GenerateDateTimeFormatData.cpp 96 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396239723982399240024012402240324042405240624072408240924102411241224132414241524162417241824192420242124222423242424252426242724282429
  1. /*
  2. * Copyright (c) 2021-2023, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "../LibUnicode/GeneratorUtil.h" // FIXME: Move this somewhere common.
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/DeprecatedString.h>
  10. #include <AK/Error.h>
  11. #include <AK/Find.h>
  12. #include <AK/Format.h>
  13. #include <AK/GenericLexer.h>
  14. #include <AK/HashFunctions.h>
  15. #include <AK/HashMap.h>
  16. #include <AK/JsonObject.h>
  17. #include <AK/JsonParser.h>
  18. #include <AK/JsonValue.h>
  19. #include <AK/LexicalPath.h>
  20. #include <AK/SourceGenerator.h>
  21. #include <AK/StringBuilder.h>
  22. #include <AK/Traits.h>
  23. #include <AK/Utf8View.h>
  24. #include <LibCore/ArgsParser.h>
  25. #include <LibCore/DirIterator.h>
  26. #include <LibCore/Stream.h>
  27. #include <LibLocale/DateTimeFormat.h>
  28. #include <LibTimeZone/TimeZone.h>
  29. struct CalendarPattern : public Locale::CalendarPattern {
  30. bool contains_only_date_fields() const
  31. {
  32. return !day_period.has_value() && !hour.has_value() && !minute.has_value() && !second.has_value() && !fractional_second_digits.has_value() && !time_zone_name.has_value();
  33. }
  34. bool contains_only_time_fields() const
  35. {
  36. return !weekday.has_value() && !era.has_value() && !year.has_value() && !month.has_value() && !day.has_value();
  37. }
  38. unsigned hash() const
  39. {
  40. auto hash = pair_int_hash(pattern_index, pattern12_index);
  41. hash = pair_int_hash(hash, skeleton_index);
  42. auto hash_field = [&](auto const& field) {
  43. if (field.has_value())
  44. hash = pair_int_hash(hash, static_cast<u8>(*field));
  45. else
  46. hash = pair_int_hash(hash, -1);
  47. };
  48. hash_field(era);
  49. hash_field(year);
  50. hash_field(month);
  51. hash_field(weekday);
  52. hash_field(day);
  53. hash_field(day_period);
  54. hash_field(hour);
  55. hash_field(minute);
  56. hash_field(second);
  57. hash_field(fractional_second_digits);
  58. hash_field(time_zone_name);
  59. return hash;
  60. }
  61. bool operator==(CalendarPattern const& other) const
  62. {
  63. return (skeleton_index == other.skeleton_index)
  64. && (pattern_index == other.pattern_index)
  65. && (pattern12_index == other.pattern12_index)
  66. && (era == other.era)
  67. && (year == other.year)
  68. && (month == other.month)
  69. && (weekday == other.weekday)
  70. && (day == other.day)
  71. && (day_period == other.day_period)
  72. && (hour == other.hour)
  73. && (minute == other.minute)
  74. && (second == other.second)
  75. && (fractional_second_digits == other.fractional_second_digits)
  76. && (time_zone_name == other.time_zone_name);
  77. }
  78. size_t skeleton_index { 0 };
  79. size_t pattern_index { 0 };
  80. size_t pattern12_index { 0 };
  81. };
  82. template<>
  83. struct AK::Formatter<CalendarPattern> : Formatter<FormatString> {
  84. ErrorOr<void> format(FormatBuilder& builder, CalendarPattern const& pattern)
  85. {
  86. auto field_to_i8 = [](auto const& field) -> i8 {
  87. if (!field.has_value())
  88. return -1;
  89. return static_cast<i8>(*field);
  90. };
  91. return Formatter<FormatString>::format(builder,
  92. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  93. pattern.skeleton_index,
  94. pattern.pattern_index,
  95. pattern.pattern12_index,
  96. field_to_i8(pattern.era),
  97. field_to_i8(pattern.year),
  98. field_to_i8(pattern.month),
  99. field_to_i8(pattern.weekday),
  100. field_to_i8(pattern.day),
  101. field_to_i8(pattern.day_period),
  102. field_to_i8(pattern.hour),
  103. field_to_i8(pattern.minute),
  104. field_to_i8(pattern.second),
  105. field_to_i8(pattern.fractional_second_digits),
  106. field_to_i8(pattern.time_zone_name));
  107. }
  108. };
  109. template<>
  110. struct AK::Traits<CalendarPattern> : public GenericTraits<CalendarPattern> {
  111. static unsigned hash(CalendarPattern const& c) { return c.hash(); }
  112. };
  113. struct CalendarRangePattern : public CalendarPattern {
  114. unsigned hash() const
  115. {
  116. auto hash = CalendarPattern::hash();
  117. if (field.has_value())
  118. hash = pair_int_hash(hash, static_cast<u8>(*field));
  119. hash = pair_int_hash(hash, start_range);
  120. hash = pair_int_hash(hash, separator);
  121. hash = pair_int_hash(hash, end_range);
  122. return hash;
  123. }
  124. bool operator==(CalendarRangePattern const& other) const
  125. {
  126. if (!CalendarPattern::operator==(other))
  127. return false;
  128. return (field == other.field)
  129. && (start_range == other.start_range)
  130. && (separator == other.separator)
  131. && (end_range == other.end_range);
  132. }
  133. Optional<Locale::CalendarRangePattern::Field> field {};
  134. size_t start_range { 0 };
  135. size_t separator { 0 };
  136. size_t end_range { 0 };
  137. };
  138. template<>
  139. struct AK::Formatter<CalendarRangePattern> : Formatter<FormatString> {
  140. ErrorOr<void> format(FormatBuilder& builder, CalendarRangePattern const& pattern)
  141. {
  142. auto field_to_i8 = [](auto const& field) -> i8 {
  143. if (!field.has_value())
  144. return -1;
  145. return static_cast<i8>(*field);
  146. };
  147. return Formatter<FormatString>::format(builder,
  148. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  149. pattern.skeleton_index,
  150. field_to_i8(pattern.field),
  151. pattern.start_range,
  152. pattern.separator,
  153. pattern.end_range,
  154. field_to_i8(pattern.era),
  155. field_to_i8(pattern.year),
  156. field_to_i8(pattern.month),
  157. field_to_i8(pattern.weekday),
  158. field_to_i8(pattern.day),
  159. field_to_i8(pattern.day_period),
  160. field_to_i8(pattern.hour),
  161. field_to_i8(pattern.minute),
  162. field_to_i8(pattern.second),
  163. field_to_i8(pattern.fractional_second_digits),
  164. field_to_i8(pattern.time_zone_name));
  165. }
  166. };
  167. template<>
  168. struct AK::Traits<CalendarRangePattern> : public GenericTraits<CalendarRangePattern> {
  169. static unsigned hash(CalendarRangePattern const& c) { return c.hash(); }
  170. };
  171. struct CalendarFormat {
  172. unsigned hash() const
  173. {
  174. auto hash = pair_int_hash(full_format, long_format);
  175. hash = pair_int_hash(hash, medium_format);
  176. hash = pair_int_hash(hash, short_format);
  177. return hash;
  178. }
  179. bool operator==(CalendarFormat const& other) const
  180. {
  181. return (full_format == other.full_format)
  182. && (long_format == other.long_format)
  183. && (medium_format == other.medium_format)
  184. && (short_format == other.short_format);
  185. }
  186. size_t full_format { 0 };
  187. size_t long_format { 0 };
  188. size_t medium_format { 0 };
  189. size_t short_format { 0 };
  190. };
  191. template<>
  192. struct AK::Formatter<CalendarFormat> : Formatter<FormatString> {
  193. ErrorOr<void> format(FormatBuilder& builder, CalendarFormat const& pattern)
  194. {
  195. return Formatter<FormatString>::format(builder,
  196. "{{ {}, {}, {}, {} }}"sv,
  197. pattern.full_format,
  198. pattern.long_format,
  199. pattern.medium_format,
  200. pattern.short_format);
  201. }
  202. };
  203. template<>
  204. struct AK::Traits<CalendarFormat> : public GenericTraits<CalendarFormat> {
  205. static unsigned hash(CalendarFormat const& c) { return c.hash(); }
  206. };
  207. using SymbolList = Vector<size_t>;
  208. struct CalendarSymbols {
  209. unsigned hash() const
  210. {
  211. auto hash = pair_int_hash(narrow_symbols, short_symbols);
  212. hash = pair_int_hash(hash, long_symbols);
  213. return hash;
  214. }
  215. bool operator==(CalendarSymbols const& other) const
  216. {
  217. return (narrow_symbols == other.narrow_symbols)
  218. && (short_symbols == other.short_symbols)
  219. && (long_symbols == other.long_symbols);
  220. }
  221. size_t narrow_symbols { 0 };
  222. size_t short_symbols { 0 };
  223. size_t long_symbols { 0 };
  224. };
  225. template<>
  226. struct AK::Formatter<CalendarSymbols> : Formatter<FormatString> {
  227. ErrorOr<void> format(FormatBuilder& builder, CalendarSymbols const& symbols)
  228. {
  229. return Formatter<FormatString>::format(builder,
  230. "{{ {}, {}, {} }}"sv,
  231. symbols.narrow_symbols,
  232. symbols.short_symbols,
  233. symbols.long_symbols);
  234. }
  235. };
  236. template<>
  237. struct AK::Traits<CalendarSymbols> : public GenericTraits<CalendarSymbols> {
  238. static unsigned hash(CalendarSymbols const& c) { return c.hash(); }
  239. };
  240. using CalendarPatternList = Vector<size_t>;
  241. using CalendarRangePatternList = Vector<size_t>;
  242. using CalendarSymbolsList = Vector<size_t>;
  243. struct Calendar {
  244. unsigned hash() const
  245. {
  246. auto hash = int_hash(date_formats);
  247. hash = pair_int_hash(hash, time_formats);
  248. hash = pair_int_hash(hash, date_time_formats);
  249. hash = pair_int_hash(hash, available_formats);
  250. hash = pair_int_hash(hash, default_range_format);
  251. hash = pair_int_hash(hash, range_formats);
  252. hash = pair_int_hash(hash, range12_formats);
  253. hash = pair_int_hash(hash, symbols);
  254. return hash;
  255. }
  256. bool operator==(Calendar const& other) const
  257. {
  258. return (date_formats == other.date_formats)
  259. && (time_formats == other.time_formats)
  260. && (date_time_formats == other.date_time_formats)
  261. && (available_formats == other.available_formats)
  262. && (default_range_format == other.default_range_format)
  263. && (range_formats == other.range_formats)
  264. && (range12_formats == other.range12_formats)
  265. && (symbols == other.symbols);
  266. }
  267. size_t date_formats { 0 };
  268. size_t time_formats { 0 };
  269. size_t date_time_formats { 0 };
  270. size_t available_formats { 0 };
  271. size_t default_range_format { 0 };
  272. size_t range_formats { 0 };
  273. size_t range12_formats { 0 };
  274. size_t symbols { 0 };
  275. };
  276. template<>
  277. struct AK::Formatter<Calendar> : Formatter<FormatString> {
  278. ErrorOr<void> format(FormatBuilder& builder, Calendar const& calendar)
  279. {
  280. return Formatter<FormatString>::format(builder,
  281. "{{ {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  282. calendar.date_formats,
  283. calendar.time_formats,
  284. calendar.date_time_formats,
  285. calendar.available_formats,
  286. calendar.default_range_format,
  287. calendar.range_formats,
  288. calendar.range12_formats,
  289. calendar.symbols);
  290. }
  291. };
  292. template<>
