GenerateDateTimeFormatData.cpp 96 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265226622672268226922702271227222732274227522762277227822792280228122822283228422852286228722882289229022912292229322942295229622972298229923002301230223032304230523062307230823092310231123122313231423152316231723182319232023212322232323242325232623272328232923302331233223332334233523362337233823392340234123422343234423452346234723482349235023512352235323542355235623572358235923602361236223632364236523662367236823692370237123722373237423752376237723782379238023812382238323842385238623872388238923902391239223932394239523962397239823992400240124022403240424052406240724082409241024112412241324142415241624172418241924202421242224232424242524262427242824292430243124322433243424352436243724382439244024412442244324442445244624472448244924502451245224532454245524562457
  1. /*
  2. * Copyright (c) 2021-2022, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "../LibUnicode/GeneratorUtil.h" // FIXME: Move this somewhere common.
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/Find.h>
  10. #include <AK/Format.h>
  11. #include <AK/GenericLexer.h>
  12. #include <AK/HashFunctions.h>
  13. #include <AK/HashMap.h>
  14. #include <AK/JsonObject.h>
  15. #include <AK/JsonParser.h>
  16. #include <AK/JsonValue.h>
  17. #include <AK/LexicalPath.h>
  18. #include <AK/SourceGenerator.h>
  19. #include <AK/String.h>
  20. #include <AK/StringBuilder.h>
  21. #include <AK/Traits.h>
  22. #include <AK/Utf8View.h>
  23. #include <LibCore/ArgsParser.h>
  24. #include <LibCore/DirIterator.h>
  25. #include <LibCore/Stream.h>
  26. #include <LibLocale/DateTimeFormat.h>
  27. #include <LibTimeZone/TimeZone.h>
  28. using StringIndexType = u16;
  29. constexpr auto s_string_index_type = "u16"sv;
  30. using CalendarPatternIndexType = u16;
  31. constexpr auto s_calendar_pattern_index_type = "u16"sv;
  32. using CalendarPatternListIndexType = u8;
  33. constexpr auto s_calendar_pattern_list_index_type = "u8"sv;
  34. using CalendarRangePatternIndexType = u16;
  35. constexpr auto s_calendar_range_pattern_index_type = "u16"sv;
  36. using CalendarRangePatternListIndexType = u16;
  37. constexpr auto s_calendar_range_pattern_list_index_type = "u16"sv;
  38. using CalendarFormatIndexType = u8;
  39. constexpr auto s_calendar_format_index_type = "u8"sv;
  40. using SymbolListIndexType = u16;
  41. constexpr auto s_symbol_list_index_type = "u16"sv;
  42. using CalendarSymbolsIndexType = u16;
  43. constexpr auto s_calendar_symbols_index_type = "u16"sv;
  44. using CalendarSymbolsListIndexType = u8;
  45. constexpr auto s_calendar_symbols_list_index_type = "u8"sv;
  46. using CalendarIndexType = u8;
  47. constexpr auto s_calendar_index_type = "u8"sv;
  48. using TimeZoneNamesIndexType = u16;
  49. constexpr auto s_time_zone_index_type = "u16"sv;
  50. using TimeZoneNamesListIndexType = u8;
  51. constexpr auto s_time_zone_list_index_type = "u8"sv;
  52. using TimeZoneFormatIndexType = u8;
  53. constexpr auto s_time_zone_format_index_type = "u8"sv;
  54. using DayPeriodIndexType = u8;
  55. constexpr auto s_day_period_index_type = "u8"sv;
  56. using DayPeriodListIndexType = u8;
  57. constexpr auto s_day_period_list_index_type = "u8"sv;
  58. using HourCycleListIndexType = u8;
  59. constexpr auto s_hour_cycle_list_index_type = "u8"sv;
  60. struct CalendarPattern : public Locale::CalendarPattern {
  61. bool contains_only_date_fields() const
  62. {
  63. return !day_period.has_value() && !hour.has_value() && !minute.has_value() && !second.has_value() && !fractional_second_digits.has_value() && !time_zone_name.has_value();
  64. }
  65. bool contains_only_time_fields() const
  66. {
  67. return !weekday.has_value() && !era.has_value() && !year.has_value() && !month.has_value() && !day.has_value();
  68. }
  69. unsigned hash() const
  70. {
  71. auto hash = pair_int_hash(pattern_index, pattern12_index);
  72. hash = pair_int_hash(hash, skeleton_index);
  73. auto hash_field = [&](auto const& field) {
  74. if (field.has_value())
  75. hash = pair_int_hash(hash, static_cast<u8>(*field));
  76. else
  77. hash = pair_int_hash(hash, -1);
  78. };
  79. hash_field(era);
  80. hash_field(year);
  81. hash_field(month);
  82. hash_field(weekday);
  83. hash_field(day);
  84. hash_field(day_period);
  85. hash_field(hour);
  86. hash_field(minute);
  87. hash_field(second);
  88. hash_field(fractional_second_digits);
  89. hash_field(time_zone_name);
  90. return hash;
  91. }
  92. bool operator==(CalendarPattern const& other) const
  93. {
  94. return (skeleton_index == other.skeleton_index)
  95. && (pattern_index == other.pattern_index)
  96. && (pattern12_index == other.pattern12_index)
  97. && (era == other.era)
  98. && (year == other.year)
  99. && (month == other.month)
  100. && (weekday == other.weekday)
  101. && (day == other.day)
  102. && (day_period == other.day_period)
  103. && (hour == other.hour)
  104. && (minute == other.minute)
  105. && (second == other.second)
  106. && (fractional_second_digits == other.fractional_second_digits)
  107. && (time_zone_name == other.time_zone_name);
  108. }
  109. StringIndexType skeleton_index { 0 };
  110. StringIndexType pattern_index { 0 };
  111. StringIndexType pattern12_index { 0 };
  112. };
  113. template<>
  114. struct AK::Formatter<CalendarPattern> : Formatter<FormatString> {
  115. ErrorOr<void> format(FormatBuilder& builder, CalendarPattern const& pattern)
  116. {
  117. auto field_to_i8 = [](auto const& field) -> i8 {
  118. if (!field.has_value())
  119. return -1;
  120. return static_cast<i8>(*field);
  121. };
  122. return Formatter<FormatString>::format(builder,
  123. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  124. pattern.skeleton_index,
  125. pattern.pattern_index,
  126. pattern.pattern12_index,
  127. field_to_i8(pattern.era),
  128. field_to_i8(pattern.year),
  129. field_to_i8(pattern.month),
  130. field_to_i8(pattern.weekday),
  131. field_to_i8(pattern.day),
  132. field_to_i8(pattern.day_period),
  133. field_to_i8(pattern.hour),
  134. field_to_i8(pattern.minute),
  135. field_to_i8(pattern.second),
  136. field_to_i8(pattern.fractional_second_digits),
  137. field_to_i8(pattern.time_zone_name));
  138. }
  139. };
  140. template<>
  141. struct AK::Traits<CalendarPattern> : public GenericTraits<CalendarPattern> {
  142. static unsigned hash(CalendarPattern const& c) { return c.hash(); }
  143. };
  144. struct CalendarRangePattern : public CalendarPattern {
  145. unsigned hash() const
  146. {
  147. auto hash = CalendarPattern::hash();
  148. if (field.has_value())
  149. hash = pair_int_hash(hash, static_cast<u8>(*field));
  150. hash = pair_int_hash(hash, start_range);
  151. hash = pair_int_hash(hash, separator);
  152. hash = pair_int_hash(hash, end_range);
  153. return hash;
  154. }
  155. bool operator==(CalendarRangePattern const& other) const
  156. {
  157. if (!CalendarPattern::operator==(other))
  158. return false;
  159. return (field == other.field)
  160. && (start_range == other.start_range)
  161. && (separator == other.separator)
  162. && (end_range == other.end_range);
  163. }
  164. Optional<Locale::CalendarRangePattern::Field> field {};
  165. StringIndexType start_range { 0 };
  166. StringIndexType separator { 0 };
  167. StringIndexType end_range { 0 };
  168. };
  169. template<>
  170. struct AK::Formatter<CalendarRangePattern> : Formatter<FormatString> {
  171. ErrorOr<void> format(FormatBuilder& builder, CalendarRangePattern const& pattern)
  172. {
  173. auto field_to_i8 = [](auto const& field) -> i8 {
  174. if (!field.has_value())
  175. return -1;
  176. return static_cast<i8>(*field);
  177. };
  178. return Formatter<FormatString>::format(builder,
  179. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  180. pattern.skeleton_index,
  181. field_to_i8(pattern.field),
  182. pattern.start_range,
  183. pattern.separator,
  184. pattern.end_range,
  185. field_to_i8(pattern.era),
  186. field_to_i8(pattern.year),
  187. field_to_i8(pattern.month),
  188. field_to_i8(pattern.weekday),
  189. field_to_i8(pattern.day),
  190. field_to_i8(pattern.day_period),
  191. field_to_i8(pattern.hour),
  192. field_to_i8(pattern.minute),
  193. field_to_i8(pattern.second),
  194. field_to_i8(pattern.fractional_second_digits),
  195. field_to_i8(pattern.time_zone_name));
  196. }
  197. };
  198. template<>
  199. struct AK::Traits<CalendarRangePattern> : public GenericTraits<CalendarRangePattern> {
  200. static unsigned hash(CalendarRangePattern const& c) { return c.hash(); }
  201. };
  202. struct CalendarFormat {
  203. unsigned hash() const
  204. {
  205. auto hash = pair_int_hash(full_format, long_format);
  206. hash = pair_int_hash(hash, medium_format);
  207. hash = pair_int_hash(hash, short_format);
  208. return hash;
  209. }
  210. bool operator==(CalendarFormat const& other) const
  211. {
  212. return (full_format == other.full_format)
  213. && (long_format == other.long_format)
  214. && (medium_format == other.medium_format)
  215. && (short_format == other.short_format);
  216. }
  217. CalendarPatternIndexType full_format { 0 };
  218. CalendarPatternIndexType long_format { 0 };
  219. CalendarPatternIndexType medium_format { 0 };
  220. CalendarPatternIndexType short_format { 0 };
  221. };
  222. template<>
  223. struct AK::Formatter<CalendarFormat> : Formatter<FormatString> {
  224. ErrorOr<void> format(FormatBuilder& builder, CalendarFormat const& pattern)
  225. {
  226. return Formatter<FormatString>::format(builder,
  227. "{{ {}, {}, {}, {} }}"sv,
  228. pattern.full_format,
  229. pattern.long_format,
  230. pattern.medium_format,
  231. pattern.short_format);
  232. }
  233. };
  234. template<>
  235. struct AK::Traits<CalendarFormat> : public GenericTraits<CalendarFormat> {
  236. static unsigned hash(CalendarFormat const& c) { return c.hash(); }
  237. };
  238. using SymbolList = Vector<StringIndexType>;
  239. struct CalendarSymbols {
  240. unsigned hash() const
  241. {
  242. auto hash = pair_int_hash(narrow_symbols, short_symbols);
  243. hash = pair_int_hash(hash, long_symbols);
  244. return hash;
  245. }
  246. bool operator==(CalendarSymbols const& other) const
  247. {
  248. return (narrow_symbols == other.narrow_symbols)
  249. && (short_symbols == other.short_symbols)
  250. && (long_symbols == other.long_symbols);
  251. }
  252. SymbolListIndexType narrow_symbols { 0 };
  253. SymbolListIndexType short_symbols { 0 };
  254. SymbolListIndexType long_symbols { 0 };
  255. };
  256. template<>
  257. struct AK::Formatter<CalendarSymbols> : Formatter<FormatString> {
  258. ErrorOr<void> format(FormatBuilder& builder, CalendarSymbols const& symbols)
  259. {
  260. return Formatter<FormatString>::format(builder,
  261. "{{ {}, {}, {} }}"sv,
  262. symbols.narrow_symbols,
  263. symbols.short_symbols,
  264. symbols.long_symbols);
  265. }
  266. };
  267. template<>
  268. struct AK::Traits<CalendarSymbols> : public GenericTraits<CalendarSymbols> {
  269. static unsigned hash(CalendarSymbols const& c) { return c.hash(); }
  270. };
  271. using CalendarPatternList = Vector<CalendarPatternIndexType>;
  272. using CalendarRangePatternList = Vector<CalendarRangePatternIndexType>;
