GenerateDateTimeFormatData.cpp 96 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373237423752376237723782379238023812382238323842385238623872388238923902391239223932394239523962397239823992400240124022403240424052406240724082409241024112412241324142415241624172418241924202421242224232424242524262427
  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/ByteString.h>
  9. #include <AK/CharacterTypes.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/Directory.h>
  26. #include <LibLocale/DateTimeFormat.h>
  27. #include <LibTimeZone/TimeZone.h>
  28. struct CalendarPattern : public Locale::CalendarPattern {
  29. bool contains_only_date_fields() const
  30. {
  31. return !day_period.has_value() && !hour.has_value() && !minute.has_value() && !second.has_value() && !fractional_second_digits.has_value() && !time_zone_name.has_value();
  32. }
  33. bool contains_only_time_fields() const
  34. {
  35. return !weekday.has_value() && !era.has_value() && !year.has_value() && !month.has_value() && !day.has_value();
  36. }
  37. unsigned hash() const
  38. {
  39. auto hash = pair_int_hash(pattern_index, pattern12_index);
  40. hash = pair_int_hash(hash, skeleton_index);
  41. auto hash_field = [&](auto const& field) {
  42. if (field.has_value())
  43. hash = pair_int_hash(hash, static_cast<u8>(*field));
  44. else
  45. hash = pair_int_hash(hash, -1);
  46. };
  47. hash_field(era);
  48. hash_field(year);
  49. hash_field(month);
  50. hash_field(weekday);
  51. hash_field(day);
  52. hash_field(day_period);
  53. hash_field(hour);
  54. hash_field(minute);
  55. hash_field(second);
  56. hash_field(fractional_second_digits);
  57. hash_field(time_zone_name);
  58. return hash;
  59. }
  60. bool operator==(CalendarPattern const& other) const
  61. {
  62. return (skeleton_index == other.skeleton_index)
  63. && (pattern_index == other.pattern_index)
  64. && (pattern12_index == other.pattern12_index)
  65. && (era == other.era)
  66. && (year == other.year)
  67. && (month == other.month)
  68. && (weekday == other.weekday)
  69. && (day == other.day)
  70. && (day_period == other.day_period)
  71. && (hour == other.hour)
  72. && (minute == other.minute)
  73. && (second == other.second)
  74. && (fractional_second_digits == other.fractional_second_digits)
  75. && (time_zone_name == other.time_zone_name);
  76. }
  77. size_t skeleton_index { 0 };
  78. size_t pattern_index { 0 };
  79. size_t pattern12_index { 0 };
  80. };
  81. template<>
  82. struct AK::Formatter<CalendarPattern> : Formatter<FormatString> {
  83. ErrorOr<void> format(FormatBuilder& builder, CalendarPattern const& pattern)
  84. {
  85. auto field_to_i8 = [](auto const& field) -> i8 {
  86. if (!field.has_value())
  87. return -1;
  88. return static_cast<i8>(*field);
  89. };
  90. return Formatter<FormatString>::format(builder,
  91. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  92. pattern.skeleton_index,
  93. pattern.pattern_index,
  94. pattern.pattern12_index,
  95. field_to_i8(pattern.era),
  96. field_to_i8(pattern.year),
  97. field_to_i8(pattern.month),
  98. field_to_i8(pattern.weekday),
  99. field_to_i8(pattern.day),
  100. field_to_i8(pattern.day_period),
  101. field_to_i8(pattern.hour),
  102. field_to_i8(pattern.minute),
  103. field_to_i8(pattern.second),
  104. field_to_i8(pattern.fractional_second_digits),
  105. field_to_i8(pattern.time_zone_name));
  106. }
  107. };
  108. template<>
  109. struct AK::Traits<CalendarPattern> : public DefaultTraits<CalendarPattern> {
  110. static unsigned hash(CalendarPattern const& c) { return c.hash(); }
  111. };
  112. struct CalendarRangePattern : public CalendarPattern {
  113. unsigned hash() const
  114. {
  115. auto hash = CalendarPattern::hash();
  116. if (field.has_value())
  117. hash = pair_int_hash(hash, static_cast<u8>(*field));
  118. hash = pair_int_hash(hash, start_range);
  119. hash = pair_int_hash(hash, separator);
  120. hash = pair_int_hash(hash, end_range);
  121. return hash;
  122. }
  123. bool operator==(CalendarRangePattern const& other) const
  124. {
  125. if (!CalendarPattern::operator==(other))
  126. return false;
  127. return (field == other.field)
  128. && (start_range == other.start_range)
  129. && (separator == other.separator)
  130. && (end_range == other.end_range);
  131. }
  132. Optional<Locale::CalendarRangePattern::Field> field {};
  133. size_t start_range { 0 };
  134. size_t separator { 0 };
  135. size_t end_range { 0 };
  136. };
  137. template<>
  138. struct AK::Formatter<CalendarRangePattern> : Formatter<FormatString> {
  139. ErrorOr<void> format(FormatBuilder& builder, CalendarRangePattern const& pattern)
  140. {
  141. auto field_to_i8 = [](auto const& field) -> i8 {
  142. if (!field.has_value())
  143. return -1;
  144. return static_cast<i8>(*field);
  145. };
  146. return Formatter<FormatString>::format(builder,
  147. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  148. pattern.skeleton_index,
  149. field_to_i8(pattern.field),
  150. pattern.start_range,
  151. pattern.separator,
  152. pattern.end_range,
  153. field_to_i8(pattern.era),
  154. field_to_i8(pattern.year),
  155. field_to_i8(pattern.month),
  156. field_to_i8(pattern.weekday),
  157. field_to_i8(pattern.day),
  158. field_to_i8(pattern.day_period),
  159. field_to_i8(pattern.hour),
  160. field_to_i8(pattern.minute),
  161. field_to_i8(pattern.second),
  162. field_to_i8(pattern.fractional_second_digits),
  163. field_to_i8(pattern.time_zone_name));
  164. }
  165. };
  166. template<>
  167. struct AK::Traits<CalendarRangePattern> : public DefaultTraits<CalendarRangePattern> {
  168. static unsigned hash(CalendarRangePattern const& c) { return c.hash(); }
  169. };
  170. struct CalendarFormat {
  171. unsigned hash() const
  172. {
  173. auto hash = pair_int_hash(full_format, long_format);
  174. hash = pair_int_hash(hash, medium_format);
  175. hash = pair_int_hash(hash, short_format);
  176. return hash;
  177. }
  178. bool operator==(CalendarFormat const& other) const
  179. {
  180. return (full_format == other.full_format)
  181. && (long_format == other.long_format)
  182. && (medium_format == other.medium_format)
  183. && (short_format == other.short_format);
  184. }
  185. size_t full_format { 0 };
  186. size_t long_format { 0 };
  187. size_t medium_format { 0 };
  188. size_t short_format { 0 };
  189. };
  190. template<>
  191. struct AK::Formatter<CalendarFormat> : Formatter<FormatString> {
  192. ErrorOr<void> format(FormatBuilder& builder, CalendarFormat const& pattern)
  193. {
  194. return Formatter<FormatString>::format(builder,
  195. "{{ {}, {}, {}, {} }}"sv,
  196. pattern.full_format,
  197. pattern.long_format,
  198. pattern.medium_format,
  199. pattern.short_format);
  200. }
  201. };
  202. template<>
  203. struct AK::Traits<CalendarFormat> : public DefaultTraits<CalendarFormat> {
  204. static unsigned hash(CalendarFormat const& c) { return c.hash(); }
  205. };
  206. using SymbolList = Vector<size_t>;
  207. struct CalendarSymbols {
  208. unsigned hash() const
  209. {
  210. auto hash = pair_int_hash(narrow_symbols, short_symbols);
  211. hash = pair_int_hash(hash, long_symbols);
  212. return hash;
  213. }
  214. bool operator==(CalendarSymbols const& other) const
  215. {
  216. return (narrow_symbols == other.narrow_symbols)
  217. && (short_symbols == other.short_symbols)
  218. && (long_symbols == other.long_symbols);
  219. }
  220. size_t narrow_symbols { 0 };
  221. size_t short_symbols { 0 };
  222. size_t long_symbols { 0 };
  223. };
  224. template<>
  225. struct AK::Formatter<CalendarSymbols> : Formatter<FormatString> {
  226. ErrorOr<void> format(FormatBuilder& builder, CalendarSymbols const& symbols)
  227. {
  228. return Formatter<FormatString>::format(builder,
  229. "{{ {}, {}, {} }}"sv,
  230. symbols.narrow_symbols,
  231. symbols.short_symbols,
  232. symbols.long_symbols);
  233. }
  234. };
  235. template<>
  236. struct AK::Traits<CalendarSymbols> : public DefaultTraits<CalendarSymbols> {
  237. static unsigned hash(CalendarSymbols const& c) { return c.hash(); }
  238. };
  239. using CalendarPatternList = Vector<size_t>;
  240. using CalendarRangePatternList = Vector<size_t>;
  241. using CalendarSymbolsList = Vector<size_t>;
  242. struct Calendar {
  243. unsigned hash() const
  244. {
  245. auto hash = int_hash(date_formats);
  246. hash = pair_int_hash(hash, time_formats);
  247. hash = pair_int_hash(hash, date_time_formats);
  248. hash = pair_int_hash(hash, available_formats);
  249. hash = pair_int_hash(hash, default_range_format);
  250. hash = pair_int_hash(hash, range_formats);
  251. hash = pair_int_hash(hash, range12_formats);
  252. hash = pair_int_hash(hash, symbols);
  253. return hash;
  254. }
  255. bool operator==(Calendar const& other) const
  256. {
  257. return (date_formats == other.date_formats)
  258. && (time_formats == other.time_formats)
  259. && (date_time_formats == other.date_time_formats)
  260. && (available_formats == other.available_formats)
  261. && (default_range_format == other.default_range_format)
  262. && (range_formats == other.range_formats)
  263. && (range12_formats == other.range12_formats)
  264. && (symbols == other.symbols);
  265. }
  266. size_t date_formats { 0 };
  267. size_t time_formats { 0 };
  268. size_t date_time_formats { 0 };
  269. size_t available_formats { 0 };
  270. size_t default_range_format { 0 };
  271. size_t range_formats { 0 };
  272. size_t range12_formats { 0 };
  273. size_t symbols { 0 };
  274. };
  275. template<>
  276. struct AK::Formatter<Calendar> : Formatter<FormatString> {
  277. ErrorOr<void> format(FormatBuilder& builder, Calendar const& calendar)
  278. {
  279. return Formatter<FormatString>::format(builder,
  280. "{{ {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  281. calendar.date_formats,
  282. calendar.time_formats,
  283. calendar.date_time_formats,
  284. calendar.available_formats,
  285. calendar.default_range_format,
  286. calendar.range_formats,
  287. calendar.range12_formats,
  288. calendar.symbols);
  289. }
  290. };
  291. template<>
  292. struct AK::Traits<Calendar> : public DefaultTraits<Calendar> {
  293. static unsigned hash(Calendar const& c) { return c.hash(); }
  294. };
  295. struct TimeZoneNames {
  296. unsigned hash() const
  297. {
  298. auto hash = pair_int_hash(short_standard_name, long_standard_name);
  299. hash = pair_int_hash(hash, short_daylight_name);