  293. struct AK::Traits<Calendar> : public GenericTraits<Calendar> {
  294. static unsigned hash(Calendar const& c) { return c.hash(); }
  295. };
  296. struct TimeZoneNames {
  297. unsigned hash() const
  298. {
  299. auto hash = pair_int_hash(short_standard_name, long_standard_name);
  300. hash = pair_int_hash(hash, short_daylight_name);
  301. hash = pair_int_hash(hash, long_daylight_name);
  302. hash = pair_int_hash(hash, short_generic_name);
  303. hash = pair_int_hash(hash, long_generic_name);
  304. return hash;
  305. }
  306. bool operator==(TimeZoneNames const& other) const
  307. {
  308. return (short_standard_name == other.short_standard_name)
  309. && (long_standard_name == other.long_standard_name)
  310. && (short_daylight_name == other.short_daylight_name)
  311. && (long_daylight_name == other.long_daylight_name)
  312. && (short_generic_name == other.short_generic_name)
  313. && (long_generic_name == other.long_generic_name);
  314. }
  315. size_t short_standard_name { 0 };
  316. size_t long_standard_name { 0 };
  317. size_t short_daylight_name { 0 };
  318. size_t long_daylight_name { 0 };
  319. size_t short_generic_name { 0 };
  320. size_t long_generic_name { 0 };
  321. };
  322. template<>
  323. struct AK::Formatter<TimeZoneNames> : Formatter<FormatString> {
  324. ErrorOr<void> format(FormatBuilder& builder, TimeZoneNames const& time_zone)
  325. {
  326. return Formatter<FormatString>::format(builder,
  327. "{{ {}, {}, {}, {}, {}, {} }}"sv,
  328. time_zone.short_standard_name,
  329. time_zone.long_standard_name,
  330. time_zone.short_daylight_name,
  331. time_zone.long_daylight_name,
  332. time_zone.short_generic_name,
  333. time_zone.long_generic_name);
  334. }
  335. };
  336. template<>
  337. struct AK::Traits<TimeZoneNames> : public GenericTraits<TimeZoneNames> {
  338. static unsigned hash(TimeZoneNames const& t) { return t.hash(); }
  339. };
  340. struct TimeZoneFormat {
  341. unsigned hash() const
  342. {
  343. auto hash = int_hash(symbol_ahead_sign);
  344. hash = pair_int_hash(hash, symbol_ahead_separator);
  345. hash = pair_int_hash(hash, symbol_behind_sign);
  346. hash = pair_int_hash(hash, symbol_behind_separator);
  347. hash = pair_int_hash(hash, gmt_format);
  348. hash = pair_int_hash(hash, gmt_zero_format);
  349. return hash;
  350. }
  351. bool operator==(TimeZoneFormat const& other) const
  352. {
  353. return (symbol_ahead_sign == other.symbol_ahead_sign)
  354. && (symbol_ahead_separator == other.symbol_ahead_separator)
  355. && (symbol_behind_sign == other.symbol_behind_sign)
  356. && (symbol_behind_separator == other.symbol_behind_separator)
  357. && (gmt_format == other.gmt_format)
  358. && (gmt_zero_format == other.gmt_zero_format);
  359. }
  360. size_t symbol_ahead_sign { 0 };
  361. size_t symbol_ahead_separator { 0 };
  362. size_t symbol_behind_sign { 0 };
  363. size_t symbol_behind_separator { 0 };
  364. size_t gmt_format { 0 };
  365. size_t gmt_zero_format { 0 };
  366. };
  367. template<>
  368. struct AK::Formatter<TimeZoneFormat> : Formatter<FormatString> {
  369. ErrorOr<void> format(FormatBuilder& builder, TimeZoneFormat const& time_zone_format)
  370. {
  371. return Formatter<FormatString>::format(builder, "{{ {}, {}, {}, {}, {}, {} }}"sv,
  372. time_zone_format.symbol_ahead_sign,
  373. time_zone_format.symbol_ahead_separator,
  374. time_zone_format.symbol_behind_sign,
  375. time_zone_format.symbol_behind_separator,
  376. time_zone_format.gmt_format,
  377. time_zone_format.gmt_zero_format);
  378. }
  379. };
  380. template<>
  381. struct AK::Traits<TimeZoneFormat> : public GenericTraits<TimeZoneFormat> {
  382. static unsigned hash(TimeZoneFormat const& t) { return t.hash(); }
  383. };
  384. struct DayPeriod {
  385. unsigned hash() const
  386. {
  387. auto hash = int_hash(static_cast<u8>(day_period));
  388. hash = pair_int_hash(hash, begin);
  389. hash = pair_int_hash(hash, end);
  390. return hash;
  391. }
  392. bool operator==(DayPeriod const& other) const
  393. {
  394. return (day_period == other.day_period)
  395. && (begin == other.begin)
  396. && (end == other.end);
  397. }
  398. Locale::DayPeriod day_period {};
  399. u8 begin { 0 };
  400. u8 end { 0 };
  401. };
  402. template<>
  403. struct AK::Formatter<DayPeriod> : Formatter<FormatString> {
  404. ErrorOr<void> format(FormatBuilder& builder, DayPeriod const& day_period)
  405. {
  406. return Formatter<FormatString>::format(builder,
  407. "{{ {}, {}, {} }}"sv,
  408. static_cast<u8>(day_period.day_period),
  409. day_period.begin,
  410. day_period.end);
  411. }
  412. };
  413. template<>
  414. struct AK::Traits<DayPeriod> : public GenericTraits<DayPeriod> {
  415. static unsigned hash(DayPeriod const& d) { return d.hash(); }
  416. };
  417. using TimeZoneNamesList = Vector<size_t>;
  418. using DayPeriodList = Vector<size_t>;
  419. using HourCycleList = Vector<Locale::HourCycle>;
  420. template<>
  421. struct AK::Formatter<Locale::HourCycle> : Formatter<FormatString> {
  422. ErrorOr<void> format(FormatBuilder& builder, Locale::HourCycle hour_cycle)
  423. {
  424. return builder.put_u64(to_underlying(hour_cycle));
  425. }
  426. };
  427. struct LocaleData {
  428. HashMap<DeprecatedString, size_t> calendars;
  429. size_t time_zones { 0 };
  430. size_t time_zone_formats { 0 };
  431. size_t day_periods { 0 };
  432. };
  433. struct CLDR {
  434. UniqueStringStorage unique_strings;
  435. UniqueStorage<CalendarPattern> unique_patterns;
  436. UniqueStorage<CalendarPatternList> unique_pattern_lists;
  437. UniqueStorage<CalendarRangePattern> unique_range_patterns;
  438. UniqueStorage<CalendarRangePatternList> unique_range_pattern_lists;
  439. UniqueStorage<CalendarFormat> unique_formats;
  440. UniqueStorage<SymbolList> unique_symbol_lists;
  441. UniqueStorage<CalendarSymbols> unique_calendar_symbols;
  442. UniqueStorage<CalendarSymbolsList> unique_calendar_symbols_lists;
  443. UniqueStorage<Calendar> unique_calendars;
  444. UniqueStorage<TimeZoneNames> unique_time_zones;
  445. UniqueStorage<TimeZoneNamesList> unique_time_zone_lists;
  446. UniqueStorage<TimeZoneFormat> unique_time_zone_formats;
  447. UniqueStorage<DayPeriod> unique_day_periods;
  448. UniqueStorage<DayPeriodList> unique_day_period_lists;
  449. UniqueStorage<HourCycleList> unique_hour_cycle_lists;
  450. HashMap<DeprecatedString, LocaleData> locales;
  451. HashMap<DeprecatedString, size_t> hour_cycles;
  452. Vector<DeprecatedString> hour_cycle_regions;
  453. HashMap<DeprecatedString, u8> minimum_days;
  454. Vector<DeprecatedString> minimum_days_regions;
  455. HashMap<DeprecatedString, Locale::Weekday> first_day;
  456. Vector<DeprecatedString> first_day_regions;
  457. HashMap<DeprecatedString, Locale::Weekday> weekend_start;
  458. Vector<DeprecatedString> weekend_start_regions;
  459. HashMap<DeprecatedString, Locale::Weekday> weekend_end;
  460. Vector<DeprecatedString> weekend_end_regions;
  461. HashMap<DeprecatedString, Vector<TimeZone::TimeZone>> meta_zones;
  462. Vector<DeprecatedString> time_zones { "UTC"sv };
  463. Vector<DeprecatedString> calendars;
  464. };
  465. static Optional<Locale::DayPeriod> day_period_from_string(StringView day_period)
  466. {
  467. if (day_period == "am"sv)
  468. return Locale::DayPeriod::AM;
  469. if (day_period == "pm"sv)
  470. return Locale::DayPeriod::PM;
  471. if (day_period == "noon"sv)
  472. return Locale::DayPeriod::Noon;
  473. if (day_period == "morning1"sv)
  474. return Locale::DayPeriod::Morning1;
  475. if (day_period == "morning2"sv)
  476. return Locale::DayPeriod::Morning2;
  477. if (day_period == "afternoon1"sv)
  478. return Locale::DayPeriod::Afternoon1;
  479. if (day_period == "afternoon2"sv)
  480. return Locale::DayPeriod::Afternoon2;
  481. if (day_period == "evening1"sv)
  482. return Locale::DayPeriod::Evening1;
  483. if (day_period == "evening2"sv)
  484. return Locale::DayPeriod::Evening2;
  485. if (day_period == "night1"sv)
  486. return Locale::DayPeriod::Night1;
  487. if (day_period == "night2"sv)
  488. return Locale::DayPeriod::Night2;
  489. return {};
  490. }
  491. static ErrorOr<void> parse_hour_cycles(DeprecatedString core_path, CLDR& cldr)
  492. {
  493. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  494. LexicalPath time_data_path(move(core_path));
  495. time_data_path = time_data_path.append("supplemental"sv);
  496. time_data_path = time_data_path.append("timeData.json"sv);
  497. auto time_data = TRY(read_json_file(time_data_path.string()));
  498. auto const& supplemental_object = time_data.as_object().get_object("supplemental"sv).value();
  499. auto const& time_data_object = supplemental_object.get_object("timeData"sv).value();
  500. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Locale::HourCycle> {
  501. if (hour_cycle == "h"sv)
  502. return Locale::HourCycle::H12;
  503. if (hour_cycle == "H"sv)
  504. return Locale::HourCycle::H23;
  505. if (hour_cycle == "K"sv)
  506. return Locale::HourCycle::H11;
  507. if (hour_cycle == "k"sv)
  508. return Locale::HourCycle::H24;
  509. return {};
  510. };
  511. time_data_object.for_each_member([&](auto const& key, JsonValue const& value) {
  512. auto allowed_hour_cycles_string = value.as_object().get_deprecated_string("_allowed"sv).value();
  513. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  514. Vector<Locale::HourCycle> hour_cycles;
  515. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  516. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  517. hour_cycles.append(*hour_cycle);
  518. }
  519. auto hour_cycles_index = cldr.unique_hour_cycle_lists.ensure(move(hour_cycles));
  520. cldr.hour_cycles.set(key, hour_cycles_index);
  521. if (!cldr.hour_cycle_regions.contains_slow(key))
  522. cldr.hour_cycle_regions.append(key);
  523. });
  524. return {};
  525. }
  526. static ErrorOr<void> parse_week_data(DeprecatedString core_path, CLDR& cldr)
  527. {
  528. // https://unicode.org/reports/tr35/tr35-dates.html#Week_Data
  529. LexicalPath week_data_path(move(core_path));
  530. week_data_path = week_data_path.append("supplemental"sv);
  531. week_data_path = week_data_path.append("weekData.json"sv);
  532. auto week_data = TRY(read_json_file(week_data_path.string()));
  533. auto const& supplemental_object = week_data.as_object().get_object("supplemental"sv).value();
  534. auto const& week_data_object = supplemental_object.get_object("weekData"sv).value();
  535. auto parse_weekday = [](StringView day) -> Locale::Weekday {
  536. if (day == "sun"sv)
  537. return Locale::Weekday::Sunday;
  538. if (day == "mon"sv)
  539. return Locale::Weekday::Monday;
  540. if (day == "tue"sv)
  541. return Locale::Weekday::Tuesday;