  273. using CalendarSymbolsList = Vector<CalendarSymbolsIndexType>;
  274. struct Calendar {
  275. unsigned hash() const
  276. {
  277. auto hash = int_hash(date_formats);
  278. hash = pair_int_hash(hash, time_formats);
  279. hash = pair_int_hash(hash, date_time_formats);
  280. hash = pair_int_hash(hash, available_formats);
  281. hash = pair_int_hash(hash, default_range_format);
  282. hash = pair_int_hash(hash, range_formats);
  283. hash = pair_int_hash(hash, range12_formats);
  284. hash = pair_int_hash(hash, symbols);
  285. return hash;
  286. }
  287. bool operator==(Calendar const& other) const
  288. {
  289. return (date_formats == other.date_formats)
  290. && (time_formats == other.time_formats)
  291. && (date_time_formats == other.date_time_formats)
  292. && (available_formats == other.available_formats)
  293. && (default_range_format == other.default_range_format)
  294. && (range_formats == other.range_formats)
  295. && (range12_formats == other.range12_formats)
  296. && (symbols == other.symbols);
  297. }
  298. CalendarFormatIndexType date_formats { 0 };
  299. CalendarFormatIndexType time_formats { 0 };
  300. CalendarFormatIndexType date_time_formats { 0 };
  301. CalendarPatternListIndexType available_formats { 0 };
  302. CalendarRangePatternIndexType default_range_format { 0 };
  303. CalendarRangePatternListIndexType range_formats { 0 };
  304. CalendarRangePatternListIndexType range12_formats { 0 };
  305. CalendarSymbolsListIndexType symbols { 0 };
  306. };
  307. template<>
  308. struct AK::Formatter<Calendar> : Formatter<FormatString> {
  309. ErrorOr<void> format(FormatBuilder& builder, Calendar const& calendar)
  310. {
  311. return Formatter<FormatString>::format(builder,
  312. "{{ {}, {}, {}, {}, {}, {}, {}, {} }}"sv,
  313. calendar.date_formats,
  314. calendar.time_formats,
  315. calendar.date_time_formats,
  316. calendar.available_formats,
  317. calendar.default_range_format,
  318. calendar.range_formats,
  319. calendar.range12_formats,
  320. calendar.symbols);
  321. }
  322. };
  323. template<>
  324. struct AK::Traits<Calendar> : public GenericTraits<Calendar> {
  325. static unsigned hash(Calendar const& c) { return c.hash(); }
  326. };
  327. struct TimeZoneNames {
  328. unsigned hash() const
  329. {
  330. auto hash = pair_int_hash(short_standard_name, long_standard_name);
  331. hash = pair_int_hash(hash, short_daylight_name);
  332. hash = pair_int_hash(hash, long_daylight_name);
  333. hash = pair_int_hash(hash, short_generic_name);
  334. hash = pair_int_hash(hash, long_generic_name);
  335. return hash;
  336. }
  337. bool operator==(TimeZoneNames const& other) const
  338. {
  339. return (short_standard_name == other.short_standard_name)
  340. && (long_standard_name == other.long_standard_name)
  341. && (short_daylight_name == other.short_daylight_name)
  342. && (long_daylight_name == other.long_daylight_name)
  343. && (short_generic_name == other.short_generic_name)
  344. && (long_generic_name == other.long_generic_name);
  345. }
  346. StringIndexType short_standard_name { 0 };
  347. StringIndexType long_standard_name { 0 };
  348. StringIndexType short_daylight_name { 0 };
  349. StringIndexType long_daylight_name { 0 };
  350. StringIndexType short_generic_name { 0 };
  351. StringIndexType long_generic_name { 0 };
  352. };
  353. template<>
  354. struct AK::Formatter<TimeZoneNames> : Formatter<FormatString> {
  355. ErrorOr<void> format(FormatBuilder& builder, TimeZoneNames const& time_zone)
  356. {
  357. return Formatter<FormatString>::format(builder,
  358. "{{ {}, {}, {}, {}, {}, {} }}"sv,
  359. time_zone.short_standard_name,
  360. time_zone.long_standard_name,
  361. time_zone.short_daylight_name,
  362. time_zone.long_daylight_name,
  363. time_zone.short_generic_name,
  364. time_zone.long_generic_name);
  365. }
  366. };
  367. template<>
  368. struct AK::Traits<TimeZoneNames> : public GenericTraits<TimeZoneNames> {
  369. static unsigned hash(TimeZoneNames const& t) { return t.hash(); }
  370. };
  371. struct TimeZoneFormat {
  372. unsigned hash() const
  373. {
  374. auto hash = int_hash(symbol_ahead_sign);
  375. hash = pair_int_hash(hash, symbol_ahead_separator);
  376. hash = pair_int_hash(hash, symbol_behind_sign);
  377. hash = pair_int_hash(hash, symbol_behind_separator);
  378. hash = pair_int_hash(hash, gmt_format);
  379. hash = pair_int_hash(hash, gmt_zero_format);
  380. return hash;
  381. }
  382. bool operator==(TimeZoneFormat const& other) const
  383. {
  384. return (symbol_ahead_sign == other.symbol_ahead_sign)
  385. && (symbol_ahead_separator == other.symbol_ahead_separator)
  386. && (symbol_behind_sign == other.symbol_behind_sign)
  387. && (symbol_behind_separator == other.symbol_behind_separator)
  388. && (gmt_format == other.gmt_format)
  389. && (gmt_zero_format == other.gmt_zero_format);
  390. }
  391. StringIndexType symbol_ahead_sign { 0 };
  392. StringIndexType symbol_ahead_separator { 0 };
  393. StringIndexType symbol_behind_sign { 0 };
  394. StringIndexType symbol_behind_separator { 0 };
  395. StringIndexType gmt_format { 0 };
  396. StringIndexType gmt_zero_format { 0 };
  397. };
  398. template<>
  399. struct AK::Formatter<TimeZoneFormat> : Formatter<FormatString> {
  400. ErrorOr<void> format(FormatBuilder& builder, TimeZoneFormat const& time_zone_format)
  401. {
  402. return Formatter<FormatString>::format(builder, "{{ {}, {}, {}, {}, {}, {} }}"sv,
  403. time_zone_format.symbol_ahead_sign,
  404. time_zone_format.symbol_ahead_separator,
  405. time_zone_format.symbol_behind_sign,
  406. time_zone_format.symbol_behind_separator,
  407. time_zone_format.gmt_format,
  408. time_zone_format.gmt_zero_format);
  409. }
  410. };
  411. template<>
  412. struct AK::Traits<TimeZoneFormat> : public GenericTraits<TimeZoneFormat> {
  413. static unsigned hash(TimeZoneFormat const& t) { return t.hash(); }
  414. };
  415. struct DayPeriod {
  416. unsigned hash() const
  417. {
  418. auto hash = int_hash(static_cast<u8>(day_period));
  419. hash = pair_int_hash(hash, begin);
  420. hash = pair_int_hash(hash, end);
  421. return hash;
  422. }
  423. bool operator==(DayPeriod const& other) const
  424. {
  425. return (day_period == other.day_period)
  426. && (begin == other.begin)
  427. && (end == other.end);
  428. }
  429. Locale::DayPeriod day_period {};
  430. u8 begin { 0 };
  431. u8 end { 0 };
  432. };
  433. template<>
  434. struct AK::Formatter<DayPeriod> : Formatter<FormatString> {
  435. ErrorOr<void> format(FormatBuilder& builder, DayPeriod const& day_period)
  436. {
  437. return Formatter<FormatString>::format(builder,
  438. "{{ {}, {}, {} }}"sv,
  439. static_cast<u8>(day_period.day_period),
  440. day_period.begin,
  441. day_period.end);
  442. }
  443. };
  444. template<>
  445. struct AK::Traits<DayPeriod> : public GenericTraits<DayPeriod> {
  446. static unsigned hash(DayPeriod const& d) { return d.hash(); }
  447. };
  448. using TimeZoneNamesList = Vector<TimeZoneNamesIndexType>;
  449. using DayPeriodList = Vector<DayPeriodIndexType>;
  450. using HourCycleList = Vector<Locale::HourCycle>;
  451. template<>
  452. struct AK::Formatter<Locale::HourCycle> : Formatter<FormatString> {
  453. ErrorOr<void> format(FormatBuilder& builder, Locale::HourCycle hour_cycle)
  454. {
  455. return builder.put_u64(to_underlying(hour_cycle));
  456. }
  457. };
  458. struct LocaleData {
  459. HashMap<String, CalendarIndexType> calendars;
  460. TimeZoneNamesListIndexType time_zones { 0 };
  461. TimeZoneFormatIndexType time_zone_formats { 0 };
  462. DayPeriodListIndexType day_periods { 0 };
  463. };
  464. struct CLDR {
  465. UniqueStringStorage<StringIndexType> unique_strings;
  466. UniqueStorage<CalendarPattern, CalendarPatternIndexType> unique_patterns;
  467. UniqueStorage<CalendarPatternList, CalendarPatternListIndexType> unique_pattern_lists;
  468. UniqueStorage<CalendarRangePattern, CalendarRangePatternIndexType> unique_range_patterns;
  469. UniqueStorage<CalendarRangePatternList, CalendarRangePatternListIndexType> unique_range_pattern_lists;
  470. UniqueStorage<CalendarFormat, CalendarFormatIndexType> unique_formats;
  471. UniqueStorage<SymbolList, SymbolListIndexType> unique_symbol_lists;
  472. UniqueStorage<CalendarSymbols, CalendarSymbolsIndexType> unique_calendar_symbols;
  473. UniqueStorage<CalendarSymbolsList, CalendarSymbolsListIndexType> unique_calendar_symbols_lists;
  474. UniqueStorage<Calendar, CalendarIndexType> unique_calendars;
  475. UniqueStorage<TimeZoneNames, TimeZoneNamesIndexType> unique_time_zones;
  476. UniqueStorage<TimeZoneNamesList, TimeZoneNamesListIndexType> unique_time_zone_lists;
  477. UniqueStorage<TimeZoneFormat, TimeZoneFormatIndexType> unique_time_zone_formats;
  478. UniqueStorage<DayPeriod, DayPeriodIndexType> unique_day_periods;
  479. UniqueStorage<DayPeriodList, DayPeriodListIndexType> unique_day_period_lists;
  480. UniqueStorage<HourCycleList, HourCycleListIndexType> unique_hour_cycle_lists;
  481. HashMap<String, LocaleData> locales;
  482. HashMap<String, HourCycleListIndexType> hour_cycles;
  483. Vector<String> hour_cycle_regions;
  484. HashMap<String, u8> minimum_days;
  485. Vector<String> minimum_days_regions;
  486. HashMap<String, Locale::Weekday> first_day;
  487. Vector<String> first_day_regions;
  488. HashMap<String, Locale::Weekday> weekend_start;
  489. Vector<String> weekend_start_regions;
  490. HashMap<String, Locale::Weekday> weekend_end;
  491. Vector<String> weekend_end_regions;
  492. HashMap<String, Vector<TimeZone::TimeZone>> meta_zones;
  493. Vector<String> time_zones { "UTC"sv };
  494. Vector<String> calendars;
  495. };
  496. static Optional<Locale::DayPeriod> day_period_from_string(StringView day_period)
  497. {
  498. if (day_period == "am"sv)
  499. return Locale::DayPeriod::AM;
  500. if (day_period == "pm"sv)
  501. return Locale::DayPeriod::PM;
  502. if (day_period == "noon"sv)
  503. return Locale::DayPeriod::Noon;
  504. if (day_period == "morning1"sv)
  505. return Locale::DayPeriod::Morning1;
  506. if (day_period == "morning2"sv)
  507. return Locale::DayPeriod::Morning2;
  508. if (day_period == "afternoon1"sv)
  509. return Locale::DayPeriod::Afternoon1;
  510. if (day_period == "afternoon2"sv)
  511. return Locale::DayPeriod::Afternoon2;
  512. if (day_period == "evening1"sv)
  513. return Locale::DayPeriod::Evening1;
  514. if (day_period == "evening2"sv)
  515. return Locale::DayPeriod::Evening2;
  516. if (day_period == "night1"sv)
  517. return Locale::DayPeriod::Night1;
  518. if (day_period == "night2"sv)
  519. return Locale::DayPeriod::Night2;
  520. return {};
  521. }
  522. static ErrorOr<void> parse_hour_cycles(String core_path, CLDR& cldr)
  523. {
  524. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  525. LexicalPath time_data_path(move(core_path));
  526. time_data_path = time_data_path.append("supplemental"sv);
  527. time_data_path = time_data_path.append("timeData.json"sv);
  528. auto time_data = TRY(read_json_file(time_data_path.string()));
  529. auto const& supplemental_object = time_data.as_object().get("supplemental"sv);
