GenerateUnicodeDateTimeFormat.cpp 89 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264
  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. Vector<Alias> calendar_aliases {
  487. // FIXME: Aliases should come from BCP47. See: https://unicode-org.atlassian.net/browse/CLDR-15158
  488. { "gregorian"sv, "gregory"sv },
  489. };
  490. };
  491. static Optional<Unicode::DayPeriod> day_period_from_string(StringView day_period)
  492. {
  493. if (day_period == "am"sv)
  494. return Unicode::DayPeriod::AM;
  495. if (day_period == "pm"sv)
  496. return Unicode::DayPeriod::PM;
  497. if (day_period == "morning1"sv)
  498. return Unicode::DayPeriod::Morning1;
  499. if (day_period == "morning2"sv)
  500. return Unicode::DayPeriod::Morning2;
  501. if (day_period == "afternoon1"sv)
  502. return Unicode::DayPeriod::Afternoon1;
  503. if (day_period == "afternoon2"sv)
  504. return Unicode::DayPeriod::Afternoon2;
  505. if (day_period == "evening1"sv)
  506. return Unicode::DayPeriod::Evening1;
  507. if (day_period == "evening2"sv)
  508. return Unicode::DayPeriod::Evening2;
  509. if (day_period == "night1"sv)
  510. return Unicode::DayPeriod::Night1;
  511. if (day_period == "night2"sv)
  512. return Unicode::DayPeriod::Night2;
  513. return {};
  514. }
  515. static ErrorOr<void> parse_hour_cycles(String core_path, UnicodeLocaleData& locale_data)
  516. {
  517. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  518. LexicalPath time_data_path(move(core_path));
  519. time_data_path = time_data_path.append("supplemental"sv);
  520. time_data_path = time_data_path.append("timeData.json"sv);
  521. auto time_data = TRY(read_json_file(time_data_path.string()));
  522. auto const& supplemental_object = time_data.as_object().get("supplemental"sv);
  523. auto const& time_data_object = supplemental_object.as_object().get("timeData"sv);
  524. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Unicode::HourCycle> {
  525. if (hour_cycle == "h"sv)
  526. return Unicode::HourCycle::H12;
  527. if (hour_cycle == "H"sv)
  528. return Unicode::HourCycle::H23;
  529. if (hour_cycle == "K"sv)
  530. return Unicode::HourCycle::H11;
  531. if (hour_cycle == "k"sv)
  532. return Unicode::HourCycle::H24;
  533. return {};
  534. };
  535. time_data_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  536. auto allowed_hour_cycles_string = value.as_object().get("_allowed"sv).as_string();
  537. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  538. Vector<Unicode::HourCycle> hour_cycles;
  539. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  540. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  541. hour_cycles.append(*hour_cycle);
  542. }
  543. auto hour_cycles_index = locale_data.unique_hour_cycle_lists.ensure(move(hour_cycles));
  544. locale_data.hour_cycles.set(key, hour_cycles_index);
  545. if (!locale_data.hour_cycle_regions.contains_slow(key))
  546. locale_data.hour_cycle_regions.append(key);
  547. });
  548. return {};
  549. }
  550. static ErrorOr<void> parse_meta_zones(String core_path, UnicodeLocaleData& locale_data)
  551. {
  552. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  553. LexicalPath meta_zone_path(move(core_path));
  554. meta_zone_path = meta_zone_path.append("supplemental"sv);
  555. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  556. auto meta_zone = TRY(read_json_file(meta_zone_path.string()));
  557. auto const& supplemental_object = meta_zone.as_object().get("supplemental"sv);
  558. auto const& meta_zone_object = supplemental_object.as_object().get("metaZones"sv);
  559. auto const& meta_zone_array = meta_zone_object.as_object().get("metazones"sv);
  560. meta_zone_array.as_array().for_each([&](JsonValue const& value) {
  561. auto const& mapping = value.as_object().get("mapZone"sv);
  562. auto const& meta_zone = mapping.as_object().get("_other"sv);
  563. auto const& golden_zone = mapping.as_object().get("_type"sv);
  564. if (auto time_zone = TimeZone::time_zone_from_string(golden_zone.as_string()); time_zone.has_value()) {
  565. auto& golden_zones = locale_data.meta_zones.ensure(meta_zone.as_string());
  566. golden_zones.append(*time_zone);
  567. }
  568. });
  569. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  570. if (auto time_zone = TimeZone::time_zone_from_string("UTC"sv); time_zone.has_value())
  571. locale_data.meta_zones.set("UTC"sv, { *time_zone });
  572. return {};
  573. }
  574. static constexpr auto is_char(char ch)
  575. {
  576. return [ch](auto c) { return c == ch; };
  577. }
  578. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  579. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  580. // extra spaces around the day period. Some example expected removals:
  581. //
  582. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  583. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  584. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  585. static String remove_period_from_pattern(String pattern)
  586. {
  587. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  588. auto index = pattern.find(remove);
  589. if (!index.has_value())
  590. continue;
  591. constexpr u32 space = ' ';
  592. constexpr u32 open = '{';
  593. constexpr u32 close = '}';
  594. Utf8View utf8_pattern { pattern };
  595. Optional<u32> before_removal;
  596. Optional<u32> after_removal;
  597. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  598. before_removal = *it;
  599. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  600. after_removal = *it;
  601. if ((before_removal == space) && (after_removal != open)) {
  602. pattern = String::formatted("{}{}",
  603. pattern.substring_view(0, *index - 1),
  604. pattern.substring_view(*index + remove.length()));
  605. } else if ((after_removal == space) && (before_removal != close)) {
  606. pattern = String::formatted("{}{}",
  607. pattern.substring_view(0, *index),
  608. pattern.substring_view(*index + remove.length() + 1));
  609. } else {
  610. pattern = String::formatted("{}{}",
  611. pattern.substring_view(0, *index),
  612. pattern.substring_view(*index + remove.length()));
  613. }
  614. }
  615. return pattern;
  616. }
  617. static Optional<CalendarPattern> parse_date_time_pattern_raw(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  618. {
  619. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  620. using Unicode::CalendarPatternStyle;
  621. CalendarPattern format {};
  622. if (!skeleton.is_empty())
  623. format.skeleton_index = locale_data.unique_strings.ensure(move(skeleton));
  624. GenericLexer lexer { pattern };
  625. StringBuilder builder;
  626. bool hour12 { false };
  627. while (!lexer.is_eof()) {
  628. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  629. // processing (this just avoids conflicts with the patterns below).
  630. if (lexer.next_is(is_quote)) {
  631. builder.append(lexer.consume_quoted_string());
  632. continue;
  633. }
  634. auto starting_char = lexer.peek();
  635. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  636. // Era
  637. if (all_of(segment, is_char('G'))) {
  638. builder.append("{era}");
  639. if (segment.length() <= 3)
  640. format.era = CalendarPatternStyle::Short;
  641. else if (segment.length() == 4)
  642. format.era = CalendarPatternStyle::Long;
  643. else
  644. format.era = CalendarPatternStyle::Narrow;
  645. }
  646. // Year
  647. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  648. builder.append("{year}");
  649. if (segment.length() == 2)
  650. format.year = CalendarPatternStyle::TwoDigit;
  651. else
  652. format.year = CalendarPatternStyle::Numeric;
  653. }
  654. // Quarter
  655. else if (all_of(segment, is_any_of("qQ"sv))) {
  656. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  657. return {};
  658. }
  659. // Month
  660. else if (all_of(segment, is_any_of("ML"sv))) {
  661. builder.append("{month}");
  662. if (segment.length() == 1)
  663. format.month = CalendarPatternStyle::Numeric;
  664. else if (segment.length() == 2)
  665. format.month = CalendarPatternStyle::TwoDigit;
  666. else if (segment.length() == 3)
  667. format.month = CalendarPatternStyle::Short;
  668. else if (segment.length() == 4)
  669. format.month = CalendarPatternStyle::Long;
  670. else if (segment.length() == 5)
  671. format.month = CalendarPatternStyle::Narrow;
  672. } else if (all_of(segment, is_char('l'))) {
  673. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  674. return {};
  675. }
  676. // Week
  677. else if (all_of(segment, is_any_of("wW"sv))) {
  678. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  679. return {};
  680. }
  681. // Day
  682. else if (all_of(segment, is_char('d'))) {
  683. builder.append("{day}");
  684. if (segment.length() == 1)
  685. format.day = CalendarPatternStyle::Numeric;
  686. else
  687. format.day = CalendarPatternStyle::TwoDigit;
  688. } else if (all_of(segment, is_any_of("DFg"sv))) {
  689. builder.append("{day}");
  690. format.day = CalendarPatternStyle::Numeric;
  691. }
  692. // Weekday
  693. else if (all_of(segment, is_char('E'))) {
  694. builder.append("{weekday}");
  695. if (segment.length() == 4)
  696. format.weekday = CalendarPatternStyle::Long;
  697. else if (segment.length() == 5)
  698. format.weekday = CalendarPatternStyle::Narrow;
  699. else
  700. format.weekday = CalendarPatternStyle::Short;
  701. } else if (all_of(segment, is_any_of("ec"sv))) {
  702. builder.append("{weekday}");
  703. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  704. // pattern styles are not supported by Intl.DateTimeFormat.