  542. if (day == "wed"sv)
  543. return Locale::Weekday::Wednesday;
  544. if (day == "thu"sv)
  545. return Locale::Weekday::Thursday;
  546. if (day == "fri"sv)
  547. return Locale::Weekday::Friday;
  548. if (day == "sat"sv)
  549. return Locale::Weekday::Saturday;
  550. VERIFY_NOT_REACHED();
  551. };
  552. auto parse_regional_weekdays = [&](auto const& region, auto const& weekday, auto& weekdays_map, auto& weekday_regions) {
  553. if (region.ends_with("alt-variant"sv))
  554. return;
  555. weekdays_map.set(region, parse_weekday(weekday));
  556. if (!weekday_regions.contains_slow(region))
  557. weekday_regions.append(region);
  558. };
  559. auto const& minimum_days_object = week_data_object.get_object("minDays"sv).value();
  560. auto const& first_day_object = week_data_object.get_object("firstDay"sv).value();
  561. auto const& weekend_start_object = week_data_object.get_object("weekendStart"sv).value();
  562. auto const& weekend_end_object = week_data_object.get_object("weekendEnd"sv).value();
  563. minimum_days_object.for_each_member([&](auto const& region, auto const& value) {
  564. auto minimum_days = value.as_string().template to_uint<u8>();
  565. cldr.minimum_days.set(region, *minimum_days);
  566. if (!cldr.minimum_days_regions.contains_slow(region))
  567. cldr.minimum_days_regions.append(region);
  568. });
  569. first_day_object.for_each_member([&](auto const& region, auto const& value) {
  570. parse_regional_weekdays(region, value.as_string(), cldr.first_day, cldr.first_day_regions);
  571. });
  572. weekend_start_object.for_each_member([&](auto const& region, auto const& value) {
  573. parse_regional_weekdays(region, value.as_string(), cldr.weekend_start, cldr.weekend_start_regions);
  574. });
  575. weekend_end_object.for_each_member([&](auto const& region, auto const& value) {
  576. parse_regional_weekdays(region, value.as_string(), cldr.weekend_end, cldr.weekend_end_regions);
  577. });
  578. return {};
  579. }
  580. static ErrorOr<void> parse_meta_zones(DeprecatedString core_path, CLDR& cldr)
  581. {
  582. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  583. LexicalPath meta_zone_path(move(core_path));
  584. meta_zone_path = meta_zone_path.append("supplemental"sv);
  585. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  586. auto meta_zone = TRY(read_json_file(meta_zone_path.string()));
  587. auto const& supplemental_object = meta_zone.as_object().get_object("supplemental"sv).value();
  588. auto const& meta_zone_object = supplemental_object.get_object("metaZones"sv).value();
  589. auto const& meta_zone_array = meta_zone_object.get_array("metazones"sv).value();
  590. meta_zone_array.for_each([&](JsonValue const& value) {
  591. auto const& mapping = value.as_object().get_object("mapZone"sv).value();
  592. auto const& meta_zone = mapping.get_deprecated_string("_other"sv).value();
  593. auto const& golden_zone = mapping.get_deprecated_string("_type"sv).value();
  594. if (auto time_zone = TimeZone::time_zone_from_string(golden_zone); time_zone.has_value()) {
  595. auto& golden_zones = cldr.meta_zones.ensure(meta_zone);
  596. golden_zones.append(*time_zone);
  597. }
  598. });
  599. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  600. if (auto time_zone = TimeZone::time_zone_from_string("UTC"sv); time_zone.has_value())
  601. cldr.meta_zones.set("UTC"sv, { *time_zone });
  602. return {};
  603. }
  604. static constexpr auto is_char(char ch)
  605. {
  606. return [ch](auto c) { return c == ch; };
  607. }
  608. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  609. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  610. // extra spaces around the day period. Some example expected removals:
  611. //
  612. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  613. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  614. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  615. static ErrorOr<String> remove_period_from_pattern(String pattern)
  616. {
  617. auto is_surrounding_space = [&](auto code_point_iterator) {
  618. if (code_point_iterator.done())
  619. return false;
  620. constexpr auto spaces = Array { static_cast<u32>(0x0020), 0x00a0, 0x2009, 0x202f };
  621. return spaces.span().contains_slow(*code_point_iterator);
  622. };
  623. auto is_opening = [&](auto code_point_iterator) {
  624. if (code_point_iterator.done())
  625. return false;
  626. return *code_point_iterator == '{';
  627. };
  628. auto is_closing = [&](auto code_point_iterator) {
  629. if (code_point_iterator.done())
  630. return false;
  631. return *code_point_iterator == '}';
  632. };
  633. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  634. auto index = pattern.find_byte_offset(remove);
  635. if (!index.has_value())
  636. continue;
  637. Utf8View utf8_pattern { pattern };
  638. Utf8CodePointIterator before_removal;
  639. Utf8CodePointIterator after_removal;
  640. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  641. before_removal = it;
  642. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  643. after_removal = it;
  644. auto pattern_view = pattern.bytes_as_string_view();
  645. if (is_surrounding_space(before_removal) && !is_opening(after_removal)) {
  646. pattern = TRY(String::formatted("{}{}",
  647. pattern_view.substring_view(0, *index - before_removal.underlying_code_point_length_in_bytes()),
  648. pattern_view.substring_view(*index + remove.length())));
  649. } else if (is_surrounding_space(after_removal) && !is_closing(before_removal)) {
  650. pattern = TRY(String::formatted("{}{}",
  651. pattern_view.substring_view(0, *index),
  652. pattern_view.substring_view(*index + remove.length() + after_removal.underlying_code_point_length_in_bytes())));
  653. } else {
  654. pattern = TRY(String::formatted("{}{}",
  655. pattern_view.substring_view(0, *index),
  656. pattern_view.substring_view(*index + remove.length())));
  657. }
  658. }
  659. return pattern;
  660. }
  661. static ErrorOr<Optional<CalendarPattern>> parse_date_time_pattern_raw(DeprecatedString pattern, DeprecatedString skeleton, CLDR& cldr)
  662. {
  663. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  664. using Locale::CalendarPatternStyle;
  665. CalendarPattern format {};
  666. if (!skeleton.is_empty())
  667. format.skeleton_index = cldr.unique_strings.ensure(move(skeleton));
  668. GenericLexer lexer { pattern };
  669. StringBuilder builder;
  670. bool hour12 { false };
  671. while (!lexer.is_eof()) {
  672. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  673. // processing (this just avoids conflicts with the patterns below).
  674. if (lexer.next_is(is_quote)) {
  675. builder.append(lexer.consume_quoted_string());
  676. continue;
  677. }
  678. auto starting_char = lexer.peek();
  679. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  680. // Era
  681. if (all_of(segment, is_char('G'))) {
  682. builder.append("{era}"sv);
  683. if (segment.length() <= 3)
  684. format.era = CalendarPatternStyle::Short;
  685. else if (segment.length() == 4)
  686. format.era = CalendarPatternStyle::Long;
  687. else
  688. format.era = CalendarPatternStyle::Narrow;
  689. }
  690. // Year
  691. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  692. builder.append("{year}"sv);
  693. if (segment.length() == 2)
  694. format.year = CalendarPatternStyle::TwoDigit;
  695. else
  696. format.year = CalendarPatternStyle::Numeric;
  697. }
  698. // Quarter
  699. else if (all_of(segment, is_any_of("qQ"sv))) {
  700. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  701. return OptionalNone {};
  702. }
  703. // Month
  704. else if (all_of(segment, is_any_of("ML"sv))) {
  705. builder.append("{month}"sv);
  706. if (segment.length() == 1)
  707. format.month = CalendarPatternStyle::Numeric;
  708. else if (segment.length() == 2)
  709. format.month = CalendarPatternStyle::TwoDigit;
  710. else if (segment.length() == 3)
  711. format.month = CalendarPatternStyle::Short;
  712. else if (segment.length() == 4)
  713. format.month = CalendarPatternStyle::Long;
  714. else if (segment.length() == 5)
  715. format.month = CalendarPatternStyle::Narrow;
  716. } else if (all_of(segment, is_char('l'))) {
  717. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  718. return OptionalNone {};
  719. }
  720. // Week
  721. else if (all_of(segment, is_any_of("wW"sv))) {
  722. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  723. return OptionalNone {};
  724. }
  725. // Day
  726. else if (all_of(segment, is_char('d'))) {
  727. builder.append("{day}"sv);
  728. if (segment.length() == 1)
  729. format.day = CalendarPatternStyle::Numeric;
  730. else
  731. format.day = CalendarPatternStyle::TwoDigit;
  732. } else if (all_of(segment, is_any_of("DFg"sv))) {
  733. builder.append("{day}"sv);
  734. format.day = CalendarPatternStyle::Numeric;
  735. }
  736. // Weekday
  737. else if (all_of(segment, is_char('E'))) {
  738. builder.append("{weekday}"sv);
  739. if (segment.length() == 4)
  740. format.weekday = CalendarPatternStyle::Long;
  741. else if (segment.length() == 5)
  742. format.weekday = CalendarPatternStyle::Narrow;
  743. else
  744. format.weekday = CalendarPatternStyle::Short;
  745. } else if (all_of(segment, is_any_of("ec"sv))) {
  746. builder.append("{weekday}"sv);
  747. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  748. // pattern styles are not supported by Intl.DateTimeFormat.
  749. if (segment.length() <= 2)
  750. return OptionalNone {};
  751. if (segment.length() == 4)
  752. format.weekday = CalendarPatternStyle::Long;
  753. else if (segment.length() == 5)
  754. format.weekday = CalendarPatternStyle::Narrow;
  755. else
  756. format.weekday = CalendarPatternStyle::Short;
  757. }
  758. // Period
  759. else if (all_of(segment, is_any_of("ab"sv))) {
  760. builder.append("{ampm}"sv);
  761. hour12 = true;
  762. } else if (all_of(segment, is_char('B'))) {
  763. builder.append("{dayPeriod}"sv);
  764. hour12 = true;
  765. if (segment.length() == 4)
  766. format.day_period = CalendarPatternStyle::Long;
  767. else if (segment.length() == 5)
  768. format.day_period = CalendarPatternStyle::Narrow;
  769. else
  770. format.day_period = CalendarPatternStyle::Short;
  771. }
  772. // Hour
  773. else if (all_of(segment, is_any_of("hHKk"sv))) {
  774. builder.append("{hour}"sv);
  775. if ((segment[0] == 'h') || (segment[0] == 'K'))
  776. hour12 = true;
  777. if (segment.length() == 1)
  778. format.hour = CalendarPatternStyle::Numeric;
  779. else
  780. format.hour = CalendarPatternStyle::TwoDigit;
  781. } else if (all_of(segment, is_any_of("jJC"sv))) {
  782. // TR-35 indicates these should not be used.