  530. auto const& time_data_object = supplemental_object.as_object().get("timeData"sv);
  531. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Locale::HourCycle> {
  532. if (hour_cycle == "h"sv)
  533. return Locale::HourCycle::H12;
  534. if (hour_cycle == "H"sv)
  535. return Locale::HourCycle::H23;
  536. if (hour_cycle == "K"sv)
  537. return Locale::HourCycle::H11;
  538. if (hour_cycle == "k"sv)
  539. return Locale::HourCycle::H24;
  540. return {};
  541. };
  542. time_data_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  543. auto allowed_hour_cycles_string = value.as_object().get("_allowed"sv).as_string();
  544. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  545. Vector<Locale::HourCycle> hour_cycles;
  546. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  547. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  548. hour_cycles.append(*hour_cycle);
  549. }
  550. auto hour_cycles_index = cldr.unique_hour_cycle_lists.ensure(move(hour_cycles));
  551. cldr.hour_cycles.set(key, hour_cycles_index);
  552. if (!cldr.hour_cycle_regions.contains_slow(key))
  553. cldr.hour_cycle_regions.append(key);
  554. });
  555. return {};
  556. }
  557. static ErrorOr<void> parse_week_data(String core_path, CLDR& cldr)
  558. {
  559. // https://unicode.org/reports/tr35/tr35-dates.html#Week_Data
  560. LexicalPath week_data_path(move(core_path));
  561. week_data_path = week_data_path.append("supplemental"sv);
  562. week_data_path = week_data_path.append("weekData.json"sv);
  563. auto week_data = TRY(read_json_file(week_data_path.string()));
  564. auto const& supplemental_object = week_data.as_object().get("supplemental"sv);
  565. auto const& week_data_object = supplemental_object.as_object().get("weekData"sv);
  566. auto parse_weekday = [](StringView day) -> Locale::Weekday {
  567. if (day == "sun"sv)
  568. return Locale::Weekday::Sunday;
  569. if (day == "mon"sv)
  570. return Locale::Weekday::Monday;
  571. if (day == "tue"sv)
  572. return Locale::Weekday::Tuesday;
  573. if (day == "wed"sv)
  574. return Locale::Weekday::Wednesday;
  575. if (day == "thu"sv)
  576. return Locale::Weekday::Thursday;
  577. if (day == "fri"sv)
  578. return Locale::Weekday::Friday;
  579. if (day == "sat"sv)
  580. return Locale::Weekday::Saturday;
  581. VERIFY_NOT_REACHED();
  582. };
  583. auto parse_regional_weekdays = [&](auto const& region, auto const& weekday, auto& weekdays_map, auto& weekday_regions) {
  584. if (region.ends_with("alt-variant"sv))
  585. return;
  586. weekdays_map.set(region, parse_weekday(weekday));
  587. if (!weekday_regions.contains_slow(region))
  588. weekday_regions.append(region);
  589. };
  590. auto const& minimum_days_object = week_data_object.as_object().get("minDays"sv);
  591. auto const& first_day_object = week_data_object.as_object().get("firstDay"sv);
  592. auto const& weekend_start_object = week_data_object.as_object().get("weekendStart"sv);
  593. auto const& weekend_end_object = week_data_object.as_object().get("weekendEnd"sv);
  594. minimum_days_object.as_object().for_each_member([&](auto const& region, auto const& value) {
  595. auto minimum_days = value.as_string().template to_uint<u8>();
  596. cldr.minimum_days.set(region, *minimum_days);
  597. if (!cldr.minimum_days_regions.contains_slow(region))
  598. cldr.minimum_days_regions.append(region);
  599. });
  600. first_day_object.as_object().for_each_member([&](auto const& region, auto const& value) {
  601. parse_regional_weekdays(region, value.as_string(), cldr.first_day, cldr.first_day_regions);
  602. });
  603. weekend_start_object.as_object().for_each_member([&](auto const& region, auto const& value) {
  604. parse_regional_weekdays(region, value.as_string(), cldr.weekend_start, cldr.weekend_start_regions);
  605. });
  606. weekend_end_object.as_object().for_each_member([&](auto const& region, auto const& value) {
  607. parse_regional_weekdays(region, value.as_string(), cldr.weekend_end, cldr.weekend_end_regions);
  608. });
  609. return {};
  610. }
  611. static ErrorOr<void> parse_meta_zones(String core_path, CLDR& cldr)
  612. {
  613. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  614. LexicalPath meta_zone_path(move(core_path));
  615. meta_zone_path = meta_zone_path.append("supplemental"sv);
  616. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  617. auto meta_zone = TRY(read_json_file(meta_zone_path.string()));
  618. auto const& supplemental_object = meta_zone.as_object().get("supplemental"sv);
  619. auto const& meta_zone_object = supplemental_object.as_object().get("metaZones"sv);
  620. auto const& meta_zone_array = meta_zone_object.as_object().get("metazones"sv);
  621. meta_zone_array.as_array().for_each([&](JsonValue const& value) {
  622. auto const& mapping = value.as_object().get("mapZone"sv);
  623. auto const& meta_zone = mapping.as_object().get("_other"sv);
  624. auto const& golden_zone = mapping.as_object().get("_type"sv);
  625. if (auto time_zone = TimeZone::time_zone_from_string(golden_zone.as_string()); time_zone.has_value()) {
  626. auto& golden_zones = cldr.meta_zones.ensure(meta_zone.as_string());
  627. golden_zones.append(*time_zone);
  628. }
  629. });
  630. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  631. if (auto time_zone = TimeZone::time_zone_from_string("UTC"sv); time_zone.has_value())
  632. cldr.meta_zones.set("UTC"sv, { *time_zone });
  633. return {};
  634. }
  635. static constexpr auto is_char(char ch)
  636. {
  637. return [ch](auto c) { return c == ch; };
  638. }
  639. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  640. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  641. // extra spaces around the day period. Some example expected removals:
  642. //
  643. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  644. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  645. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  646. static String remove_period_from_pattern(String pattern)
  647. {
  648. auto is_surrounding_space = [&](auto code_point_iterator) {
  649. if (code_point_iterator.done())
  650. return false;
  651. constexpr auto spaces = Array { static_cast<u32>(0x0020), 0x00a0, 0x2009, 0x202f };
  652. return spaces.span().contains_slow(*code_point_iterator);
  653. };
  654. auto is_opening = [&](auto code_point_iterator) {
  655. if (code_point_iterator.done())
  656. return false;
  657. return *code_point_iterator == '{';
  658. };
  659. auto is_closing = [&](auto code_point_iterator) {
  660. if (code_point_iterator.done())
  661. return false;
  662. return *code_point_iterator == '}';
  663. };
  664. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  665. auto index = pattern.find(remove);
  666. if (!index.has_value())
  667. continue;
  668. Utf8View utf8_pattern { pattern };
  669. Utf8CodePointIterator before_removal;
  670. Utf8CodePointIterator after_removal;
  671. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  672. before_removal = it;
  673. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  674. after_removal = it;
  675. if (is_surrounding_space(before_removal) && !is_opening(after_removal)) {
  676. pattern = String::formatted("{}{}",
  677. pattern.substring_view(0, *index - before_removal.underlying_code_point_length_in_bytes()),
  678. pattern.substring_view(*index + remove.length()));
  679. } else if (is_surrounding_space(after_removal) && !is_closing(before_removal)) {
  680. pattern = String::formatted("{}{}",
  681. pattern.substring_view(0, *index),
  682. pattern.substring_view(*index + remove.length() + after_removal.underlying_code_point_length_in_bytes()));
  683. } else {
  684. pattern = String::formatted("{}{}",
  685. pattern.substring_view(0, *index),
  686. pattern.substring_view(*index + remove.length()));
  687. }
  688. }
  689. return pattern;
  690. }
  691. static Optional<CalendarPattern> parse_date_time_pattern_raw(String pattern, String skeleton, CLDR& cldr)
  692. {
  693. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  694. using Locale::CalendarPatternStyle;
  695. CalendarPattern format {};
  696. if (!skeleton.is_empty())
  697. format.skeleton_index = cldr.unique_strings.ensure(move(skeleton));
  698. GenericLexer lexer { pattern };
  699. StringBuilder builder;
  700. bool hour12 { false };
  701. while (!lexer.is_eof()) {
  702. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  703. // processing (this just avoids conflicts with the patterns below).
  704. if (lexer.next_is(is_quote)) {
  705. builder.append(lexer.consume_quoted_string());
  706. continue;
  707. }
  708. auto starting_char = lexer.peek();
  709. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  710. // Era
  711. if (all_of(segment, is_char('G'))) {
  712. builder.append("{era}"sv);
  713. if (segment.length() <= 3)
  714. format.era = CalendarPatternStyle::Short;
  715. else if (segment.length() == 4)
  716. format.era = CalendarPatternStyle::Long;
  717. else
  718. format.era = CalendarPatternStyle::Narrow;
  719. }
  720. // Year
  721. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  722. builder.append("{year}"sv);
  723. if (segment.length() == 2)
  724. format.year = CalendarPatternStyle::TwoDigit;
  725. else
  726. format.year = CalendarPatternStyle::Numeric;
  727. }
  728. // Quarter
  729. else if (all_of(segment, is_any_of("qQ"sv))) {
  730. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  731. return {};
  732. }
  733. // Month
  734. else if (all_of(segment, is_any_of("ML"sv))) {
  735. builder.append("{month}"sv);
  736. if (segment.length() == 1)
  737. format.month = CalendarPatternStyle::Numeric;
  738. else if (segment.length() == 2)
  739. format.month = CalendarPatternStyle::TwoDigit;
  740. else if (segment.length() == 3)
  741. format.month = CalendarPatternStyle::Short;
  742. else if (segment.length() == 4)
  743. format.month = CalendarPatternStyle::Long;
  744. else if (segment.length() == 5)
  745. format.month = CalendarPatternStyle::Narrow;
  746. } else if (all_of(segment, is_char('l'))) {
  747. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  748. return {};
  749. }
  750. // Week
  751. else if (all_of(segment, is_any_of("wW"sv))) {
  752. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  753. return {};
  754. }
  755. // Day
  756. else if (all_of(segment, is_char('d'))) {
  757. builder.append("{day}"sv);
  758. if (segment.length() == 1)
  759. format.day = CalendarPatternStyle::Numeric;
  760. else
  761. format.day = CalendarPatternStyle::TwoDigit;
  762. } else if (all_of(segment, is_any_of("DFg"sv))) {
  763. builder.append("{day}"sv);
  764. format.day = CalendarPatternStyle::Numeric;
  765. }
  766. // Weekday
  767. else if (all_of(segment, is_char('E'))) {
  768. builder.append("{weekday}"sv);
  769. if (segment.length() == 4)
  770. format.weekday = CalendarPatternStyle::Long;
  771. else if (segment.length() == 5)
  772. format.weekday = CalendarPatternStyle::Narrow;
  773. else
  774. format.weekday = CalendarPatternStyle::Short;
  775. } else if (all_of(segment, is_any_of("ec"sv))) {
  776. builder.append("{weekday}"sv);
  777. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  778. // pattern styles are not supported by Intl.DateTimeFormat.