  705. if (segment.length() <= 2)
  706. return {};
  707. if (segment.length() == 4)
  708. format.weekday = CalendarPatternStyle::Long;
  709. else if (segment.length() == 5)
  710. format.weekday = CalendarPatternStyle::Narrow;
  711. else
  712. format.weekday = CalendarPatternStyle::Short;
  713. }
  714. // Period
  715. else if (all_of(segment, is_any_of("ab"sv))) {
  716. builder.append("{ampm}");
  717. hour12 = true;
  718. } else if (all_of(segment, is_char('B'))) {
  719. builder.append("{dayPeriod}");
  720. hour12 = true;
  721. if (segment.length() == 4)
  722. format.day_period = CalendarPatternStyle::Long;
  723. else if (segment.length() == 5)
  724. format.day_period = CalendarPatternStyle::Narrow;
  725. else
  726. format.day_period = CalendarPatternStyle::Short;
  727. }
  728. // Hour
  729. else if (all_of(segment, is_any_of("hHKk"sv))) {
  730. builder.append("{hour}");
  731. if ((segment[0] == 'h') || (segment[0] == 'K'))
  732. hour12 = true;
  733. if (segment.length() == 1)
  734. format.hour = CalendarPatternStyle::Numeric;
  735. else
  736. format.hour = CalendarPatternStyle::TwoDigit;
  737. } else if (all_of(segment, is_any_of("jJC"sv))) {
  738. // TR-35 indicates these should not be used.
  739. return {};
  740. }
  741. // Minute
  742. else if (all_of(segment, is_char('m'))) {
  743. builder.append("{minute}");
  744. if (segment.length() == 1)
  745. format.minute = CalendarPatternStyle::Numeric;
  746. else
  747. format.minute = CalendarPatternStyle::TwoDigit;
  748. }
  749. // Second
  750. else if (all_of(segment, is_char('s'))) {
  751. builder.append("{second}");
  752. if (segment.length() == 1)
  753. format.second = CalendarPatternStyle::Numeric;
  754. else
  755. format.second = CalendarPatternStyle::TwoDigit;
  756. } else if (all_of(segment, is_char('S'))) {
  757. builder.append("{fractionalSecondDigits}");
  758. VERIFY(segment.length() <= 3);
  759. format.fractional_second_digits = static_cast<u8>(segment.length());
  760. } else if (all_of(segment, is_char('A'))) {
  761. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  762. return {};
  763. }
  764. // Zone
  765. else if (all_of(segment, is_any_of("zV"sv))) {
  766. builder.append("{timeZoneName}");
  767. if (segment.length() < 4)
  768. format.time_zone_name = CalendarPatternStyle::Short;
  769. else
  770. format.time_zone_name = CalendarPatternStyle::Long;
  771. } else if (all_of(segment, is_any_of("ZOXx"sv))) {
  772. builder.append("{timeZoneName}");
  773. if (segment.length() < 4)
  774. format.time_zone_name = CalendarPatternStyle::ShortOffset;
  775. else
  776. format.time_zone_name = CalendarPatternStyle::LongOffset;
  777. } else if (all_of(segment, is_char('v'))) {
  778. builder.append("{timeZoneName}");
  779. if (segment.length() < 4)
  780. format.time_zone_name = CalendarPatternStyle::ShortGeneric;
  781. else
  782. format.time_zone_name = CalendarPatternStyle::LongGeneric;
  783. }
  784. // Non-patterns
  785. else {
  786. builder.append(segment);
  787. }
  788. }
  789. pattern = builder.build();
  790. if (hour12) {
  791. format.pattern = remove_period_from_pattern(pattern);
  792. format.pattern12 = move(pattern);
  793. } else {
  794. format.pattern = move(pattern);
  795. }
  796. return format;
  797. }
  798. static Optional<CalendarPatternIndexType> parse_date_time_pattern(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  799. {
  800. auto format = parse_date_time_pattern_raw(move(pattern), move(skeleton), locale_data);
  801. if (!format.has_value())
  802. return {};
  803. format->pattern_index = locale_data.unique_strings.ensure(move(format->pattern));
  804. if (format->pattern12.has_value())
  805. format->pattern12_index = locale_data.unique_strings.ensure(format->pattern12.release_value());
  806. return locale_data.unique_patterns.ensure(format.release_value());
  807. }
  808. template<typename... Chars>
  809. static constexpr bool char_is_one_of(char ch, Chars&&... chars)
  810. {
  811. return ((ch == chars) || ...);
  812. }
  813. static void parse_interval_patterns(Calendar& calendar, JsonObject const& interval_formats_object, UnicodeLocaleData& locale_data)
  814. {
  815. // https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats
  816. CalendarRangePatternList range_formats {};
  817. CalendarRangePatternList range12_formats {};
  818. auto name_of_field = [&](char field) {
  819. if (char_is_one_of(field, 'G'))
  820. return Unicode::CalendarRangePattern::Field::Era;
  821. if (char_is_one_of(field, 'y', 'Y', 'u', 'U', 'r'))
  822. return Unicode::CalendarRangePattern::Field::Year;
  823. if (char_is_one_of(field, 'M', 'L'))
  824. return Unicode::CalendarRangePattern::Field::Month;
  825. if (char_is_one_of(field, 'd', 'D', 'F', 'g'))
  826. return Unicode::CalendarRangePattern::Field::Day;
  827. if (char_is_one_of(field, 'a', 'b'))
  828. return Unicode::CalendarRangePattern::Field::AmPm;
  829. if (char_is_one_of(field, 'B'))
  830. return Unicode::CalendarRangePattern::Field::DayPeriod;
  831. if (char_is_one_of(field, 'h', 'H', 'K', 'k'))
  832. return Unicode::CalendarRangePattern::Field::Hour;
  833. if (char_is_one_of(field, 'm'))
  834. return Unicode::CalendarRangePattern::Field::Minute;
  835. if (char_is_one_of(field, 's'))
  836. return Unicode::CalendarRangePattern::Field::Second;
  837. if (char_is_one_of(field, 'S'))
  838. return Unicode::CalendarRangePattern::Field::FractionalSecondDigits;
  839. VERIFY_NOT_REACHED();
  840. };
  841. auto split_default_range_pattern = [&](auto skeleton, auto const& pattern) {
  842. auto start_range_end = pattern.find('}').value() + 1;
  843. auto end_range_begin = pattern.find_last('{').value();
  844. auto start_range = pattern.substring_view(0, start_range_end);
  845. auto separator = pattern.substring_view(start_range_end, end_range_begin - start_range_end);
  846. auto end_range = pattern.substring_view(end_range_begin);
  847. CalendarRangePattern format {};
  848. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  849. format.start_range = locale_data.unique_strings.ensure(start_range);
  850. format.separator = locale_data.unique_strings.ensure(separator);
  851. format.end_range = locale_data.unique_strings.ensure(end_range);
  852. return format;
  853. };
  854. auto split_range_pattern = [&](auto skeleton, auto field, auto const& pattern, auto const& parsed_fields) {
  855. HashMap<StringView, size_t> partitions;
  856. StringView last_partition;
  857. auto begin_index = pattern.find('{');
  858. size_t end_index = 0;
  859. while (begin_index.has_value()) {
  860. end_index = pattern.find('}', *begin_index).value();
  861. auto partition = pattern.substring_view(*begin_index, end_index - *begin_index);
  862. if (partitions.contains(partition))
  863. break;
  864. partitions.set(partition, *begin_index);
  865. last_partition = partition;
  866. begin_index = pattern.find('{', end_index + 1);
  867. }
  868. VERIFY(begin_index.has_value() && !last_partition.is_empty());
  869. auto start_range_end = partitions.get(last_partition).value() + last_partition.length() + 1;
  870. auto start_range = pattern.substring_view(0, start_range_end);
  871. auto separator = pattern.substring_view(start_range_end, *begin_index - start_range_end);
  872. auto end_range = pattern.substring_view(*begin_index);
  873. CalendarRangePattern format {};
  874. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  875. format.field = field;
  876. format.start_range = locale_data.unique_strings.ensure(start_range);
  877. format.separator = locale_data.unique_strings.ensure(separator);
  878. format.end_range = locale_data.unique_strings.ensure(end_range);
  879. format.for_each_calendar_field_zipped_with(parsed_fields, [](auto& format_field, auto const& parsed_field, auto) {
  880. format_field = parsed_field;
  881. });
  882. return format;
  883. };
  884. interval_formats_object.for_each_member([&](auto const& skeleton, auto const& value) {
  885. if (skeleton == "intervalFormatFallback"sv) {
  886. auto range_format = split_default_range_pattern(skeleton, value.as_string());
  887. calendar.default_range_format = locale_data.unique_range_patterns.ensure(move(range_format));