  300. hash = pair_int_hash(hash, long_daylight_name);
  301. hash = pair_int_hash(hash, short_generic_name);
  302. hash = pair_int_hash(hash, long_generic_name);
  303. return hash;
  304. }
  305. bool operator==(TimeZoneNames const& other) const
  306. {
  307. return (short_standard_name == other.short_standard_name)
  308. && (long_standard_name == other.long_standard_name)
  309. && (short_daylight_name == other.short_daylight_name)
  310. && (long_daylight_name == other.long_daylight_name)
  311. && (short_generic_name == other.short_generic_name)
  312. && (long_generic_name == other.long_generic_name);
  313. }
  314. size_t short_standard_name { 0 };
  315. size_t long_standard_name { 0 };
  316. size_t short_daylight_name { 0 };
  317. size_t long_daylight_name { 0 };
  318. size_t short_generic_name { 0 };
  319. size_t long_generic_name { 0 };
  320. };
  321. template<>
  322. struct AK::Formatter<TimeZoneNames> : Formatter<FormatString> {
  323. ErrorOr<void> format(FormatBuilder& builder, TimeZoneNames const& time_zone)
  324. {
  325. return Formatter<FormatString>::format(builder,
  326. "{{ {}, {}, {}, {}, {}, {} }}"sv,
  327. time_zone.short_standard_name,
  328. time_zone.long_standard_name,
  329. time_zone.short_daylight_name,
  330. time_zone.long_daylight_name,
  331. time_zone.short_generic_name,
  332. time_zone.long_generic_name);
  333. }
  334. };
  335. template<>
  336. struct AK::Traits<TimeZoneNames> : public DefaultTraits<TimeZoneNames> {
  337. static unsigned hash(TimeZoneNames const& t) { return t.hash(); }
  338. };
  339. struct TimeZoneFormat {
  340. unsigned hash() const
  341. {
  342. auto hash = int_hash(symbol_ahead_sign);
  343. hash = pair_int_hash(hash, symbol_ahead_separator);
  344. hash = pair_int_hash(hash, symbol_behind_sign);
  345. hash = pair_int_hash(hash, symbol_behind_separator);
  346. hash = pair_int_hash(hash, gmt_format);
  347. hash = pair_int_hash(hash, gmt_zero_format);
  348. return hash;
  349. }
  350. bool operator==(TimeZoneFormat const& other) const
  351. {
  352. return (symbol_ahead_sign == other.symbol_ahead_sign)
  353. && (symbol_ahead_separator == other.symbol_ahead_separator)
  354. && (symbol_behind_sign == other.symbol_behind_sign)
  355. && (symbol_behind_separator == other.symbol_behind_separator)
  356. && (gmt_format == other.gmt_format)
  357. && (gmt_zero_format == other.gmt_zero_format);
  358. }
  359. size_t symbol_ahead_sign { 0 };
  360. size_t symbol_ahead_separator { 0 };
  361. size_t symbol_behind_sign { 0 };
  362. size_t symbol_behind_separator { 0 };
  363. size_t gmt_format { 0 };
  364. size_t gmt_zero_format { 0 };
  365. };
  366. template<>
  367. struct AK::Formatter<TimeZoneFormat> : Formatter<FormatString> {
  368. ErrorOr<void> format(FormatBuilder& builder, TimeZoneFormat const& time_zone_format)
  369. {
  370. return Formatter<FormatString>::format(builder, "{{ {}, {}, {}, {}, {}, {} }}"sv,
  371. time_zone_format.symbol_ahead_sign,
  372. time_zone_format.symbol_ahead_separator,
  373. time_zone_format.symbol_behind_sign,
  374. time_zone_format.symbol_behind_separator,
  375. time_zone_format.gmt_format,
  376. time_zone_format.gmt_zero_format);
  377. }
  378. };
  379. template<>
  380. struct AK::Traits<TimeZoneFormat> : public DefaultTraits<TimeZoneFormat> {
  381. static unsigned hash(TimeZoneFormat const& t) { return t.hash(); }
  382. };
  383. struct DayPeriod {
  384. unsigned hash() const
  385. {
  386. auto hash = int_hash(static_cast<u8>(day_period));
  387. hash = pair_int_hash(hash, begin);
  388. hash = pair_int_hash(hash, end);
  389. return hash;
  390. }
  391. bool operator==(DayPeriod const& other) const
  392. {
  393. return (day_period == other.day_period)
  394. && (begin == other.begin)
  395. && (end == other.end);
  396. }
  397. Locale::DayPeriod day_period {};
  398. u8 begin { 0 };
  399. u8 end { 0 };
  400. };
  401. template<>
  402. struct AK::Formatter<DayPeriod> : Formatter<FormatString> {
  403. ErrorOr<void> format(FormatBuilder& builder, DayPeriod const& day_period)
  404. {
  405. return Formatter<FormatString>::format(builder,
  406. "{{ {}, {}, {} }}"sv,
  407. static_cast<u8>(day_period.day_period),
  408. day_period.begin,
  409. day_period.end);
  410. }
  411. };
  412. template<>
  413. struct AK::Traits<DayPeriod> : public DefaultTraits<DayPeriod> {
  414. static unsigned hash(DayPeriod const& d) { return d.hash(); }
  415. };
  416. using TimeZoneNamesList = Vector<size_t>;
  417. using DayPeriodList = Vector<size_t>;
  418. using HourCycleList = Vector<Locale::HourCycle>;
  419. template<>
  420. struct AK::Formatter<Locale::HourCycle> : Formatter<FormatString> {
  421. ErrorOr<void> format(FormatBuilder& builder, Locale::HourCycle hour_cycle)
  422. {
  423. return builder.put_u64(to_underlying(hour_cycle));
  424. }
  425. };
  426. struct LocaleData {
  427. HashMap<ByteString, size_t> calendars;
  428. size_t time_zones { 0 };
  429. size_t time_zone_formats { 0 };
  430. size_t day_periods { 0 };
  431. };
  432. struct CLDR {
  433. UniqueStringStorage unique_strings;
  434. UniqueStorage<CalendarPattern> unique_patterns;
  435. UniqueStorage<CalendarPatternList> unique_pattern_lists;
  436. UniqueStorage<CalendarRangePattern> unique_range_patterns;
  437. UniqueStorage<CalendarRangePatternList> unique_range_pattern_lists;
  438. UniqueStorage<CalendarFormat> unique_formats;
  439. UniqueStorage<SymbolList> unique_symbol_lists;
  440. UniqueStorage<CalendarSymbols> unique_calendar_symbols;
  441. UniqueStorage<CalendarSymbolsList> unique_calendar_symbols_lists;
  442. UniqueStorage<Calendar> unique_calendars;
  443. UniqueStorage<TimeZoneNames> unique_time_zones;
  444. UniqueStorage<TimeZoneNamesList> unique_time_zone_lists;
  445. UniqueStorage<TimeZoneFormat> unique_time_zone_formats;
  446. UniqueStorage<DayPeriod> unique_day_periods;
  447. UniqueStorage<DayPeriodList> unique_day_period_lists;
  448. UniqueStorage<HourCycleList> unique_hour_cycle_lists;
  449. HashMap<ByteString, LocaleData> locales;
  450. HashMap<ByteString, size_t> hour_cycles;
  451. Vector<ByteString> hour_cycle_regions;
  452. HashMap<ByteString, u8> minimum_days;
  453. Vector<ByteString> minimum_days_regions;
  454. HashMap<ByteString, Locale::Weekday> first_day;
  455. Vector<ByteString> first_day_regions;
  456. HashMap<ByteString, Locale::Weekday> weekend_start;
  457. Vector<ByteString> weekend_start_regions;
  458. HashMap<ByteString, Locale::Weekday> weekend_end;
  459. Vector<ByteString> weekend_end_regions;
  460. HashMap<ByteString, Vector<TimeZone::TimeZone>> meta_zones;
  461. Vector<ByteString> time_zones { "UTC"sv };
  462. Vector<ByteString> calendars;
  463. };
  464. static Optional<Locale::DayPeriod> day_period_from_string(StringView day_period)
  465. {
  466. if (day_period == "am"sv)
  467. return Locale::DayPeriod::AM;
  468. if (day_period == "pm"sv)
  469. return Locale::DayPeriod::PM;
  470. if (day_period == "noon"sv)
  471. return Locale::DayPeriod::Noon;
  472. if (day_period == "morning1"sv)
  473. return Locale::DayPeriod::Morning1;
  474. if (day_period == "morning2"sv)
  475. return Locale::DayPeriod::Morning2;
  476. if (day_period == "afternoon1"sv)
  477. return Locale::DayPeriod::Afternoon1;
  478. if (day_period == "afternoon2"sv)
  479. return Locale::DayPeriod::Afternoon2;
  480. if (day_period == "evening1"sv)
  481. return Locale::DayPeriod::Evening1;
  482. if (day_period == "evening2"sv)
  483. return Locale::DayPeriod::Evening2;
  484. if (day_period == "night1"sv)
  485. return Locale::DayPeriod::Night1;
  486. if (day_period == "night2"sv)
  487. return Locale::DayPeriod::Night2;
  488. return {};
  489. }
  490. static ErrorOr<void> parse_hour_cycles(ByteString core_path, CLDR& cldr)
  491. {
  492. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  493. LexicalPath time_data_path(move(core_path));
  494. time_data_path = time_data_path.append("supplemental"sv);
  495. time_data_path = time_data_path.append("timeData.json"sv);
  496. auto time_data = TRY(read_json_file(time_data_path.string()));
  497. auto const& supplemental_object = time_data.as_object().get_object("supplemental"sv).value();
  498. auto const& time_data_object = supplemental_object.get_object("timeData"sv).value();
  499. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Locale::HourCycle> {
  500. if (hour_cycle.is_one_of("h"sv, "hb"sv, "hB"sv))
  501. return Locale::HourCycle::H12;
  502. if (hour_cycle.is_one_of("H"sv, "Hb"sv, "HB"sv))
  503. return Locale::HourCycle::H23;
  504. if (hour_cycle == "K"sv)
  505. return Locale::HourCycle::H11;
  506. if (hour_cycle == "k"sv)
  507. return Locale::HourCycle::H24;
  508. return {};
  509. };
  510. time_data_object.for_each_member([&](auto const& key, JsonValue const& value) {
  511. auto allowed_hour_cycles_string = value.as_object().get_byte_string("_allowed"sv).value();
  512. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  513. Vector<Locale::HourCycle> hour_cycles;
  514. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  515. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  516. hour_cycles.append(*hour_cycle);
  517. }
  518. auto hour_cycles_index = cldr.unique_hour_cycle_lists.ensure(move(hour_cycles));
  519. cldr.hour_cycles.set(key, hour_cycles_index);
  520. if (!cldr.hour_cycle_regions.contains_slow(key))
  521. cldr.hour_cycle_regions.append(key);
  522. });
  523. return {};
  524. }
  525. static ErrorOr<void> parse_week_data(ByteString core_path, CLDR& cldr)
  526. {
  527. // https://unicode.org/reports/tr35/tr35-dates.html#Week_Data
  528. LexicalPath week_data_path(move(core_path));
  529. week_data_path = week_data_path.append("supplemental"sv);
  530. week_data_path = week_data_path.append("weekData.json"sv);
  531. auto week_data = TRY(read_json_file(week_data_path.string()));
  532. auto const& supplemental_object = week_data.as_object().get_object("supplemental"sv).value();
  533. auto const& week_data_object = supplemental_object.get_object("weekData"sv).value();
  534. auto parse_weekday = [](StringView day) -> Locale::Weekday {
  535. if (day == "sun"sv)
  536. return Locale::Weekday::Sunday;
  537. if (day == "mon"sv)
  538. return Locale::Weekday::Monday;
  539. if (day == "tue"sv)
  540. return Locale::Weekday::Tuesday;
  541. if (day == "wed"sv)
  542. return Locale::Weekday::Wednesday;
  543. if (day == "thu"sv)
  544. return Locale::Weekday::Thursday;
  545. if (day == "fri"sv)