  779. if (segment.length() <= 2)
  780. return {};
  781. if (segment.length() == 4)
  782. format.weekday = CalendarPatternStyle::Long;
  783. else if (segment.length() == 5)
  784. format.weekday = CalendarPatternStyle::Narrow;
  785. else
  786. format.weekday = CalendarPatternStyle::Short;
  787. }
  788. // Period
  789. else if (all_of(segment, is_any_of("ab"sv))) {
  790. builder.append("{ampm}"sv);
  791. hour12 = true;
  792. } else if (all_of(segment, is_char('B'))) {
  793. builder.append("{dayPeriod}"sv);
  794. hour12 = true;
  795. if (segment.length() == 4)
  796. format.day_period = CalendarPatternStyle::Long;
  797. else if (segment.length() == 5)
  798. format.day_period = CalendarPatternStyle::Narrow;
  799. else
  800. format.day_period = CalendarPatternStyle::Short;
  801. }
  802. // Hour
  803. else if (all_of(segment, is_any_of("hHKk"sv))) {
  804. builder.append("{hour}"sv);
  805. if ((segment[0] == 'h') || (segment[0] == 'K'))
  806. hour12 = true;
  807. if (segment.length() == 1)
  808. format.hour = CalendarPatternStyle::Numeric;
  809. else
  810. format.hour = CalendarPatternStyle::TwoDigit;
  811. } else if (all_of(segment, is_any_of("jJC"sv))) {
  812. // TR-35 indicates these should not be used.
  813. return {};
  814. }
  815. // Minute
  816. else if (all_of(segment, is_char('m'))) {
  817. builder.append("{minute}"sv);
  818. if (segment.length() == 1)
  819. format.minute = CalendarPatternStyle::Numeric;
  820. else
  821. format.minute = CalendarPatternStyle::TwoDigit;
  822. }
  823. // Second
  824. else if (all_of(segment, is_char('s'))) {
  825. builder.append("{second}"sv);
  826. if (segment.length() == 1)
  827. format.second = CalendarPatternStyle::Numeric;
  828. else
  829. format.second = CalendarPatternStyle::TwoDigit;
  830. } else if (all_of(segment, is_char('S'))) {
  831. builder.append("{fractionalSecondDigits}"sv);
  832. VERIFY(segment.length() <= 3);
  833. format.fractional_second_digits = static_cast<u8>(segment.length());
  834. } else if (all_of(segment, is_char('A'))) {
  835. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  836. return {};
  837. }
  838. // Zone
  839. else if (all_of(segment, is_any_of("zV"sv))) {
  840. builder.append("{timeZoneName}"sv);
  841. if (segment.length() < 4)
  842. format.time_zone_name = CalendarPatternStyle::Short;
  843. else
  844. format.time_zone_name = CalendarPatternStyle::Long;
  845. } else if (all_of(segment, is_any_of("ZOXx"sv))) {
  846. builder.append("{timeZoneName}"sv);
  847. if (segment.length() < 4)
  848. format.time_zone_name = CalendarPatternStyle::ShortOffset;
  849. else
  850. format.time_zone_name = CalendarPatternStyle::LongOffset;
  851. } else if (all_of(segment, is_char('v'))) {
  852. builder.append("{timeZoneName}"sv);
  853. if (segment.length() < 4)
  854. format.time_zone_name = CalendarPatternStyle::ShortGeneric;
  855. else
  856. format.time_zone_name = CalendarPatternStyle::LongGeneric;
  857. }
  858. // Non-patterns
  859. else {
  860. builder.append(segment);
  861. }
  862. }
  863. pattern = builder.build();
  864. if (hour12) {
  865. format.pattern = remove_period_from_pattern(pattern);
  866. format.pattern12 = move(pattern);
  867. } else {
  868. format.pattern = move(pattern);
  869. }
  870. return format;
  871. }
  872. static Optional<CalendarPatternIndexType> parse_date_time_pattern(String pattern, String skeleton, CLDR& cldr)
  873. {
  874. auto format = parse_date_time_pattern_raw(move(pattern), move(skeleton), cldr);
  875. if (!format.has_value())
  876. return {};
  877. format->pattern_index = cldr.unique_strings.ensure(move(format->pattern));
  878. if (format->pattern12.has_value())
  879. format->pattern12_index = cldr.unique_strings.ensure(format->pattern12.release_value());
  880. return cldr.unique_patterns.ensure(format.release_value());
  881. }
  882. template<typename... Chars>
  883. static constexpr bool char_is_one_of(char ch, Chars&&... chars)
  884. {
  885. return ((ch == chars) || ...);
  886. }
  887. static void parse_interval_patterns(Calendar& calendar, JsonObject const& interval_formats_object, CLDR& cldr)
  888. {
  889. // https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats
  890. CalendarRangePatternList range_formats {};
  891. CalendarRangePatternList range12_formats {};
  892. auto name_of_field = [&](char field) {
  893. if (char_is_one_of(field, 'G'))
  894. return Locale::CalendarRangePattern::Field::Era;
  895. if (char_is_one_of(field, 'y', 'Y', 'u', 'U', 'r'))
  896. return Locale::CalendarRangePattern::Field::Year;
  897. if (char_is_one_of(field, 'M', 'L'))
  898. return Locale::CalendarRangePattern::Field::Month;
  899. if (char_is_one_of(field, 'd', 'D', 'F', 'g'))
  900. return Locale::CalendarRangePattern::Field::Day;
  901. if (char_is_one_of(field, 'a', 'b'))
  902. return Locale::CalendarRangePattern::Field::AmPm;
  903. if (char_is_one_of(field, 'B'))
  904. return Locale::CalendarRangePattern::Field::DayPeriod;
  905. if (char_is_one_of(field, 'h', 'H', 'K', 'k'))
  906. return Locale::CalendarRangePattern::Field::Hour;
  907. if (char_is_one_of(field, 'm'))
  908. return Locale::CalendarRangePattern::Field::Minute;
  909. if (char_is_one_of(field, 's'))
  910. return Locale::CalendarRangePattern::Field::Second;
  911. if (char_is_one_of(field, 'S'))
  912. return Locale::CalendarRangePattern::Field::FractionalSecondDigits;
  913. VERIFY_NOT_REACHED();
  914. };
  915. auto split_default_range_pattern = [&](auto skeleton, auto const& pattern) {
  916. auto start_range_end = pattern.find('}').value() + 1;
  917. auto end_range_begin = pattern.find_last('{').value();
  918. auto start_range = pattern.substring_view(0, start_range_end);
  919. auto separator = pattern.substring_view(start_range_end, end_range_begin - start_range_end);
  920. auto end_range = pattern.substring_view(end_range_begin);
  921. CalendarRangePattern format {};
  922. format.skeleton_index = cldr.unique_strings.ensure(skeleton);
  923. format.start_range = cldr.unique_strings.ensure(start_range);
  924. format.separator = cldr.unique_strings.ensure(separator);
  925. format.end_range = cldr.unique_strings.ensure(end_range);
  926. return format;
  927. };
  928. auto split_range_pattern = [&](auto skeleton, auto field, auto const& pattern, auto const& parsed_fields) {
  929. HashMap<StringView, size_t> partitions;
  930. StringView last_partition;
  931. auto begin_index = pattern.find('{');
  932. size_t end_index = 0;
  933. while (begin_index.has_value()) {
  934. end_index = pattern.find('}', *begin_index).value();
  935. auto partition = pattern.substring_view(*begin_index, end_index - *begin_index);
  936. if (partitions.contains(partition))
  937. break;
  938. partitions.set(partition, *begin_index);
  939. last_partition = partition;
  940. begin_index = pattern.find('{', end_index + 1);
  941. }
  942. VERIFY(begin_index.has_value() && !last_partition.is_empty());
  943. auto start_range_end = partitions.get(last_partition).value() + last_partition.length() + 1;
  944. auto start_range = pattern.substring_view(0, start_range_end);
  945. auto separator = pattern.substring_view(start_range_end, *begin_index - start_range_end);
  946. auto end_range = pattern.substring_view(*begin_index);
  947. CalendarRangePattern format {};
  948. format.skeleton_index = cldr.unique_strings.ensure(skeleton);
  949. format.field = field;
  950. format.start_range = cldr.unique_strings.ensure(start_range);
  951. format.separator = cldr.unique_strings.ensure(separator);
  952. format.end_range = cldr.unique_strings.ensure(end_range);
  953. format.for_each_calendar_field_zipped_with(parsed_fields, [](auto& format_field, auto const& parsed_field, auto) {
  954. format_field = parsed_field;
  955. });
  956. return format;
  957. };
  958. interval_formats_object.for_each_member([&](auto const& skeleton, auto const& value) {
  959. if (skeleton == "intervalFormatFallback"sv) {
  960. auto range_format = split_default_range_pattern(skeleton, value.as_string());
  961. calendar.default_range_format = cldr.unique_range_patterns.ensure(move(range_format));
  962. return;
  963. }
  964. value.as_object().for_each_member([&](auto const& field, auto const& pattern) {
  965. if (field.ends_with("alt-variant"sv))
  966. return;
  967. VERIFY(field.length() == 1);
  968. auto name = name_of_field(field[0]);
  969. auto format = parse_date_time_pattern_raw(pattern.as_string(), skeleton, cldr).release_value();
  970. auto range_format = split_range_pattern(skeleton, name, format.pattern, format);
  971. range_formats.append(cldr.unique_range_patterns.ensure(move(range_format)));
  972. if (format.pattern12.has_value()) {
  973. auto range12_pattern = split_range_pattern(skeleton, name, *format.pattern12, format);
  974. range12_formats.append(cldr.unique_range_patterns.ensure(move(range12_pattern)));
  975. } else {
  976. range12_formats.append(range_formats.last());
  977. }
  978. });
  979. });
  980. calendar.range_formats = cldr.unique_range_pattern_lists.ensure(move(range_formats));
  981. calendar.range12_formats = cldr.unique_range_pattern_lists.ensure(move(range12_formats));
  982. }
  983. static void generate_default_patterns(CalendarPatternList& formats, CLDR& cldr)
  984. {
  985. // For compatibility with ICU, we generate a list of default patterns for every locale:
  986. // https://github.com/unicode-org/icu/blob/release-71-1/icu4c/source/i18n/dtptngen.cpp#L1343-L1354=
  987. 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 };
  988. for (auto pattern : default_patterns) {
  989. auto index = parse_date_time_pattern(pattern, pattern, cldr);
  990. VERIFY(index.has_value());
  991. if (!formats.contains_slow(*index))
  992. formats.append(*index);
  993. }
  994. }
  995. static void generate_missing_patterns(Calendar& calendar, CalendarPatternList& formats, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, CLDR& cldr)
  996. {
  997. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  998. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  999. auto pattern = cldr.unique_strings.get(format);
  1000. auto time_pattern = cldr.unique_strings.get(time_format);
  1001. auto date_pattern = cldr.unique_strings.get(date_format);
  1002. auto new_pattern = pattern.replace("{0}"sv, time_pattern, ReplaceMode::FirstOnly).replace("{1}"sv, date_pattern, ReplaceMode::FirstOnly);
  1003. return cldr.unique_strings.ensure(move(new_pattern));
  1004. };
  1005. auto inject_fractional_second_digits = [&](auto format) {
  1006. auto pattern = cldr.unique_strings.get(format);
  1007. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv, ReplaceMode::FirstOnly);
  1008. return cldr.unique_strings.ensure(move(new_pattern));
  1009. };
  1010. auto append_if_unique = [&](auto format) {
  1011. auto format_index = cldr.unique_patterns.ensure(move(format));
  1012. if (!formats.contains_slow(format_index))
  1013. formats.append(format_index);
  1014. };
  1015. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  1016. for (auto const& format : date_formats)
  1017. append_if_unique(format);
  1018. for (auto const& format : time_formats) {
  1019. append_if_unique(format);
  1020. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  1021. auto new_format = format;
  1022. new_format.fractional_second_digits = 2;
  1023. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  1024. if (new_format.pattern12_index != 0)
  1025. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  1026. time_formats_with_fractional_second_digits.append(new_format);
  1027. append_if_unique(move(new_format));
  1028. }
  1029. }
  1030. time_formats.extend(move(time_formats_with_fractional_second_digits));
  1031. for (auto const& date_format : date_formats) {
  1032. auto const& date_time_formats = cldr.unique_formats.get(calendar.date_time_formats);
  1033. CalendarPatternIndexType date_time_format_index = 0;
  1034. if (date_format.month == Locale::CalendarPatternStyle::Long) {
  1035. if (date_format.weekday.has_value())
  1036. date_time_format_index = date_time_formats.full_format;
  1037. else
  1038. date_time_format_index = date_time_formats.long_format;
  1039. } else if (date_format.month == Locale::CalendarPatternStyle::Short) {
  1040. date_time_format_index = date_time_formats.medium_format;