  888. return;
  889. }
  890. value.as_object().for_each_member([&](auto const& field, auto const& pattern) {
  891. if (field.ends_with("alt-variant"sv))
  892. return;
  893. VERIFY(field.length() == 1);
  894. auto name = name_of_field(field[0]);
  895. auto format = parse_date_time_pattern_raw(pattern.as_string(), skeleton, locale_data).release_value();
  896. auto range_format = split_range_pattern(skeleton, name, format.pattern, format);
  897. range_formats.append(locale_data.unique_range_patterns.ensure(move(range_format)));
  898. if (format.pattern12.has_value()) {
  899. auto range12_pattern = split_range_pattern(skeleton, name, *format.pattern12, format);
  900. range12_formats.append(locale_data.unique_range_patterns.ensure(move(range12_pattern)));
  901. } else {
  902. range12_formats.append(range_formats.last());
  903. }
  904. });
  905. });
  906. calendar.range_formats = locale_data.unique_range_pattern_lists.ensure(move(range_formats));
  907. calendar.range12_formats = locale_data.unique_range_pattern_lists.ensure(move(range12_formats));
  908. }
  909. static void generate_missing_patterns(Calendar& calendar, CalendarPatternList& formats, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, UnicodeLocaleData& locale_data)
  910. {
  911. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  912. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  913. auto pattern = locale_data.unique_strings.get(format);
  914. auto time_pattern = locale_data.unique_strings.get(time_format);
  915. auto date_pattern = locale_data.unique_strings.get(date_format);
  916. auto new_pattern = pattern.replace("{0}", time_pattern).replace("{1}", date_pattern);
  917. return locale_data.unique_strings.ensure(move(new_pattern));
  918. };
  919. auto inject_fractional_second_digits = [&](auto format) {
  920. auto pattern = locale_data.unique_strings.get(format);
  921. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv);
  922. return locale_data.unique_strings.ensure(move(new_pattern));
  923. };
  924. auto append_if_unique = [&](auto format) {
  925. auto format_index = locale_data.unique_patterns.ensure(move(format));
  926. if (!formats.contains_slow(format_index))
  927. formats.append(format_index);
  928. };
  929. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  930. for (auto const& format : date_formats)
  931. append_if_unique(format);
  932. for (auto const& format : time_formats) {
  933. append_if_unique(format);
  934. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  935. auto new_format = format;
  936. new_format.fractional_second_digits = 2;
  937. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  938. if (new_format.pattern12_index != 0)
  939. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  940. time_formats_with_fractional_second_digits.append(new_format);
  941. append_if_unique(move(new_format));
  942. }
  943. }
  944. time_formats.extend(move(time_formats_with_fractional_second_digits));
  945. for (auto const& date_format : date_formats) {
  946. auto const& date_time_formats = locale_data.unique_formats.get(calendar.date_time_formats);
  947. CalendarPatternIndexType date_time_format_index = 0;
  948. if (date_format.month == Unicode::CalendarPatternStyle::Long) {
  949. if (date_format.weekday.has_value())
  950. date_time_format_index = date_time_formats.full_format;
  951. else
  952. date_time_format_index = date_time_formats.long_format;
  953. } else if (date_format.month == Unicode::CalendarPatternStyle::Short) {
  954. date_time_format_index = date_time_formats.medium_format;
  955. } else {
  956. date_time_format_index = date_time_formats.short_format;
  957. }
  958. for (auto const& time_format : time_formats) {
  959. auto format = locale_data.unique_patterns.get(date_time_format_index);
  960. if (time_format.pattern12_index != 0)
  961. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  962. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  963. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  964. if (date_field.has_value())
  965. field = date_field;
  966. });
  967. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  968. if (time_field.has_value())
  969. field = time_field;
  970. });
  971. append_if_unique(move(format));
  972. }
  973. }
  974. }
  975. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, UnicodeLocaleData& locale_data)
  976. {
  977. auto create_symbol_lists = [](size_t size) {
  978. SymbolList narrow_symbol_list;
  979. SymbolList short_symbol_list;
  980. SymbolList long_symbol_list;
  981. narrow_symbol_list.resize(size);
  982. short_symbol_list.resize(size);
  983. long_symbol_list.resize(size);
  984. return Array<SymbolList, 3> { {
  985. move(narrow_symbol_list),
  986. move(short_symbol_list),
  987. move(long_symbol_list),
  988. } };
  989. };
  990. CalendarSymbolsList symbols_list {};
  991. auto store_symbol_lists = [&](auto symbol, auto symbol_lists) {
  992. auto symbol_index = to_underlying(symbol);
  993. if (symbol_index >= symbols_list.size())
  994. symbols_list.resize(symbol_index + 1);
  995. CalendarSymbols symbols {};
  996. symbols.narrow_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[0]));
  997. symbols.short_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[1]));
  998. symbols.long_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[2]));
  999. auto calendar_symbols_index = locale_data.unique_calendar_symbols.ensure(move(symbols));
  1000. symbols_list[symbol_index] = calendar_symbols_index;
  1001. };
  1002. auto parse_era_symbols = [&](auto const& symbols_object) {
  1003. auto const& narrow_symbols = symbols_object.get("eraNarrow"sv).as_object();
  1004. auto const& short_symbols = symbols_object.get("eraAbbr"sv).as_object();
  1005. auto const& long_symbols = symbols_object.get("eraNames"sv).as_object();
  1006. auto symbol_lists = create_symbol_lists(2);
  1007. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1008. if (auto key_index = key.to_uint(); key_index.has_value())
  1009. symbols[*key_index] = locale_data.unique_strings.ensure(move(symbol));
  1010. };
  1011. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1012. append_symbol(symbol_lists[0], key, value.as_string());
  1013. });
  1014. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1015. append_symbol(symbol_lists[1], key, value.as_string());
  1016. });
  1017. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1018. append_symbol(symbol_lists[2], key, value.as_string());
  1019. });
  1020. store_symbol_lists(Unicode::CalendarSymbol::Era, move(symbol_lists));
  1021. };
  1022. auto parse_month_symbols = [&](auto const& symbols_object) {
  1023. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1024. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1025. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1026. auto symbol_lists = create_symbol_lists(12);
  1027. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1028. auto key_index = key.to_uint().value() - 1;
  1029. symbols[key_index] = locale_data.unique_strings.ensure(move(symbol));
  1030. };
  1031. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1032. append_symbol(symbol_lists[0], key, value.as_string());
  1033. });
  1034. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1035. append_symbol(symbol_lists[1], key, value.as_string());
  1036. });
  1037. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1038. append_symbol(symbol_lists[2], key, value.as_string());
  1039. });
  1040. store_symbol_lists(Unicode::CalendarSymbol::Month, move(symbol_lists));
  1041. };
  1042. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  1043. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1044. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1045. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1046. auto symbol_lists = create_symbol_lists(7);
  1047. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1048. if (key == "sun"sv)
  1049. symbols[to_underlying(Unicode::Weekday::Sunday)] = locale_data.unique_strings.ensure(move(symbol));
  1050. else if (key == "mon"sv)