  546. return Locale::Weekday::Friday;
  547. if (day == "sat"sv)
  548. return Locale::Weekday::Saturday;
  549. VERIFY_NOT_REACHED();
  550. };
  551. auto parse_regional_weekdays = [&](auto const& region, auto const& weekday, auto& weekdays_map, auto& weekday_regions) {
  552. if (region.ends_with("alt-variant"sv))
  553. return;
  554. weekdays_map.set(region, parse_weekday(weekday));
  555. if (!weekday_regions.contains_slow(region))
  556. weekday_regions.append(region);
  557. };
  558. auto const& minimum_days_object = week_data_object.get_object("minDays"sv).value();
  559. auto const& first_day_object = week_data_object.get_object("firstDay"sv).value();
  560. auto const& weekend_start_object = week_data_object.get_object("weekendStart"sv).value();
  561. auto const& weekend_end_object = week_data_object.get_object("weekendEnd"sv).value();
  562. minimum_days_object.for_each_member([&](auto const& region, auto const& value) {
  563. auto minimum_days = value.as_string().template to_number<u8>();
  564. cldr.minimum_days.set(region, *minimum_days);
  565. if (!cldr.minimum_days_regions.contains_slow(region))
  566. cldr.minimum_days_regions.append(region);
  567. });
  568. first_day_object.for_each_member([&](auto const& region, auto const& value) {
  569. parse_regional_weekdays(region, value.as_string(), cldr.first_day, cldr.first_day_regions);
  570. });
  571. weekend_start_object.for_each_member([&](auto const& region, auto const& value) {
  572. parse_regional_weekdays(region, value.as_string(), cldr.weekend_start, cldr.weekend_start_regions);
  573. });
  574. weekend_end_object.for_each_member([&](auto const& region, auto const& value) {
  575. parse_regional_weekdays(region, value.as_string(), cldr.weekend_end, cldr.weekend_end_regions);
  576. });
  577. return {};
  578. }
  579. static ErrorOr<void> parse_meta_zones(ByteString core_path, CLDR& cldr)
  580. {
  581. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  582. LexicalPath meta_zone_path(move(core_path));
  583. meta_zone_path = meta_zone_path.append("supplemental"sv);
  584. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  585. auto meta_zone = TRY(read_json_file(meta_zone_path.string()));
  586. auto const& supplemental_object = meta_zone.as_object().get_object("supplemental"sv).value();
  587. auto const& meta_zone_object = supplemental_object.get_object("metaZones"sv).value();
  588. auto const& meta_zone_array = meta_zone_object.get_array("metazones"sv).value();
  589. meta_zone_array.for_each([&](JsonValue const& value) {
  590. auto const& mapping = value.as_object().get_object("mapZone"sv).value();
  591. auto const& meta_zone = mapping.get_byte_string("_other"sv).value();
  592. auto const& golden_zone = mapping.get_byte_string("_type"sv).value();
  593. if (auto time_zone = TimeZone::time_zone_from_string(golden_zone); time_zone.has_value()) {
  594. auto& golden_zones = cldr.meta_zones.ensure(meta_zone);
  595. golden_zones.append(*time_zone);
  596. }
  597. });
  598. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  599. if (auto time_zone = TimeZone::time_zone_from_string("UTC"sv); time_zone.has_value())
  600. cldr.meta_zones.set("UTC"sv, { *time_zone });
  601. return {};
  602. }
  603. static constexpr auto is_char(char ch)
  604. {
  605. return [ch](auto c) { return c == ch; };
  606. }
  607. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  608. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  609. // extra spaces around the day period. Some example expected removals:
  610. //
  611. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  612. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  613. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  614. static ErrorOr<String> remove_period_from_pattern(String pattern)
  615. {
  616. auto is_surrounding_space = [&](auto code_point_iterator) {
  617. if (code_point_iterator.done())
  618. return false;
  619. constexpr auto spaces = Array { static_cast<u32>(0x0020), 0x00a0, 0x2009, 0x202f };
  620. return spaces.span().contains_slow(*code_point_iterator);
  621. };
  622. auto is_opening = [&](auto code_point_iterator) {
  623. if (code_point_iterator.done())
  624. return false;
  625. return *code_point_iterator == '{';
  626. };
  627. auto is_closing = [&](auto code_point_iterator) {
  628. if (code_point_iterator.done())
  629. return false;
  630. return *code_point_iterator == '}';
  631. };
  632. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  633. auto index = pattern.find_byte_offset(remove);
  634. if (!index.has_value())
  635. continue;
  636. Utf8View utf8_pattern { pattern };
  637. Utf8CodePointIterator before_removal;
  638. Utf8CodePointIterator after_removal;
  639. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  640. before_removal = it;
  641. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  642. after_removal = it;
  643. auto pattern_view = pattern.bytes_as_string_view();
  644. if (is_surrounding_space(before_removal) && !is_opening(after_removal)) {
  645. pattern = TRY(String::formatted("{}{}",
  646. pattern_view.substring_view(0, *index - before_removal.underlying_code_point_length_in_bytes()),
  647. pattern_view.substring_view(*index + remove.length())));
  648. } else if (is_surrounding_space(after_removal) && !is_closing(before_removal)) {
  649. pattern = TRY(String::formatted("{}{}",
  650. pattern_view.substring_view(0, *index),
  651. pattern_view.substring_view(*index + remove.length() + after_removal.underlying_code_point_length_in_bytes())));
  652. } else {
  653. pattern = TRY(String::formatted("{}{}",
  654. pattern_view.substring_view(0, *index),
  655. pattern_view.substring_view(*index + remove.length())));
  656. }
  657. }
  658. return pattern;
  659. }
  660. static ErrorOr<Optional<CalendarPattern>> parse_date_time_pattern_raw(ByteString pattern, ByteString skeleton, CLDR& cldr)
  661. {
  662. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  663. using Locale::CalendarPatternStyle;
  664. CalendarPattern format {};
  665. if (!skeleton.is_empty())
  666. format.skeleton_index = cldr.unique_strings.ensure(move(skeleton));
  667. GenericLexer lexer { pattern };
  668. StringBuilder builder;
  669. bool hour12 { false };
  670. while (!lexer.is_eof()) {
  671. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  672. // processing (this just avoids conflicts with the patterns below).
  673. if (lexer.next_is(is_quote)) {
  674. builder.append(lexer.consume_quoted_string());
  675. continue;
  676. }
  677. auto starting_char = lexer.peek();
  678. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  679. // Era
  680. if (all_of(segment, is_char('G'))) {
  681. builder.append("{era}"sv);
  682. if (segment.length() <= 3)
  683. format.era = CalendarPatternStyle::Short;
  684. else if (segment.length() == 4)
  685. format.era = CalendarPatternStyle::Long;
  686. else
  687. format.era = CalendarPatternStyle::Narrow;
  688. }
  689. // Year
  690. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  691. builder.append("{year}"sv);
  692. if (segment.length() == 2)
  693. format.year = CalendarPatternStyle::TwoDigit;
  694. else
  695. format.year = CalendarPatternStyle::Numeric;
  696. }
  697. // Quarter
  698. else if (all_of(segment, is_any_of("qQ"sv))) {
  699. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  700. return OptionalNone {};
  701. }
  702. // Month
  703. else if (all_of(segment, is_any_of("ML"sv))) {
  704. builder.append("{month}"sv);
  705. if (segment.length() == 1)
  706. format.month = CalendarPatternStyle::Numeric;
  707. else if (segment.length() == 2)
  708. format.month = CalendarPatternStyle::TwoDigit;
  709. else if (segment.length() == 3)
  710. format.month = CalendarPatternStyle::Short;
  711. else if (segment.length() == 4)
  712. format.month = CalendarPatternStyle::Long;
  713. else if (segment.length() == 5)
  714. format.month = CalendarPatternStyle::Narrow;
  715. } else if (all_of(segment, is_char('l'))) {
  716. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  717. return OptionalNone {};
  718. }
  719. // Week
  720. else if (all_of(segment, is_any_of("wW"sv))) {
  721. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  722. return OptionalNone {};
  723. }
  724. // Day
  725. else if (all_of(segment, is_char('d'))) {
  726. builder.append("{day}"sv);
  727. if (segment.length() == 1)
  728. format.day = CalendarPatternStyle::Numeric;
  729. else
  730. format.day = CalendarPatternStyle::TwoDigit;
  731. } else if (all_of(segment, is_any_of("DFg"sv))) {
  732. builder.append("{day}"sv);
  733. format.day = CalendarPatternStyle::Numeric;
  734. }
  735. // Weekday
  736. else if (all_of(segment, is_char('E'))) {
  737. builder.append("{weekday}"sv);
  738. if (segment.length() == 4)
  739. format.weekday = CalendarPatternStyle::Long;
  740. else if (segment.length() == 5)
  741. format.weekday = CalendarPatternStyle::Narrow;
  742. else
  743. format.weekday = CalendarPatternStyle::Short;
  744. } else if (all_of(segment, is_any_of("ec"sv))) {
  745. builder.append("{weekday}"sv);
  746. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  747. // pattern styles are not supported by Intl.DateTimeFormat.
  748. if (segment.length() <= 2)
  749. return OptionalNone {};
  750. if (segment.length() == 4)
  751. format.weekday = CalendarPatternStyle::Long;
  752. else if (segment.length() == 5)
  753. format.weekday = CalendarPatternStyle::Narrow;
  754. else
  755. format.weekday = CalendarPatternStyle::Short;
  756. }
  757. // Period
  758. else if (all_of(segment, is_any_of("ab"sv))) {
  759. builder.append("{ampm}"sv);
  760. hour12 = true;
  761. } else if (all_of(segment, is_char('B'))) {
  762. builder.append("{dayPeriod}"sv);
  763. hour12 = true;
  764. if (segment.length() == 4)
  765. format.day_period = CalendarPatternStyle::Long;
  766. else if (segment.length() == 5)
  767. format.day_period = CalendarPatternStyle::Narrow;
  768. else
  769. format.day_period = CalendarPatternStyle::Short;
  770. }
  771. // Hour
  772. else if (all_of(segment, is_any_of("hHKk"sv))) {
  773. builder.append("{hour}"sv);
  774. if ((segment[0] == 'h') || (segment[0] == 'K'))
  775. hour12 = true;
  776. if (segment.length() == 1)
  777. format.hour = CalendarPatternStyle::Numeric;
  778. else
  779. format.hour = CalendarPatternStyle::TwoDigit;
  780. } else if (all_of(segment, is_any_of("jJC"sv))) {
  781. // TR-35 indicates these should not be used.