  1041. } else {
  1042. date_time_format_index = date_time_formats.short_format;
  1043. }
  1044. for (auto const& time_format : time_formats) {
  1045. auto format = cldr.unique_patterns.get(date_time_format_index);
  1046. if (time_format.pattern12_index != 0)
  1047. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  1048. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  1049. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  1050. if (date_field.has_value())
  1051. field = date_field;
  1052. });
  1053. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  1054. if (time_field.has_value())
  1055. field = time_field;
  1056. });
  1057. append_if_unique(move(format));
  1058. }
  1059. }
  1060. }
  1061. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, CLDR& cldr)
  1062. {
  1063. auto create_symbol_lists = [](size_t size) {
  1064. SymbolList narrow_symbol_list;
  1065. SymbolList short_symbol_list;
  1066. SymbolList long_symbol_list;
  1067. narrow_symbol_list.resize(size);
  1068. short_symbol_list.resize(size);
  1069. long_symbol_list.resize(size);
  1070. return Array<SymbolList, 3> { {
  1071. move(narrow_symbol_list),
  1072. move(short_symbol_list),
  1073. move(long_symbol_list),
  1074. } };
  1075. };
  1076. CalendarSymbolsList symbols_list {};
  1077. auto store_symbol_lists = [&](auto symbol, auto symbol_lists) {
  1078. auto symbol_index = to_underlying(symbol);
  1079. if (symbol_index >= symbols_list.size())
  1080. symbols_list.resize(symbol_index + 1);
  1081. CalendarSymbols symbols {};
  1082. symbols.narrow_symbols = cldr.unique_symbol_lists.ensure(move(symbol_lists[0]));
  1083. symbols.short_symbols = cldr.unique_symbol_lists.ensure(move(symbol_lists[1]));
  1084. symbols.long_symbols = cldr.unique_symbol_lists.ensure(move(symbol_lists[2]));
  1085. auto calendar_symbols_index = cldr.unique_calendar_symbols.ensure(move(symbols));
  1086. symbols_list[symbol_index] = calendar_symbols_index;
  1087. };
  1088. auto parse_era_symbols = [&](auto const& symbols_object) {
  1089. auto const& narrow_symbols = symbols_object.get("eraNarrow"sv).as_object();
  1090. auto const& short_symbols = symbols_object.get("eraAbbr"sv).as_object();
  1091. auto const& long_symbols = symbols_object.get("eraNames"sv).as_object();
  1092. auto symbol_lists = create_symbol_lists(2);
  1093. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1094. if (auto key_index = key.to_uint(); key_index.has_value())
  1095. symbols[*key_index] = cldr.unique_strings.ensure(move(symbol));
  1096. };
  1097. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1098. append_symbol(symbol_lists[0], key, value.as_string());
  1099. });
  1100. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1101. append_symbol(symbol_lists[1], key, value.as_string());
  1102. });
  1103. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1104. append_symbol(symbol_lists[2], key, value.as_string());
  1105. });
  1106. store_symbol_lists(Locale::CalendarSymbol::Era, move(symbol_lists));
  1107. };
  1108. auto parse_month_symbols = [&](auto const& symbols_object) {
  1109. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1110. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1111. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1112. auto symbol_lists = create_symbol_lists(12);
  1113. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1114. auto key_index = key.to_uint().value() - 1;
  1115. symbols[key_index] = cldr.unique_strings.ensure(move(symbol));
  1116. };
  1117. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1118. append_symbol(symbol_lists[0], key, value.as_string());
  1119. });
  1120. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1121. append_symbol(symbol_lists[1], key, value.as_string());
  1122. });
  1123. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1124. append_symbol(symbol_lists[2], key, value.as_string());
  1125. });
  1126. store_symbol_lists(Locale::CalendarSymbol::Month, move(symbol_lists));
  1127. };
  1128. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  1129. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1130. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1131. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1132. auto symbol_lists = create_symbol_lists(7);
  1133. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1134. if (key == "sun"sv)
  1135. symbols[to_underlying(Locale::Weekday::Sunday)] = cldr.unique_strings.ensure(move(symbol));
  1136. else if (key == "mon"sv)
  1137. symbols[to_underlying(Locale::Weekday::Monday)] = cldr.unique_strings.ensure(move(symbol));
  1138. else if (key == "tue"sv)
  1139. symbols[to_underlying(Locale::Weekday::Tuesday)] = cldr.unique_strings.ensure(move(symbol));
  1140. else if (key == "wed"sv)
  1141. symbols[to_underlying(Locale::Weekday::Wednesday)] = cldr.unique_strings.ensure(move(symbol));
  1142. else if (key == "thu"sv)
  1143. symbols[to_underlying(Locale::Weekday::Thursday)] = cldr.unique_strings.ensure(move(symbol));
  1144. else if (key == "fri"sv)
  1145. symbols[to_underlying(Locale::Weekday::Friday)] = cldr.unique_strings.ensure(move(symbol));
  1146. else if (key == "sat"sv)
  1147. symbols[to_underlying(Locale::Weekday::Saturday)] = cldr.unique_strings.ensure(move(symbol));
  1148. };
  1149. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1150. append_symbol(symbol_lists[0], key, value.as_string());
  1151. });
  1152. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1153. append_symbol(symbol_lists[1], key, value.as_string());
  1154. });
  1155. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1156. append_symbol(symbol_lists[2], key, value.as_string());
  1157. });
  1158. store_symbol_lists(Locale::CalendarSymbol::Weekday, move(symbol_lists));
  1159. };
  1160. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  1161. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1162. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1163. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1164. auto symbol_lists = create_symbol_lists(11);
  1165. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1166. if (auto day_period = day_period_from_string(key); day_period.has_value())
  1167. symbols[to_underlying(*day_period)] = cldr.unique_strings.ensure(move(symbol));
  1168. };
  1169. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1170. append_symbol(symbol_lists[0], key, value.as_string());
  1171. });
  1172. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1173. append_symbol(symbol_lists[1], key, value.as_string());
  1174. });
  1175. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1176. append_symbol(symbol_lists[2], key, value.as_string());
  1177. });
  1178. store_symbol_lists(Locale::CalendarSymbol::DayPeriod, move(symbol_lists));
  1179. };
  1180. parse_era_symbols(calendar_object.get("eras"sv).as_object());
  1181. parse_month_symbols(calendar_object.get("months"sv).as_object().get("format"sv).as_object());
  1182. parse_weekday_symbols(calendar_object.get("days"sv).as_object().get("format"sv).as_object());
  1183. parse_day_period_symbols(calendar_object.get("dayPeriods"sv).as_object().get("format"sv).as_object());
  1184. calendar.symbols = cldr.unique_calendar_symbols_lists.ensure(move(symbols_list));
  1185. }
  1186. static ErrorOr<void> parse_calendars(String locale_calendars_path, CLDR& cldr, LocaleData& locale)
  1187. {
  1188. LexicalPath calendars_path(move(locale_calendars_path));
  1189. if (!calendars_path.basename().starts_with("ca-"sv))
  1190. return {};
  1191. auto calendars = TRY(read_json_file(calendars_path.string()));
  1192. auto const& main_object = calendars.as_object().get("main"sv);
  1193. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  1194. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1195. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  1196. auto parse_patterns = [&](auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) {
  1197. auto parse_pattern = [&](auto name) {
  1198. auto format = patterns_object.get(name);
  1199. auto skeleton = skeletons_object.get(name);
  1200. auto format_index = parse_date_time_pattern(format.as_string(), skeleton.as_string_or(String::empty()), cldr).value();
  1201. if (patterns)
  1202. patterns->append(cldr.unique_patterns.get(format_index));
  1203. return format_index;
  1204. };
  1205. CalendarFormat formats {};
  1206. formats.full_format = parse_pattern("full"sv);
  1207. formats.long_format = parse_pattern("long"sv);
  1208. formats.medium_format = parse_pattern("medium"sv);
  1209. formats.short_format = parse_pattern("short"sv);
  1210. return cldr.unique_formats.ensure(move(formats));
  1211. };
  1212. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  1213. // The generic calendar is not a supported Unicode calendar key, so skip it:
  1214. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  1215. if (calendar_name == "generic"sv)
  1216. return;
  1217. Calendar calendar {};
  1218. CalendarPatternList available_formats {};
  1219. if (!cldr.calendars.contains_slow(calendar_name))
  1220. cldr.calendars.append(calendar_name);
  1221. Vector<CalendarPattern> date_formats;
  1222. Vector<CalendarPattern> time_formats;
  1223. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  1224. auto const& date_skeletons_object = value.as_object().get("dateSkeletons"sv);
  1225. calendar.date_formats = parse_patterns(date_formats_object.as_object(), date_skeletons_object.as_object(), &date_formats);
  1226. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  1227. auto const& time_skeletons_object = value.as_object().get("timeSkeletons"sv);
  1228. calendar.time_formats = parse_patterns(time_formats_object.as_object(), time_skeletons_object.as_object(), &time_formats);
  1229. auto const& standard_date_time_formats_object = value.as_object().get("dateTimeFormats-atTime"sv).as_object().get("standard"sv);
  1230. calendar.date_time_formats = parse_patterns(standard_date_time_formats_object.as_object(), JsonObject {}, nullptr);
  1231. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  1232. auto const& available_formats_object = date_time_formats_object.as_object().get("availableFormats"sv);
  1233. available_formats_object.as_object().for_each_member([&](auto const& skeleton, JsonValue const& pattern) {
  1234. auto pattern_index = parse_date_time_pattern(pattern.as_string(), skeleton, cldr);
  1235. if (!pattern_index.has_value())
  1236. return;
  1237. auto const& format = cldr.unique_patterns.get(*pattern_index);
  1238. if (format.contains_only_date_fields())
  1239. date_formats.append(format);
  1240. else if (format.contains_only_time_fields())
  1241. time_formats.append(format);
  1242. if (!available_formats.contains_slow(*pattern_index))
  1243. available_formats.append(*pattern_index);
  1244. });
  1245. auto const& interval_formats_object = date_time_formats_object.as_object().get("intervalFormats"sv);
  1246. parse_interval_patterns(calendar, interval_formats_object.as_object(), cldr);
  1247. generate_default_patterns(available_formats, cldr);
  1248. generate_missing_patterns(calendar, available_formats, move(date_formats), move(time_formats), cldr);
  1249. parse_calendar_symbols(calendar, value.as_object(), cldr);
  1250. calendar.available_formats = cldr.unique_pattern_lists.ensure(move(available_formats));
  1251. locale.calendars.set(calendar_name, cldr.unique_calendars.ensure(move(calendar)));
  1252. });
  1253. return {};
  1254. }
  1255. static ErrorOr<void> parse_time_zone_names(String locale_time_zone_names_path, CLDR& cldr, LocaleData& locale)
  1256. {
  1257. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  1258. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  1259. auto time_zone_names = TRY(read_json_file(time_zone_names_path.string()));
  1260. auto const& main_object = time_zone_names.as_object().get("main"sv);
  1261. auto const& locale_object = main_object.as_object().get(time_zone_names_path.parent().basename());
  1262. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1263. auto const& time_zone_names_object = dates_object.as_object().get("timeZoneNames"sv);
  1264. auto const& meta_zone_object = time_zone_names_object.as_object().get("metazone"sv);
  1265. auto const& hour_format_string = time_zone_names_object.as_object().get("hourFormat"sv);