  1051. symbols[to_underlying(Unicode::Weekday::Monday)] = locale_data.unique_strings.ensure(move(symbol));
  1052. else if (key == "tue"sv)
  1053. symbols[to_underlying(Unicode::Weekday::Tuesday)] = locale_data.unique_strings.ensure(move(symbol));
  1054. else if (key == "wed"sv)
  1055. symbols[to_underlying(Unicode::Weekday::Wednesday)] = locale_data.unique_strings.ensure(move(symbol));
  1056. else if (key == "thu"sv)
  1057. symbols[to_underlying(Unicode::Weekday::Thursday)] = locale_data.unique_strings.ensure(move(symbol));
  1058. else if (key == "fri"sv)
  1059. symbols[to_underlying(Unicode::Weekday::Friday)] = locale_data.unique_strings.ensure(move(symbol));
  1060. else if (key == "sat"sv)
  1061. symbols[to_underlying(Unicode::Weekday::Saturday)] = locale_data.unique_strings.ensure(move(symbol));
  1062. };
  1063. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1064. append_symbol(symbol_lists[0], key, value.as_string());
  1065. });
  1066. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1067. append_symbol(symbol_lists[1], key, value.as_string());
  1068. });
  1069. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1070. append_symbol(symbol_lists[2], key, value.as_string());
  1071. });
  1072. store_symbol_lists(Unicode::CalendarSymbol::Weekday, move(symbol_lists));
  1073. };
  1074. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  1075. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1076. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1077. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1078. auto symbol_lists = create_symbol_lists(10);
  1079. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1080. if (auto day_period = day_period_from_string(key); day_period.has_value())
  1081. symbols[to_underlying(*day_period)] = locale_data.unique_strings.ensure(move(symbol));
  1082. };
  1083. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1084. append_symbol(symbol_lists[0], key, value.as_string());
  1085. });
  1086. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1087. append_symbol(symbol_lists[1], key, value.as_string());
  1088. });
  1089. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1090. append_symbol(symbol_lists[2], key, value.as_string());
  1091. });
  1092. store_symbol_lists(Unicode::CalendarSymbol::DayPeriod, move(symbol_lists));
  1093. };
  1094. parse_era_symbols(calendar_object.get("eras"sv).as_object());
  1095. parse_month_symbols(calendar_object.get("months"sv).as_object().get("format"sv).as_object());
  1096. parse_weekday_symbols(calendar_object.get("days"sv).as_object().get("format"sv).as_object());
  1097. parse_day_period_symbols(calendar_object.get("dayPeriods"sv).as_object().get("format"sv).as_object());
  1098. calendar.symbols = locale_data.unique_calendar_symbols_lists.ensure(move(symbols_list));
  1099. }
  1100. static ErrorOr<void> parse_calendars(String locale_calendars_path, UnicodeLocaleData& locale_data, Locale& locale)
  1101. {
  1102. LexicalPath calendars_path(move(locale_calendars_path));
  1103. if (!calendars_path.basename().starts_with("ca-"sv))
  1104. return {};
  1105. auto calendars = TRY(read_json_file(calendars_path.string()));
  1106. auto const& main_object = calendars.as_object().get("main"sv);
  1107. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  1108. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1109. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  1110. auto parse_patterns = [&](auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) {
  1111. auto parse_pattern = [&](auto name) {
  1112. auto format = patterns_object.get(name);
  1113. auto skeleton = skeletons_object.get(name);
  1114. auto format_index = parse_date_time_pattern(format.as_string(), skeleton.as_string_or(String::empty()), locale_data).value();
  1115. if (patterns)
  1116. patterns->append(locale_data.unique_patterns.get(format_index));
  1117. return format_index;
  1118. };
  1119. CalendarFormat formats {};
  1120. formats.full_format = parse_pattern("full"sv);
  1121. formats.long_format = parse_pattern("long"sv);
  1122. formats.medium_format = parse_pattern("medium"sv);
  1123. formats.short_format = parse_pattern("short"sv);
  1124. return locale_data.unique_formats.ensure(move(formats));
  1125. };
  1126. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  1127. // The generic calendar is not a supported Unicode calendar key, so skip it:
  1128. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  1129. if (calendar_name == "generic"sv)
  1130. return;
  1131. Calendar calendar {};
  1132. CalendarPatternList available_formats {};
  1133. if (!locale_data.calendars.contains_slow(calendar_name))
  1134. locale_data.calendars.append(calendar_name);
  1135. Vector<CalendarPattern> date_formats;
  1136. Vector<CalendarPattern> time_formats;
  1137. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  1138. auto const& date_skeletons_object = value.as_object().get("dateSkeletons"sv);
  1139. calendar.date_formats = parse_patterns(date_formats_object.as_object(), date_skeletons_object.as_object(), &date_formats);
  1140. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  1141. auto const& time_skeletons_object = value.as_object().get("timeSkeletons"sv);
  1142. calendar.time_formats = parse_patterns(time_formats_object.as_object(), time_skeletons_object.as_object(), &time_formats);
  1143. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  1144. calendar.date_time_formats = parse_patterns(date_time_formats_object.as_object(), JsonObject {}, nullptr);
  1145. auto const& available_formats_object = date_time_formats_object.as_object().get("availableFormats"sv);
  1146. available_formats_object.as_object().for_each_member([&](auto const& skeleton, JsonValue const& pattern) {
  1147. auto pattern_index = parse_date_time_pattern(pattern.as_string(), skeleton, locale_data);
  1148. if (!pattern_index.has_value())
  1149. return;
  1150. auto const& format = locale_data.unique_patterns.get(*pattern_index);
  1151. if (format.contains_only_date_fields())
  1152. date_formats.append(format);
  1153. else if (format.contains_only_time_fields())
  1154. time_formats.append(format);
  1155. if (!available_formats.contains_slow(*pattern_index))
  1156. available_formats.append(*pattern_index);
  1157. });
  1158. auto const& interval_formats_object = date_time_formats_object.as_object().get("intervalFormats"sv);
  1159. parse_interval_patterns(calendar, interval_formats_object.as_object(), locale_data);
  1160. generate_missing_patterns(calendar, available_formats, move(date_formats), move(time_formats), locale_data);
  1161. parse_calendar_symbols(calendar, value.as_object(), locale_data);
  1162. calendar.available_formats = locale_data.unique_pattern_lists.ensure(move(available_formats));
  1163. locale.calendars.set(calendar_name, locale_data.unique_calendars.ensure(move(calendar)));
  1164. });
  1165. return {};
  1166. }
  1167. static ErrorOr<void> parse_time_zone_names(String locale_time_zone_names_path, UnicodeLocaleData& locale_data, Locale& locale)
  1168. {
  1169. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  1170. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  1171. auto time_zone_names = TRY(read_json_file(time_zone_names_path.string()));
  1172. auto const& main_object = time_zone_names.as_object().get("main"sv);
  1173. auto const& locale_object = main_object.as_object().get(time_zone_names_path.parent().basename());
  1174. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1175. auto const& time_zone_names_object = dates_object.as_object().get("timeZoneNames"sv);
  1176. auto const& meta_zone_object = time_zone_names_object.as_object().get("metazone"sv);
  1177. auto const& hour_format_string = time_zone_names_object.as_object().get("hourFormat"sv);
  1178. auto const& gmt_format_string = time_zone_names_object.as_object().get("gmtFormat"sv);
  1179. auto const& gmt_zero_format_string = time_zone_names_object.as_object().get("gmtZeroFormat"sv);
  1180. if (meta_zone_object.is_null())
  1181. return {};
  1182. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object, StringView key) -> Optional<StringIndexType> {
  1183. auto const& names = meta_zone_object.get(type);
