/Optional.h>
  1364. #include <AK/StringView.h>
  1365. #include <LibTimeZone/TimeZone.h>
  1366. #include <LibUnicode/DateTimeFormat.h>
  1367. #include <LibUnicode/Locale.h>
  1368. #include <LibUnicode/UnicodeDateTimeFormat.h>
  1369. #include <LibUnicode/UnicodeLocale.h>
  1370. namespace Unicode {
  1371. )~~~");
  1372. locale_data.unique_strings.generate(generator);
  1373. generator.append(R"~~~(
  1374. template <typename SourceType, typename TargetType>
  1375. static void convert_calendar_fields(SourceType const& source, TargetType& target)
  1376. {
  1377. if (source.era != -1)
  1378. target.era = static_cast<CalendarPatternStyle>(source.era);
  1379. if (source.year != -1)
  1380. target.year = static_cast<CalendarPatternStyle>(source.year);
  1381. if (source.month != -1)
  1382. target.month = static_cast<CalendarPatternStyle>(source.month);
  1383. if (source.weekday != -1)
  1384. target.weekday = static_cast<CalendarPatternStyle>(source.weekday);
  1385. if (source.day != -1)
  1386. target.day = static_cast<CalendarPatternStyle>(source.day);
  1387. if (source.day_period != -1)
  1388. target.day_period = static_cast<CalendarPatternStyle>(source.day_period);
  1389. if (source.hour != -1)
  1390. target.hour = static_cast<CalendarPatternStyle>(source.hour);
  1391. if (source.minute != -1)
  1392. target.minute = static_cast<CalendarPatternStyle>(source.minute);
  1393. if (source.second != -1)
  1394. target.second = static_cast<CalendarPatternStyle>(source.second);
  1395. if (source.fractional_second_digits != -1)
  1396. target.fractional_second_digits = static_cast<u8>(source.fractional_second_digits);
  1397. if (source.time_zone_name != -1)
  1398. target.time_zone_name = static_cast<CalendarPatternStyle>(source.time_zone_name);
  1399. }
  1400. struct CalendarPatternImpl {
  1401. CalendarPattern to_unicode_calendar_pattern() const {
  1402. CalendarPattern calendar_pattern {};
  1403. calendar_pattern.skeleton = s_string_list[skeleton];
  1404. calendar_pattern.pattern = s_string_list[pattern];
  1405. if (pattern12 != 0)
  1406. calendar_pattern.pattern12 = s_string_list[pattern12];
  1407. convert_calendar_fields(*this, calendar_pattern);
  1408. return calendar_pattern;
  1409. }
  1410. @string_index_type@ skeleton { 0 };
  1411. @string_index_type@ pattern { 0 };
  1412. @string_index_type@ pattern12 { 0 };
  1413. i8 era { -1 };
  1414. i8 year { -1 };
  1415. i8 month { -1 };
  1416. i8 weekday { -1 };
  1417. i8 day { -1 };
  1418. i8 day_period { -1 };
  1419. i8 hour { -1 };
  1420. i8 minute { -1 };
  1421. i8 second { -1 };
  1422. i8 fractional_second_digits { -1 };
  1423. i8 time_zone_name { -1 };
  1424. };
  1425. struct CalendarRangePatternImpl {
  1426. CalendarRangePattern to_unicode_calendar_range_pattern() const {
  1427. CalendarRangePattern calendar_range_pattern {};
  1428. if (field != -1)
  1429. calendar_range_pattern.field = static_cast<CalendarRangePattern::Field>(field);
  1430. calendar_range_pattern.start_range = s_string_list[start_range];
  1431. calendar_range_pattern.separator = s_string_list[separator];
  1432. calendar_range_pattern.end_range = s_string_list[end_range];
  1433. convert_calendar_fields(*this, calendar_range_pattern);
  1434. return calendar_range_pattern;
  1435. }
  1436. @string_index_type@ skeleton { 0 };
  1437. i8 field { -1 };
  1438. @string_index_type@ start_range { 0 };
  1439. @string_index_type@ separator { 0 };
  1440. @string_index_type@ end_range { 0 };
  1441. i8 era { -1 };
  1442. i8 year { -1 };
  1443. i8 month { -1 };
  1444. i8 weekday { -1 };
  1445. i8 day { -1 };
  1446. i8 day_period { -1 };
  1447. i8 hour { -1 };
  1448. i8 minute { -1 };
  1449. i8 second { -1 };
  1450. i8 fractional_second_digits { -1 };
  1451. i8 time_zone_name { -1 };
  1452. };
  1453. )~~~");
  1454. locale_data.unique_patterns.generate(generator, "CalendarPatternImpl"sv, "s_calendar_patterns"sv, 10);
  1455. locale_data.unique_pattern_lists.generate(generator, s_calendar_pattern_index_type, "s_calendar_pattern_lists"sv);
  1456. locale_data.unique_range_patterns.generate(generator, "CalendarRangePatternImpl"sv, "s_calendar_range_patterns"sv, 10);
  1457. locale_data.unique_range_pattern_lists.generate(generator, s_calendar_range_pattern_index_type, "s_calendar_range_pattern_lists"sv);
  1458. generator.append(R"~~~(
  1459. struct CalendarFormatImpl {
  1460. CalendarFormat to_unicode_calendar_format() const {
  1461. CalendarFormat calendar_format {};
  1462. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  1463. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  1464. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  1465. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  1466. return calendar_format;