  782. return OptionalNone {};
  783. }
  784. // Minute
  785. else if (all_of(segment, is_char('m'))) {
  786. builder.append("{minute}"sv);
  787. if (segment.length() == 1)
  788. format.minute = CalendarPatternStyle::Numeric;
  789. else
  790. format.minute = CalendarPatternStyle::TwoDigit;
  791. }
  792. // Second
  793. else if (all_of(segment, is_char('s'))) {
  794. builder.append("{second}"sv);
  795. if (segment.length() == 1)
  796. format.second = CalendarPatternStyle::Numeric;
  797. else
  798. format.second = CalendarPatternStyle::TwoDigit;
  799. } else if (all_of(segment, is_char('S'))) {
  800. builder.append("{fractionalSecondDigits}"sv);
  801. VERIFY(segment.length() <= 3);
  802. format.fractional_second_digits = static_cast<u8>(segment.length());
  803. } else if (all_of(segment, is_char('A'))) {
  804. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  805. return OptionalNone {};
  806. }
  807. // Zone
  808. else if (all_of(segment, is_any_of("zV"sv))) {
  809. builder.append("{timeZoneName}"sv);
  810. if (segment.length() < 4)
  811. format.time_zone_name = CalendarPatternStyle::Short;
  812. else
  813. format.time_zone_name = CalendarPatternStyle::Long;
  814. } else if (all_of(segment, is_any_of("ZOXx"sv))) {
  815. builder.append("{timeZoneName}"sv);
  816. if (segment.length() < 4)
  817. format.time_zone_name = CalendarPatternStyle::ShortOffset;
  818. else
  819. format.time_zone_name = CalendarPatternStyle::LongOffset;
  820. } else if (all_of(segment, is_char('v'))) {
  821. builder.append("{timeZoneName}"sv);
  822. if (segment.length() < 4)
  823. format.time_zone_name = CalendarPatternStyle::ShortGeneric;
  824. else
  825. format.time_zone_name = CalendarPatternStyle::LongGeneric;
  826. }
  827. // Non-patterns
  828. else {
  829. builder.append(segment);
  830. }
  831. }
  832. auto parsed_pattern = TRY(builder.to_string());
  833. if (hour12) {
  834. format.pattern = TRY(remove_period_from_pattern(parsed_pattern));
  835. format.pattern12 = move(parsed_pattern);
  836. } else {
  837. format.pattern = move(parsed_pattern);
  838. }
  839. return format;
  840. }
  841. static ErrorOr<Optional<size_t>> parse_date_time_pattern(ByteString pattern, ByteString skeleton, CLDR& cldr)
  842. {
  843. auto format = TRY(parse_date_time_pattern_raw(move(pattern), move(skeleton), cldr));
  844. if (!format.has_value())
  845. return OptionalNone {};
  846. format->pattern_index = cldr.unique_strings.ensure(format->pattern.to_byte_string());
  847. if (format->pattern12.has_value())
  848. format->pattern12_index = cldr.unique_strings.ensure(format->pattern12->to_byte_string());
  849. return Optional<size_t> { cldr.unique_patterns.ensure(format.release_value()) };
  850. }
  851. template<typename... Chars>
  852. static constexpr bool char_is_one_of(char ch, Chars&&... chars)
  853. {
  854. return ((ch == chars) || ...);
  855. }
  856. static ErrorOr<void> parse_interval_patterns(Calendar& calendar, JsonObject const& interval_formats_object, CLDR& cldr)
  857. {
  858. // https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats
  859. CalendarRangePatternList range_formats {};
  860. CalendarRangePatternList range12_formats {};
  861. auto name_of_field = [&](char field) {
  862. if (char_is_one_of(field, 'G'))
  863. return Locale::CalendarRangePattern::Field::Era;
  864. if (char_is_one_of(field, 'y', 'Y', 'u', 'U', 'r'))
  865. return Locale::CalendarRangePattern::Field::Year;
  866. if (char_is_one_of(field, 'M', 'L'))
  867. return Locale::CalendarRangePattern::Field::Month;
  868. if (char_is_one_of(field, 'd', 'D', 'F', 'g'))
  869. return Locale::CalendarRangePattern::Field::Day;
  870. if (char_is_one_of(field, 'a', 'b'))
  871. return Locale::CalendarRangePattern::Field::AmPm;
  872. if (char_is_one_of(field, 'B'))
  873. return Locale::CalendarRangePattern::Field::DayPeriod;
  874. if (char_is_one_of(field, 'h', 'H', 'K', 'k'))
  875. return Locale::CalendarRangePattern::Field::Hour;
  876. if (char_is_one_of(field, 'm'))
  877. return Locale::CalendarRangePattern::Field::Minute;
  878. if (char_is_one_of(field, 's'))
  879. return Locale::CalendarRangePattern::Field::Second;
  880. if (char_is_one_of(field, 'S'))
  881. return Locale::CalendarRangePattern::Field::FractionalSecondDigits;
  882. VERIFY_NOT_REACHED();
  883. };
  884. auto split_default_range_pattern = [&](auto skeleton, auto const& pattern) {
  885. auto start_range_end = pattern.find('}').value() + 1;
  886. auto end_range_begin = pattern.find_last('{').value();
  887. auto start_range = pattern.substring_view(0, start_range_end);
  888. auto separator = pattern.substring_view(start_range_end, end_range_begin - start_range_end);
  889. auto end_range = pattern.substring_view(end_range_begin);
  890. CalendarRangePattern format {};
  891. format.skeleton_index = cldr.unique_strings.ensure(skeleton);
  892. format.start_range = cldr.unique_strings.ensure(start_range);
  893. format.separator = cldr.unique_strings.ensure(separator);
  894. format.end_range = cldr.unique_strings.ensure(end_range);
  895. return format;
  896. };
  897. auto split_range_pattern = [&](auto skeleton, auto field, auto const& pattern, auto const& parsed_fields) {
  898. HashMap<StringView, size_t> partitions;
  899. StringView last_partition;
  900. auto begin_index = pattern.find_byte_offset('{');
  901. size_t end_index = 0;
  902. while (begin_index.has_value()) {
  903. end_index = pattern.find_byte_offset('}', *begin_index).value();
  904. auto partition = pattern.bytes_as_string_view().substring_view(*begin_index, end_index - *begin_index);
  905. if (partitions.contains(partition))
  906. break;
  907. partitions.set(partition, *begin_index);
  908. last_partition = partition;
  909. begin_index = pattern.find_byte_offset('{', end_index + 1);
  910. }
  911. VERIFY(begin_index.has_value() && !last_partition.is_empty());
  912. auto start_range_end = partitions.get(last_partition).value() + last_partition.length() + 1;
  913. auto pattern_view = pattern.bytes_as_string_view();
  914. auto start_range = pattern_view.substring_view(0, start_range_end);
  915. auto separator = pattern_view.substring_view(start_range_end, *begin_index - start_range_end);
  916. auto end_range = pattern_view.substring_view(*begin_index);
  917. CalendarRangePattern format {};
  918. format.skeleton_index = cldr.unique_strings.ensure(skeleton);
  919. format.field = field;
  920. format.start_range = cldr.unique_strings.ensure(start_range);
  921. format.separator = cldr.unique_strings.ensure(separator);
  922. format.end_range = cldr.unique_strings.ensure(end_range);
  923. format.for_each_calendar_field_zipped_with(parsed_fields, [](auto& format_field, auto const& parsed_field, auto) {
  924. format_field = parsed_field;
  925. });
  926. return format;
  927. };
  928. TRY(interval_formats_object.try_for_each_member([&](auto const& skeleton, auto const& value) -> ErrorOr<void> {
  929. if (skeleton == "intervalFormatFallback"sv) {
  930. auto range_format = split_default_range_pattern(skeleton, value.as_string());
  931. calendar.default_range_format = cldr.unique_range_patterns.ensure(move(range_format));
  932. return {};
  933. }
  934. TRY(value.as_object().try_for_each_member([&](auto const& field, auto const& pattern) -> ErrorOr<void> {
  935. if (field.ends_with("alt-variant"sv))
  936. return {};
  937. VERIFY(field.length() == 1);
  938. auto name = name_of_field(field[0]);
  939. auto format = TRY(parse_date_time_pattern_raw(pattern.as_string(), skeleton, cldr)).release_value();
  940. auto range_format = split_range_pattern(skeleton, name, format.pattern, format);
  941. range_formats.append(cldr.unique_range_patterns.ensure(move(range_format)));
  942. if (format.pattern12.has_value()) {
  943. auto range12_pattern = split_range_pattern(skeleton, name, *format.pattern12, format);
  944. range12_formats.append(cldr.unique_range_patterns.ensure(move(range12_pattern)));
  945. } else {
  946. range12_formats.append(range_formats.last());
  947. }
  948. return {};
  949. }));
  950. return {};
  951. }));
  952. calendar.range_formats = cldr.unique_range_pattern_lists.ensure(move(range_formats));
  953. calendar.range12_formats = cldr.unique_range_pattern_lists.ensure(move(range12_formats));
  954. return {};
  955. }
  956. static ErrorOr<void> generate_default_patterns(CalendarPatternList& formats, CLDR& cldr)
  957. {
  958. // For compatibility with ICU, we generate a list of default patterns for every locale:
  959. // https://github.com/unicode-org/icu/blob/release-71-1/icu4c/source/i18n/dtptngen.cpp#L1343-L1354=
  960. 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 };
  961. for (auto pattern : default_patterns) {
  962. auto index = TRY(parse_date_time_pattern(pattern, pattern, cldr));
  963. VERIFY(index.has_value());
  964. if (!formats.contains_slow(*index))
  965. formats.append(*index);
  966. }
  967. return {};
  968. }
  969. static void generate_missing_patterns(Calendar& calendar, CalendarPatternList& formats, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, CLDR& cldr)
  970. {
  971. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  972. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  973. auto pattern = cldr.unique_strings.get(format);
  974. auto time_pattern = cldr.unique_strings.get(time_format);
  975. auto date_pattern = cldr.unique_strings.get(date_format);
  976. auto new_pattern = pattern.replace("{0}"sv, time_pattern, ReplaceMode::FirstOnly).replace("{1}"sv, date_pattern, ReplaceMode::FirstOnly);
  977. return cldr.unique_strings.ensure(move(new_pattern));
  978. };
  979. auto inject_fractional_second_digits = [&](auto format) {
  980. auto pattern = cldr.unique_strings.get(format);
  981. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv, ReplaceMode::FirstOnly);
  982. return cldr.unique_strings.ensure(move(new_pattern));
  983. };
  984. auto append_if_unique = [&](auto format) {
  985. auto format_index = cldr.unique_patterns.ensure(move(format));
  986. if (!formats.contains_slow(format_index))
  987. formats.append(format_index);
  988. };
  989. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  990. for (auto const& format : date_formats)
  991. append_if_unique(format);
  992. for (auto const& format : time_formats) {
  993. append_if_unique(format);
  994. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  995. auto new_format = format;
  996. new_format.fractional_second_digits = 2;
  997. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  998. if (new_format.pattern12_index != 0)
  999. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  1000. time_formats_with_fractional_second_digits.append(new_format);
  1001. append_if_unique(move(new_format));
  1002. }
  1003. }
  1004. time_formats.extend(move(time_formats_with_fractional_second_digits));
  1005. for (auto const& date_format : date_formats) {
  1006. auto const& date_time_formats = cldr.unique_formats.get(calendar.date_time_formats);
  1007. size_t date_time_format_index = 0;
  1008. if (date_format.month == Locale::CalendarPatternStyle::Long) {
  1009. if (date_format.weekday.has_value())
  1010. date_time_format_index = date_time_formats.full_format;
  1011. else
  1012. date_time_format_index = date_time_formats.long_format;
  1013. } else if (date_format.month == Locale::CalendarPatternStyle::Short) {
  1014. date_time_format_index = date_time_formats.medium_format;
  1015. } else {
  1016. date_time_format_index = date_time_formats.short_format;
  1017. }
  1018. for (auto const& time_format : time_formats) {
  1019. auto format = cldr.unique_patterns.get(date_time_format_index);
  1020. if (time_format.pattern12_index != 0)
  1021. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  1022. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  1023. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  1024. if (date_field.has_value())
  1025. field = date_field;
  1026. });
  1027. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  1028. if (time_field.has_value())
  1029. field = time_field;
  1030. });
  1031. append_if_unique(move(format));
  1032. }
  1033. }
  1034. }
  1035. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, CLDR& cldr)
  1036. {
  1037. auto create_symbol_lists = [](size_t size) {
  1038. SymbolList narrow_symbol_list;
  1039. SymbolList short_symbol_list;