  1266. auto const& gmt_format_string = time_zone_names_object.as_object().get("gmtFormat"sv);
  1267. auto const& gmt_zero_format_string = time_zone_names_object.as_object().get("gmtZeroFormat"sv);
  1268. if (meta_zone_object.is_null())
  1269. return {};
  1270. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object, StringView key) -> Optional<StringIndexType> {
  1271. auto const& names = meta_zone_object.get(type);
  1272. if (!names.is_object())
  1273. return {};
  1274. auto const& name = names.as_object().get(key);
  1275. if (name.is_string())
  1276. return cldr.unique_strings.ensure(name.as_string());
  1277. return {};
  1278. };
  1279. auto parse_hour_format = [&](auto const& format, auto& time_zone_formats) {
  1280. auto hour_formats = format.split_view(';');
  1281. auto hour_format_ahead_start = hour_formats[0].find('H').value();
  1282. auto separator_ahead_start = hour_formats[0].find_last('H').value() + 1;
  1283. auto separator_ahead_end = hour_formats[0].find('m').value();
  1284. auto hour_format_behind_start = hour_formats[1].find('H').value();
  1285. auto separator_behind_start = hour_formats[1].find_last('H').value() + 1;
  1286. auto separator_behind_end = hour_formats[1].find('m').value();
  1287. auto symbol_ahead_sign = hour_formats[0].substring_view(0, hour_format_ahead_start);
  1288. auto symbol_ahead_separator = hour_formats[0].substring_view(separator_ahead_start, separator_ahead_end - separator_ahead_start);
  1289. auto symbol_behind_sign = hour_formats[1].substring_view(0, hour_format_behind_start);
  1290. auto symbol_behind_separator = hour_formats[1].substring_view(separator_behind_start, separator_behind_end - separator_behind_start);
  1291. time_zone_formats.symbol_ahead_sign = cldr.unique_strings.ensure(symbol_ahead_sign);
  1292. time_zone_formats.symbol_ahead_separator = cldr.unique_strings.ensure(symbol_ahead_separator);
  1293. time_zone_formats.symbol_behind_sign = cldr.unique_strings.ensure(symbol_behind_sign);
  1294. time_zone_formats.symbol_behind_separator = cldr.unique_strings.ensure(symbol_behind_separator);
  1295. };
  1296. TimeZoneNamesList time_zones;
  1297. TimeZoneFormat time_zone_formats {};
  1298. parse_hour_format(hour_format_string.as_string(), time_zone_formats);
  1299. time_zone_formats.gmt_format = cldr.unique_strings.ensure(gmt_format_string.as_string());
  1300. time_zone_formats.gmt_zero_format = cldr.unique_strings.ensure(gmt_zero_format_string.as_string());
  1301. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  1302. auto golden_zones = cldr.meta_zones.find(meta_zone);
  1303. if (golden_zones == cldr.meta_zones.end())
  1304. return;
  1305. TimeZoneNames time_zone_names {};
  1306. if (auto name = parse_name("long"sv, meta_zone_object, "standard"sv); name.has_value())
  1307. time_zone_names.long_standard_name = name.value();
  1308. if (auto name = parse_name("short"sv, meta_zone_object, "standard"sv); name.has_value())
  1309. time_zone_names.short_standard_name = name.value();
  1310. if (auto name = parse_name("long"sv, meta_zone_object, "daylight"sv); name.has_value())
  1311. time_zone_names.long_daylight_name = name.value();
  1312. if (auto name = parse_name("short"sv, meta_zone_object, "daylight"sv); name.has_value())
  1313. time_zone_names.short_daylight_name = name.value();
  1314. if (auto name = parse_name("long"sv, meta_zone_object, "generic"sv); name.has_value())
  1315. time_zone_names.long_generic_name = name.value();
  1316. if (auto name = parse_name("short"sv, meta_zone_object, "generic"sv); name.has_value())
  1317. time_zone_names.short_generic_name = name.value();
  1318. auto time_zone_index = cldr.unique_time_zones.ensure(move(time_zone_names));
  1319. for (auto golden_zone : golden_zones->value) {
  1320. auto time_zone = to_underlying(golden_zone);
  1321. if (time_zone >= time_zones.size())
  1322. time_zones.resize(time_zone + 1);
  1323. time_zones[time_zone] = time_zone_index;
  1324. }
  1325. };
  1326. meta_zone_object.as_object().for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  1327. parse_time_zone(meta_zone, value.as_object());
  1328. });
  1329. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  1330. auto const& zone_object = time_zone_names_object.as_object().get("zone"sv);
  1331. auto const& etc_object = zone_object.as_object().get("Etc"sv);
  1332. auto const& utc_object = etc_object.as_object().get("UTC"sv);
  1333. parse_time_zone("UTC"sv, utc_object.as_object());
  1334. locale.time_zones = cldr.unique_time_zone_lists.ensure(move(time_zones));
  1335. locale.time_zone_formats = cldr.unique_time_zone_formats.ensure(move(time_zone_formats));
  1336. return {};
  1337. }
  1338. static ErrorOr<void> parse_day_periods(String core_path, CLDR& cldr)
  1339. {
  1340. // https://unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rule_Sets
  1341. LexicalPath day_periods_path(move(core_path));
  1342. day_periods_path = day_periods_path.append("supplemental"sv);
  1343. day_periods_path = day_periods_path.append("dayPeriods.json"sv);
  1344. auto locale_day_periods = TRY(read_json_file(day_periods_path.string()));
  1345. auto const& supplemental_object = locale_day_periods.as_object().get("supplemental"sv);
  1346. auto const& day_periods_object = supplemental_object.as_object().get("dayPeriodRuleSet"sv);
  1347. auto parse_hour = [](auto const& time) {
  1348. auto hour_end_index = time.find(':').value();
  1349. // The times are in the form "hours:minutes", but we only need the hour segment.
  1350. // TR-35 explicitly states that minutes other than :00 must not be used.
  1351. VERIFY(time.substring_view(hour_end_index) == ":00"sv);
  1352. auto hour = time.substring_view(0, hour_end_index);
  1353. return hour.template to_uint<u8>().value();
  1354. };
  1355. auto parse_day_period = [&](auto const& symbol, auto const& ranges) -> Optional<DayPeriod> {
  1356. if (!ranges.has("_from"sv))
  1357. return {};
  1358. auto day_period = day_period_from_string(symbol);
  1359. if (!day_period.has_value())
  1360. return {};
  1361. auto begin = parse_hour(ranges.get("_from"sv).as_string());
  1362. auto end = parse_hour(ranges.get("_before"sv).as_string());
  1363. return DayPeriod { *day_period, begin, end };
  1364. };
  1365. day_periods_object.as_object().for_each_member([&](auto const& language, JsonValue const& value) {
  1366. auto locale = cldr.locales.find(language);
  1367. if (locale == cldr.locales.end())
  1368. return;
  1369. DayPeriodList day_periods;
  1370. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& ranges) {
  1371. if (auto day_period = parse_day_period(symbol, ranges.as_object()); day_period.has_value()) {
  1372. auto day_period_index = cldr.unique_day_periods.ensure(day_period.release_value());
  1373. day_periods.append(day_period_index);
  1374. }
  1375. });
  1376. locale->value.day_periods = cldr.unique_day_period_lists.ensure(move(day_periods));
  1377. });
  1378. return {};
  1379. }
  1380. static ErrorOr<void> parse_all_locales(String core_path, String dates_path, CLDR& cldr)
  1381. {
  1382. TRY(parse_hour_cycles(core_path, cldr));
  1383. TRY(parse_week_data(core_path, cldr));
  1384. TRY(parse_meta_zones(core_path, cldr));
  1385. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  1386. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  1387. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(cldr.unique_strings, LexicalPath::basename(path)));
  1388. StringBuilder builder;
  1389. builder.append(cldr.unique_strings.get(parsed_locale.language));
  1390. if (auto script = cldr.unique_strings.get(parsed_locale.script); !script.is_empty())
  1391. builder.appendff("-{}", script);
  1392. if (auto region = cldr.unique_strings.get(parsed_locale.region); !region.is_empty())
  1393. builder.appendff("-{}", region);
  1394. return builder.build();
  1395. };
  1396. while (dates_iterator.has_next()) {
  1397. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  1398. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  1399. auto language = TRY(remove_variants_from_path(dates_path));
  1400. auto& locale = cldr.locales.ensure(language);
  1401. while (calendars_iterator.has_next()) {
  1402. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  1403. TRY(parse_calendars(move(calendars_path), cldr, locale));
  1404. }
  1405. TRY(parse_time_zone_names(move(dates_path), cldr, locale));
  1406. }
  1407. TRY(parse_day_periods(move(core_path), cldr));
  1408. return {};
  1409. }
  1410. static String format_identifier(StringView owner, String identifier)
  1411. {
  1412. identifier = identifier.replace("-"sv, "_"sv, ReplaceMode::All);
  1413. identifier = identifier.replace("/"sv, "_"sv, ReplaceMode::All);
  1414. if (all_of(identifier, is_ascii_digit))
  1415. return String::formatted("{}_{}", owner[0], identifier);
  1416. if (is_ascii_lower_alpha(identifier[0]))
  1417. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  1418. return identifier;
  1419. }
  1420. static ErrorOr<void> generate_unicode_locale_header(Core::Stream::BufferedFile& file, CLDR& cldr)
  1421. {
  1422. StringBuilder builder;
  1423. SourceGenerator generator { builder };
  1424. generator.append(R"~~~(
  1425. #pragma once
  1426. #include <AK/Types.h>
  1427. namespace Locale {
  1428. )~~~");
  1429. generate_enum(generator, format_identifier, "Calendar"sv, {}, cldr.calendars);
  1430. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, cldr.hour_cycle_regions);
  1431. generate_enum(generator, format_identifier, "MinimumDaysRegion"sv, {}, cldr.minimum_days_regions);
  1432. generate_enum(generator, format_identifier, "FirstDayRegion"sv, {}, cldr.first_day_regions);
  1433. generate_enum(generator, format_identifier, "WeekendStartRegion"sv, {}, cldr.weekend_start_regions);
  1434. generate_enum(generator, format_identifier, "WeekendEndRegion"sv, {}, cldr.weekend_end_regions);
  1435. generator.append(R"~~~(
  1436. }
  1437. )~~~");
  1438. TRY(file.write(generator.as_string_view().bytes()));
  1439. return {};
  1440. }
  1441. static ErrorOr<void> generate_unicode_locale_implementation(Core::Stream::BufferedFile& file, CLDR& cldr)
  1442. {
  1443. StringBuilder builder;
  1444. SourceGenerator generator { builder };
  1445. generator.set("string_index_type"sv, s_string_index_type);
  1446. generator.set("calendar_pattern_index_type"sv, s_calendar_pattern_index_type);
  1447. generator.set("calendar_pattern_list_index_type"sv, s_calendar_pattern_list_index_type);
  1448. generator.set("calendar_range_pattern_index_type"sv, s_calendar_range_pattern_index_type);
  1449. generator.set("calendar_range_pattern_list_index_type"sv, s_calendar_range_pattern_list_index_type);
  1450. generator.set("calendar_format_index_type"sv, s_calendar_format_index_type);
  1451. generator.set("symbol_list_index_type"sv, s_symbol_list_index_type);
  1452. generator.set("calendar_symbols_index_type"sv, s_calendar_symbols_index_type);
  1453. generator.set("calendar_symbols_list_index_type"sv, s_calendar_symbols_list_index_type);
  1454. generator.set("calendar_index_type"sv, s_calendar_index_type);
  1455. generator.set("time_zone_index_type"sv, s_time_zone_index_type);
  1456. generator.set("time_zone_list_index_type"sv, s_time_zone_list_index_type);
  1457. generator.set("day_period_index_type"sv, s_day_period_index_type);
  1458. generator.set("day_period_list_index_type"sv, s_day_period_list_index_type);
  1459. generator.append(R"~~~(
  1460. #include <AK/Array.h>
  1461. #include <AK/BinarySearch.h>
  1462. #include <AK/Optional.h>
  1463. #include <AK/StringView.h>
  1464. #include <LibLocale/DateTimeFormat.h>
  1465. #include <LibLocale/DateTimeFormatData.h>
  1466. #include <LibLocale/Locale.h>
  1467. #include <LibLocale/LocaleData.h>
  1468. #include <LibTimeZone/TimeZone.h>
  1469. namespace Locale {
  1470. )~~~");
  1471. cldr.unique_strings.generate(generator);
  1472. generator.append(R"~~~(
  1473. template <typename SourceType, typename TargetType>
  1474. static void convert_calendar_fields(SourceType const& source, TargetType& target)
  1475. {
  1476. if (source.era != -1)
  1477. target.era = static_cast<CalendarPatternStyle>(source.era);
  1478. if (source.year != -1)
  1479. target.year = static_cast<CalendarPatternStyle>(source.year);