  1467. }
  1468. @calendar_pattern_index_type@ full_format { 0 };
  1469. @calendar_pattern_index_type@ long_format { 0 };
  1470. @calendar_pattern_index_type@ medium_format { 0 };
  1471. @calendar_pattern_index_type@ short_format { 0 };
  1472. };
  1473. struct CalendarSymbols {
  1474. @symbol_list_index_type@ narrow_symbols { 0 };
  1475. @symbol_list_index_type@ short_symbols { 0 };
  1476. @symbol_list_index_type@ long_symbols { 0 };
  1477. };
  1478. struct CalendarData {
  1479. @calendar_format_index_type@ date_formats { 0 };
  1480. @calendar_format_index_type@ time_formats { 0 };
  1481. @calendar_format_index_type@ date_time_formats { 0 };
  1482. @calendar_pattern_list_index_type@ available_formats { 0 };
  1483. @calendar_range_pattern_index_type@ default_range_format { 0 };
  1484. @calendar_range_pattern_list_index_type@ range_formats { 0 };
  1485. @calendar_range_pattern_list_index_type@ range12_formats { 0 };
  1486. @calendar_symbols_list_index_type@ symbols { 0 };
  1487. };
  1488. struct TimeZoneNames {
  1489. @string_index_type@ short_standard_name { 0 };
  1490. @string_index_type@ long_standard_name { 0 };
  1491. @string_index_type@ short_daylight_name { 0 };
  1492. @string_index_type@ long_daylight_name { 0 };
  1493. @string_index_type@ short_generic_name { 0 };
  1494. @string_index_type@ long_generic_name { 0 };
  1495. };
  1496. struct TimeZoneFormatImpl {
  1497. TimeZoneFormat to_time_zone_format() const {
  1498. TimeZoneFormat time_zone_format {};
  1499. time_zone_format.symbol_ahead_sign = s_string_list[symbol_ahead_sign];
  1500. time_zone_format.symbol_ahead_separator = s_string_list[symbol_ahead_separator];
  1501. time_zone_format.symbol_behind_sign = s_string_list[symbol_behind_sign];
  1502. time_zone_format.symbol_behind_separator = s_string_list[symbol_behind_separator];
  1503. time_zone_format.gmt_format = s_string_list[gmt_format];
  1504. time_zone_format.gmt_zero_format = s_string_list[gmt_zero_format];
  1505. return time_zone_format;
  1506. }
  1507. @string_index_type@ symbol_ahead_sign { 0 };
  1508. @string_index_type@ symbol_ahead_separator { 0 };
  1509. @string_index_type@ symbol_behind_sign { 0 };
  1510. @string_index_type@ symbol_behind_separator { 0 };
  1511. @string_index_type@ gmt_format { 0 };
  1512. @string_index_type@ gmt_zero_format { 0 };
  1513. };
  1514. struct DayPeriodData {
  1515. u8 day_period { 0 };
  1516. u8 begin { 0 };
  1517. u8 end { 0 };
  1518. };
  1519. )~~~");
  1520. locale_data.unique_formats.generate(generator, "CalendarFormatImpl"sv, "s_calendar_formats"sv, 10);
  1521. locale_data.unique_symbol_lists.generate(generator, s_string_index_type, "s_symbol_lists"sv);
  1522. locale_data.unique_calendar_symbols.generate(generator, "CalendarSymbols"sv, "s_calendar_symbols"sv, 10);
  1523. locale_data.unique_calendar_symbols_lists.generate(generator, s_calendar_symbols_index_type, "s_calendar_symbol_lists"sv);
  1524. locale_data.unique_calendars.generate(generator, "CalendarData"sv, "s_calendars"sv, 10);
  1525. locale_data.unique_time_zones.generate(generator, "TimeZoneNames"sv, "s_time_zones"sv, 30);
  1526. locale_data.unique_time_zone_lists.generate(generator, s_time_zone_index_type, "s_time_zone_lists"sv);
  1527. locale_data.unique_time_zone_formats.generate(generator, "TimeZoneFormatImpl"sv, "s_time_zone_formats"sv, 30);
  1528. locale_data.unique_day_periods.generate(generator, "DayPeriodData"sv, "s_day_periods"sv, 30);
  1529. locale_data.unique_day_period_lists.generate(generator, s_day_period_index_type, "s_day_period_lists"sv);
  1530. locale_data.unique_hour_cycle_lists.generate(generator, "u8"sv, "s_hour_cycle_lists"sv);
  1531. auto append_calendars = [&](String name, auto const& calendars) {
  1532. generator.set("name", name);
  1533. generator.set("size", String::number(calendars.size()));
  1534. generator.append(R"~~~(
  1535. static constexpr Array<@calendar_index_type@, @size@> @name@ { {)~~~");
  1536. bool first = true;
  1537. for (auto const& calendar_key : locale_data.calendars) {
  1538. auto calendar = calendars.find(calendar_key)->value;
  1539. generator.append(first ? " " : ", ");
  1540. generator.append(String::number(calendar));
  1541. first = false;
  1542. }
  1543. generator.append(" } };");
  1544. };
  1545. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  1546. generator.set("type", type);
  1547. generator.set("name", name);
  1548. generator.set("size", String::number(keys.size()));
  1549. generator.append(R"~~~(
  1550. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1551. bool first = true;
  1552. for (auto const& key : keys) {