  1184. if (!names.is_object())
  1185. return {};
  1186. auto const& name = names.as_object().get(key);
  1187. if (name.is_string())
  1188. return locale_data.unique_strings.ensure(name.as_string());
  1189. return {};
  1190. };
  1191. auto parse_hour_format = [&](auto const& format, auto& time_zone_formats) {
  1192. auto hour_formats = format.split_view(';');
  1193. auto hour_format_ahead_start = hour_formats[0].find('H').value();
  1194. auto separator_ahead_start = hour_formats[0].find_last('H').value() + 1;
  1195. auto separator_ahead_end = hour_formats[0].find('m').value();
  1196. auto hour_format_behind_start = hour_formats[1].find('H').value();
  1197. auto separator_behind_start = hour_formats[1].find_last('H').value() + 1;
  1198. auto separator_behind_end = hour_formats[1].find('m').value();
  1199. auto symbol_ahead_sign = hour_formats[0].substring_view(0, hour_format_ahead_start);
  1200. auto symbol_ahead_separator = hour_formats[0].substring_view(separator_ahead_start, separator_ahead_end - separator_ahead_start);
  1201. auto symbol_behind_sign = hour_formats[1].substring_view(0, hour_format_behind_start);
  1202. auto symbol_behind_separator = hour_formats[1].substring_view(separator_behind_start, separator_behind_end - separator_behind_start);
  1203. time_zone_formats.symbol_ahead_sign = locale_data.unique_strings.ensure(symbol_ahead_sign);
  1204. time_zone_formats.symbol_ahead_separator = locale_data.unique_strings.ensure(symbol_ahead_separator);
  1205. time_zone_formats.symbol_behind_sign = locale_data.unique_strings.ensure(symbol_behind_sign);
  1206. time_zone_formats.symbol_behind_separator = locale_data.unique_strings.ensure(symbol_behind_separator);
  1207. };
  1208. TimeZoneNamesList time_zones;
  1209. TimeZoneFormat time_zone_formats {};
  1210. parse_hour_format(hour_format_string.as_string(), time_zone_formats);
  1211. time_zone_formats.gmt_format = locale_data.unique_strings.ensure(gmt_format_string.as_string());
  1212. time_zone_formats.gmt_zero_format = locale_data.unique_strings.ensure(gmt_zero_format_string.as_string());
  1213. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  1214. auto golden_zones = locale_data.meta_zones.find(meta_zone);
  1215. if (golden_zones == locale_data.meta_zones.end())
  1216. return;
  1217. TimeZoneNames time_zone_names {};
  1218. if (auto name = parse_name("long"sv, meta_zone_object, "standard"sv); name.has_value())
  1219. time_zone_names.long_standard_name = name.value();
  1220. if (auto name = parse_name("short"sv, meta_zone_object, "standard"sv); name.has_value())
  1221. time_zone_names.short_standard_name = name.value();
  1222. if (auto name = parse_name("long"sv, meta_zone_object, "daylight"sv); name.has_value())
  1223. time_zone_names.long_daylight_name = name.value();
  1224. if (auto name = parse_name("short"sv, meta_zone_object, "daylight"sv); name.has_value())
  1225. time_zone_names.short_daylight_name = name.value();
  1226. if (auto name = parse_name("long"sv, meta_zone_object, "generic"sv); name.has_value())
  1227. time_zone_names.long_generic_name = name.value();
  1228. if (auto name = parse_name("short"sv, meta_zone_object, "generic"sv); name.has_value())
  1229. time_zone_names.short_generic_name = name.value();
  1230. auto time_zone_index = locale_data.unique_time_zones.ensure(move(time_zone_names));
  1231. for (auto golden_zone : golden_zones->value) {
  1232. auto time_zone = to_underlying(golden_zone);
  1233. if (time_zone >= time_zones.size())
  1234. time_zones.resize(time_zone + 1);
  1235. time_zones[time_zone] = time_zone_index;
  1236. }
  1237. };
  1238. meta_zone_object.as_object().for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  1239. parse_time_zone(meta_zone, value.as_object());
  1240. });
  1241. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  1242. auto const& zone_object = time_zone_names_object.as_object().get("zone"sv);
  1243. auto const& etc_object = zone_object.as_object().get("Etc"sv);
  1244. auto const& utc_object = etc_object.as_object().get("UTC"sv);
  1245. parse_time_zone("UTC"sv, utc_object.as_object());
  1246. locale.time_zones = locale_data.unique_time_zone_lists.ensure(move(time_zones));
  1247. locale.time_zone_formats = locale_data.unique_time_zone_formats.ensure(move(time_zone_formats));
  1248. return {};
  1249. }
  1250. static ErrorOr<void> parse_day_periods(String core_path, UnicodeLocaleData& locale_data)
  1251. {
  1252. // https://unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rule_Sets
  1253. LexicalPath day_periods_path(move(core_path));
  1254. day_periods_path = day_periods_path.append("supplemental"sv);
  1255. day_periods_path = day_periods_path.append("dayPeriods.json"sv);
  1256. auto locale_day_periods = TRY(read_json_file(day_periods_path.string()));
  1257. auto const& supplemental_object = locale_day_periods.as_object().get("supplemental"sv);
  1258. auto const& day_periods_object = supplemental_object.as_object().get("dayPeriodRuleSet"sv);
  1259. auto parse_hour = [](auto const& time) {
  1260. auto hour_end_index = time.find(':').value();
  1261. // The times are in the form "hours:minutes", but we only need the hour segment.
  1262. // TR-35 explicitly states that minutes other than :00 must not be used.
  1263. VERIFY(time.substring_view(hour_end_index) == ":00"sv);
  1264. auto hour = time.substring_view(0, hour_end_index);
  1265. return hour.template to_uint<u8>().value();
  1266. };
  1267. auto parse_day_period = [&](auto const& symbol, auto const& ranges) -> Optional<DayPeriod> {
  1268. auto day_period = day_period_from_string(symbol);
  1269. if (!day_period.has_value())
  1270. return {};
  1271. auto begin = parse_hour(ranges.get("_from"sv).as_string());
  1272. auto end = parse_hour(ranges.get("_before"sv).as_string());
  1273. return DayPeriod { *day_period, begin, end };
  1274. };
  1275. day_periods_object.as_object().for_each_member([&](auto const& language, JsonValue const& value) {
  1276. auto locale = locale_data.locales.find(language);
  1277. if (locale == locale_data.locales.end())
  1278. return;
  1279. DayPeriodList day_periods;
  1280. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& ranges) {
  1281. if (auto day_period = parse_day_period(symbol, ranges.as_object()); day_period.has_value()) {
  1282. auto day_period_index = locale_data.unique_day_periods.ensure(day_period.release_value());
  1283. day_periods.append(day_period_index);
  1284. }
  1285. });
  1286. locale->value.day_periods = locale_data.unique_day_period_lists.ensure(move(day_periods));
  1287. });
  1288. return {};
  1289. }
  1290. static ErrorOr<void> parse_all_locales(String core_path, String dates_path, UnicodeLocaleData& locale_data)
  1291. {
  1292. TRY(parse_hour_cycles(core_path, locale_data));
  1293. TRY(parse_meta_zones(core_path, locale_data));
  1294. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  1295. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  1296. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  1297. StringBuilder builder;
  1298. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  1299. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  1300. builder.appendff("-{}", script);
  1301. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  1302. builder.appendff("-{}", region);
  1303. return builder.build();
  1304. };
  1305. while (dates_iterator.has_next()) {
  1306. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  1307. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  1308. auto language = TRY(remove_variants_from_path(dates_path));
  1309. auto& locale = locale_data.locales.ensure(language);
  1310. while (calendars_iterator.has_next()) {
  1311. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  1312. TRY(parse_calendars(move(calendars_path), locale_data, locale));
  1313. }
  1314. TRY(parse_time_zone_names(move(dates_path), locale_data, locale));
  1315. }
  1316. TRY(parse_day_periods(move(core_path), locale_data));
  1317. return {};
  1318. }
  1319. static String format_identifier(StringView owner, String identifier)
  1320. {
  1321. identifier = identifier.replace("-"sv, "_"sv, true);