  783. return OptionalNone {};
  784. }
  785. // Minute
  786. else if (all_of(segment, is_char('m'))) {
  787. builder.append("{minute}"sv);
  788. if (segment.length() == 1)
  789. format.minute = CalendarPatternStyle::Numeric;
  790. else
  791. format.minute = CalendarPatternStyle::TwoDigit;
  792. }
  793. // Second
  794. else if (all_of(segment, is_char('s'))) {
  795. builder.append("{second}"sv);
  796. if (segment.length() == 1)
  797. format.second = CalendarPatternStyle::Numeric;
  798. else
  799. format.second = CalendarPatternStyle::TwoDigit;
  800. } else if (all_of(segment, is_char('S'))) {
  801. builder.append("{fractionalSecondDigits}"sv);
  802. VERIFY(segment.length() <= 3);
  803. format.fractional_second_digits = static_cast<u8>(segment.length());
  804. } else if (all_of(segment, is_char('A'))) {
  805. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  806. return OptionalNone {};
  807. }
  808. // Zone
  809. else if (all_of(segment, is_any_of("zV"sv))) {
  810. builder.append("{timeZoneName}"sv);
  811. if (segment.length() < 4)
  812. format.time_zone_name = CalendarPatternStyle::Short;
  813. else
  814. format.time_zone_name = CalendarPatternStyle::Long;
  815. } else if (all_of(segment, is_any_of("ZOXx"sv))) {
  816. builder.append("{timeZoneName}"sv);
  817. if (segment.length() < 4)
  818. format.time_zone_name = CalendarPatternStyle::ShortOffset;
  819. else
  820. format.time_zone_name = CalendarPatternStyle::LongOffset;
  821. } else if (all_of(segment, is_char('v'))) {
  822. builder.append("{timeZoneName}"sv);
  823. if (segment.length() < 4)
  824. format.time_zone_name = CalendarPatternStyle::ShortGeneric;
  825. else
  826. format.time_zone_name = CalendarPatternStyle::LongGeneric;
  827. }
  828. // Non-patterns
  829. else {
  830. builder.append(segment);
  831. }
  832. }
  833. auto parsed_pattern = TRY(builder.to_string());
  834. if (hour12) {
  835. format.pattern = TRY(remove_period_from_pattern(parsed_pattern));
  836. format.pattern12 = move(parsed_pattern);
  837. } else {
  838. format.pattern = move(parsed_pattern);
  839. }
  840. return format;
  841. }
  842. static ErrorOr<Optional<size_t>> parse_date_time_pattern(DeprecatedString pattern, DeprecatedString skeleton, CLDR& cldr)
  843. {
  844. auto format = TRY(parse_date_time_pattern_raw(move(pattern), move(skeleton), cldr));
  845. if (!format.has_value())
  846. return OptionalNone {};
  847. format->pattern_index = cldr.unique_strings.ensure(format->pattern.to_deprecated_string());
  848. if (format->pattern12.has_value())
  849. format->pattern12_index = cldr.unique_strings.ensure(format->pattern12->to_deprecated_string());
  850. return Optional<size_t> { cldr.unique_patterns.ensure(format.release_value()) };
  851. }
  852. template<typename... Chars>
  853. static constexpr bool char_is_one_of(char ch, Chars&&... chars)
  854. {
  855. return ((ch == chars) || ...);
  856. }
  857. static ErrorOr<void> parse_interval_patterns(Calendar& calendar, JsonObject const& interval_formats_object, CLDR& cldr)
  858. {
  859. // https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats
  860. CalendarRangePatternList range_formats {};
  861. CalendarRangePatternList range12_formats {};
  862. auto name_of_field = [&](char field) {
  863. if (char_is_one_of(field, 'G'))
  864. return Locale::CalendarRangePattern::Field::Era;
  865. if (char_is_one_of(field, 'y', 'Y', 'u', 'U', 'r'))
  866. return Locale::CalendarRangePattern::Field::Year;
  867. if (char_is_one_of(field, 'M', 'L'))
  868. return Locale::CalendarRangePattern::Field::Month;
  869. if (char_is_one_of(field, 'd', 'D', 'F', 'g'))
  870. return Locale::CalendarRangePattern::Field::Day;
  871. if (char_is_one_of(field, 'a', 'b'))
  872. return Locale::CalendarRangePattern::Field::AmPm;
  873. if (char_is_one_of(field, 'B'))
  874. return Locale::CalendarRangePattern::Field::DayPeriod;
  875. if (char_is_one_of(field, 'h', 'H', 'K', 'k'))
  876. return Locale::CalendarRangePattern::Field::Hour;
  877. if (char_is_one_of(field, 'm'))
  878. return Locale::CalendarRangePattern::Field::Minute;
  879. if (char_is_one_of(field, 's'))
  880. return Locale::CalendarRangePattern::Field::Second;
  881. if (char_is_one_of(field, 'S'))
  882. return Locale::CalendarRangePattern::Field::FractionalSecondDigits;
  883. VERIFY_NOT_REACHED();
  884. };
  885. auto split_default_range_pattern = [&](auto skeleton, auto const& pattern) {
  886. auto start_range_end = pattern.find('}').value() + 1;
  887. auto end_range_begin = pattern.find_last('{').value();
  888. auto start_range = pattern.substring_view(0, start_range_end);
  889. auto separator = pattern.substring_view(start_range_end, end_range_begin - start_range_end);
  890. auto end_range = pattern.substring_view(end_range_begin);
  891. CalendarRangePattern format {};
  892. format.skeleton_index = cldr.unique_strings.ensure(skeleton);
  893. format.start_range = cldr.unique_strings.ensure(start_range);
  894. format.separator = cldr.unique_strings.ensure(separator);
  895. format.end_range = cldr.unique_strings.ensure(end_range);
  896. return format;
  897. };
  898. auto split_range_pattern = [&](auto skeleton, auto field, auto const& pattern, auto const& parsed_fields) {
  899. HashMap<StringView, size_t> partitions;
  900. StringView last_partition;
  901. auto begin_index = pattern.find_byte_offset('{');
  902. size_t end_index = 0;
  903. while (begin_index.has_value()) {
  904. end_index = pattern.find_byte_offset('}', *begin_index).value();
  905. auto partition = pattern.bytes_as_string_view().substring_view(*begin_index, end_index - *begin_index);
  906. if (partitions.contains(partition))
  907. break;
  908. partitions.set(partition, *begin_index);
  909. last_partition = partition;
  910. begin_index = pattern.find_byte_offset('{', end_index + 1);
  911. }
  912. VERIFY(begin_index.has_value() && !last_partition.is_empty());
  913. auto start_range_end = partitions.get(last_partition).value() + last_partition.length() + 1;
  914. auto pattern_view = pattern.bytes_as_string_view();
  915. auto start_range = pattern_view.substring_view(0, start_range_end);
  916. auto separator = pattern_view.substring_view(start_range_end, *begin_index - start_range_end);
  917. auto end_range = pattern_view.substring_view(*begin_index);
  918. CalendarRangePattern format {};
  919. format.skeleton_index = cldr.unique_strings.ensure(skeleton);
  920. format.field = field;
  921. format.start_range = cldr.unique_strings.ensure(start_range);
  922. format.separator = cldr.unique_strings.ensure(separator);
  923. format.end_range = cldr.unique_strings.ensure(end_range);
  924. format.for_each_calendar_field_zipped_with(parsed_fields, [](auto& format_field, auto const& parsed_field, auto) {
  925. format_field = parsed_field;
  926. });
  927. return format;
  928. };
  929. TRY(interval_formats_object.try_for_each_member([&](auto const& skeleton, auto const& value) -> ErrorOr<void> {
  930. if (skeleton == "intervalFormatFallback"sv) {
  931. auto range_format = split_default_range_pattern(skeleton, value.as_string());
  932. calendar.default_range_format = cldr.unique_range_patterns.ensure(move(range_format));
  933. return {};
  934. }
  935. TRY(value.as_object().try_for_each_member([&](auto const& field, auto const& pattern) -> ErrorOr<void> {
  936. if (field.ends_with("alt-variant"sv))
  937. return {};
  938. VERIFY(field.length() == 1);
  939. auto name = name_of_field(field[0]);
  940. auto format = TRY(parse_date_time_pattern_raw(pattern.as_string(), skeleton, cldr)).release_value();
  941. auto range_format = split_range_pattern(skeleton, name, format.pattern, format);
  942. range_formats.append(cldr.unique_range_patterns.ensure(move(range_format)));
  943. if (format.pattern12.has_value()) {
  944. auto range12_pattern = split_range_pattern(skeleton, name, *format.pattern12, format);
  945. range12_formats.append(cldr.unique_range_patterns.ensure(move(range12_pattern)));
  946. } else {
  947. range12_formats.append(range_formats.last());
  948. }
  949. return {};
  950. }));
  951. return {};
  952. }));
  953. calendar.range_formats = cldr.unique_range_pattern_lists.ensure(move(range_formats));
  954. calendar.range12_formats = cldr.unique_range_pattern_lists.ensure(move(range12_formats));
  955. return {};
  956. }
  957. static ErrorOr<void> generate_default_patterns(CalendarPatternList& formats, CLDR& cldr)
  958. {
  959. // For compatibility with ICU, we generate a list of default patterns for every locale:
  960. // https://github.com/unicode-org/icu/blob/release-71-1/icu4c/source/i18n/dtptngen.cpp#L1343-L1354=
  961. static constexpr auto default_patterns = Array { "G"sv, "y"sv, "M"sv, "E"sv, "D"sv, "F"sv, "d"sv, "a"sv, "B"sv, "H"sv, "mm"sv, "ss"sv, "SS"sv, "v"sv };
  962. for (auto pattern : default_patterns) {
  963. auto index = TRY(parse_date_time_pattern(pattern, pattern, cldr));
  964. VERIFY(index.has_value());
  965. if (!formats.contains_slow(*index))
  966. formats.append(*index);
  967. }
  968. return {};
  969. }
  970. static void generate_missing_patterns(Calendar& calendar, CalendarPatternList& formats, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, CLDR& cldr)
  971. {
  972. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  973. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  974. auto pattern = cldr.unique_strings.get(format);
  975. auto time_pattern = cldr.unique_strings.get(time_format);
  976. auto date_pattern = cldr.unique_strings.get(date_format);
  977. auto new_pattern = pattern.replace("{0}"sv, time_pattern, ReplaceMode::FirstOnly).replace("{1}"sv, date_pattern, ReplaceMode::FirstOnly);
  978. return cldr.unique_strings.ensure(move(new_pattern));
  979. };
  980. auto inject_fractional_second_digits = [&](auto format) {
  981. auto pattern = cldr.unique_strings.get(format);
  982. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv, ReplaceMode::FirstOnly);
  983. return cldr.unique_strings.ensure(move(new_pattern));
  984. };
  985. auto append_if_unique = [&](auto format) {
  986. auto format_index = cldr.unique_patterns.ensure(move(format));
  987. if (!formats.contains_slow(format_index))
  988. formats.append(format_index);
  989. };
  990. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  991. for (auto const& format : date_formats)
  992. append_if_unique(format);
  993. for (auto const& format : time_formats) {
  994. append_if_unique(format);
  995. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  996. auto new_format = format;
  997. new_format.fractional_second_digits = 2;
  998. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  999. if (new_format.pattern12_index != 0)
  1000. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  1001. time_formats_with_fractional_second_digits.append(new_format);
  1002. append_if_unique(move(new_format));
  1003. }
  1004. }
  1005. time_formats.extend(move(time_formats_with_fractional_second_digits));
  1006. for (auto const& date_format : date_formats) {
  1007. auto const& date_time_formats = cldr.unique_formats.get(calendar.date_time_formats);
  1008. size_t date_time_format_index = 0;
  1009. if (date_format.month == Locale::CalendarPatternStyle::Long) {
  1010. if (date_format.weekday.has_value())
  1011. date_time_format_index = date_time_formats.full_format;
  1012. else
  1013. date_time_format_index = date_time_formats.long_format;
  1014. } else if (date_format.month == Locale::CalendarPatternStyle::Short) {
  1015. date_time_format_index = date_time_formats.medium_format;
  1016. } else {
  1017. date_time_format_index = date_time_formats.short_format;
  1018. }
  1019. for (auto const& time_format : time_formats) {
  1020. auto format = cldr.unique_patterns.get(date_time_format_index);
  1021. if (time_format.pattern12_index != 0)
  1022. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  1023. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  1024. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  1025. if (date_field.has_value())
  1026. field = date_field;
  1027. });
  1028. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  1029. if (time_field.has_value())
  1030. field = time_field;
  1031. });
  1032. append_if_unique(move(format));
  1033. }
  1034. }
  1035. }
  1036. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, CLDR& cldr)
  1037. {
  1038. auto create_symbol_lists = [](size_t size) {
  1039. SymbolList narrow_symbol_list;
  1040. SymbolList short_symbol_list;
  1041. SymbolList long_symbol_list;
  1042. narrow_symbol_list.resize(size);
  1043. short_symbol_list.resize(size);
  1044. long_symbol_list.resize(size);
  1045. return Array<SymbolList, 3> { {
  1046. move(narrow_symbol_list),
  1047. move(short_symbol_list),
  1048. move(long_symbol_list),
  1049. } };
  1050. };
  1051. CalendarSymbolsList symbols_list {};
  1052. auto store_symbol_lists = [&](auto symbol, auto symbol_lists) {
  1053. auto symbol_index = to_underlying(symbol);
  1054. if (symbol_index >= symbols_list.size())
  1055. symbols_list.resize(symbol_index + 1);
  1056. CalendarSymbols symbols {};
  1057. symbols.narrow_symbols = cldr.unique_symbol_lists.ensure(move(symbol_lists[0]));
  1058. symbols.short_symbols = cldr.unique_symbol_lists.ensure(move(symbol_lists[1]));
  1059. symbols.long_symbols = cldr.unique_symbol_lists.ensure(move(symbol_lists[2]));
  1060. auto calendar_symbols_index = cldr.unique_calendar_symbols.ensure(move(symbols));
  1061. symbols_list[symbol_index] = calendar_symbols_index;
  1062. };
  1063. auto parse_era_symbols = [&](auto const& symbols_object) {
  1064. auto const& narrow_symbols = symbols_object.get_object("eraNarrow"sv).value();
  1065. auto const& short_symbols = symbols_object.get_object("eraAbbr"sv).value();
  1066. auto const& long_symbols = symbols_object.get_object("eraNames"sv).value();
  1067. auto symbol_lists = create_symbol_lists(2);
  1068. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1069. if (auto key_index = key.to_uint(); key_index.has_value())
  1070. symbols[*key_index] = cldr.unique_strings.ensure(move(symbol));
  1071. };
  1072. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1073. append_symbol(symbol_lists[0], key, value.as_string());
  1074. });
  1075. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1076. append_symbol(symbol_lists[1], key, value.as_string());
  1077. });
  1078. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1079. append_symbol(symbol_lists[2], key, value.as_string());
  1080. });
  1081. store_symbol_lists(Locale::CalendarSymbol::Era, move(symbol_lists));
  1082. };
  1083. auto parse_month_symbols = [&](auto const& symbols_object) {
  1084. auto const& narrow_symbols = symbols_object.get_object("narrow"sv).value();
  1085. auto const& short_symbols = symbols_object.get_object("abbreviated"sv).value();
  1086. auto const& long_symbols = symbols_object.get_object("wide"sv).value();
  1087. auto symbol_lists = create_symbol_lists(12);
  1088. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1089. auto key_index = key.to_uint().value() - 1;