  1553. auto const& value = map.find(key)->value;
  1554. auto mapping = mapping_getter(value);
  1555. generator.append(first ? " " : ", ");
  1556. generator.append(String::number(mapping));
  1557. first = false;
  1558. }
  1559. generator.append(" } };");
  1560. };
  1561. auto locales = locale_data.locales.keys();
  1562. quick_sort(locales);
  1563. generate_mapping(generator, locale_data.locales, s_calendar_index_type, "s_locale_calendars"sv, "s_calendars_{}", format_identifier, [&](auto const& name, auto const& value) { append_calendars(name, value.calendars); });
  1564. append_mapping(locales, locale_data.locales, s_time_zone_index_type, "s_locale_time_zones"sv, [](auto const& locale) { return locale.time_zones; });
  1565. 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; });
  1566. append_mapping(locales, locale_data.locales, s_day_period_index_type, "s_locale_day_periods"sv, [](auto const& locale) { return locale.day_periods; });
  1567. 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; });
  1568. generator.append("\n");
  1569. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1570. HashValueMap<String> hashes;
  1571. hashes.ensure_capacity(values.size());
  1572. for (auto const& value : values)
  1573. hashes.set(value.hash(), format_identifier(enum_title, value));
  1574. for (auto const& alias : aliases)
  1575. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1576. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1577. };
  1578. append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, locale_data.hour_cycle_regions);
  1579. generator.append(R"~~~(
  1580. static Optional<Calendar> keyword_to_calendar(KeywordCalendar keyword)
  1581. {
  1582. switch (keyword) {)~~~");
  1583. for (auto const& calendar : locale_data.calendars) {
  1584. generator.set("name"sv, format_identifier({}, calendar));
  1585. generator.append(R"~~~(
  1586. case KeywordCalendar::@name@:
  1587. return Calendar::@name@;)~~~");
  1588. }
  1589. generator.append(R"~~~(
  1590. default:
  1591. return {};
  1592. }
  1593. }
  1594. Vector<HourCycle> get_regional_hour_cycles(StringView region)
  1595. {
  1596. auto region_value = hour_cycle_region_from_string(region);
  1597. if (!region_value.has_value())
  1598. return {};
  1599. auto region_index = to_underlying(*region_value);
  1600. auto regional_hour_cycles_index = s_hour_cycles.at(region_index);
  1601. auto const& regional_hour_cycles = s_hour_cycle_lists.at(regional_hour_cycles_index);
  1602. Vector<HourCycle> hour_cycles;
  1603. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1604. for (auto hour_cycle : regional_hour_cycles)
  1605. hour_cycles.unchecked_append(static_cast<HourCycle>(hour_cycle));
  1606. return hour_cycles;
  1607. }
  1608. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1609. {
  1610. auto locale_value = locale_from_string(locale);
  1611. if (!locale_value.has_value())
  1612. return nullptr;
  1613. auto calendar_keyword = keyword_ca_from_string(calendar);
  1614. if (!calendar_keyword.has_value())
  1615. return {};
  1616. auto calendar_value = keyword_to_calendar(*calendar_keyword);
  1617. if (!calendar_value.has_value())
  1618. return {};
  1619. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1620. size_t calendar_index = to_underlying(*calendar_value);
  1621. auto const& calendar_indices = s_locale_calendars.at(locale_index);
  1622. calendar_index = calendar_indices[calendar_index];
  1623. return &s_calendars[calendar_index];
  1624. }
  1625. Optional<CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1626. {
  1627. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1628. auto const& formats = s_calendar_formats.at(data->date_formats);
  1629. return formats.to_unicode_calendar_format();
  1630. }
  1631. return {};
  1632. }
  1633. Optional<CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1634. {
  1635. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1636. auto const& formats = s_calendar_formats.at(data->time_formats);
  1637. return formats.to_unicode_calendar_format();
  1638. }
  1639. return {};
  1640. }
  1641. Optional<CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1642. {
  1643. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1644. auto const& formats = s_calendar_formats.at(data->date_time_formats);
  1645. return formats.to_unicode_calendar_format();
  1646. }
  1647. return {};
  1648. }
  1649. Vector<CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1650. {
  1651. Vector<CalendarPattern> result {};
  1652. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1653. auto const& available_formats = s_calendar_pattern_lists.at(data->available_formats);