  1480. if (source.month != -1)
  1481. target.month = static_cast<CalendarPatternStyle>(source.month);
  1482. if (source.weekday != -1)
  1483. target.weekday = static_cast<CalendarPatternStyle>(source.weekday);
  1484. if (source.day != -1)
  1485. target.day = static_cast<CalendarPatternStyle>(source.day);
  1486. if (source.day_period != -1)
  1487. target.day_period = static_cast<CalendarPatternStyle>(source.day_period);
  1488. if (source.hour != -1)
  1489. target.hour = static_cast<CalendarPatternStyle>(source.hour);
  1490. if (source.minute != -1)
  1491. target.minute = static_cast<CalendarPatternStyle>(source.minute);
  1492. if (source.second != -1)
  1493. target.second = static_cast<CalendarPatternStyle>(source.second);
  1494. if (source.fractional_second_digits != -1)
  1495. target.fractional_second_digits = static_cast<u8>(source.fractional_second_digits);
  1496. if (source.time_zone_name != -1)
  1497. target.time_zone_name = static_cast<CalendarPatternStyle>(source.time_zone_name);
  1498. }
  1499. struct CalendarPatternImpl {
  1500. CalendarPattern to_unicode_calendar_pattern() const {
  1501. CalendarPattern calendar_pattern {};
  1502. calendar_pattern.skeleton = decode_string(skeleton);
  1503. calendar_pattern.pattern = decode_string(pattern);
  1504. if (pattern12 != 0)
  1505. calendar_pattern.pattern12 = decode_string(pattern12);
  1506. convert_calendar_fields(*this, calendar_pattern);
  1507. return calendar_pattern;
  1508. }
  1509. @string_index_type@ skeleton { 0 };
  1510. @string_index_type@ pattern { 0 };
  1511. @string_index_type@ pattern12 { 0 };
  1512. i8 era { -1 };
  1513. i8 year { -1 };
  1514. i8 month { -1 };
  1515. i8 weekday { -1 };
  1516. i8 day { -1 };
  1517. i8 day_period { -1 };
  1518. i8 hour { -1 };
  1519. i8 minute { -1 };
  1520. i8 second { -1 };
  1521. i8 fractional_second_digits { -1 };
  1522. i8 time_zone_name { -1 };
  1523. };
  1524. struct CalendarRangePatternImpl {
  1525. CalendarRangePattern to_unicode_calendar_range_pattern() const {
  1526. CalendarRangePattern calendar_range_pattern {};
  1527. if (field != -1)
  1528. calendar_range_pattern.field = static_cast<CalendarRangePattern::Field>(field);
  1529. calendar_range_pattern.start_range = decode_string(start_range);
  1530. calendar_range_pattern.separator = decode_string(separator);
  1531. calendar_range_pattern.end_range = decode_string(end_range);
  1532. convert_calendar_fields(*this, calendar_range_pattern);
  1533. return calendar_range_pattern;
  1534. }
  1535. @string_index_type@ skeleton { 0 };
  1536. i8 field { -1 };
  1537. @string_index_type@ start_range { 0 };
  1538. @string_index_type@ separator { 0 };
  1539. @string_index_type@ end_range { 0 };
  1540. i8 era { -1 };
  1541. i8 year { -1 };
  1542. i8 month { -1 };
  1543. i8 weekday { -1 };
  1544. i8 day { -1 };
  1545. i8 day_period { -1 };
  1546. i8 hour { -1 };
  1547. i8 minute { -1 };
  1548. i8 second { -1 };
  1549. i8 fractional_second_digits { -1 };
  1550. i8 time_zone_name { -1 };
  1551. };
  1552. )~~~");
  1553. cldr.unique_patterns.generate(generator, "CalendarPatternImpl"sv, "s_calendar_patterns"sv, 10);
  1554. cldr.unique_pattern_lists.generate(generator, s_calendar_pattern_index_type, "s_calendar_pattern_lists"sv);
  1555. cldr.unique_range_patterns.generate(generator, "CalendarRangePatternImpl"sv, "s_calendar_range_patterns"sv, 10);
  1556. cldr.unique_range_pattern_lists.generate(generator, s_calendar_range_pattern_index_type, "s_calendar_range_pattern_lists"sv);
  1557. generator.append(R"~~~(
  1558. struct CalendarFormatImpl {
  1559. CalendarFormat to_unicode_calendar_format() const {
  1560. CalendarFormat calendar_format {};
  1561. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  1562. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  1563. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  1564. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  1565. return calendar_format;
  1566. }
  1567. @calendar_pattern_index_type@ full_format { 0 };
  1568. @calendar_pattern_index_type@ long_format { 0 };
  1569. @calendar_pattern_index_type@ medium_format { 0 };
  1570. @calendar_pattern_index_type@ short_format { 0 };
  1571. };
  1572. struct CalendarSymbols {
  1573. @symbol_list_index_type@ narrow_symbols { 0 };
  1574. @symbol_list_index_type@ short_symbols { 0 };
  1575. @symbol_list_index_type@ long_symbols { 0 };
  1576. };
  1577. struct CalendarData {
  1578. @calendar_format_index_type@ date_formats { 0 };
  1579. @calendar_format_index_type@ time_formats { 0 };
  1580. @calendar_format_index_type@ date_time_formats { 0 };
  1581. @calendar_pattern_list_index_type@ available_formats { 0 };
  1582. @calendar_range_pattern_index_type@ default_range_format { 0 };
  1583. @calendar_range_pattern_list_index_type@ range_formats { 0 };
  1584. @calendar_range_pattern_list_index_type@ range12_formats { 0 };
  1585. @calendar_symbols_list_index_type@ symbols { 0 };
  1586. };
  1587. struct TimeZoneNames {
  1588. @string_index_type@ short_standard_name { 0 };
  1589. @string_index_type@ long_standard_name { 0 };
  1590. @string_index_type@ short_daylight_name { 0 };
  1591. @string_index_type@ long_daylight_name { 0 };
  1592. @string_index_type@ short_generic_name { 0 };
  1593. @string_index_type@ long_generic_name { 0 };
  1594. };
  1595. struct TimeZoneFormatImpl {
  1596. TimeZoneFormat to_time_zone_format() const {
  1597. TimeZoneFormat time_zone_format {};
  1598. time_zone_format.symbol_ahead_sign = decode_string(symbol_ahead_sign);
  1599. time_zone_format.symbol_ahead_separator = decode_string(symbol_ahead_separator);
  1600. time_zone_format.symbol_behind_sign = decode_string(symbol_behind_sign);
  1601. time_zone_format.symbol_behind_separator = decode_string(symbol_behind_separator);
  1602. time_zone_format.gmt_format = decode_string(gmt_format);
  1603. time_zone_format.gmt_zero_format = decode_string(gmt_zero_format);
  1604. return time_zone_format;
  1605. }
  1606. @string_index_type@ symbol_ahead_sign { 0 };
  1607. @string_index_type@ symbol_ahead_separator { 0 };
  1608. @string_index_type@ symbol_behind_sign { 0 };
  1609. @string_index_type@ symbol_behind_separator { 0 };
  1610. @string_index_type@ gmt_format { 0 };
  1611. @string_index_type@ gmt_zero_format { 0 };
  1612. };
  1613. struct DayPeriodData {
  1614. u8 day_period { 0 };
  1615. u8 begin { 0 };
  1616. u8 end { 0 };
  1617. };
  1618. )~~~");
  1619. cldr.unique_formats.generate(generator, "CalendarFormatImpl"sv, "s_calendar_formats"sv, 10);
  1620. cldr.unique_symbol_lists.generate(generator, s_string_index_type, "s_symbol_lists"sv);
  1621. cldr.unique_calendar_symbols.generate(generator, "CalendarSymbols"sv, "s_calendar_symbols"sv, 10);
  1622. cldr.unique_calendar_symbols_lists.generate(generator, s_calendar_symbols_index_type, "s_calendar_symbol_lists"sv);
  1623. cldr.unique_calendars.generate(generator, "CalendarData"sv, "s_calendars"sv, 10);
  1624. cldr.unique_time_zones.generate(generator, "TimeZoneNames"sv, "s_time_zones"sv, 30);
  1625. cldr.unique_time_zone_lists.generate(generator, s_time_zone_index_type, "s_time_zone_lists"sv);
  1626. cldr.unique_time_zone_formats.generate(generator, "TimeZoneFormatImpl"sv, "s_time_zone_formats"sv, 30);
  1627. cldr.unique_day_periods.generate(generator, "DayPeriodData"sv, "s_day_periods"sv, 30);
  1628. cldr.unique_day_period_lists.generate(generator, s_day_period_index_type, "s_day_period_lists"sv);
  1629. cldr.unique_hour_cycle_lists.generate(generator, "u8"sv, "s_hour_cycle_lists"sv);
  1630. auto append_calendars = [&](String name, auto const& calendars) {
  1631. generator.set("name", name);
  1632. generator.set("size", String::number(calendars.size()));
  1633. generator.append(R"~~~(
  1634. static constexpr Array<@calendar_index_type@, @size@> @name@ { {)~~~");
  1635. bool first = true;
  1636. for (auto const& calendar_key : cldr.calendars) {
  1637. auto calendar = calendars.find(calendar_key)->value;
  1638. generator.append(first ? " "sv : ", "sv);
  1639. generator.append(String::number(calendar));
  1640. first = false;
  1641. }
  1642. generator.append(" } };");
  1643. };
  1644. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  1645. generator.set("type", type);
  1646. generator.set("name", name);
  1647. generator.set("size", String::number(keys.size()));
  1648. generator.append(R"~~~(
  1649. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1650. bool first = true;
  1651. for (auto const& key : keys) {
  1652. auto const& value = map.find(key)->value;
  1653. auto mapping = mapping_getter(value);
  1654. generator.append(first ? " "sv : ", "sv);
  1655. generator.append(String::number(mapping));
  1656. first = false;
  1657. }
  1658. generator.append(" } };");
  1659. };
  1660. auto locales = cldr.locales.keys();
  1661. quick_sort(locales);
  1662. generate_mapping(generator, cldr.locales, s_calendar_index_type, "s_locale_calendars"sv, "s_calendars_{}"sv, format_identifier, [&](auto const& name, auto const& value) { append_calendars(name, value.calendars); });
  1663. append_mapping(locales, cldr.locales, s_time_zone_index_type, "s_locale_time_zones"sv, [](auto const& locale) { return locale.time_zones; });
  1664. append_mapping(locales, cldr.locales, s_time_zone_format_index_type, "s_locale_time_zone_formats"sv, [](auto const& locale) { return locale.time_zone_formats; });
  1665. append_mapping(locales, cldr.locales, s_day_period_index_type, "s_locale_day_periods"sv, [](auto const& locale) { return locale.day_periods; });
  1666. append_mapping(cldr.hour_cycle_regions, cldr.hour_cycles, s_hour_cycle_list_index_type, "s_hour_cycles"sv, [](auto const& hour_cycles) { return hour_cycles; });
  1667. append_mapping(cldr.minimum_days_regions, cldr.minimum_days, "u8"sv, "s_minimum_days"sv, [](auto minimum_days) { return minimum_days; });
  1668. append_mapping(cldr.first_day_regions, cldr.first_day, "u8"sv, "s_first_day"sv, [](auto first_day) { return to_underlying(first_day); });
  1669. append_mapping(cldr.weekend_start_regions, cldr.weekend_start, "u8"sv, "s_weekend_start"sv, [](auto weekend_start) { return to_underlying(weekend_start); });
  1670. append_mapping(cldr.weekend_end_regions, cldr.weekend_end, "u8"sv, "s_weekend_end"sv, [](auto weekend_end) { return to_underlying(weekend_end); });
  1671. generator.append("\n");
  1672. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1673. HashValueMap<String> hashes;
  1674. hashes.ensure_capacity(values.size());
  1675. for (auto const& value : values)
  1676. hashes.set(value.hash(), format_identifier(enum_title, value));
  1677. for (auto const& alias : aliases)
  1678. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1679. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1680. };
  1681. append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, cldr.hour_cycle_regions);
  1682. append_from_string("MinimumDaysRegion"sv, "minimum_days_region"sv, cldr.minimum_days_regions);
  1683. append_from_string("FirstDayRegion"sv, "first_day_region"sv, cldr.first_day_regions);
  1684. append_from_string("WeekendStartRegion"sv, "weekend_start_region"sv, cldr.weekend_start_regions);
  1685. append_from_string("WeekendEndRegion"sv, "weekend_end_region"sv, cldr.weekend_end_regions);
  1686. generator.append(R"~~~(
  1687. static Optional<Calendar> keyword_to_calendar(KeywordCalendar keyword)
  1688. {
  1689. switch (keyword) {)~~~");
  1690. for (auto const& calendar : cldr.calendars) {
  1691. generator.set("name"sv, format_identifier({}, calendar));
  1692. generator.append(R"~~~(
  1693. case KeywordCalendar::@name@:
  1694. return Calendar::@name@;)~~~");
  1695. }
  1696. generator.append(R"~~~(
  1697. default:
  1698. return {};
  1699. }
  1700. }
  1701. Vector<HourCycle> get_regional_hour_cycles(StringView region)
  1702. {
  1703. auto region_value = hour_cycle_region_from_string(region);
  1704. if (!region_value.has_value())
  1705. return {};
  1706. auto region_index = to_underlying(*region_value);
  1707. auto regional_hour_cycles_index = s_hour_cycles.at(region_index);
  1708. auto const& regional_hour_cycles = s_hour_cycle_lists.at(regional_hour_cycles_index);
  1709. Vector<HourCycle> hour_cycles;
  1710. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1711. for (auto hour_cycle : regional_hour_cycles)