  1090. symbols[key_index] = cldr.unique_strings.ensure(move(symbol));
  1091. };
  1092. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1093. append_symbol(symbol_lists[0], key, value.as_string());
  1094. });
  1095. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1096. append_symbol(symbol_lists[1], key, value.as_string());
  1097. });
  1098. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1099. append_symbol(symbol_lists[2], key, value.as_string());
  1100. });
  1101. store_symbol_lists(Locale::CalendarSymbol::Month, move(symbol_lists));
  1102. };
  1103. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  1104. auto const& narrow_symbols = symbols_object.get_object("narrow"sv).value();
  1105. auto const& short_symbols = symbols_object.get_object("abbreviated"sv).value();
  1106. auto const& long_symbols = symbols_object.get_object("wide"sv).value();
  1107. auto symbol_lists = create_symbol_lists(7);
  1108. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1109. if (key == "sun"sv)
  1110. symbols[to_underlying(Locale::Weekday::Sunday)] = cldr.unique_strings.ensure(move(symbol));
  1111. else if (key == "mon"sv)
  1112. symbols[to_underlying(Locale::Weekday::Monday)] = cldr.unique_strings.ensure(move(symbol));
  1113. else if (key == "tue"sv)
  1114. symbols[to_underlying(Locale::Weekday::Tuesday)] = cldr.unique_strings.ensure(move(symbol));
  1115. else if (key == "wed"sv)
  1116. symbols[to_underlying(Locale::Weekday::Wednesday)] = cldr.unique_strings.ensure(move(symbol));
  1117. else if (key == "thu"sv)
  1118. symbols[to_underlying(Locale::Weekday::Thursday)] = cldr.unique_strings.ensure(move(symbol));
  1119. else if (key == "fri"sv)
  1120. symbols[to_underlying(Locale::Weekday::Friday)] = cldr.unique_strings.ensure(move(symbol));
  1121. else if (key == "sat"sv)
  1122. symbols[to_underlying(Locale::Weekday::Saturday)] = cldr.unique_strings.ensure(move(symbol));
  1123. };
  1124. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1125. append_symbol(symbol_lists[0], key, value.as_string());
  1126. });
  1127. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1128. append_symbol(symbol_lists[1], key, value.as_string());
  1129. });
  1130. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1131. append_symbol(symbol_lists[2], key, value.as_string());
  1132. });
  1133. store_symbol_lists(Locale::CalendarSymbol::Weekday, move(symbol_lists));
  1134. };
  1135. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  1136. auto const& narrow_symbols = symbols_object.get_object("narrow"sv).value();
  1137. auto const& short_symbols = symbols_object.get_object("abbreviated"sv).value();
  1138. auto const& long_symbols = symbols_object.get_object("wide"sv).value();
  1139. auto symbol_lists = create_symbol_lists(11);
  1140. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1141. if (auto day_period = day_period_from_string(key); day_period.has_value())
  1142. symbols[to_underlying(*day_period)] = cldr.unique_strings.ensure(move(symbol));
  1143. };
  1144. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1145. append_symbol(symbol_lists[0], key, value.as_string());
  1146. });
  1147. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1148. append_symbol(symbol_lists[1], key, value.as_string());
  1149. });
  1150. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1151. append_symbol(symbol_lists[2], key, value.as_string());
  1152. });
  1153. store_symbol_lists(Locale::CalendarSymbol::DayPeriod, move(symbol_lists));
  1154. };
  1155. parse_era_symbols(calendar_object.get_object("eras"sv).value());
  1156. parse_month_symbols(calendar_object.get_object("months"sv)->get_object("format"sv).value());
  1157. parse_weekday_symbols(calendar_object.get_object("days"sv)->get_object("format"sv).value());
  1158. parse_day_period_symbols(calendar_object.get_object("dayPeriods"sv)->get_object("format"sv).value());
  1159. calendar.symbols = cldr.unique_calendar_symbols_lists.ensure(move(symbols_list));
  1160. }
  1161. static ErrorOr<void> parse_calendars(DeprecatedString locale_calendars_path, CLDR& cldr, LocaleData& locale)
  1162. {
  1163. LexicalPath calendars_path(move(locale_calendars_path));
  1164. if (!calendars_path.basename().starts_with("ca-"sv))
  1165. return {};
  1166. auto calendars = TRY(read_json_file(calendars_path.string()));
  1167. auto const& main_object = calendars.as_object().get_object("main"sv).value();
  1168. auto const& locale_object = main_object.get_object(calendars_path.parent().basename()).value();
  1169. auto const& dates_object = locale_object.get_object("dates"sv).value();
  1170. auto const& calendars_object = dates_object.get_object("calendars"sv).value();
  1171. auto parse_patterns = [&](auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) -> ErrorOr<size_t> {
  1172. auto parse_pattern = [&](auto name) -> ErrorOr<size_t> {
  1173. auto format = patterns_object.get_deprecated_string(name);
  1174. auto skeleton = skeletons_object.get_deprecated_string(name);
  1175. auto format_index = TRY(parse_date_time_pattern(format.value(), skeleton.value_or(DeprecatedString::empty()), cldr)).value();
  1176. if (patterns)
  1177. patterns->append(cldr.unique_patterns.get(format_index));
  1178. return format_index;
  1179. };
  1180. CalendarFormat formats {};
  1181. formats.full_format = TRY(parse_pattern("full"sv));
  1182. formats.long_format = TRY(parse_pattern("long"sv));
  1183. formats.medium_format = TRY(parse_pattern("medium"sv));
  1184. formats.short_format = TRY(parse_pattern("short"sv));
  1185. return cldr.unique_formats.ensure(move(formats));
  1186. };
  1187. TRY(calendars_object.try_for_each_member([&](auto const& calendar_name, JsonValue const& value) -> ErrorOr<void> {
  1188. // The generic calendar is not a supported Unicode calendar key, so skip it:
  1189. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  1190. if (calendar_name == "generic"sv)
  1191. return {};
  1192. Calendar calendar {};
  1193. CalendarPatternList available_formats {};
  1194. if (!cldr.calendars.contains_slow(calendar_name))
  1195. cldr.calendars.append(calendar_name);
  1196. Vector<CalendarPattern> date_formats;
  1197. Vector<CalendarPattern> time_formats;
  1198. auto const& date_formats_object = value.as_object().get_object("dateFormats"sv).value();
  1199. auto const& date_skeletons_object = value.as_object().get_object("dateSkeletons"sv).value();
  1200. calendar.date_formats = TRY(parse_patterns(date_formats_object, date_skeletons_object, &date_formats));
  1201. auto const& time_formats_object = value.as_object().get_object("timeFormats"sv).value();
  1202. auto const& time_skeletons_object = value.as_object().get_object("timeSkeletons"sv).value();
  1203. calendar.time_formats = TRY(parse_patterns(time_formats_object, time_skeletons_object, &time_formats));
  1204. auto const& standard_date_time_formats_object = value.as_object().get_object("dateTimeFormats-atTime"sv)->get_object("standard"sv).value();
  1205. calendar.date_time_formats = TRY(parse_patterns(standard_date_time_formats_object, JsonObject {}, nullptr));
  1206. auto const& date_time_formats_object = value.as_object().get_object("dateTimeFormats"sv).value();
  1207. auto const& available_formats_object = date_time_formats_object.get_object("availableFormats"sv).value();
  1208. TRY(available_formats_object.try_for_each_member([&](auto const& skeleton, JsonValue const& pattern) -> ErrorOr<void> {
  1209. auto pattern_index = TRY(parse_date_time_pattern(pattern.as_string(), skeleton, cldr));
  1210. if (!pattern_index.has_value())
  1211. return {};
  1212. auto const& format = cldr.unique_patterns.get(*pattern_index);
  1213. if (format.contains_only_date_fields())
  1214. date_formats.append(format);
  1215. else if (format.contains_only_time_fields())
  1216. time_formats.append(format);
  1217. if (!available_formats.contains_slow(*pattern_index))
  1218. available_formats.append(*pattern_index);
  1219. return {};
  1220. }));
  1221. auto const& interval_formats_object = date_time_formats_object.get_object("intervalFormats"sv).value();
  1222. TRY(parse_interval_patterns(calendar, interval_formats_object, cldr));
  1223. TRY(generate_default_patterns(available_formats, cldr));
  1224. generate_missing_patterns(calendar, available_formats, move(date_formats), move(time_formats), cldr);
  1225. parse_calendar_symbols(calendar, value.as_object(), cldr);
  1226. calendar.available_formats = cldr.unique_pattern_lists.ensure(move(available_formats));
  1227. locale.calendars.set(calendar_name, cldr.unique_calendars.ensure(move(calendar)));
  1228. return {};
  1229. }));
  1230. return {};
  1231. }
  1232. static ErrorOr<void> parse_time_zone_names(DeprecatedString locale_time_zone_names_path, CLDR& cldr, LocaleData& locale)
  1233. {
  1234. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  1235. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  1236. auto time_zone_names = TRY(read_json_file(time_zone_names_path.string()));
  1237. auto const& main_object = time_zone_names.as_object().get_object("main"sv).value();
  1238. auto const& locale_object = main_object.get_object(time_zone_names_path.parent().basename()).value();
  1239. auto const& dates_object = locale_object.get_object("dates"sv).value();
  1240. auto const& time_zone_names_object = dates_object.get_object("timeZoneNames"sv).value();
  1241. auto const& meta_zone_object = time_zone_names_object.get_object("metazone"sv);
  1242. auto const& hour_format_string = time_zone_names_object.get_deprecated_string("hourFormat"sv).value();
  1243. auto const& gmt_format_string = time_zone_names_object.get_deprecated_string("gmtFormat"sv).value();
  1244. auto const& gmt_zero_format_string = time_zone_names_object.get_deprecated_string("gmtZeroFormat"sv).value();
  1245. if (!meta_zone_object.has_value())
  1246. return {};
  1247. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object, StringView key) -> Optional<size_t> {
  1248. auto const& names = meta_zone_object.get_object(type);
  1249. if (!names.has_value())
  1250. return {};
  1251. auto const& name = names->get_deprecated_string(key);
  1252. if (name.has_value())
  1253. return cldr.unique_strings.ensure(name.value());
  1254. return {};
  1255. };
  1256. auto parse_hour_format = [&](auto const& format, auto& time_zone_formats) {
  1257. auto hour_formats = format.split_view(';');
  1258. auto hour_format_ahead_start = hour_formats[0].find('H').value();
  1259. auto separator_ahead_start = hour_formats[0].find_last('H').value() + 1;
  1260. auto separator_ahead_end = hour_formats[0].find('m').value();
  1261. auto hour_format_behind_start = hour_formats[1].find('H').value();
  1262. auto separator_behind_start = hour_formats[1].find_last('H').value() + 1;
  1263. auto separator_behind_end = hour_formats[1].find('m').value();
  1264. auto symbol_ahead_sign = hour_formats[0].substring_view(0, hour_format_ahead_start);
  1265. auto symbol_ahead_separator = hour_formats[0].substring_view(separator_ahead_start, separator_ahead_end - separator_ahead_start);
  1266. auto symbol_behind_sign = hour_formats[1].substring_view(0, hour_format_behind_start);
  1267. auto symbol_behind_separator = hour_formats[1].substring_view(separator_behind_start, separator_behind_end - separator_behind_start);
  1268. time_zone_formats.symbol_ahead_sign = cldr.unique_strings.ensure(symbol_ahead_sign);
  1269. time_zone_formats.symbol_ahead_separator = cldr.unique_strings.ensure(symbol_ahead_separator);
  1270. time_zone_formats.symbol_behind_sign = cldr.unique_strings.ensure(symbol_behind_sign);
  1271. time_zone_formats.symbol_behind_separator = cldr.unique_strings.ensure(symbol_behind_separator);
  1272. };
  1273. TimeZoneNamesList time_zones;
  1274. TimeZoneFormat time_zone_formats {};
  1275. parse_hour_format(hour_format_string, time_zone_formats);
  1276. time_zone_formats.gmt_format = cldr.unique_strings.ensure(gmt_format_string);
  1277. time_zone_formats.gmt_zero_format = cldr.unique_strings.ensure(gmt_zero_format_string);
  1278. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  1279. auto golden_zones = cldr.meta_zones.find(meta_zone);
  1280. if (golden_zones == cldr.meta_zones.end())
  1281. return;
  1282. TimeZoneNames time_zone_names {};
  1283. if (auto name = parse_name("long"sv, meta_zone_object, "standard"sv); name.has_value())
  1284. time_zone_names.long_standard_name = name.value();
  1285. if (auto name = parse_name("short"sv, meta_zone_object, "standard"sv); name.has_value())
  1286. time_zone_names.short_standard_name = name.value();
  1287. if (auto name = parse_name("long"sv, meta_zone_object, "daylight"sv); name.has_value())
  1288. time_zone_names.long_daylight_name = name.value();
  1289. if (auto name = parse_name("short"sv, meta_zone_object, "daylight"sv); name.has_value())
  1290. time_zone_names.short_daylight_name = name.value();
  1291. if (auto name = parse_name("long"sv, meta_zone_object, "generic"sv); name.has_value())
  1292. time_zone_names.long_generic_name = name.value();
  1293. if (auto name = parse_name("short"sv, meta_zone_object, "generic"sv); name.has_value())
  1294. time_zone_names.short_generic_name = name.value();
  1295. auto time_zone_index = cldr.unique_time_zones.ensure(move(time_zone_names));
  1296. for (auto golden_zone : golden_zones->value) {
  1297. auto time_zone = to_underlying(golden_zone);
  1298. if (time_zone >= time_zones.size())
  1299. time_zones.resize(time_zone + 1);
  1300. time_zones[time_zone] = time_zone_index;
  1301. }
  1302. };
  1303. meta_zone_object->for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  1304. parse_time_zone(meta_zone, value.as_object());
  1305. });
  1306. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  1307. auto const& zone_object = time_zone_names_object.get_object("zone"sv).value();
  1308. auto const& etc_object = zone_object.get_object("Etc"sv).value();
  1309. auto const& utc_object = etc_object.get_object("UTC"sv).value();
  1310. parse_time_zone("UTC"sv, utc_object);
  1311. locale.time_zones = cldr.unique_time_zone_lists.ensure(move(time_zones));
  1312. locale.time_zone_formats = cldr.unique_time_zone_formats.ensure(move(time_zone_formats));
  1313. return {};
  1314. }
  1315. static ErrorOr<void> parse_day_periods(DeprecatedString core_path, CLDR& cldr)
  1316. {
  1317. // https://unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rule_Sets
  1318. LexicalPath day_periods_path(move(core_path));
  1319. day_periods_path = day_periods_path.append("supplemental"sv);
  1320. day_periods_path = day_periods_path.append("dayPeriods.json"sv);
  1321. auto locale_day_periods = TRY(read_json_file(day_periods_path.string()));
  1322. auto const& supplemental_object = locale_day_periods.as_object().get_object("supplemental"sv).value();
  1323. auto const& day_periods_object = supplemental_object.get_object("dayPeriodRuleSet"sv).value();
  1324. auto parse_hour = [](auto const& time) {
  1325. auto hour_end_index = time.find(':').value();
  1326. // The times are in the form "hours:minutes", but we only need the hour segment.