  1654. result.ensure_capacity(available_formats.size());
  1655. for (auto const& format : available_formats)
  1656. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1657. }
  1658. return result;
  1659. }
  1660. Optional<CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar)
  1661. {
  1662. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1663. auto const& pattern = s_calendar_range_patterns[data->default_range_format];
  1664. return pattern.to_unicode_calendar_range_pattern();
  1665. }
  1666. return {};
  1667. }
  1668. Vector<CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton)
  1669. {
  1670. Vector<CalendarRangePattern> result {};
  1671. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1672. auto const& range_formats = s_calendar_range_pattern_lists.at(data->range_formats);
  1673. for (auto format : range_formats) {
  1674. auto const& pattern = s_calendar_range_patterns[format];
  1675. if (skeleton == s_string_list[pattern.skeleton])
  1676. result.append(pattern.to_unicode_calendar_range_pattern());
  1677. }
  1678. }
  1679. return result;
  1680. }
  1681. Vector<CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton)
  1682. {
  1683. Vector<CalendarRangePattern> result {};
  1684. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1685. auto const& range12_formats = s_calendar_range_pattern_lists.at(data->range12_formats);
  1686. for (auto format : range12_formats) {
  1687. auto const& pattern = s_calendar_range_patterns[format];
  1688. if (skeleton == s_string_list[pattern.skeleton])
  1689. result.append(pattern.to_unicode_calendar_range_pattern());
  1690. }
  1691. }
  1692. return result;
  1693. }
  1694. static Span<@string_index_type@ const> find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1695. {
  1696. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1697. auto const& symbols_list = s_calendar_symbol_lists[data->symbols];
  1698. auto symbol_index = to_underlying(symbol);
  1699. auto calendar_symbols_index = symbols_list.at(symbol_index);
  1700. auto const& symbols = s_calendar_symbols.at(calendar_symbols_index);
  1701. @symbol_list_index_type@ symbol_list_index = 0;
  1702. switch (style) {
  1703. case CalendarPatternStyle::Narrow:
  1704. symbol_list_index = symbols.narrow_symbols;
  1705. break;
  1706. case CalendarPatternStyle::Short:
  1707. symbol_list_index = symbols.short_symbols;
  1708. break;
  1709. case CalendarPatternStyle::Long:
  1710. symbol_list_index = symbols.long_symbols;
  1711. break;
  1712. default:
  1713. VERIFY_NOT_REACHED();
  1714. }
  1715. return s_symbol_lists.at(symbol_list_index);
  1716. }
  1717. return {};
  1718. }
  1719. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Era value)
  1720. {
  1721. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1722. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1723. return s_string_list[symbols.at(value_index)];
  1724. return {};
  1725. }
  1726. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Month value)
  1727. {
  1728. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1729. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1730. return s_string_list[symbols.at(value_index)];
  1731. return {};
  1732. }
  1733. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Weekday value)
  1734. {
  1735. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1736. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1737. return s_string_list[symbols.at(value_index)];
  1738. return {};
  1739. }
  1740. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, DayPeriod value)
  1741. {
  1742. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1743. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1744. return s_string_list[symbols.at(value_index)];
  1745. return {};
  1746. }
  1747. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour)
  1748. {
  1749. auto locale_value = locale_from_string(locale);
  1750. if (!locale_value.has_value())
  1751. return {};
  1752. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1753. auto day_periods_index = s_locale_day_periods[locale_index];
  1754. auto day_periods = s_day_period_lists[day_periods_index];
  1755. for (auto day_period_index : day_periods) {
  1756. auto day_period = s_day_periods[day_period_index];
  1757. auto h = hour;
  1758. if (day_period.begin > day_period.end) {
  1759. day_period.end += 24;
  1760. h += 24;
  1761. }
  1762. if ((day_period.begin <= h) && (h < day_period.end)) {
  1763. auto period = static_cast<DayPeriod>(day_period.day_period);
  1764. return get_calendar_day_period_symbol(locale, calendar, style, period);
  1765. }
  1766. }
  1767. // Fallback to fixed periods if the locale does not have flexible day periods.