  1040. SymbolList long_symbol_list;
  1041. narrow_symbol_list.resize(size);
  1042. short_symbol_list.resize(size);
  1043. long_symbol_list.resize(size);
  1044. return Array<SymbolList, 3> { {
  1045. move(narrow_symbol_list),
  1046. move(short_symbol_list),
  1047. move(long_symbol_list),
  1048. } };
  1049. };
  1050. CalendarSymbolsList symbols_list {};
  1051. auto store_symbol_lists = [&](auto symbol, auto symbol_lists) {
  1052. auto symbol_index = to_underlying(symbol);
  1053. if (symbol_index >= symbols_list.size())
  1054. symbols_list.resize(symbol_index + 1);
  1055. CalendarSymbols symbols {};
  1056. symbols.narrow_symbols = cldr.unique_symbol_lists.ensure(move(symbol_lists[0]));
  1057. symbols.short_symbols = cldr.unique_symbol_lists.ensure(move(symbol_lists[1]));
  1058. symbols.long_symbols = cldr.unique_symbol_lists.ensure(move(symbol_lists[2]));
  1059. auto calendar_symbols_index = cldr.unique_calendar_symbols.ensure(move(symbols));
  1060. symbols_list[symbol_index] = calendar_symbols_index;
  1061. };
  1062. auto parse_era_symbols = [&](auto const& symbols_object) {
  1063. auto const& narrow_symbols = symbols_object.get_object("eraNarrow"sv).value();
  1064. auto const& short_symbols = symbols_object.get_object("eraAbbr"sv).value();
  1065. auto const& long_symbols = symbols_object.get_object("eraNames"sv).value();
  1066. auto symbol_lists = create_symbol_lists(2);
  1067. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1068. if (auto key_index = key.template to_number<unsigned>(); key_index.has_value())
  1069. symbols[*key_index] = cldr.unique_strings.ensure(move(symbol));
  1070. };
  1071. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1072. append_symbol(symbol_lists[0], key, value.as_string());
  1073. });
  1074. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1075. append_symbol(symbol_lists[1], key, value.as_string());
  1076. });
  1077. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1078. append_symbol(symbol_lists[2], key, value.as_string());
  1079. });
  1080. store_symbol_lists(Locale::CalendarSymbol::Era, move(symbol_lists));
  1081. };
  1082. auto parse_month_symbols = [&](auto const& symbols_object) {
  1083. auto const& narrow_symbols = symbols_object.get_object("narrow"sv).value();
  1084. auto const& short_symbols = symbols_object.get_object("abbreviated"sv).value();
  1085. auto const& long_symbols = symbols_object.get_object("wide"sv).value();
  1086. auto symbol_lists = create_symbol_lists(12);
  1087. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1088. auto key_index = key.template to_number<unsigned>().value() - 1;
  1089. symbols[key_index] = cldr.unique_strings.ensure(move(symbol));
  1090. };
  1091. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1092. append_symbol(symbol_lists[0], key, value.as_string());
  1093. });
  1094. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1095. append_symbol(symbol_lists[1], key, value.as_string());
  1096. });
  1097. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1098. append_symbol(symbol_lists[2], key, value.as_string());
  1099. });
  1100. store_symbol_lists(Locale::CalendarSymbol::Month, move(symbol_lists));
  1101. };
  1102. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  1103. auto const& narrow_symbols = symbols_object.get_object("narrow"sv).value();
  1104. auto const& short_symbols = symbols_object.get_object("abbreviated"sv).value();
  1105. auto const& long_symbols = symbols_object.get_object("wide"sv).value();
  1106. auto symbol_lists = create_symbol_lists(7);
  1107. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1108. if (key == "sun"sv)
  1109. symbols[to_underlying(Locale::Weekday::Sunday)] = cldr.unique_strings.ensure(move(symbol));
  1110. else if (key == "mon"sv)
  1111. symbols[to_underlying(Locale::Weekday::Monday)] = cldr.unique_strings.ensure(move(symbol));
  1112. else if (key == "tue"sv)
  1113. symbols[to_underlying(Locale::Weekday::Tuesday)] = cldr.unique_strings.ensure(move(symbol));
  1114. else if (key == "wed"sv)
  1115. symbols[to_underlying(Locale::Weekday::Wednesday)] = cldr.unique_strings.ensure(move(symbol));
  1116. else if (key == "thu"sv)
  1117. symbols[to_underlying(Locale::Weekday::Thursday)] = cldr.unique_strings.ensure(move(symbol));
  1118. else if (key == "fri"sv)
  1119. symbols[to_underlying(Locale::Weekday::Friday)] = cldr.unique_strings.ensure(move(symbol));
  1120. else if (key == "sat"sv)
  1121. symbols[to_underlying(Locale::Weekday::Saturday)] = cldr.unique_strings.ensure(move(symbol));
  1122. };
  1123. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1124. append_symbol(symbol_lists[0], key, value.as_string());
  1125. });
  1126. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1127. append_symbol(symbol_lists[1], key, value.as_string());
  1128. });
  1129. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1130. append_symbol(symbol_lists[2], key, value.as_string());
  1131. });
  1132. store_symbol_lists(Locale::CalendarSymbol::Weekday, move(symbol_lists));
  1133. };
  1134. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  1135. auto const& narrow_symbols = symbols_object.get_object("narrow"sv).value();
  1136. auto const& short_symbols = symbols_object.get_object("abbreviated"sv).value();
  1137. auto const& long_symbols = symbols_object.get_object("wide"sv).value();
  1138. auto symbol_lists = create_symbol_lists(11);
  1139. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1140. if (auto day_period = day_period_from_string(key); day_period.has_value())
  1141. symbols[to_underlying(*day_period)] = cldr.unique_strings.ensure(move(symbol));
  1142. };
  1143. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1144. append_symbol(symbol_lists[0], key, value.as_string());
  1145. });
  1146. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1147. append_symbol(symbol_lists[1], key, value.as_string());
  1148. });
  1149. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1150. append_symbol(symbol_lists[2], key, value.as_string());
  1151. });
  1152. store_symbol_lists(Locale::CalendarSymbol::DayPeriod, move(symbol_lists));
  1153. };
  1154. parse_era_symbols(calendar_object.get_object("eras"sv).value());
  1155. parse_month_symbols(calendar_object.get_object("months"sv)->get_object("format"sv).value());
  1156. parse_weekday_symbols(calendar_object.get_object("days"sv)->get_object("format"sv).value());
  1157. parse_day_period_symbols(calendar_object.get_object("dayPeriods"sv)->get_object("format"sv).value());
  1158. calendar.symbols = cldr.unique_calendar_symbols_lists.ensure(move(symbols_list));
  1159. }
  1160. static ErrorOr<void> parse_calendars(ByteString locale_calendars_path, CLDR& cldr, LocaleData& locale)
  1161. {
  1162. LexicalPath calendars_path(move(locale_calendars_path));
  1163. if (!calendars_path.basename().starts_with("ca-"sv))
  1164. return {};
  1165. auto calendars = TRY(read_json_file(calendars_path.string()));
  1166. auto const& main_object = calendars.as_object().get_object("main"sv).value();
  1167. auto const& locale_object = main_object.get_object(calendars_path.parent().basename()).value();
  1168. auto const& dates_object = locale_object.get_object("dates"sv).value();
  1169. auto const& calendars_object = dates_object.get_object("calendars"sv).value();
  1170. auto parse_patterns = [&](auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) -> ErrorOr<size_t> {
  1171. auto parse_pattern = [&](auto name) -> ErrorOr<size_t> {
  1172. auto format = patterns_object.get_byte_string(name);
  1173. auto skeleton = skeletons_object.get_byte_string(name);
  1174. auto format_index = TRY(parse_date_time_pattern(format.value(), skeleton.value_or(ByteString::empty()), cldr)).value();
  1175. if (patterns)
  1176. patterns->append(cldr.unique_patterns.get(format_index));
  1177. return format_index;
  1178. };
  1179. CalendarFormat formats {};
  1180. formats.full_format = TRY(parse_pattern("full"sv));
  1181. formats.long_format = TRY(parse_pattern("long"sv));
  1182. formats.medium_format = TRY(parse_pattern("medium"sv));
  1183. formats.short_format = TRY(parse_pattern("short"sv));
  1184. return cldr.unique_formats.ensure(move(formats));
  1185. };
  1186. TRY(calendars_object.try_for_each_member([&](auto const& calendar_name, JsonValue const& value) -> ErrorOr<void> {
  1187. // The generic calendar is not a supported Unicode calendar key, so skip it:
  1188. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  1189. if (calendar_name == "generic"sv)
  1190. return {};
  1191. Calendar calendar {};
  1192. CalendarPatternList available_formats {};
  1193. if (!cldr.calendars.contains_slow(calendar_name))
  1194. cldr.calendars.append(calendar_name);
  1195. Vector<CalendarPattern> date_formats;
  1196. Vector<CalendarPattern> time_formats;
  1197. auto const& date_formats_object = value.as_object().get_object("dateFormats"sv).value();
  1198. auto const& date_skeletons_object = value.as_object().get_object("dateSkeletons"sv).value();
  1199. calendar.date_formats = TRY(parse_patterns(date_formats_object, date_skeletons_object, &date_formats));
  1200. auto const& time_formats_object = value.as_object().get_object("timeFormats"sv).value();
  1201. auto const& time_skeletons_object = value.as_object().get_object("timeSkeletons"sv).value();
  1202. calendar.time_formats = TRY(parse_patterns(time_formats_object, time_skeletons_object, &time_formats));
  1203. auto const& standard_date_time_formats_object = value.as_object().get_object("dateTimeFormats-atTime"sv)->get_object("standard"sv).value();
  1204. calendar.date_time_formats = TRY(parse_patterns(standard_date_time_formats_object, JsonObject {}, nullptr));
  1205. auto const& date_time_formats_object = value.as_object().get_object("dateTimeFormats"sv).value();
  1206. auto const& available_formats_object = date_time_formats_object.get_object("availableFormats"sv).value();
  1207. TRY(available_formats_object.try_for_each_member([&](auto const& skeleton, JsonValue const& pattern) -> ErrorOr<void> {
  1208. auto pattern_index = TRY(parse_date_time_pattern(pattern.as_string(), skeleton, cldr));
  1209. if (!pattern_index.has_value())
  1210. return {};
  1211. auto const& format = cldr.unique_patterns.get(*pattern_index);
  1212. if (format.contains_only_date_fields())
  1213. date_formats.append(format);
  1214. else if (format.contains_only_time_fields())
  1215. time_formats.append(format);
  1216. if (!available_formats.contains_slow(*pattern_index))
  1217. available_formats.append(*pattern_index);
  1218. return {};
  1219. }));
  1220. auto const& interval_formats_object = date_time_formats_object.get_object("intervalFormats"sv).value();
  1221. TRY(parse_interval_patterns(calendar, interval_formats_object, cldr));
  1222. TRY(generate_default_patterns(available_formats, cldr));
  1223. generate_missing_patterns(calendar, available_formats, move(date_formats), move(time_formats), cldr);
  1224. parse_calendar_symbols(calendar, value.as_object(), cldr);
  1225. calendar.available_formats = cldr.unique_pattern_lists.ensure(move(available_formats));
  1226. locale.calendars.set(calendar_name, cldr.unique_calendars.ensure(move(calendar)));
  1227. return {};
  1228. }));
  1229. return {};
  1230. }
  1231. static ErrorOr<void> parse_time_zone_names(ByteString locale_time_zone_names_path, CLDR& cldr, LocaleData& locale)
  1232. {
  1233. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  1234. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  1235. auto time_zone_names = TRY(read_json_file(time_zone_names_path.string()));
  1236. auto const& main_object = time_zone_names.as_object().get_object("main"sv).value();
  1237. auto const& locale_object = main_object.get_object(time_zone_names_path.parent().basename()).value();
  1238. auto const& dates_object = locale_object.get_object("dates"sv).value();
  1239. auto const& time_zone_names_object = dates_object.get_object("timeZoneNames"sv).value();
  1240. auto const& meta_zone_object = time_zone_names_object.get_object("metazone"sv);
  1241. auto const& hour_format_string = time_zone_names_object.get_byte_string("hourFormat"sv).value();
  1242. auto const& gmt_format_string = time_zone_names_object.get_byte_string("gmtFormat"sv).value();
  1243. auto const& gmt_zero_format_string = time_zone_names_object.get_byte_string("gmtZeroFormat"sv).value();
  1244. if (!meta_zone_object.has_value())