  1327. // TR-35 explicitly states that minutes other than :00 must not be used.
  1328. VERIFY(time.substring_view(hour_end_index) == ":00"sv);
  1329. auto hour = time.substring_view(0, hour_end_index);
  1330. return hour.template to_uint<u8>().value();
  1331. };
  1332. auto parse_day_period = [&](auto const& symbol, auto const& ranges) -> Optional<DayPeriod> {
  1333. if (!ranges.has("_from"sv))
  1334. return {};
  1335. auto day_period = day_period_from_string(symbol);
  1336. if (!day_period.has_value())
  1337. return {};
  1338. auto begin = parse_hour(ranges.get_deprecated_string("_from"sv).value());
  1339. auto end = parse_hour(ranges.get_deprecated_string("_before"sv).value());
  1340. return DayPeriod { *day_period, begin, end };
  1341. };
  1342. day_periods_object.for_each_member([&](auto const& language, JsonValue const& value) {
  1343. auto locale = cldr.locales.find(language);
  1344. if (locale == cldr.locales.end())
  1345. return;
  1346. DayPeriodList day_periods;
  1347. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& ranges) {
  1348. if (auto day_period = parse_day_period(symbol, ranges.as_object()); day_period.has_value()) {
  1349. auto day_period_index = cldr.unique_day_periods.ensure(day_period.release_value());
  1350. day_periods.append(day_period_index);
  1351. }
  1352. });
  1353. locale->value.day_periods = cldr.unique_day_period_lists.ensure(move(day_periods));
  1354. });
  1355. return {};
  1356. }
  1357. static ErrorOr<void> parse_all_locales(DeprecatedString core_path, DeprecatedString dates_path, CLDR& cldr)
  1358. {
  1359. TRY(parse_hour_cycles(core_path, cldr));
  1360. TRY(parse_week_data(core_path, cldr));
  1361. TRY(parse_meta_zones(core_path, cldr));
  1362. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  1363. auto remove_variants_from_path = [&](DeprecatedString path) -> ErrorOr<DeprecatedString> {
  1364. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, LexicalPath::basename(path)));
  1365. StringBuilder builder;
  1366. builder.append(cldr.unique_strings.get(parsed_locale.language));
  1367. if (auto script = cldr.unique_strings.get(parsed_locale.script); !script.is_empty())
  1368. builder.appendff("-{}", script);
  1369. if (auto region = cldr.unique_strings.get(parsed_locale.region); !region.is_empty())
  1370. builder.appendff("-{}", region);
  1371. return builder.to_deprecated_string();
  1372. };
  1373. while (dates_iterator.has_next()) {
  1374. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  1375. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  1376. auto language = TRY(remove_variants_from_path(dates_path));
  1377. auto& locale = cldr.locales.ensure(language);
  1378. while (calendars_iterator.has_next()) {
  1379. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  1380. TRY(parse_calendars(move(calendars_path), cldr, locale));
  1381. }
  1382. TRY(parse_time_zone_names(move(dates_path), cldr, locale));
  1383. }
  1384. TRY(parse_day_periods(move(core_path), cldr));
  1385. return {};
  1386. }
  1387. static DeprecatedString format_identifier(StringView owner, DeprecatedString identifier)
  1388. {
  1389. identifier = identifier.replace("-"sv, "_"sv, ReplaceMode::All);
  1390. identifier = identifier.replace("/"sv, "_"sv, ReplaceMode::All);
  1391. if (all_of(identifier, is_ascii_digit))
  1392. return DeprecatedString::formatted("{}_{}", owner[0], identifier);
  1393. if (is_ascii_lower_alpha(identifier[0]))
  1394. return DeprecatedString::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  1395. return identifier;
  1396. }
  1397. static ErrorOr<void> generate_unicode_locale_header(Core::Stream::BufferedFile& file, CLDR& cldr)
  1398. {
  1399. StringBuilder builder;
  1400. SourceGenerator generator { builder };
  1401. generator.append(R"~~~(
  1402. #pragma once
  1403. #include <AK/Types.h>
  1404. namespace Locale {
  1405. )~~~");
  1406. generate_enum(generator, format_identifier, "Calendar"sv, {}, cldr.calendars);
  1407. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, cldr.hour_cycle_regions);
  1408. generate_enum(generator, format_identifier, "MinimumDaysRegion"sv, {}, cldr.minimum_days_regions);
  1409. generate_enum(generator, format_identifier, "FirstDayRegion"sv, {}, cldr.first_day_regions);
  1410. generate_enum(generator, format_identifier, "WeekendStartRegion"sv, {}, cldr.weekend_start_regions);
  1411. generate_enum(generator, format_identifier, "WeekendEndRegion"sv, {}, cldr.weekend_end_regions);
  1412. generator.append(R"~~~(
  1413. }
  1414. )~~~");
  1415. TRY(file.write(generator.as_string_view().bytes()));
  1416. return {};
  1417. }
  1418. static ErrorOr<void> generate_unicode_locale_implementation(Core::Stream::BufferedFile& file, CLDR& cldr)
  1419. {
  1420. StringBuilder builder;
  1421. SourceGenerator generator { builder };
  1422. generator.set("string_index_type"sv, cldr.unique_strings.type_that_fits());
  1423. generator.set("calendar_pattern_index_type"sv, cldr.unique_patterns.type_that_fits());
  1424. generator.set("calendar_pattern_list_index_type"sv, cldr.unique_pattern_lists.type_that_fits());
  1425. generator.set("calendar_range_pattern_index_type"sv, cldr.unique_range_patterns.type_that_fits());
  1426. generator.set("calendar_range_pattern_list_index_type"sv, cldr.unique_range_pattern_lists.type_that_fits());
  1427. generator.set("calendar_format_index_type"sv, cldr.unique_formats.type_that_fits());
  1428. generator.set("symbol_list_index_type"sv, cldr.unique_symbol_lists.type_that_fits());
  1429. generator.set("calendar_symbols_index_type"sv, cldr.unique_calendar_symbols.type_that_fits());
  1430. generator.set("calendar_symbols_list_index_type"sv, cldr.unique_calendar_symbols_lists.type_that_fits());
  1431. generator.set("calendar_index_type"sv, cldr.unique_calendars.type_that_fits());
  1432. generator.set("time_zone_index_type"sv, cldr.unique_time_zones.type_that_fits());
  1433. generator.set("time_zone_list_index_type"sv, cldr.unique_time_zone_lists.type_that_fits());
  1434. generator.set("day_period_index_type"sv, cldr.unique_day_periods.type_that_fits());
  1435. generator.set("day_period_list_index_type"sv, cldr.unique_day_period_lists.type_that_fits());
  1436. generator.append(R"~~~(
  1437. #include <AK/Array.h>
  1438. #include <AK/BinarySearch.h>
  1439. #include <AK/Error.h>
  1440. #include <AK/Optional.h>
  1441. #include <AK/String.h>
  1442. #include <AK/StringView.h>
  1443. #include <LibLocale/DateTimeFormat.h>
  1444. #include <LibLocale/DateTimeFormatData.h>
  1445. #include <LibLocale/Locale.h>
  1446. #include <LibLocale/LocaleData.h>
  1447. #include <LibTimeZone/TimeZone.h>
  1448. namespace Locale {
  1449. )~~~");
  1450. cldr.unique_strings.generate(generator);
  1451. generator.append(R"~~~(
  1452. template <typename SourceType, typename TargetType>
  1453. static void convert_calendar_fields(SourceType const& source, TargetType& target)
  1454. {
  1455. if (source.era != -1)
  1456. target.era = static_cast<CalendarPatternStyle>(source.era);
  1457. if (source.year != -1)
  1458. target.year = static_cast<CalendarPatternStyle>(source.year);
  1459. if (source.month != -1)
  1460. target.month = static_cast<CalendarPatternStyle>(source.month);
  1461. if (source.weekday != -1)
  1462. target.weekday = static_cast<CalendarPatternStyle>(source.weekday);
  1463. if (source.day != -1)
  1464. target.day = static_cast<CalendarPatternStyle>(source.day);
  1465. if (source.day_period != -1)
  1466. target.day_period = static_cast<CalendarPatternStyle>(source.day_period);
  1467. if (source.hour != -1)
  1468. target.hour = static_cast<CalendarPatternStyle>(source.hour);
  1469. if (source.minute != -1)
  1470. target.minute = static_cast<CalendarPatternStyle>(source.minute);
  1471. if (source.second != -1)
  1472. target.second = static_cast<CalendarPatternStyle>(source.second);
  1473. if (source.fractional_second_digits != -1)
  1474. target.fractional_second_digits = static_cast<u8>(source.fractional_second_digits);
  1475. if (source.time_zone_name != -1)
  1476. target.time_zone_name = static_cast<CalendarPatternStyle>(source.time_zone_name);
  1477. }
  1478. struct CalendarPatternImpl {
  1479. ErrorOr<CalendarPattern> to_unicode_calendar_pattern() const {
  1480. CalendarPattern calendar_pattern {};
  1481. calendar_pattern.skeleton = TRY(String::from_utf8(decode_string(skeleton)));
  1482. calendar_pattern.pattern = TRY(String::from_utf8(decode_string(pattern)));
  1483. if (pattern12 != 0)
  1484. calendar_pattern.pattern12 = TRY(String::from_utf8(decode_string(pattern12)));
  1485. convert_calendar_fields(*this, calendar_pattern);
  1486. return calendar_pattern;
  1487. }
  1488. @string_index_type@ skeleton { 0 };
  1489. @string_index_type@ pattern { 0 };
  1490. @string_index_type@ pattern12 { 0 };
  1491. i8 era { -1 };
  1492. i8 year { -1 };
  1493. i8 month { -1 };
  1494. i8 weekday { -1 };
  1495. i8 day { -1 };
  1496. i8 day_period { -1 };
  1497. i8 hour { -1 };
  1498. i8 minute { -1 };
  1499. i8 second { -1 };
  1500. i8 fractional_second_digits { -1 };
  1501. i8 time_zone_name { -1 };
  1502. };
  1503. struct CalendarRangePatternImpl {
  1504. ErrorOr<CalendarRangePattern> to_unicode_calendar_range_pattern() const {
  1505. CalendarRangePattern calendar_range_pattern {};
  1506. if (field != -1)
  1507. calendar_range_pattern.field = static_cast<CalendarRangePattern::Field>(field);
  1508. calendar_range_pattern.start_range = TRY(String::from_utf8(decode_string(start_range)));
  1509. calendar_range_pattern.separator = decode_string(separator);
  1510. calendar_range_pattern.end_range = TRY(String::from_utf8(decode_string(end_range)));
  1511. convert_calendar_fields(*this, calendar_range_pattern);
  1512. return calendar_range_pattern;
  1513. }
  1514. @string_index_type@ skeleton { 0 };
  1515. i8 field { -1 };
  1516. @string_index_type@ start_range { 0 };
  1517. @string_index_type@ separator { 0 };
  1518. @string_index_type@ end_range { 0 };
  1519. i8 era { -1 };
  1520. i8 year { -1 };
  1521. i8 month { -1 };
  1522. i8 weekday { -1 };
  1523. i8 day { -1 };
  1524. i8 day_period { -1 };
  1525. i8 hour { -1 };
  1526. i8 minute { -1 };
  1527. i8 second { -1 };
  1528. i8 fractional_second_digits { -1 };
  1529. i8 time_zone_name { -1 };
  1530. };
  1531. )~~~");
  1532. cldr.unique_patterns.generate(generator, "CalendarPatternImpl"sv, "s_calendar_patterns"sv, 10);
  1533. cldr.unique_pattern_lists.generate(generator, cldr.unique_patterns.type_that_fits(), "s_calendar_pattern_lists"sv);
  1534. cldr.unique_range_patterns.generate(generator, "CalendarRangePatternImpl"sv, "s_calendar_range_patterns"sv, 10);
  1535. cldr.unique_range_pattern_lists.generate(generator, cldr.unique_range_patterns.type_that_fits(), "s_calendar_range_pattern_lists"sv);
  1536. generator.append(R"~~~(
  1537. struct CalendarFormatImpl {
  1538. ErrorOr<CalendarFormat> to_unicode_calendar_format() const {
  1539. CalendarFormat calendar_format {};
  1540. calendar_format.full_format = TRY(s_calendar_patterns[full_format].to_unicode_calendar_pattern());
  1541. calendar_format.long_format = TRY(s_calendar_patterns[long_format].to_unicode_calendar_pattern());