  1322. identifier = identifier.replace("/"sv, "_"sv, true);
  1323. if (all_of(identifier, is_ascii_digit))
  1324. return String::formatted("{}_{}", owner[0], identifier);
  1325. if (is_ascii_lower_alpha(identifier[0]))
  1326. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  1327. return identifier;
  1328. }
  1329. static ErrorOr<void> generate_unicode_locale_header(Core::Stream::BufferedFile& file, UnicodeLocaleData& locale_data)
  1330. {
  1331. StringBuilder builder;
  1332. SourceGenerator generator { builder };
  1333. generator.append(R"~~~(
  1334. #pragma once
  1335. #include <AK/Types.h>
  1336. namespace Unicode {
  1337. )~~~");
  1338. generate_enum(generator, format_identifier, "Calendar"sv, {}, locale_data.calendars, locale_data.calendar_aliases);
  1339. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, locale_data.hour_cycle_regions);
  1340. generator.append(R"~~~(
  1341. }
  1342. )~~~");
  1343. TRY(file.write(generator.as_string_view().bytes()));
  1344. return {};
  1345. }
  1346. static ErrorOr<void> generate_unicode_locale_implementation(Core::Stream::BufferedFile& file, UnicodeLocaleData& locale_data)
  1347. {
  1348. StringBuilder builder;
  1349. SourceGenerator generator { builder };
  1350. generator.set("string_index_type"sv, s_string_index_type);
  1351. generator.set("calendar_pattern_index_type"sv, s_calendar_pattern_index_type);
  1352. generator.set("calendar_pattern_list_index_type"sv, s_calendar_pattern_list_index_type);
  1353. generator.set("calendar_range_pattern_index_type"sv, s_calendar_range_pattern_index_type);
  1354. generator.set("calendar_range_pattern_list_index_type"sv, s_calendar_range_pattern_list_index_type);
  1355. generator.set("calendar_format_index_type"sv, s_calendar_format_index_type);
  1356. generator.set("symbol_list_index_type"sv, s_symbol_list_index_type);
  1357. generator.set("calendar_symbols_index_type"sv, s_calendar_symbols_index_type);
  1358. generator.set("calendar_symbols_list_index_type"sv, s_calendar_symbols_list_index_type);
  1359. generator.set("calendar_index_type"sv, s_calendar_index_type);
  1360. generator.set("time_zone_index_type"sv, s_time_zone_index_type);
  1361. generator.set("time_zone_list_index_type"sv, s_time_zone_list_index_type);
  1362. generator.set("day_period_index_type"sv, s_day_period_index_type);
  1363. generator.set("day_period_list_index_type"sv, s_day_period_list_index_type);
  1364. generator.append(R"~~~(
  1365. #include <AK/Array.h>
  1366. #include <AK/BinarySearch.h>
  1367. #include <AK/Optional.h>
  1368. #include <AK/StringView.h>
  1369. #include <LibTimeZone/TimeZone.h>
  1370. #include <LibUnicode/DateTimeFormat.h>
  1371. #include <LibUnicode/Locale.h>
  1372. #include <LibUnicode/UnicodeDateTimeFormat.h>
  1373. namespace Unicode {
  1374. )~~~");
  1375. locale_data.unique_strings.generate(generator);
  1376. generator.append(R"~~~(
  1377. template <typename SourceType, typename TargetType>
  1378. static void convert_calendar_fields(SourceType const& source, TargetType& target)
  1379. {
  1380. if (source.era != -1)
  1381. target.era = static_cast<CalendarPatternStyle>(source.era);
  1382. if (source.year != -1)
  1383. target.year = static_cast<CalendarPatternStyle>(source.year);
  1384. if (source.month != -1)
  1385. target.month = static_cast<CalendarPatternStyle>(source.month);
  1386. if (source.weekday != -1)
  1387. target.weekday = static_cast<CalendarPatternStyle>(source.weekday);
  1388. if (source.day != -1)
  1389. target.day = static_cast<CalendarPatternStyle>(source.day);
  1390. if (source.day_period != -1)
  1391. target.day_period = static_cast<CalendarPatternStyle>(source.day_period);
  1392. if (source.hour != -1)
  1393. target.hour = static_cast<CalendarPatternStyle>(source.hour);
  1394. if (source.minute != -1)
  1395. target.minute = static_cast<CalendarPatternStyle>(source.minute);
  1396. if (source.second != -1)
  1397. target.second = static_cast<CalendarPatternStyle>(source.second);
  1398. if (source.fractional_second_digits != -1)
  1399. target.fractional_second_digits = static_cast<u8>(source.fractional_second_digits);
  1400. if (source.time_zone_name != -1)
  1401. target.time_zone_name = static_cast<CalendarPatternStyle>(source.time_zone_name);
  1402. }
  1403. struct CalendarPatternImpl {
  1404. CalendarPattern to_unicode_calendar_pattern() const {
  1405. CalendarPattern calendar_pattern {};
  1406. calendar_pattern.skeleton = s_string_list[skeleton];
  1407. calendar_pattern.pattern = s_string_list[pattern];
  1408. if (pattern12 != 0)
  1409. calendar_pattern.pattern12 = s_string_list[pattern12];
  1410. convert_calendar_fields(*this, calendar_pattern);
  1411. return calendar_pattern;
  1412. }
  1413. @string_index_type@ skeleton { 0 };
  1414. @string_index_type@ pattern { 0 };
  1415. @string_index_type@ pattern12 { 0 };
  1416. i8 era { -1 };
  1417. i8 year { -1 };
  1418. i8 month { -1 };
  1419. i8 weekday { -1 };
  1420. i8 day { -1 };
  1421. i8 day_period { -1 };
  1422. i8 hour { -1 };
  1423. i8 minute { -1 };
  1424. i8 second { -1 };
  1425. i8 fractional_second_digits { -1 };
  1426. i8 time_zone_name { -1 };
  1427. };
  1428. struct CalendarRangePatternImpl {
  1429. CalendarRangePattern to_unicode_calendar_range_pattern() const {
  1430. CalendarRangePattern calendar_range_pattern {};
  1431. if (field != -1)
  1432. calendar_range_pattern.field = static_cast<CalendarRangePattern::Field>(field);
  1433. calendar_range_pattern.start_range = s_string_list[start_range];
  1434. calendar_range_pattern.separator = s_string_list[separator];
  1435. calendar_range_pattern.end_range = s_string_list[end_range];
  1436. convert_calendar_fields(*this, calendar_range_pattern);
  1437. return calendar_range_pattern;
  1438. }
  1439. @string_index_type@ skeleton { 0 };
  1440. i8 field { -1 };
  1441. @string_index_type@ start_range { 0 };
  1442. @string_index_type@ separator { 0 };
  1443. @string_index_type@ end_range { 0 };
  1444. i8 era { -1 };
  1445. i8 year { -1 };
  1446. i8 month { -1 };
  1447. i8 weekday { -1 };
  1448. i8 day { -1 };
  1449. i8 day_period { -1 };
  1450. i8 hour { -1 };
  1451. i8 minute { -1 };
  1452. i8 second { -1 };
  1453. i8 fractional_second_digits { -1 };
  1454. i8 time_zone_name { -1 };
  1455. };
  1456. )~~~");
  1457. locale_data.unique_patterns.generate(generator, "CalendarPatternImpl"sv, "s_calendar_patterns"sv, 10);
  1458. locale_data.unique_pattern_lists.generate(generator, s_calendar_pattern_index_type, "s_calendar_pattern_lists"sv);
  1459. locale_data.unique_range_patterns.generate(generator, "CalendarRangePatternImpl"sv, "s_calendar_range_patterns"sv, 10);
  1460. locale_data.unique_range_pattern_lists.generate(generator, s_calendar_range_pattern_index_type, "s_calendar_range_pattern_lists"sv);
  1461. generator.append(R"~~~(
  1462. struct CalendarFormatImpl {
  1463. CalendarFormat to_unicode_calendar_format() const {
  1464. CalendarFormat calendar_format {};
  1465. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  1466. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  1467. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  1468. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  1469. return calendar_format;
  1470. }
  1471. @calendar_pattern_index_type@ full_format { 0 };
  1472. @calendar_pattern_index_type@ long_format { 0 };
  1473. @calendar_pattern_index_type@ medium_format { 0 };
  1474. @calendar_pattern_index_type@ short_format { 0 };
  1475. };
  1476. struct CalendarSymbols {
  1477. @symbol_list_index_type@ narrow_symbols { 0 };
  1478. @symbol_list_index_type@ short_symbols { 0 };
  1479. @symbol_list_index_type@ long_symbols { 0 };
  1480. };
  1481. struct CalendarData {
  1482. @calendar_format_index_type@ date_formats { 0 };
  1483. @calendar_format_index_type@ time_formats { 0 };
  1484. @calendar_format_index_type@ date_time_formats { 0 };
  1485. @calendar_pattern_list_index_type@ available_formats { 0 };
  1486. @calendar_range_pattern_index_type@ default_range_format { 0 };
  1487. @calendar_range_pattern_list_index_type@ range_formats { 0 };