  1245. return {};
  1246. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object, StringView key) -> Optional<size_t> {
  1247. auto const& names = meta_zone_object.get_object(type);
  1248. if (!names.has_value())
  1249. return {};
  1250. auto const& name = names->get_byte_string(key);
  1251. if (name.has_value())
  1252. return cldr.unique_strings.ensure(name.value());
  1253. return {};
  1254. };
  1255. auto parse_hour_format = [&](auto const& format, auto& time_zone_formats) {
  1256. auto hour_formats = format.split_view(';');
  1257. auto hour_format_ahead_start = hour_formats[0].find('H').value();
  1258. auto separator_ahead_start = hour_formats[0].find_last('H').value() + 1;
  1259. auto separator_ahead_end = hour_formats[0].find('m').value();
  1260. auto hour_format_behind_start = hour_formats[1].find('H').value();
  1261. auto separator_behind_start = hour_formats[1].find_last('H').value() + 1;
  1262. auto separator_behind_end = hour_formats[1].find('m').value();
  1263. auto symbol_ahead_sign = hour_formats[0].substring_view(0, hour_format_ahead_start);
  1264. auto symbol_ahead_separator = hour_formats[0].substring_view(separator_ahead_start, separator_ahead_end - separator_ahead_start);
  1265. auto symbol_behind_sign = hour_formats[1].substring_view(0, hour_format_behind_start);
  1266. auto symbol_behind_separator = hour_formats[1].substring_view(separator_behind_start, separator_behind_end - separator_behind_start);
  1267. time_zone_formats.symbol_ahead_sign = cldr.unique_strings.ensure(symbol_ahead_sign);
  1268. time_zone_formats.symbol_ahead_separator = cldr.unique_strings.ensure(symbol_ahead_separator);
  1269. time_zone_formats.symbol_behind_sign = cldr.unique_strings.ensure(symbol_behind_sign);
  1270. time_zone_formats.symbol_behind_separator = cldr.unique_strings.ensure(symbol_behind_separator);
  1271. };
  1272. TimeZoneNamesList time_zones;
  1273. TimeZoneFormat time_zone_formats {};
  1274. parse_hour_format(hour_format_string, time_zone_formats);
  1275. time_zone_formats.gmt_format = cldr.unique_strings.ensure(gmt_format_string);
  1276. time_zone_formats.gmt_zero_format = cldr.unique_strings.ensure(gmt_zero_format_string);
  1277. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  1278. auto golden_zones = cldr.meta_zones.find(meta_zone);
  1279. if (golden_zones == cldr.meta_zones.end())
  1280. return;
  1281. TimeZoneNames time_zone_names {};
  1282. if (auto name = parse_name("long"sv, meta_zone_object, "standard"sv); name.has_value())
  1283. time_zone_names.long_standard_name = name.value();
  1284. if (auto name = parse_name("short"sv, meta_zone_object, "standard"sv); name.has_value())
  1285. time_zone_names.short_standard_name = name.value();
  1286. if (auto name = parse_name("long"sv, meta_zone_object, "daylight"sv); name.has_value())
  1287. time_zone_names.long_daylight_name = name.value();
  1288. if (auto name = parse_name("short"sv, meta_zone_object, "daylight"sv); name.has_value())
  1289. time_zone_names.short_daylight_name = name.value();
  1290. if (auto name = parse_name("long"sv, meta_zone_object, "generic"sv); name.has_value())
  1291. time_zone_names.long_generic_name = name.value();
  1292. if (auto name = parse_name("short"sv, meta_zone_object, "generic"sv); name.has_value())
  1293. time_zone_names.short_generic_name = name.value();
  1294. auto time_zone_index = cldr.unique_time_zones.ensure(move(time_zone_names));
  1295. for (auto golden_zone : golden_zones->value) {
  1296. auto time_zone = to_underlying(golden_zone);
  1297. if (time_zone >= time_zones.size())
  1298. time_zones.resize(time_zone + 1);
  1299. time_zones[time_zone] = time_zone_index;
  1300. }
  1301. };
  1302. meta_zone_object->for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  1303. parse_time_zone(meta_zone, value.as_object());
  1304. });
  1305. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  1306. auto const& zone_object = time_zone_names_object.get_object("zone"sv).value();
  1307. auto const& etc_object = zone_object.get_object("Etc"sv).value();
  1308. auto const& utc_object = etc_object.get_object("UTC"sv).value();
  1309. parse_time_zone("UTC"sv, utc_object);
  1310. locale.time_zones = cldr.unique_time_zone_lists.ensure(move(time_zones));
  1311. locale.time_zone_formats = cldr.unique_time_zone_formats.ensure(move(time_zone_formats));
  1312. return {};
  1313. }
  1314. static ErrorOr<void> parse_day_periods(ByteString core_path, CLDR& cldr)
  1315. {
  1316. // https://unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rule_Sets
  1317. LexicalPath day_periods_path(move(core_path));
  1318. day_periods_path = day_periods_path.append("supplemental"sv);
  1319. day_periods_path = day_periods_path.append("dayPeriods.json"sv);
  1320. auto locale_day_periods = TRY(read_json_file(day_periods_path.string()));
  1321. auto const& supplemental_object = locale_day_periods.as_object().get_object("supplemental"sv).value();
  1322. auto const& day_periods_object = supplemental_object.get_object("dayPeriodRuleSet"sv).value();
  1323. auto parse_hour = [](auto const& time) {
  1324. auto hour_end_index = time.find(':').value();
  1325. // The times are in the form "hours:minutes", but we only need the hour segment.
  1326. // TR-35 explicitly states that minutes other than :00 must not be used.
  1327. VERIFY(time.substring_view(hour_end_index) == ":00"sv);
  1328. auto hour = time.substring_view(0, hour_end_index);
  1329. return hour.template to_number<u8>().value();
  1330. };
  1331. auto parse_day_period = [&](auto const& symbol, auto const& ranges) -> Optional<DayPeriod> {
  1332. if (!ranges.has("_from"sv))
  1333. return {};
  1334. auto day_period = day_period_from_string(symbol);
  1335. if (!day_period.has_value())
  1336. return {};
  1337. auto begin = parse_hour(ranges.get_byte_string("_from"sv).value());
  1338. auto end = parse_hour(ranges.get_byte_string("_before"sv).value());
  1339. return DayPeriod { *day_period, begin, end };
  1340. };
  1341. day_periods_object.for_each_member([&](auto const& language, JsonValue const& value) {
  1342. auto locale = cldr.locales.find(language);
  1343. if (locale == cldr.locales.end())
  1344. return;
  1345. DayPeriodList day_periods;
  1346. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& ranges) {
  1347. if (auto day_period = parse_day_period(symbol, ranges.as_object()); day_period.has_value()) {
  1348. auto day_period_index = cldr.unique_day_periods.ensure(day_period.release_value());
  1349. day_periods.append(day_period_index);
  1350. }
  1351. });
  1352. locale->value.day_periods = cldr.unique_day_period_lists.ensure(move(day_periods));
  1353. });
  1354. return {};
  1355. }
  1356. static ErrorOr<void> parse_all_locales(ByteString core_path, ByteString dates_path, CLDR& cldr)
  1357. {
  1358. TRY(parse_hour_cycles(core_path, cldr));
  1359. TRY(parse_week_data(core_path, cldr));
  1360. TRY(parse_meta_zones(core_path, cldr));
  1361. auto remove_variants_from_path = [&](ByteString path) -> ErrorOr<ByteString> {
  1362. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, LexicalPath::basename(path)));
  1363. StringBuilder builder;
  1364. builder.append(cldr.unique_strings.get(parsed_locale.language));
  1365. if (auto script = cldr.unique_strings.get(parsed_locale.script); !script.is_empty())
  1366. builder.appendff("-{}", script);
  1367. if (auto region = cldr.unique_strings.get(parsed_locale.region); !region.is_empty())
  1368. builder.appendff("-{}", region);
  1369. return builder.to_byte_string();
  1370. };
  1371. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", dates_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  1372. auto dates_path = LexicalPath::join(directory.path().string(), entry.name).string();
  1373. auto language = TRY(remove_variants_from_path(dates_path));
  1374. auto& locale = cldr.locales.ensure(language);
  1375. TRY(Core::Directory::for_each_entry(dates_path, Core::DirIterator::SkipParentAndBaseDir, [&](auto& dates_entry, auto& dates_directory) -> ErrorOr<IterationDecision> {
  1376. auto calendars_path = LexicalPath::join(dates_directory.path().string(), dates_entry.name).string();
  1377. TRY(parse_calendars(move(calendars_path), cldr, locale));
  1378. return IterationDecision::Continue;
  1379. }));
  1380. TRY(parse_time_zone_names(move(dates_path), cldr, locale));
  1381. return IterationDecision::Continue;
  1382. }));
  1383. TRY(parse_day_periods(move(core_path), cldr));
  1384. return {};
  1385. }
  1386. static ByteString format_identifier(StringView owner, ByteString identifier)
  1387. {
  1388. identifier = identifier.replace("-"sv, "_"sv, ReplaceMode::All);
  1389. identifier = identifier.replace("/"sv, "_"sv, ReplaceMode::All);
  1390. if (all_of(identifier, is_ascii_digit))
  1391. return ByteString::formatted("{}_{}", owner[0], identifier);
  1392. if (is_ascii_lower_alpha(identifier[0]))
  1393. return ByteString::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  1394. return identifier;
  1395. }
  1396. static ErrorOr<void> generate_unicode_locale_header(Core::InputBufferedFile& file, CLDR& cldr)
  1397. {
  1398. StringBuilder builder;
  1399. SourceGenerator generator { builder };
  1400. generator.append(R"~~~(
  1401. #pragma once
  1402. #include <AK/Types.h>
  1403. namespace Locale {
  1404. )~~~");
  1405. generate_enum(generator, format_identifier, "Calendar"sv, {}, cldr.calendars);
  1406. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, cldr.hour_cycle_regions);
  1407. generate_enum(generator, format_identifier, "MinimumDaysRegion"sv, {}, cldr.minimum_days_regions);
  1408. generate_enum(generator, format_identifier, "FirstDayRegion"sv, {}, cldr.first_day_regions);
  1409. generate_enum(generator, format_identifier, "WeekendStartRegion"sv, {}, cldr.weekend_start_regions);
  1410. generate_enum(generator, format_identifier, "WeekendEndRegion"sv, {}, cldr.weekend_end_regions);
  1411. generator.append(R"~~~(
  1412. }
  1413. )~~~");
  1414. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  1415. return {};
  1416. }
  1417. static ErrorOr<void> generate_unicode_locale_implementation(Core::InputBufferedFile& file, CLDR& cldr)
  1418. {
  1419. StringBuilder builder;
  1420. SourceGenerator generator { builder };
  1421. generator.set("string_index_type"sv, cldr.unique_strings.type_that_fits());
  1422. generator.set("calendar_pattern_index_type"sv, cldr.unique_patterns.type_that_fits());
  1423. generator.set("calendar_pattern_list_index_type"sv, cldr.unique_pattern_lists.type_that_fits());
  1424. generator.set("calendar_range_pattern_index_type"sv, cldr.unique_range_patterns.type_that_fits());
  1425. generator.set("calendar_range_pattern_list_index_type"sv, cldr.unique_range_pattern_lists.type_that_fits());
  1426. generator.set("calendar_format_index_type"sv, cldr.unique_formats.type_that_fits());
  1427. generator.set("symbol_list_index_type"sv, cldr.unique_symbol_lists.type_that_fits());
  1428. generator.set("calendar_symbols_index_type"sv, cldr.unique_calendar_symbols.type_that_fits());
  1429. generator.set("calendar_symbols_list_index_type"sv, cldr.unique_calendar_symbols_lists.type_that_fits());
  1430. generator.set("calendar_index_type"sv, cldr.unique_calendars.type_that_fits());
  1431. generator.set("time_zone_index_type"sv, cldr.unique_time_zones.type_that_fits());
  1432. generator.set("time_zone_list_index_type"sv, cldr.unique_time_zone_lists.type_that_fits());
  1433. generator.set("day_period_index_type"sv, cldr.unique_day_periods.type_that_fits());
  1434. generator.set("day_period_list_index_type"sv, cldr.unique_day_period_lists.type_that_fits());
  1435. generator.append(R"~~~(
  1436. #include <AK/Array.h>
  1437. #include <AK/BinarySearch.h>
  1438. #include <AK/Error.h>
  1439. #include <AK/Optional.h>
  1440. #include <AK/String.h>
  1441. #include <AK/StringView.h>
  1442. #include <LibLocale/DateTimeFormat.h>
  1443. #include <LibLocale/DateTimeFormatData.h>
  1444. #include <LibLocale/Locale.h>
  1445. #include <LibLocale/LocaleData.h>
  1446. #include <LibTimeZone/TimeZone.h>
  1447. namespace Locale {
  1448. )~~~");
  1449. cldr.unique_strings.generate(generator);
  1450. generator.append(R"~~~(
  1451. template <typename SourceType, typename TargetType>
  1452. static void convert_calendar_fields(SourceType const& source, TargetType& target)
  1453. {