  1542. calendar_format.medium_format = TRY(s_calendar_patterns[medium_format].to_unicode_calendar_pattern());
  1543. calendar_format.short_format = TRY(s_calendar_patterns[short_format].to_unicode_calendar_pattern());
  1544. return calendar_format;
  1545. }
  1546. @calendar_pattern_index_type@ full_format { 0 };
  1547. @calendar_pattern_index_type@ long_format { 0 };
  1548. @calendar_pattern_index_type@ medium_format { 0 };
  1549. @calendar_pattern_index_type@ short_format { 0 };
  1550. };
  1551. struct CalendarSymbols {
  1552. @symbol_list_index_type@ narrow_symbols { 0 };
  1553. @symbol_list_index_type@ short_symbols { 0 };
  1554. @symbol_list_index_type@ long_symbols { 0 };
  1555. };
  1556. struct CalendarData {
  1557. @calendar_format_index_type@ date_formats { 0 };
  1558. @calendar_format_index_type@ time_formats { 0 };
  1559. @calendar_format_index_type@ date_time_formats { 0 };
  1560. @calendar_pattern_list_index_type@ available_formats { 0 };
  1561. @calendar_range_pattern_index_type@ default_range_format { 0 };
  1562. @calendar_range_pattern_list_index_type@ range_formats { 0 };
  1563. @calendar_range_pattern_list_index_type@ range12_formats { 0 };
  1564. @calendar_symbols_list_index_type@ symbols { 0 };
  1565. };
  1566. struct TimeZoneNames {
  1567. @string_index_type@ short_standard_name { 0 };
  1568. @string_index_type@ long_standard_name { 0 };
  1569. @string_index_type@ short_daylight_name { 0 };
  1570. @string_index_type@ long_daylight_name { 0 };
  1571. @string_index_type@ short_generic_name { 0 };
  1572. @string_index_type@ long_generic_name { 0 };
  1573. };
  1574. struct TimeZoneFormatImpl {
  1575. TimeZoneFormat to_time_zone_format() const {
  1576. TimeZoneFormat time_zone_format {};
  1577. time_zone_format.symbol_ahead_sign = decode_string(symbol_ahead_sign);
  1578. time_zone_format.symbol_ahead_separator = decode_string(symbol_ahead_separator);
  1579. time_zone_format.symbol_behind_sign = decode_string(symbol_behind_sign);
  1580. time_zone_format.symbol_behind_separator = decode_string(symbol_behind_separator);
  1581. time_zone_format.gmt_format = decode_string(gmt_format);
  1582. time_zone_format.gmt_zero_format = decode_string(gmt_zero_format);
  1583. return time_zone_format;
  1584. }
  1585. @string_index_type@ symbol_ahead_sign { 0 };
  1586. @string_index_type@ symbol_ahead_separator { 0 };
  1587. @string_index_type@ symbol_behind_sign { 0 };
  1588. @string_index_type@ symbol_behind_separator { 0 };
  1589. @string_index_type@ gmt_format { 0 };
  1590. @string_index_type@ gmt_zero_format { 0 };
  1591. };
  1592. struct DayPeriodData {
  1593. u8 day_period { 0 };
  1594. u8 begin { 0 };
  1595. u8 end { 0 };
  1596. };
  1597. )~~~");
  1598. cldr.unique_formats.generate(generator, "CalendarFormatImpl"sv, "s_calendar_formats"sv, 10);
  1599. cldr.unique_symbol_lists.generate(generator, cldr.unique_strings.type_that_fits(), "s_symbol_lists"sv);
  1600. cldr.unique_calendar_symbols.generate(generator, "CalendarSymbols"sv, "s_calendar_symbols"sv, 10);
  1601. cldr.unique_calendar_symbols_lists.generate(generator, cldr.unique_calendar_symbols.type_that_fits(), "s_calendar_symbol_lists"sv);
  1602. cldr.unique_calendars.generate(generator, "CalendarData"sv, "s_calendars"sv, 10);
  1603. cldr.unique_time_zones.generate(generator, "TimeZoneNames"sv, "s_time_zones"sv, 30);
  1604. cldr.unique_time_zone_lists.generate(generator, cldr.unique_time_zones.type_that_fits(), "s_time_zone_lists"sv);
  1605. cldr.unique_time_zone_formats.generate(generator, "TimeZoneFormatImpl"sv, "s_time_zone_formats"sv, 30);
  1606. cldr.unique_day_periods.generate(generator, "DayPeriodData"sv, "s_day_periods"sv, 30);
  1607. cldr.unique_day_period_lists.generate(generator, cldr.unique_day_periods.type_that_fits(), "s_day_period_lists"sv);
  1608. cldr.unique_hour_cycle_lists.generate(generator, cldr.unique_hour_cycle_lists.type_that_fits(), "s_hour_cycle_lists"sv);
  1609. auto append_calendars = [&](DeprecatedString name, auto const& calendars) {
  1610. generator.set("name", name);
  1611. generator.set("size", DeprecatedString::number(calendars.size()));
  1612. generator.append(R"~~~(
  1613. static constexpr Array<@calendar_index_type@, @size@> @name@ { {)~~~");
  1614. bool first = true;
  1615. for (auto const& calendar_key : cldr.calendars) {
  1616. auto calendar = calendars.find(calendar_key)->value;
  1617. generator.append(first ? " "sv : ", "sv);
  1618. generator.append(DeprecatedString::number(calendar));
  1619. first = false;
  1620. }
  1621. generator.append(" } };");
  1622. };
  1623. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  1624. generator.set("type", type);
  1625. generator.set("name", name);
  1626. generator.set("size", DeprecatedString::number(keys.size()));
  1627. generator.append(R"~~~(
  1628. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1629. bool first = true;
  1630. for (auto const& key : keys) {
  1631. auto const& value = map.find(key)->value;
  1632. auto mapping = mapping_getter(value);
  1633. generator.append(first ? " "sv : ", "sv);
  1634. generator.append(DeprecatedString::number(mapping));
  1635. first = false;
  1636. }
  1637. generator.append(" } };");
  1638. };
  1639. auto locales = cldr.locales.keys();
  1640. quick_sort(locales);
  1641. generate_mapping(generator, cldr.locales, cldr.unique_calendars.type_that_fits(), "s_locale_calendars"sv, "s_calendars_{}"sv, format_identifier, [&](auto const& name, auto const& value) { append_calendars(name, value.calendars); });
  1642. append_mapping(locales, cldr.locales, cldr.unique_time_zones.type_that_fits(), "s_locale_time_zones"sv, [](auto const& locale) { return locale.time_zones; });
  1643. append_mapping(locales, cldr.locales, cldr.unique_time_zone_formats.type_that_fits(), "s_locale_time_zone_formats"sv, [](auto const& locale) { return locale.time_zone_formats; });
  1644. append_mapping(locales, cldr.locales, cldr.unique_day_periods.type_that_fits(), "s_locale_day_periods"sv, [](auto const& locale) { return locale.day_periods; });
  1645. append_mapping(cldr.hour_cycle_regions, cldr.hour_cycles, cldr.unique_hour_cycle_lists.type_that_fits(), "s_hour_cycles"sv, [](auto const& hour_cycles) { return hour_cycles; });
  1646. append_mapping(cldr.minimum_days_regions, cldr.minimum_days, "u8"sv, "s_minimum_days"sv, [](auto minimum_days) { return minimum_days; });
  1647. append_mapping(cldr.first_day_regions, cldr.first_day, "u8"sv, "s_first_day"sv, [](auto first_day) { return to_underlying(first_day); });
  1648. append_mapping(cldr.weekend_start_regions, cldr.weekend_start, "u8"sv, "s_weekend_start"sv, [](auto weekend_start) { return to_underlying(weekend_start); });
  1649. append_mapping(cldr.weekend_end_regions, cldr.weekend_end, "u8"sv, "s_weekend_end"sv, [](auto weekend_end) { return to_underlying(weekend_end); });
  1650. generator.append("\n");
  1651. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) -> ErrorOr<void> {
  1652. HashValueMap<DeprecatedString> hashes;
  1653. TRY(hashes.try_ensure_capacity(values.size()));
  1654. for (auto const& value : values)
  1655. hashes.set(value.hash(), format_identifier(enum_title, value));
  1656. for (auto const& alias : aliases)
  1657. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1658. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1659. return {};
  1660. };
  1661. TRY(append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, cldr.hour_cycle_regions));
  1662. TRY(append_from_string("MinimumDaysRegion"sv, "minimum_days_region"sv, cldr.minimum_days_regions));
  1663. TRY(append_from_string("FirstDayRegion"sv, "first_day_region"sv, cldr.first_day_regions));
  1664. TRY(append_from_string("WeekendStartRegion"sv, "weekend_start_region"sv, cldr.weekend_start_regions));
  1665. TRY(append_from_string("WeekendEndRegion"sv, "weekend_end_region"sv, cldr.weekend_end_regions));
  1666. generator.append(R"~~~(
  1667. static Optional<Calendar> keyword_to_calendar(KeywordCalendar keyword)
  1668. {
  1669. switch (keyword) {)~~~");
  1670. for (auto const& calendar : cldr.calendars) {
  1671. generator.set("name"sv, format_identifier({}, calendar));
  1672. generator.append(R"~~~(
  1673. case KeywordCalendar::@name@:
  1674. return Calendar::@name@;)~~~");
  1675. }
  1676. generator.append(R"~~~(
  1677. default:
  1678. return {};
  1679. }
  1680. }
  1681. ErrorOr<Vector<HourCycle>> get_regional_hour_cycles(StringView region)
  1682. {
  1683. auto region_value = hour_cycle_region_from_string(region);
  1684. if (!region_value.has_value())
  1685. return Vector<HourCycle> {};
  1686. auto region_index = to_underlying(*region_value);
  1687. auto regional_hour_cycles_index = s_hour_cycles.at(region_index);
  1688. auto const& regional_hour_cycles = s_hour_cycle_lists.at(regional_hour_cycles_index);
  1689. Vector<HourCycle> hour_cycles;
  1690. TRY(hour_cycles.try_ensure_capacity(regional_hour_cycles.size()));
  1691. for (auto hour_cycle : regional_hour_cycles)
  1692. hour_cycles.unchecked_append(static_cast<HourCycle>(hour_cycle));
  1693. return hour_cycles;
  1694. }
  1695. )~~~");
  1696. auto append_regional_lookup = [&](StringView return_type, StringView lookup_type) {
  1697. generator.set("return_type", return_type);
  1698. generator.set("lookup_type", lookup_type);
  1699. generator.append(R"~~~(
  1700. Optional<@return_type@> get_regional_@lookup_type@(StringView region)
  1701. {
  1702. auto region_value = @lookup_type@_region_from_string(region);
  1703. if (!region_value.has_value())
  1704. return {};
  1705. auto region_index = to_underlying(*region_value);
  1706. auto @lookup_type@ = s_@lookup_type@.at(region_index);
  1707. return static_cast<@return_type@>(@lookup_type@);
  1708. }
  1709. )~~~");
  1710. };
  1711. append_regional_lookup("u8"sv, "minimum_days"sv);
  1712. append_regional_lookup("Weekday"sv, "first_day"sv);
  1713. append_regional_lookup("Weekday"sv, "weekend_start"sv);
  1714. append_regional_lookup("Weekday"sv, "weekend_end"sv);
  1715. generator.append(R"~~~(
  1716. static ErrorOr<CalendarData const*> find_calendar_data(StringView locale, StringView calendar)
  1717. {
  1718. auto locale_value = locale_from_string(locale);
  1719. if (!locale_value.has_value())
  1720. return nullptr;
  1721. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1722. auto const& calendar_indices = s_locale_calendars.at(locale_index);
  1723. auto lookup_calendar = [&](auto calendar_name) -> CalendarData const* {
  1724. auto calendar_keyword = keyword_ca_from_string(calendar_name);
  1725. if (!calendar_keyword.has_value())
  1726. return nullptr;
  1727. auto calendar_value = keyword_to_calendar(*calendar_keyword);
  1728. if (!calendar_value.has_value())
  1729. return nullptr;
  1730. size_t calendar_index = to_underlying(*calendar_value);
  1731. calendar_index = calendar_indices[calendar_index];
  1732. return &s_calendars[calendar_index];
  1733. };
  1734. if (auto const* calendar_data = lookup_calendar(calendar))
  1735. return calendar_data;
  1736. auto default_calendar = TRY(get_preferred_keyword_value_for_locale(locale, "ca"sv));
  1737. if (!default_calendar.has_value())
  1738. return nullptr;
  1739. return lookup_calendar(*default_calendar);