  1488. @calendar_range_pattern_list_index_type@ range12_formats { 0 };
  1489. @calendar_symbols_list_index_type@ symbols { 0 };
  1490. };
  1491. struct TimeZoneNames {
  1492. @string_index_type@ short_standard_name { 0 };
  1493. @string_index_type@ long_standard_name { 0 };
  1494. @string_index_type@ short_daylight_name { 0 };
  1495. @string_index_type@ long_daylight_name { 0 };
  1496. @string_index_type@ short_generic_name { 0 };
  1497. @string_index_type@ long_generic_name { 0 };
  1498. };
  1499. struct TimeZoneFormatImpl {
  1500. TimeZoneFormat to_time_zone_format() const {
  1501. TimeZoneFormat time_zone_format {};
  1502. time_zone_format.symbol_ahead_sign = s_string_list[symbol_ahead_sign];
  1503. time_zone_format.symbol_ahead_separator = s_string_list[symbol_ahead_separator];
  1504. time_zone_format.symbol_behind_sign = s_string_list[symbol_behind_sign];
  1505. time_zone_format.symbol_behind_separator = s_string_list[symbol_behind_separator];
  1506. time_zone_format.gmt_format = s_string_list[gmt_format];
  1507. time_zone_format.gmt_zero_format = s_string_list[gmt_zero_format];
  1508. return time_zone_format;
  1509. }
  1510. @string_index_type@ symbol_ahead_sign { 0 };
  1511. @string_index_type@ symbol_ahead_separator { 0 };
  1512. @string_index_type@ symbol_behind_sign { 0 };
  1513. @string_index_type@ symbol_behind_separator { 0 };
  1514. @string_index_type@ gmt_format { 0 };
  1515. @string_index_type@ gmt_zero_format { 0 };
  1516. };
  1517. struct DayPeriodData {
  1518. u8 day_period { 0 };
  1519. u8 begin { 0 };
  1520. u8 end { 0 };
  1521. };
  1522. )~~~");
  1523. generate_available_values(generator, "get_available_calendars"sv, locale_data.calendars, locale_data.calendar_aliases);
  1524. locale_data.unique_formats.generate(generator, "CalendarFormatImpl"sv, "s_calendar_formats"sv, 10);
  1525. locale_data.unique_symbol_lists.generate(generator, s_string_index_type, "s_symbol_lists"sv);
  1526. locale_data.unique_calendar_symbols.generate(generator, "CalendarSymbols"sv, "s_calendar_symbols"sv, 10);
  1527. locale_data.unique_calendar_symbols_lists.generate(generator, s_calendar_symbols_index_type, "s_calendar_symbol_lists"sv);
  1528. locale_data.unique_calendars.generate(generator, "CalendarData"sv, "s_calendars"sv, 10);
  1529. locale_data.unique_time_zones.generate(generator, "TimeZoneNames"sv, "s_time_zones"sv, 30);
  1530. locale_data.unique_time_zone_lists.generate(generator, s_time_zone_index_type, "s_time_zone_lists"sv);
  1531. locale_data.unique_time_zone_formats.generate(generator, "TimeZoneFormatImpl"sv, "s_time_zone_formats"sv, 30);
  1532. locale_data.unique_day_periods.generate(generator, "DayPeriodData"sv, "s_day_periods"sv, 30);
  1533. locale_data.unique_day_period_lists.generate(generator, s_day_period_index_type, "s_day_period_lists"sv);
  1534. locale_data.unique_hour_cycle_lists.generate(generator, "u8"sv, "s_hour_cycle_lists"sv);
  1535. auto append_calendars = [&](String name, auto const& calendars) {
  1536. generator.set("name", name);
  1537. generator.set("size", String::number(calendars.size()));
  1538. generator.append(R"~~~(
  1539. static constexpr Array<@calendar_index_type@, @size@> @name@ { {)~~~");
  1540. bool first = true;
  1541. for (auto const& calendar_key : locale_data.calendars) {
  1542. auto calendar = calendars.find(calendar_key)->value;
  1543. generator.append(first ? " " : ", ");
  1544. generator.append(String::number(calendar));
  1545. first = false;
  1546. }
  1547. generator.append(" } };");
  1548. };
  1549. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  1550. generator.set("type", type);
  1551. generator.set("name", name);
  1552. generator.set("size", String::number(keys.size()));
  1553. generator.append(R"~~~(
  1554. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1555. bool first = true;
  1556. for (auto const& key : keys) {
  1557. auto const& value = map.find(key)->value;
  1558. auto mapping = mapping_getter(value);
  1559. generator.append(first ? " " : ", ");
  1560. generator.append(String::number(mapping));
  1561. first = false;
  1562. }
  1563. generator.append(" } };");
  1564. };
  1565. auto locales = locale_data.locales.keys();
  1566. quick_sort(locales);
  1567. 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); });
  1568. append_mapping(locales, locale_data.locales, s_time_zone_index_type, "s_locale_time_zones"sv, [](auto const& locale) { return locale.time_zones; });
  1569. 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; });
  1570. append_mapping(locales, locale_data.locales, s_day_period_index_type, "s_locale_day_periods"sv, [](auto const& locale) { return locale.day_periods; });
  1571. 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; });
  1572. generator.append("\n");
  1573. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1574. HashValueMap<String> hashes;
  1575. hashes.ensure_capacity(values.size());
  1576. for (auto const& value : values)
  1577. hashes.set(value.hash(), format_identifier(enum_title, value));
  1578. for (auto const& alias : aliases)
  1579. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1580. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1581. };
  1582. append_from_string("Calendar"sv, "calendar"sv, locale_data.calendars, locale_data.calendar_aliases);
  1583. append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, locale_data.hour_cycle_regions);
  1584. generator.append(R"~~~(
  1585. Vector<HourCycle> get_regional_hour_cycles(StringView region)
  1586. {
  1587. auto region_value = hour_cycle_region_from_string(region);
  1588. if (!region_value.has_value())
  1589. return {};
  1590. auto region_index = to_underlying(*region_value);
  1591. auto regional_hour_cycles_index = s_hour_cycles.at(region_index);
  1592. auto const& regional_hour_cycles = s_hour_cycle_lists.at(regional_hour_cycles_index);
  1593. Vector<HourCycle> hour_cycles;
  1594. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1595. for (auto hour_cycle : regional_hour_cycles)
  1596. hour_cycles.unchecked_append(static_cast<HourCycle>(hour_cycle));
  1597. return hour_cycles;
  1598. }
  1599. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1600. {
  1601. auto locale_value = locale_from_string(locale);
  1602. if (!locale_value.has_value())
  1603. return nullptr;
  1604. auto calendar_value = calendar_from_string(calendar);
  1605. if (!calendar_value.has_value())
  1606. return nullptr;
  1607. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1608. size_t calendar_index = to_underlying(*calendar_value);
  1609. auto const& calendar_indices = s_locale_calendars.at(locale_index);
  1610. calendar_index = calendar_indices[calendar_index];
  1611. return &s_calendars[calendar_index];
  1612. }
  1613. Optional<CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1614. {
  1615. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1616. auto const& formats = s_calendar_formats.at(data->date_formats);
  1617. return formats.to_unicode_calendar_format();
  1618. }
  1619. return {};
  1620. }
  1621. Optional<CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1622. {
  1623. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1624. auto const& formats = s_calendar_formats.at(data->time_formats);
  1625. return formats.to_unicode_calendar_format();
  1626. }
  1627. return {};
  1628. }
  1629. Optional<CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1630. {
  1631. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1632. auto const& formats = s_calendar_formats.at(data->date_time_formats);
  1633. return formats.to_unicode_calendar_format();
  1634. }
  1635. return {};
  1636. }
  1637. Vector<CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1638. {
  1639. Vector<CalendarPattern> result {};
  1640. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1641. auto const& available_formats = s_calendar_pattern_lists.at(data->available_formats);
  1642. result.ensure_capacity(available_formats.size());
  1643. for (auto const& format : available_formats)
  1644. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1645. }
  1646. return result;
  1647. }