  1454. if (source.era != -1)
  1455. target.era = static_cast<CalendarPatternStyle>(source.era);
  1456. if (source.year != -1)
  1457. target.year = static_cast<CalendarPatternStyle>(source.year);
  1458. if (source.month != -1)
  1459. target.month = static_cast<CalendarPatternStyle>(source.month);
  1460. if (source.weekday != -1)
  1461. target.weekday = static_cast<CalendarPatternStyle>(source.weekday);
  1462. if (source.day != -1)
  1463. target.day = static_cast<CalendarPatternStyle>(source.day);
  1464. if (source.day_period != -1)
  1465. target.day_period = static_cast<CalendarPatternStyle>(source.day_period);
  1466. if (source.hour != -1)
  1467. target.hour = static_cast<CalendarPatternStyle>(source.hour);
  1468. if (source.minute != -1)
  1469. target.minute = static_cast<CalendarPatternStyle>(source.minute);
  1470. if (source.second != -1)
  1471. target.second = static_cast<CalendarPatternStyle>(source.second);
  1472. if (source.fractional_second_digits != -1)
  1473. target.fractional_second_digits = static_cast<u8>(source.fractional_second_digits);
  1474. if (source.time_zone_name != -1)
  1475. target.time_zone_name = static_cast<CalendarPatternStyle>(source.time_zone_name);
  1476. }
  1477. struct CalendarPatternImpl {
  1478. CalendarPattern to_unicode_calendar_pattern() const {
  1479. CalendarPattern calendar_pattern {};
  1480. calendar_pattern.skeleton = MUST(String::from_utf8(decode_string(skeleton)));
  1481. calendar_pattern.pattern = MUST(String::from_utf8(decode_string(pattern)));
  1482. if (pattern12 != 0)
  1483. calendar_pattern.pattern12 = MUST(String::from_utf8(decode_string(pattern12)));
  1484. convert_calendar_fields(*this, calendar_pattern);
  1485. return calendar_pattern;
  1486. }
  1487. @string_index_type@ skeleton { 0 };
  1488. @string_index_type@ pattern { 0 };
  1489. @string_index_type@ pattern12 { 0 };
  1490. i8 era { -1 };
  1491. i8 year { -1 };
  1492. i8 month { -1 };
  1493. i8 weekday { -1 };
  1494. i8 day { -1 };
  1495. i8 day_period { -1 };
  1496. i8 hour { -1 };
  1497. i8 minute { -1 };
  1498. i8 second { -1 };
  1499. i8 fractional_second_digits { -1 };
  1500. i8 time_zone_name { -1 };
  1501. };
  1502. struct CalendarRangePatternImpl {
  1503. CalendarRangePattern to_unicode_calendar_range_pattern() const {
  1504. CalendarRangePattern calendar_range_pattern {};
  1505. if (field != -1)
  1506. calendar_range_pattern.field = static_cast<CalendarRangePattern::Field>(field);
  1507. calendar_range_pattern.start_range = MUST(String::from_utf8(decode_string(start_range)));
  1508. calendar_range_pattern.separator = decode_string(separator);
  1509. calendar_range_pattern.end_range = MUST(String::from_utf8(decode_string(end_range)));
  1510. convert_calendar_fields(*this, calendar_range_pattern);
  1511. return calendar_range_pattern;
  1512. }
  1513. @string_index_type@ skeleton { 0 };
  1514. i8 field { -1 };
  1515. @string_index_type@ start_range { 0 };
  1516. @string_index_type@ separator { 0 };
  1517. @string_index_type@ end_range { 0 };
  1518. i8 era { -1 };
  1519. i8 year { -1 };
  1520. i8 month { -1 };
  1521. i8 weekday { -1 };
  1522. i8 day { -1 };
  1523. i8 day_period { -1 };
  1524. i8 hour { -1 };
  1525. i8 minute { -1 };
  1526. i8 second { -1 };
  1527. i8 fractional_second_digits { -1 };
  1528. i8 time_zone_name { -1 };
  1529. };
  1530. )~~~");
  1531. cldr.unique_patterns.generate(generator, "CalendarPatternImpl"sv, "s_calendar_patterns"sv, 10);
  1532. cldr.unique_pattern_lists.generate(generator, cldr.unique_patterns.type_that_fits(), "s_calendar_pattern_lists"sv);
  1533. cldr.unique_range_patterns.generate(generator, "CalendarRangePatternImpl"sv, "s_calendar_range_patterns"sv, 10);
  1534. cldr.unique_range_pattern_lists.generate(generator, cldr.unique_range_patterns.type_that_fits(), "s_calendar_range_pattern_lists"sv);
  1535. generator.append(R"~~~(
  1536. struct CalendarFormatImpl {
  1537. CalendarFormat to_unicode_calendar_format() const {
  1538. CalendarFormat calendar_format {};
  1539. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  1540. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  1541. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  1542. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  1543. return calendar_format;
  1544. }
  1545. @calendar_pattern_index_type@ full_format { 0 };
  1546. @calendar_pattern_index_type@ long_format { 0 };
  1547. @calendar_pattern_index_type@ medium_format { 0 };
  1548. @calendar_pattern_index_type@ short_format { 0 };
  1549. };
  1550. struct CalendarSymbols {
  1551. @symbol_list_index_type@ narrow_symbols { 0 };
  1552. @symbol_list_index_type@ short_symbols { 0 };
  1553. @symbol_list_index_type@ long_symbols { 0 };
  1554. };
  1555. struct CalendarData {
  1556. @calendar_format_index_type@ date_formats { 0 };
  1557. @calendar_format_index_type@ time_formats { 0 };
  1558. @calendar_format_index_type@ date_time_formats { 0 };
  1559. @calendar_pattern_list_index_type@ available_formats { 0 };
  1560. @calendar_range_pattern_index_type@ default_range_format { 0 };
  1561. @calendar_range_pattern_list_index_type@ range_formats { 0 };
  1562. @calendar_range_pattern_list_index_type@ range12_formats { 0 };
  1563. @calendar_symbols_list_index_type@ symbols { 0 };
  1564. };
  1565. struct TimeZoneNames {
  1566. @string_index_type@ short_standard_name { 0 };
  1567. @string_index_type@ long_standard_name { 0 };
  1568. @string_index_type@ short_daylight_name { 0 };
  1569. @string_index_type@ long_daylight_name { 0 };
  1570. @string_index_type@ short_generic_name { 0 };
  1571. @string_index_type@ long_generic_name { 0 };
  1572. };
  1573. struct TimeZoneFormatImpl {
  1574. TimeZoneFormat to_time_zone_format() const {
  1575. TimeZoneFormat time_zone_format {};
  1576. time_zone_format.symbol_ahead_sign = decode_string(symbol_ahead_sign);
  1577. time_zone_format.symbol_ahead_separator = decode_string(symbol_ahead_separator);
  1578. time_zone_format.symbol_behind_sign = decode_string(symbol_behind_sign);
  1579. time_zone_format.symbol_behind_separator = decode_string(symbol_behind_separator);
  1580. time_zone_format.gmt_format = decode_string(gmt_format);
  1581. time_zone_format.gmt_zero_format = decode_string(gmt_zero_format);
  1582. return time_zone_format;
  1583. }
  1584. @string_index_type@ symbol_ahead_sign { 0 };
  1585. @string_index_type@ symbol_ahead_separator { 0 };
  1586. @string_index_type@ symbol_behind_sign { 0 };
  1587. @string_index_type@ symbol_behind_separator { 0 };
  1588. @string_index_type@ gmt_format { 0 };
  1589. @string_index_type@ gmt_zero_format { 0 };
  1590. };
  1591. struct DayPeriodData {
  1592. u8 day_period { 0 };
  1593. u8 begin { 0 };
  1594. u8 end { 0 };
  1595. };
  1596. )~~~");
  1597. cldr.unique_formats.generate(generator, "CalendarFormatImpl"sv, "s_calendar_formats"sv, 10);
  1598. cldr.unique_symbol_lists.generate(generator, cldr.unique_strings.type_that_fits(), "s_symbol_lists"sv);
  1599. cldr.unique_calendar_symbols.generate(generator, "CalendarSymbols"sv, "s_calendar_symbols"sv, 10);
  1600. cldr.unique_calendar_symbols_lists.generate(generator, cldr.unique_calendar_symbols.type_that_fits(), "s_calendar_symbol_lists"sv);
  1601. cldr.unique_calendars.generate(generator, "CalendarData"sv, "s_calendars"sv, 10);
  1602. cldr.unique_time_zones.generate(generator, "TimeZoneNames"sv, "s_time_zones"sv, 30);
  1603. cldr.unique_time_zone_lists.generate(generator, cldr.unique_time_zones.type_that_fits(), "s_time_zone_lists"sv);
  1604. cldr.unique_time_zone_formats.generate(generator, "TimeZoneFormatImpl"sv, "s_time_zone_formats"sv, 30);
  1605. cldr.unique_day_periods.generate(generator, "DayPeriodData"sv, "s_day_periods"sv, 30);
  1606. cldr.unique_day_period_lists.generate(generator, cldr.unique_day_periods.type_that_fits(), "s_day_period_lists"sv);
  1607. cldr.unique_hour_cycle_lists.generate(generator, cldr.unique_hour_cycle_lists.type_that_fits(), "s_hour_cycle_lists"sv);
  1608. auto append_calendars = [&](ByteString name, auto const& calendars) {
  1609. generator.set("name", name);
  1610. generator.set("size", ByteString::number(calendars.size()));
  1611. generator.append(R"~~~(
  1612. static constexpr Array<@calendar_index_type@, @size@> @name@ { {)~~~");
  1613. bool first = true;
  1614. for (auto const& calendar_key : cldr.calendars) {
  1615. auto calendar = calendars.find(calendar_key)->value;
  1616. generator.append(first ? " "sv : ", "sv);
  1617. generator.append(ByteString::number(calendar));
  1618. first = false;
  1619. }
  1620. generator.append(" } };");
  1621. };
  1622. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  1623. generator.set("type", type);
  1624. generator.set("name", name);
  1625. generator.set("size", ByteString::number(keys.size()));
  1626. generator.append(R"~~~(
  1627. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1628. bool first = true;
  1629. for (auto const& key : keys) {
  1630. auto const& value = map.find(key)->value;
  1631. auto mapping = mapping_getter(value);
  1632. generator.append(first ? " "sv : ", "sv);
  1633. generator.append(ByteString::number(mapping));
  1634. first = false;
  1635. }
  1636. generator.append(" } };");
  1637. };
  1638. auto locales = cldr.locales.keys();
  1639. quick_sort(locales);
  1640. 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); });
  1641. append_mapping(locales, cldr.locales, cldr.unique_time_zones.type_that_fits(), "s_locale_time_zones"sv, [](auto const& locale) { return locale.time_zones; });
  1642. 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; });
  1643. append_mapping(locales, cldr.locales, cldr.unique_day_periods.type_that_fits(), "s_locale_day_periods"sv, [](auto const& locale) { return locale.day_periods; });
  1644. 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; });
  1645. append_mapping(cldr.minimum_days_regions, cldr.minimum_days, "u8"sv, "s_minimum_days"sv, [](auto minimum_days) { return minimum_days; });
  1646. append_mapping(cldr.first_day_regions, cldr.first_day, "u8"sv, "s_first_day"sv, [](auto first_day) { return to_underlying(first_day); });
  1647. append_mapping(cldr.weekend_start_regions, cldr.weekend_start, "u8"sv, "s_weekend_start"sv, [](auto weekend_start) { return to_underlying(weekend_start); });
  1648. append_mapping(cldr.weekend_end_regions, cldr.weekend_end, "u8"sv, "s_weekend_end"sv, [](auto weekend_end) { return to_underlying(weekend_end); });
  1649. generator.append("\n");
  1650. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) -> ErrorOr<void> {
  1651. HashValueMap<ByteString> hashes;
  1652. TRY(hashes.try_ensure_capacity(values.size()));
  1653. for (auto const& value : values)
  1654. hashes.set(value.hash(), format_identifier(enum_title, value));
  1655. for (auto const& alias : aliases)
  1656. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1657. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1658. return {};
  1659. };
  1660. TRY(append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, cldr.hour_cycle_regions));
  1661. TRY(append_from_string("MinimumDaysRegion"sv, "minimum_days_region"sv, cldr.minimum_days_regions));
  1662. TRY(append_from_string("FirstDayRegion"sv, "first_day_region"sv, cldr.first_day_regions));
  1663. TRY(append_from_string("WeekendStartRegion"sv, "weekend_start_region"sv, cldr.weekend_start_regions));
  1664. TRY(append_from_string("WeekendEndRegion"sv, "weekend_end_region"sv, cldr.weekend_end_regions));
  1665. generator.append(R"~~~(
  1666. static Optional<Calendar> keyword_to_calendar(KeywordCalendar keyword)
  1667. {
  1668. switch (keyword) {)~~~");
  1669. for (auto const& calendar : cldr.calendars) {
  1670. generator.set("name"sv, format_identifier({}, calendar));
  1671. generator.append(R"~~~(
  1672. case KeywordCalendar::@name@:
  1673. return Calendar::@name@;)~~~");
  1674. }
  1675. generator.append(R"~~~(
  1676. default:
  1677. return {};
  1678. }
  1679. }
  1680. Vector<HourCycle> get_regional_hour_cycles(StringView region)
  1681. {
  1682. auto region_value = hour_cycle_region_from_string(region);
  1683. if (!region_value.has_value())
  1684. return {};
  1685. auto region_index = to_underlying(*region_value);
  1686. auto regional_hour_cycles_index = s_hour_cycles.at(region_index);
  1687. auto const& regional_hour_cycles = s_hour_cycle_lists.at(regional_hour_cycles_index);