  1740. }
  1741. ErrorOr<Optional<CalendarFormat>> get_calendar_date_format(StringView locale, StringView calendar)
  1742. {
  1743. if (auto const* data = TRY(find_calendar_data(locale, calendar)); data != nullptr) {
  1744. auto const& formats = s_calendar_formats.at(data->date_formats);
  1745. return TRY(formats.to_unicode_calendar_format());
  1746. }
  1747. return OptionalNone {};
  1748. }
  1749. ErrorOr<Optional<CalendarFormat>> get_calendar_time_format(StringView locale, StringView calendar)
  1750. {
  1751. if (auto const* data = TRY(find_calendar_data(locale, calendar)); data != nullptr) {
  1752. auto const& formats = s_calendar_formats.at(data->time_formats);
  1753. return TRY(formats.to_unicode_calendar_format());
  1754. }
  1755. return OptionalNone {};
  1756. }
  1757. ErrorOr<Optional<CalendarFormat>> get_calendar_date_time_format(StringView locale, StringView calendar)
  1758. {
  1759. if (auto const* data = TRY(find_calendar_data(locale, calendar)); data != nullptr) {
  1760. auto const& formats = s_calendar_formats.at(data->date_time_formats);
  1761. return TRY(formats.to_unicode_calendar_format());
  1762. }
  1763. return OptionalNone {};
  1764. }
  1765. ErrorOr<Vector<CalendarPattern>> get_calendar_available_formats(StringView locale, StringView calendar)
  1766. {
  1767. Vector<CalendarPattern> result {};
  1768. if (auto const* data = TRY(find_calendar_data(locale, calendar)); data != nullptr) {
  1769. auto const& available_formats = s_calendar_pattern_lists.at(data->available_formats);
  1770. TRY(result.try_ensure_capacity(available_formats.size()));
  1771. for (auto const& format : available_formats)
  1772. result.unchecked_append(TRY(s_calendar_patterns[format].to_unicode_calendar_pattern()));
  1773. }
  1774. return result;
  1775. }
  1776. ErrorOr<Optional<CalendarRangePattern>> get_calendar_default_range_format(StringView locale, StringView calendar)
  1777. {
  1778. if (auto const* data = TRY(find_calendar_data(locale, calendar)); data != nullptr) {
  1779. auto const& pattern = s_calendar_range_patterns[data->default_range_format];
  1780. return TRY(pattern.to_unicode_calendar_range_pattern());
  1781. }
  1782. return OptionalNone {};
  1783. }
  1784. ErrorOr<Vector<CalendarRangePattern>> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton)
  1785. {
  1786. Vector<CalendarRangePattern> result {};
  1787. if (auto const* data = TRY(find_calendar_data(locale, calendar)); data != nullptr) {
  1788. auto const& range_formats = s_calendar_range_pattern_lists.at(data->range_formats);
  1789. for (auto format : range_formats) {
  1790. auto const& pattern = s_calendar_range_patterns[format];
  1791. if (skeleton == decode_string(pattern.skeleton))
  1792. TRY(result.try_append(TRY(pattern.to_unicode_calendar_range_pattern())));
  1793. }
  1794. }
  1795. return result;
  1796. }
  1797. ErrorOr<Vector<CalendarRangePattern>> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton)
  1798. {
  1799. Vector<CalendarRangePattern> result {};
  1800. if (auto const* data = TRY(find_calendar_data(locale, calendar)); data != nullptr) {
  1801. auto const& range12_formats = s_calendar_range_pattern_lists.at(data->range12_formats);
  1802. for (auto format : range12_formats) {
  1803. auto const& pattern = s_calendar_range_patterns[format];
  1804. if (skeleton == decode_string(pattern.skeleton))
  1805. TRY(result.try_append(TRY(pattern.to_unicode_calendar_range_pattern())));
  1806. }
  1807. }
  1808. return result;
  1809. }
  1810. static ErrorOr<Span<@string_index_type@ const>> find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1811. {
  1812. if (auto const* data = TRY(find_calendar_data(locale, calendar)); data != nullptr) {
  1813. auto const& symbols_list = s_calendar_symbol_lists[data->symbols];
  1814. auto symbol_index = to_underlying(symbol);
  1815. auto calendar_symbols_index = symbols_list.at(symbol_index);
  1816. auto const& symbols = s_calendar_symbols.at(calendar_symbols_index);
  1817. @symbol_list_index_type@ symbol_list_index = 0;
  1818. switch (style) {
  1819. case CalendarPatternStyle::Narrow:
  1820. symbol_list_index = symbols.narrow_symbols;
  1821. break;
  1822. case CalendarPatternStyle::Short:
  1823. symbol_list_index = symbols.short_symbols;
  1824. break;
  1825. case CalendarPatternStyle::Long:
  1826. symbol_list_index = symbols.long_symbols;
  1827. break;
  1828. default:
  1829. VERIFY_NOT_REACHED();
  1830. }
  1831. return s_symbol_lists.at(symbol_list_index);
  1832. }
  1833. return Span<@string_index_type@ const> {};
  1834. }
  1835. ErrorOr<Optional<StringView>> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Era value)
  1836. {
  1837. auto symbols = TRY(find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style));
  1838. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1839. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1840. return Optional<StringView> { decode_string(symbol_index) };
  1841. }
  1842. return OptionalNone {};
  1843. }
  1844. ErrorOr<Optional<StringView>> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Month value)
  1845. {
  1846. auto symbols = TRY(find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style));
  1847. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1848. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1849. return Optional<StringView> { decode_string(symbol_index) };
  1850. }
  1851. return OptionalNone {};
  1852. }
  1853. ErrorOr<Optional<StringView>> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Weekday value)
  1854. {
  1855. auto symbols = TRY(find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style));
  1856. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1857. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1858. return Optional<StringView> { decode_string(symbol_index) };
  1859. }
  1860. return OptionalNone {};
  1861. }
  1862. ErrorOr<Optional<StringView>> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, DayPeriod value)
  1863. {
  1864. auto symbols = TRY(find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style));
  1865. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1866. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1867. return Optional<StringView> { decode_string(symbol_index) };
  1868. }
  1869. return OptionalNone {};
  1870. }
  1871. ErrorOr<Optional<StringView>> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour)
  1872. {
  1873. auto locale_value = locale_from_string(locale);
  1874. if (!locale_value.has_value())
  1875. return OptionalNone {};
  1876. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1877. auto day_periods_index = s_locale_day_periods[locale_index];
  1878. auto day_periods = s_day_period_lists[day_periods_index];
  1879. for (auto day_period_index : day_periods) {
  1880. auto day_period = s_day_periods[day_period_index];
  1881. bool hour_falls_within_day_period = false;
  1882. if (day_period.begin > day_period.end) {
  1883. if (hour >= day_period.begin)
  1884. hour_falls_within_day_period = true;
  1885. else if (hour <= day_period.end)
  1886. hour_falls_within_day_period = true;
  1887. } else if ((day_period.begin <= hour) && (hour < day_period.end)) {
  1888. hour_falls_within_day_period = true;
  1889. }
  1890. if (hour_falls_within_day_period) {
  1891. auto period = static_cast<DayPeriod>(day_period.day_period);
  1892. return get_calendar_day_period_symbol(locale, calendar, style, period);
  1893. }
  1894. }
  1895. // Fallback to fixed periods if the locale does not have flexible day periods.
  1896. // TR-35 states that the meaning of AM and PM does not change with locale.
  1897. if (hour < 12)
  1898. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::AM);
  1899. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::PM);
  1900. }
  1901. Optional<TimeZoneFormat> get_time_zone_format(StringView locale)
  1902. {
  1903. auto locale_value = locale_from_string(locale);
  1904. if (!locale_value.has_value())
  1905. return {};
  1906. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1907. auto time_zone_format_index = s_locale_time_zone_formats.at(locale_index);
  1908. auto const& time_zone_format = s_time_zone_formats.at(time_zone_format_index);
  1909. return time_zone_format.to_time_zone_format();
  1910. }
  1911. static TimeZoneNames const* find_time_zone_names(StringView locale, StringView time_zone)
  1912. {
  1913. auto locale_value = locale_from_string(locale);
  1914. if (!locale_value.has_value())
  1915. return nullptr;
  1916. auto time_zone_value = ::TimeZone::time_zone_from_string(time_zone);
  1917. if (!time_zone_value.has_value())
  1918. return nullptr;
  1919. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1920. size_t time_zone_index = to_underlying(*time_zone_value);
  1921. auto time_zone_list_index = s_locale_time_zones.at(locale_index);
  1922. auto const& time_zone_list = s_time_zone_lists.at(time_zone_list_index);
  1923. if (time_zone_list.size() <= time_zone_index)
  1924. return nullptr;
  1925. time_zone_index = time_zone_list.at(time_zone_index);
  1926. return &s_time_zones[time_zone_index];
  1927. }
  1928. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style, TimeZone::InDST in_dst)
  1929. {
  1930. if (auto const* data = find_time_zone_names(locale, time_zone); data != nullptr) {
  1931. size_t name_index = 0;
  1932. switch (style) {
  1933. case CalendarPatternStyle::Short:
  1934. name_index = (in_dst == TimeZone::InDST::No) ? data->short_standard_name : data->short_daylight_name;
  1935. break;
  1936. case CalendarPatternStyle::Long:
  1937. name_index = (in_dst == TimeZone::InDST::No) ? data->long_standard_name : data->long_daylight_name;
  1938. break;
  1939. case CalendarPatternStyle::ShortGeneric:
  1940. name_index = data->short_generic_name;
  1941. break;
  1942. case CalendarPatternStyle::LongGeneric:
  1943. name_index = data->long_generic_name;
  1944. break;
  1945. default:
  1946. VERIFY_NOT_REACHED();
  1947. }
  1948. if (name_index != 0)
  1949. return decode_string(name_index);
  1950. }
  1951. return {};
  1952. }
  1953. }
  1954. )~~~");
  1955. TRY(file.write(generator.as_string_view().bytes()));
  1956. return {};
  1957. }
  1958. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1959. {
  1960. StringView generated_header_path;
  1961. StringView generated_implementation_path;
  1962. StringView core_path;
  1963. StringView dates_path;
  1964. Core::ArgsParser args_parser;
  1965. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1966. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1967. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1968. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1969. args_parser.parse(arguments);
  1970. auto generated_header_file = TRY(open_file(generated_header_path, Core::Stream::OpenMode::Write));
  1971. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::Stream::OpenMode::Write));
  1972. CLDR cldr;
  1973. TRY(parse_all_locales(core_path, dates_path, cldr));
  1974. TRY(generate_unicode_locale_header(*generated_header_file, cldr));
  1975. TRY(generate_unicode_locale_implementation(*generated_implementation_file, cldr));
  1976. return 0;
  1977. }