  1648. Optional<CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar)
  1649. {
  1650. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1651. auto const& pattern = s_calendar_range_patterns[data->default_range_format];
  1652. return pattern.to_unicode_calendar_range_pattern();
  1653. }
  1654. return {};
  1655. }
  1656. Vector<CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton)
  1657. {
  1658. Vector<CalendarRangePattern> result {};
  1659. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1660. auto const& range_formats = s_calendar_range_pattern_lists.at(data->range_formats);
  1661. for (auto format : range_formats) {
  1662. auto const& pattern = s_calendar_range_patterns[format];
  1663. if (skeleton == s_string_list[pattern.skeleton])
  1664. result.append(pattern.to_unicode_calendar_range_pattern());
  1665. }
  1666. }
  1667. return result;
  1668. }
  1669. Vector<CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton)
  1670. {
  1671. Vector<CalendarRangePattern> result {};
  1672. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1673. auto const& range12_formats = s_calendar_range_pattern_lists.at(data->range12_formats);
  1674. for (auto format : range12_formats) {
  1675. auto const& pattern = s_calendar_range_patterns[format];
  1676. if (skeleton == s_string_list[pattern.skeleton])
  1677. result.append(pattern.to_unicode_calendar_range_pattern());
  1678. }
  1679. }
  1680. return result;
  1681. }
  1682. static Span<@string_index_type@ const> find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1683. {
  1684. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1685. auto const& symbols_list = s_calendar_symbol_lists[data->symbols];
  1686. auto symbol_index = to_underlying(symbol);
  1687. auto calendar_symbols_index = symbols_list.at(symbol_index);
  1688. auto const& symbols = s_calendar_symbols.at(calendar_symbols_index);
  1689. @symbol_list_index_type@ symbol_list_index = 0;
  1690. switch (style) {
  1691. case CalendarPatternStyle::Narrow:
  1692. symbol_list_index = symbols.narrow_symbols;
  1693. break;
  1694. case CalendarPatternStyle::Short:
  1695. symbol_list_index = symbols.short_symbols;
  1696. break;
  1697. case CalendarPatternStyle::Long:
  1698. symbol_list_index = symbols.long_symbols;
  1699. break;
  1700. default:
  1701. VERIFY_NOT_REACHED();
  1702. }
  1703. return s_symbol_lists.at(symbol_list_index);
  1704. }
  1705. return {};
  1706. }
  1707. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Era value)
  1708. {
  1709. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1710. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1711. return s_string_list[symbols.at(value_index)];
  1712. return {};
  1713. }
  1714. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Month value)
  1715. {
  1716. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1717. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1718. return s_string_list[symbols.at(value_index)];
  1719. return {};
  1720. }
  1721. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Weekday value)
  1722. {
  1723. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1724. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1725. return s_string_list[symbols.at(value_index)];
  1726. return {};
  1727. }
  1728. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, DayPeriod value)
  1729. {
  1730. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1731. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1732. return s_string_list[symbols.at(value_index)];
  1733. return {};
  1734. }
  1735. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour)
  1736. {
  1737. auto locale_value = locale_from_string(locale);
  1738. if (!locale_value.has_value())
  1739. return {};
  1740. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1741. auto day_periods_index = s_locale_day_periods[locale_index];
  1742. auto day_periods = s_day_period_lists[day_periods_index];
  1743. for (auto day_period_index : day_periods) {
  1744. auto day_period = s_day_periods[day_period_index];
  1745. auto h = hour;
  1746. if (day_period.begin > day_period.end) {
  1747. day_period.end += 24;
  1748. h += 24;
  1749. }
  1750. if ((day_period.begin <= h) && (h < day_period.end)) {
  1751. auto period = static_cast<DayPeriod>(day_period.day_period);
  1752. return get_calendar_day_period_symbol(locale, calendar, style, period);
  1753. }
  1754. }
  1755. // Fallback to fixed periods if the locale does not have flexible day periods.
  1756. // TR-35 states that the meaning of AM and PM does not change with locale.
  1757. if (hour < 12)
  1758. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::AM);
  1759. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::PM);
  1760. }
  1761. Optional<TimeZoneFormat> get_time_zone_format(StringView locale)
  1762. {
  1763. auto locale_value = locale_from_string(locale);
  1764. if (!locale_value.has_value())
  1765. return {};
  1766. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1767. auto time_zone_format_index = s_locale_time_zone_formats.at(locale_index);
  1768. auto const& time_zone_format = s_time_zone_formats.at(time_zone_format_index);
  1769. return time_zone_format.to_time_zone_format();
  1770. }
  1771. static TimeZoneNames const* find_time_zone_names(StringView locale, StringView time_zone)
  1772. {
  1773. auto locale_value = locale_from_string(locale);
  1774. if (!locale_value.has_value())
  1775. return nullptr;
  1776. auto time_zone_value = ::TimeZone::time_zone_from_string(time_zone);
  1777. if (!time_zone_value.has_value())
  1778. return nullptr;
  1779. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1780. size_t time_zone_index = to_underlying(*time_zone_value);
  1781. auto time_zone_list_index = s_locale_time_zones.at(locale_index);
  1782. auto const& time_zone_list = s_time_zone_lists.at(time_zone_list_index);
  1783. if (time_zone_list.size() <= time_zone_index)
  1784. return nullptr;
  1785. time_zone_index = time_zone_list.at(time_zone_index);
  1786. return &s_time_zones[time_zone_index];
  1787. }
  1788. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style, TimeZone::InDST in_dst)
  1789. {
  1790. if (auto const* data = find_time_zone_names(locale, time_zone); data != nullptr) {
  1791. size_t name_index = 0;
  1792. switch (style) {
  1793. case CalendarPatternStyle::Short:
  1794. name_index = (in_dst == TimeZone::InDST::No) ? data->short_standard_name : data->short_daylight_name;
  1795. break;
  1796. case CalendarPatternStyle::Long:
  1797. name_index = (in_dst == TimeZone::InDST::No) ? data->long_standard_name : data->long_daylight_name;
  1798. break;
  1799. case CalendarPatternStyle::ShortGeneric:
  1800. name_index = data->short_generic_name;
  1801. break;
  1802. case CalendarPatternStyle::LongGeneric:
  1803. name_index = data->long_generic_name;
  1804. break;
  1805. default:
  1806. VERIFY_NOT_REACHED();
  1807. }
  1808. if (name_index != 0)
  1809. return s_string_list[name_index];
  1810. }
  1811. return {};
  1812. }
  1813. }
  1814. )~~~");
  1815. TRY(file.write(generator.as_string_view().bytes()));
  1816. return {};
  1817. }
  1818. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1819. {
  1820. StringView generated_header_path;
  1821. StringView generated_implementation_path;
  1822. StringView core_path;
  1823. StringView dates_path;
  1824. Core::ArgsParser args_parser;
  1825. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1826. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1827. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1828. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1829. args_parser.parse(arguments);
  1830. auto generated_header_file = TRY(open_file(generated_header_path, Core::Stream::OpenMode::Write));
  1831. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::Stream::OpenMode::Write));
  1832. UnicodeLocaleData locale_data;
  1833. TRY(parse_all_locales(core_path, dates_path, locale_data));
  1834. TRY(generate_unicode_locale_header(*generated_header_file, locale_data));
  1835. TRY(generate_unicode_locale_implementation(*generated_implementation_file, locale_data));
  1836. return 0;
  1837. }