  1688. Vector<HourCycle> hour_cycles;
  1689. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1690. for (auto hour_cycle : regional_hour_cycles)
  1691. hour_cycles.unchecked_append(static_cast<HourCycle>(hour_cycle));
  1692. return hour_cycles;
  1693. }
  1694. )~~~");
  1695. auto append_regional_lookup = [&](StringView return_type, StringView lookup_type) {
  1696. generator.set("return_type", return_type);
  1697. generator.set("lookup_type", lookup_type);
  1698. generator.append(R"~~~(
  1699. Optional<@return_type@> get_regional_@lookup_type@(StringView region)
  1700. {
  1701. auto region_value = @lookup_type@_region_from_string(region);
  1702. if (!region_value.has_value())
  1703. return {};
  1704. auto region_index = to_underlying(*region_value);
  1705. auto @lookup_type@ = s_@lookup_type@.at(region_index);
  1706. return static_cast<@return_type@>(@lookup_type@);
  1707. }
  1708. )~~~");
  1709. };
  1710. append_regional_lookup("u8"sv, "minimum_days"sv);
  1711. append_regional_lookup("Weekday"sv, "first_day"sv);
  1712. append_regional_lookup("Weekday"sv, "weekend_start"sv);
  1713. append_regional_lookup("Weekday"sv, "weekend_end"sv);
  1714. generator.append(R"~~~(
  1715. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1716. {
  1717. auto locale_value = locale_from_string(locale);
  1718. if (!locale_value.has_value())
  1719. return nullptr;
  1720. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1721. auto const& calendar_indices = s_locale_calendars.at(locale_index);
  1722. auto lookup_calendar = [&](auto calendar_name) -> CalendarData const* {
  1723. auto calendar_keyword = keyword_ca_from_string(calendar_name);
  1724. if (!calendar_keyword.has_value())
  1725. return nullptr;
  1726. auto calendar_value = keyword_to_calendar(*calendar_keyword);
  1727. if (!calendar_value.has_value())
  1728. return nullptr;
  1729. size_t calendar_index = to_underlying(*calendar_value);
  1730. calendar_index = calendar_indices[calendar_index];
  1731. return &s_calendars[calendar_index];
  1732. };
  1733. if (auto const* calendar_data = lookup_calendar(calendar))
  1734. return calendar_data;
  1735. auto default_calendar = get_preferred_keyword_value_for_locale(locale, "ca"sv);
  1736. if (!default_calendar.has_value())
  1737. return nullptr;
  1738. return lookup_calendar(*default_calendar);
  1739. }
  1740. Optional<CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1741. {
  1742. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1743. auto const& formats = s_calendar_formats.at(data->date_formats);
  1744. return formats.to_unicode_calendar_format();
  1745. }
  1746. return {};
  1747. }
  1748. Optional<CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1749. {
  1750. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1751. auto const& formats = s_calendar_formats.at(data->time_formats);
  1752. return formats.to_unicode_calendar_format();
  1753. }
  1754. return {};
  1755. }
  1756. Optional<CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1757. {
  1758. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1759. auto const& formats = s_calendar_formats.at(data->date_time_formats);
  1760. return formats.to_unicode_calendar_format();
  1761. }
  1762. return {};
  1763. }
  1764. Vector<CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1765. {
  1766. Vector<CalendarPattern> result {};
  1767. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1768. auto const& available_formats = s_calendar_pattern_lists.at(data->available_formats);
  1769. result.ensure_capacity(available_formats.size());
  1770. for (auto const& format : available_formats)
  1771. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1772. }
  1773. return result;
  1774. }
  1775. Optional<CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar)
  1776. {
  1777. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1778. auto const& pattern = s_calendar_range_patterns[data->default_range_format];
  1779. return pattern.to_unicode_calendar_range_pattern();
  1780. }
  1781. return {};
  1782. }
  1783. Vector<CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton)
  1784. {
  1785. Vector<CalendarRangePattern> result {};
  1786. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1787. auto const& range_formats = s_calendar_range_pattern_lists.at(data->range_formats);
  1788. for (auto format : range_formats) {
  1789. auto const& pattern = s_calendar_range_patterns[format];
  1790. if (skeleton == decode_string(pattern.skeleton))
  1791. result.append(pattern.to_unicode_calendar_range_pattern());
  1792. }
  1793. }
  1794. return result;
  1795. }
  1796. Vector<CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton)
  1797. {
  1798. Vector<CalendarRangePattern> result {};
  1799. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1800. auto const& range12_formats = s_calendar_range_pattern_lists.at(data->range12_formats);
  1801. for (auto format : range12_formats) {
  1802. auto const& pattern = s_calendar_range_patterns[format];
  1803. if (skeleton == decode_string(pattern.skeleton))
  1804. result.append(pattern.to_unicode_calendar_range_pattern());
  1805. }
  1806. }
  1807. return result;
  1808. }
  1809. static ReadonlySpan<@string_index_type@> find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1810. {
  1811. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1812. auto const& symbols_list = s_calendar_symbol_lists[data->symbols];
  1813. auto symbol_index = to_underlying(symbol);
  1814. auto calendar_symbols_index = symbols_list.at(symbol_index);
  1815. auto const& symbols = s_calendar_symbols.at(calendar_symbols_index);
  1816. @symbol_list_index_type@ symbol_list_index = 0;
  1817. switch (style) {
  1818. case CalendarPatternStyle::Narrow:
  1819. symbol_list_index = symbols.narrow_symbols;
  1820. break;
  1821. case CalendarPatternStyle::Short:
  1822. symbol_list_index = symbols.short_symbols;
  1823. break;
  1824. case CalendarPatternStyle::Long:
  1825. symbol_list_index = symbols.long_symbols;
  1826. break;
  1827. default:
  1828. VERIFY_NOT_REACHED();
  1829. }
  1830. return s_symbol_lists.at(symbol_list_index);
  1831. }
  1832. return {};
  1833. }
  1834. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Era value)
  1835. {
  1836. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1837. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1838. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1839. return decode_string(symbol_index);
  1840. }
  1841. return {};
  1842. }
  1843. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Month value)
  1844. {
  1845. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1846. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1847. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1848. return decode_string(symbol_index);
  1849. }
  1850. return {};
  1851. }
  1852. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Weekday value)
  1853. {
  1854. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1855. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1856. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1857. return decode_string(symbol_index);
  1858. }
  1859. return {};
  1860. }
  1861. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, DayPeriod value)
  1862. {
  1863. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1864. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1865. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1866. return decode_string(symbol_index);
  1867. }
  1868. return {};
  1869. }
  1870. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour)
  1871. {
  1872. auto locale_value = locale_from_string(locale);
  1873. if (!locale_value.has_value())
  1874. return {};
  1875. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1876. auto day_periods_index = s_locale_day_periods[locale_index];
  1877. auto day_periods = s_day_period_lists[day_periods_index];
  1878. for (auto day_period_index : day_periods) {
  1879. auto day_period = s_day_periods[day_period_index];
  1880. bool hour_falls_within_day_period = false;
  1881. if (day_period.begin > day_period.end) {
  1882. if (hour >= day_period.begin)
  1883. hour_falls_within_day_period = true;
  1884. else if (hour <= day_period.end)
  1885. hour_falls_within_day_period = true;
  1886. } else if ((day_period.begin <= hour) && (hour < day_period.end)) {
  1887. hour_falls_within_day_period = true;
  1888. }
  1889. if (hour_falls_within_day_period) {
  1890. auto period = static_cast<DayPeriod>(day_period.day_period);
  1891. return get_calendar_day_period_symbol(locale, calendar, style, period);
  1892. }
  1893. }
  1894. // Fallback to fixed periods if the locale does not have flexible day periods.
  1895. // TR-35 states that the meaning of AM and PM does not change with locale.
  1896. if (hour < 12)
  1897. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::AM);
  1898. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::PM);
  1899. }
  1900. Optional<TimeZoneFormat> get_time_zone_format(StringView locale)
  1901. {
  1902. auto locale_value = locale_from_string(locale);
  1903. if (!locale_value.has_value())
  1904. return {};
  1905. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1906. auto time_zone_format_index = s_locale_time_zone_formats.at(locale_index);
  1907. auto const& time_zone_format = s_time_zone_formats.at(time_zone_format_index);
  1908. return time_zone_format.to_time_zone_format();
  1909. }
  1910. static TimeZoneNames const* find_time_zone_names(StringView locale, StringView time_zone)
  1911. {
  1912. auto locale_value = locale_from_string(locale);
  1913. if (!locale_value.has_value())
  1914. return nullptr;
  1915. auto time_zone_value = ::TimeZone::time_zone_from_string(time_zone);
  1916. if (!time_zone_value.has_value())
  1917. return nullptr;
  1918. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1919. size_t time_zone_index = to_underlying(*time_zone_value);
  1920. auto time_zone_list_index = s_locale_time_zones.at(locale_index);
  1921. auto const& time_zone_list = s_time_zone_lists.at(time_zone_list_index);
  1922. if (time_zone_list.size() <= time_zone_index)
  1923. return nullptr;
  1924. time_zone_index = time_zone_list.at(time_zone_index);
  1925. return &s_time_zones[time_zone_index];
  1926. }
  1927. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style, TimeZone::InDST in_dst)
  1928. {
  1929. if (auto const* data = find_time_zone_names(locale, time_zone); data != nullptr) {
  1930. size_t name_index = 0;
  1931. switch (style) {
  1932. case CalendarPatternStyle::Short:
  1933. name_index = (in_dst == TimeZone::InDST::No) ? data->short_standard_name : data->short_daylight_name;
  1934. break;
  1935. case CalendarPatternStyle::Long:
  1936. name_index = (in_dst == TimeZone::InDST::No) ? data->long_standard_name : data->long_daylight_name;
  1937. break;
  1938. case CalendarPatternStyle::ShortGeneric:
  1939. name_index = data->short_generic_name;
  1940. break;
  1941. case CalendarPatternStyle::LongGeneric:
  1942. name_index = data->long_generic_name;
  1943. break;
  1944. default:
  1945. VERIFY_NOT_REACHED();
  1946. }
  1947. if (name_index != 0)
  1948. return decode_string(name_index);
  1949. }
  1950. return {};
  1951. }
  1952. }
  1953. )~~~");
  1954. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  1955. return {};
  1956. }
  1957. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1958. {
  1959. StringView generated_header_path;
  1960. StringView generated_implementation_path;
  1961. StringView core_path;
  1962. StringView dates_path;
  1963. Core::ArgsParser args_parser;
  1964. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1965. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1966. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1967. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1968. args_parser.parse(arguments);
  1969. auto generated_header_file = TRY(open_file(generated_header_path, Core::File::OpenMode::Write));
  1970. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::File::OpenMode::Write));
  1971. CLDR cldr;
  1972. TRY(parse_all_locales(core_path, dates_path, cldr));
  1973. TRY(generate_unicode_locale_header(*generated_header_file, cldr));
  1974. TRY(generate_unicode_locale_implementation(*generated_implementation_file, cldr));
  1975. return 0;
  1976. }