  1768. // TR-35 states that the meaning of AM and PM does not change with locale.
  1769. if (hour < 12)
  1770. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::AM);
  1771. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::PM);
  1772. }
  1773. Optional<TimeZoneFormat> get_time_zone_format(StringView locale)
  1774. {
  1775. auto locale_value = locale_from_string(locale);
  1776. if (!locale_value.has_value())
  1777. return {};
  1778. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1779. auto time_zone_format_index = s_locale_time_zone_formats.at(locale_index);
  1780. auto const& time_zone_format = s_time_zone_formats.at(time_zone_format_index);
  1781. return time_zone_format.to_time_zone_format();
  1782. }
  1783. static TimeZoneNames const* find_time_zone_names(StringView locale, StringView time_zone)
  1784. {
  1785. auto locale_value = locale_from_string(locale);
  1786. if (!locale_value.has_value())
  1787. return nullptr;
  1788. auto time_zone_value = ::TimeZone::time_zone_from_string(time_zone);
  1789. if (!time_zone_value.has_value())
  1790. return nullptr;
  1791. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1792. size_t time_zone_index = to_underlying(*time_zone_value);
  1793. auto time_zone_list_index = s_locale_time_zones.at(locale_index);
  1794. auto const& time_zone_list = s_time_zone_lists.at(time_zone_list_index);
  1795. if (time_zone_list.size() <= time_zone_index)
  1796. return nullptr;
  1797. time_zone_index = time_zone_list.at(time_zone_index);
  1798. return &s_time_zones[time_zone_index];
  1799. }
  1800. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style, TimeZone::InDST in_dst)
  1801. {
  1802. if (auto const* data = find_time_zone_names(locale, time_zone); data != nullptr) {
  1803. size_t name_index = 0;
  1804. switch (style) {
  1805. case CalendarPatternStyle::Short:
  1806. name_index = (in_dst == TimeZone::InDST::No) ? data->short_standard_name : data->short_daylight_name;
  1807. break;
  1808. case CalendarPatternStyle::Long:
  1809. name_index = (in_dst == TimeZone::InDST::No) ? data->long_standard_name : data->long_daylight_name;
  1810. break;
  1811. case CalendarPatternStyle::ShortGeneric:
  1812. name_index = data->short_generic_name;
  1813. break;
  1814. case CalendarPatternStyle::LongGeneric:
  1815. name_index = data->long_generic_name;
  1816. break;
  1817. default:
  1818. VERIFY_NOT_REACHED();
  1819. }
  1820. if (name_index != 0)
  1821. return s_string_list[name_index];
  1822. }
  1823. return {};
  1824. }
  1825. }
  1826. )~~~");
  1827. TRY(file.write(generator.as_string_view().bytes()));
  1828. return {};
  1829. }
  1830. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1831. {
  1832. StringView generated_header_path;
  1833. StringView generated_implementation_path;
  1834. StringView core_path;
  1835. StringView dates_path;
  1836. Core::ArgsParser args_parser;
  1837. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1838. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1839. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1840. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1841. args_parser.parse(arguments);
  1842. auto generated_header_file = TRY(open_file(generated_header_path, Core::Stream::OpenMode::Write));
  1843. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::Stream::OpenMode::Write));
  1844. UnicodeLocaleData locale_data;
  1845. TRY(parse_all_locales(core_path, dates_path, locale_data));
  1846. TRY(generate_unicode_locale_header(*generated_header_file, locale_data));
  1847. TRY(generate_unicode_locale_implementation(*generated_implementation_file, locale_data));
  1848. return 0;
  1849. }