  1712. hour_cycles.unchecked_append(static_cast<HourCycle>(hour_cycle));
  1713. return hour_cycles;
  1714. }
  1715. )~~~");
  1716. auto append_regional_lookup = [&](StringView return_type, StringView lookup_type) {
  1717. generator.set("return_type", return_type);
  1718. generator.set("lookup_type", lookup_type);
  1719. generator.append(R"~~~(
  1720. Optional<@return_type@> get_regional_@lookup_type@(StringView region)
  1721. {
  1722. auto region_value = @lookup_type@_region_from_string(region);
  1723. if (!region_value.has_value())
  1724. return {};
  1725. auto region_index = to_underlying(*region_value);
  1726. auto @lookup_type@ = s_@lookup_type@.at(region_index);
  1727. return static_cast<@return_type@>(@lookup_type@);
  1728. }
  1729. )~~~");
  1730. };
  1731. append_regional_lookup("u8"sv, "minimum_days"sv);
  1732. append_regional_lookup("Weekday"sv, "first_day"sv);
  1733. append_regional_lookup("Weekday"sv, "weekend_start"sv);
  1734. append_regional_lookup("Weekday"sv, "weekend_end"sv);
  1735. generator.append(R"~~~(
  1736. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1737. {
  1738. auto locale_value = locale_from_string(locale);
  1739. if (!locale_value.has_value())
  1740. return nullptr;
  1741. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1742. auto const& calendar_indices = s_locale_calendars.at(locale_index);
  1743. auto lookup_calendar = [&](auto calendar_name) -> CalendarData const* {
  1744. auto calendar_keyword = keyword_ca_from_string(calendar_name);
  1745. if (!calendar_keyword.has_value())
  1746. return nullptr;
  1747. auto calendar_value = keyword_to_calendar(*calendar_keyword);
  1748. if (!calendar_value.has_value())
  1749. return nullptr;
  1750. size_t calendar_index = to_underlying(*calendar_value);
  1751. calendar_index = calendar_indices[calendar_index];
  1752. return &s_calendars[calendar_index];
  1753. };
  1754. if (auto const* calendar_data = lookup_calendar(calendar))
  1755. return calendar_data;
  1756. auto default_calendar = get_preferred_keyword_value_for_locale(locale, "ca"sv);
  1757. if (!default_calendar.has_value())
  1758. return nullptr;
  1759. return lookup_calendar(*default_calendar);
  1760. }
  1761. Optional<CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1762. {
  1763. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1764. auto const& formats = s_calendar_formats.at(data->date_formats);
  1765. return formats.to_unicode_calendar_format();
  1766. }
  1767. return {};
  1768. }
  1769. Optional<CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1770. {
  1771. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1772. auto const& formats = s_calendar_formats.at(data->time_formats);
  1773. return formats.to_unicode_calendar_format();
  1774. }
  1775. return {};
  1776. }
  1777. Optional<CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1778. {
  1779. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1780. auto const& formats = s_calendar_formats.at(data->date_time_formats);
  1781. return formats.to_unicode_calendar_format();
  1782. }
  1783. return {};
  1784. }
  1785. Vector<CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1786. {
  1787. Vector<CalendarPattern> result {};
  1788. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1789. auto const& available_formats = s_calendar_pattern_lists.at(data->available_formats);
  1790. result.ensure_capacity(available_formats.size());
  1791. for (auto const& format : available_formats)
  1792. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1793. }
  1794. return result;
  1795. }
  1796. Optional<CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar)
  1797. {
  1798. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1799. auto const& pattern = s_calendar_range_patterns[data->default_range_format];
  1800. return pattern.to_unicode_calendar_range_pattern();
  1801. }
  1802. return {};
  1803. }
  1804. Vector<CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton)
  1805. {
  1806. Vector<CalendarRangePattern> result {};
  1807. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1808. auto const& range_formats = s_calendar_range_pattern_lists.at(data->range_formats);
  1809. for (auto format : range_formats) {
  1810. auto const& pattern = s_calendar_range_patterns[format];
  1811. if (skeleton == decode_string(pattern.skeleton))
  1812. result.append(pattern.to_unicode_calendar_range_pattern());
  1813. }
  1814. }
  1815. return result;
  1816. }
  1817. Vector<CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton)
  1818. {
  1819. Vector<CalendarRangePattern> result {};
  1820. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1821. auto const& range12_formats = s_calendar_range_pattern_lists.at(data->range12_formats);
  1822. for (auto format : range12_formats) {
  1823. auto const& pattern = s_calendar_range_patterns[format];
  1824. if (skeleton == decode_string(pattern.skeleton))
  1825. result.append(pattern.to_unicode_calendar_range_pattern());
  1826. }
  1827. }
  1828. return result;
  1829. }
  1830. static Span<@string_index_type@ const> find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1831. {
  1832. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1833. auto const& symbols_list = s_calendar_symbol_lists[data->symbols];
  1834. auto symbol_index = to_underlying(symbol);
  1835. auto calendar_symbols_index = symbols_list.at(symbol_index);
  1836. auto const& symbols = s_calendar_symbols.at(calendar_symbols_index);
  1837. @symbol_list_index_type@ symbol_list_index = 0;
  1838. switch (style) {
  1839. case CalendarPatternStyle::Narrow:
  1840. symbol_list_index = symbols.narrow_symbols;
  1841. break;
  1842. case CalendarPatternStyle::Short:
  1843. symbol_list_index = symbols.short_symbols;
  1844. break;
  1845. case CalendarPatternStyle::Long:
  1846. symbol_list_index = symbols.long_symbols;
  1847. break;
  1848. default:
  1849. VERIFY_NOT_REACHED();
  1850. }
  1851. return s_symbol_lists.at(symbol_list_index);
  1852. }
  1853. return {};
  1854. }
  1855. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Era value)
  1856. {
  1857. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1858. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1859. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1860. return decode_string(symbol_index);
  1861. }
  1862. return {};
  1863. }
  1864. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Month value)
  1865. {
  1866. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1867. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1868. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1869. return decode_string(symbol_index);
  1870. }
  1871. return {};
  1872. }
  1873. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Weekday value)
  1874. {
  1875. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1876. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1877. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1878. return decode_string(symbol_index);
  1879. }
  1880. return {};
  1881. }
  1882. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, DayPeriod value)
  1883. {
  1884. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1885. if (auto value_index = to_underlying(value); value_index < symbols.size()) {
  1886. if (auto symbol_index = symbols.at(value_index); symbol_index != 0)
  1887. return decode_string(symbol_index);
  1888. }
  1889. return {};
  1890. }
  1891. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour)
  1892. {
  1893. auto locale_value = locale_from_string(locale);
  1894. if (!locale_value.has_value())
  1895. return {};
  1896. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1897. auto day_periods_index = s_locale_day_periods[locale_index];
  1898. auto day_periods = s_day_period_lists[day_periods_index];
  1899. for (auto day_period_index : day_periods) {
  1900. auto day_period = s_day_periods[day_period_index];
  1901. bool hour_falls_within_day_period = false;
  1902. if (day_period.begin > day_period.end) {
  1903. if (hour >= day_period.begin)
  1904. hour_falls_within_day_period = true;
  1905. else if (hour <= day_period.end)
  1906. hour_falls_within_day_period = true;
  1907. } else if ((day_period.begin <= hour) && (hour < day_period.end)) {
  1908. hour_falls_within_day_period = true;
  1909. }
  1910. if (hour_falls_within_day_period) {
  1911. auto period = static_cast<DayPeriod>(day_period.day_period);
  1912. return get_calendar_day_period_symbol(locale, calendar, style, period);
  1913. }
  1914. }
  1915. // Fallback to fixed periods if the locale does not have flexible day periods.
  1916. // TR-35 states that the meaning of AM and PM does not change with locale.
  1917. if (hour < 12)
  1918. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::AM);
  1919. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::PM);
  1920. }
  1921. Optional<TimeZoneFormat> get_time_zone_format(StringView locale)
  1922. {
  1923. auto locale_value = locale_from_string(locale);
  1924. if (!locale_value.has_value())
  1925. return {};
  1926. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1927. auto time_zone_format_index = s_locale_time_zone_formats.at(locale_index);
  1928. auto const& time_zone_format = s_time_zone_formats.at(time_zone_format_index);
  1929. return time_zone_format.to_time_zone_format();
  1930. }
  1931. static TimeZoneNames const* find_time_zone_names(StringView locale, StringView time_zone)
  1932. {
  1933. auto locale_value = locale_from_string(locale);
  1934. if (!locale_value.has_value())
  1935. return nullptr;
  1936. auto time_zone_value = ::TimeZone::time_zone_from_string(time_zone);
  1937. if (!time_zone_value.has_value())
  1938. return nullptr;
  1939. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1940. size_t time_zone_index = to_underlying(*time_zone_value);
  1941. auto time_zone_list_index = s_locale_time_zones.at(locale_index);
  1942. auto const& time_zone_list = s_time_zone_lists.at(time_zone_list_index);
  1943. if (time_zone_list.size() <= time_zone_index)
  1944. return nullptr;
  1945. time_zone_index = time_zone_list.at(time_zone_index);
  1946. return &s_time_zones[time_zone_index];
  1947. }
  1948. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style, TimeZone::InDST in_dst)
  1949. {
  1950. if (auto const* data = find_time_zone_names(locale, time_zone); data != nullptr) {
  1951. size_t name_index = 0;
  1952. switch (style) {
  1953. case CalendarPatternStyle::Short:
  1954. name_index = (in_dst == TimeZone::InDST::No) ? data->short_standard_name : data->short_daylight_name;
  1955. break;
  1956. case CalendarPatternStyle::Long:
  1957. name_index = (in_dst == TimeZone::InDST::No) ? data->long_standard_name : data->long_daylight_name;
  1958. break;
  1959. case CalendarPatternStyle::ShortGeneric:
  1960. name_index = data->short_generic_name;
  1961. break;
  1962. case CalendarPatternStyle::LongGeneric:
  1963. name_index = data->long_generic_name;
  1964. break;
  1965. default:
  1966. VERIFY_NOT_REACHED();
  1967. }
  1968. if (name_index != 0)
  1969. return decode_string(name_index);
  1970. }
  1971. return {};
  1972. }
  1973. }
  1974. )~~~");
  1975. TRY(file.write(generator.as_string_view().bytes()));
  1976. return {};
  1977. }
  1978. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1979. {
  1980. StringView generated_header_path;
  1981. StringView generated_implementation_path;
  1982. StringView core_path;
  1983. StringView dates_path;
  1984. Core::ArgsParser args_parser;
  1985. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1986. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1987. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1988. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1989. args_parser.parse(arguments);
  1990. auto generated_header_file = TRY(open_file(generated_header_path, Core::Stream::OpenMode::Write));
  1991. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::Stream::OpenMode::Write));
  1992. CLDR cldr;
  1993. TRY(parse_all_locales(core_path, dates_path, cldr));
  1994. TRY(generate_unicode_locale_header(*generated_header_file, cldr));
  1995. TRY(generate_unicode_locale_implementation(*generated_implementation_file, cldr));
  1996. return 0;
  1997. }