GenerateUnicodeDateTimeFormat.cpp 89 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905190619071908190919101911191219131914191519161917191819191920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950195119521953195419551956195719581959196019611962196319641965196619671968196919701971197219731974197519761977197819791980198119821983198419851986198719881989199019911992199319941995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019202020212022202320242025202620272028202920302031203220332034203520362037203820392040204120422043204420452046204720482049205020512052205320542055205620572058205920602061206220632064206520662067206820692070207120722073207420752076207720782079208020812082208320842085208620872088208920902091209220932094209520962097209820992100210121022103210421052106210721082109211021112112211321142115211621172118211921202121212221232124212521262127212821292130213121322133213421352136213721382139214021412142214321442145214621472148214921502151215221532154215521562157215821592160216121622163216421652166216721682169217021712172217321742175217621772178217921802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202220322042205220622072208220922102211221222132214221522162217221822192220222122222223222422252226222722282229223022312232223322342235223622372238223922402241224222432244224522462247224822492250225122522253225422552256225722582259226022612262226322642265
  1. /*
  2. * Copyright (c) 2021-2022, Tim Flynn <trflynn89@pm.me>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "GeneratorUtil.h"
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/Format.h>
  10. #include <AK/GenericLexer.h>
  11. #include <AK/HashFunctions.h>
  12. #include <AK/HashMap.h>
  13. #include <AK/JsonObject.h>
  14. #include <AK/JsonParser.h>
  15. #include <AK/JsonValue.h>
  16. #include <AK/LexicalPath.h>
  17. #include <AK/SourceGenerator.h>
  18. #include <AK/String.h>
  19. #include <AK/StringBuilder.h>
  20. #include <AK/Traits.h>
  21. #include <AK/Utf8View.h>
  22. #include <LibCore/ArgsParser.h>
  23. #include <LibCore/DirIterator.h>
  24. #include <LibCore/File.h>
  25. #include <LibTimeZone/TimeZone.h>
  26. #include <LibUnicode/DateTimeFormat.h>
  27. using StringIndexType = u16;
  28. constexpr auto s_string_index_type = "u16"sv;
  29. using CalendarPatternIndexType = u16;
  30. constexpr auto s_calendar_pattern_index_type = "u16"sv;
  31. using CalendarPatternListIndexType = u8;
  32. constexpr auto s_calendar_pattern_list_index_type = "u8"sv;
  33. using CalendarRangePatternIndexType = u16;
  34. constexpr auto s_calendar_range_pattern_index_type = "u16"sv;
  35. using CalendarRangePatternListIndexType = u8;
  36. constexpr auto s_calendar_range_pattern_list_index_type = "u8"sv;
  37. using CalendarFormatIndexType = u8;
  38. constexpr auto s_calendar_format_index_type = "u8"sv;
  39. using SymbolListIndexType = u16;
  40. constexpr auto s_symbol_list_index_type = "u16"sv;
  41. using CalendarSymbolsIndexType = u16;
  42. constexpr auto s_calendar_symbols_index_type = "u16"sv;
  43. using CalendarSymbolsListIndexType = u8;
  44. constexpr auto s_calendar_symbols_list_index_type = "u8"sv;
  45. using CalendarIndexType = u8;
  46. constexpr auto s_calendar_index_type = "u8"sv;
  47. using TimeZoneNamesIndexType = u16;
  48. constexpr auto s_time_zone_index_type = "u16"sv;
  49. using TimeZoneNamesListIndexType = u8;
  50. constexpr auto s_time_zone_list_index_type = "u8"sv;
  51. using TimeZoneFormatIndexType = u8;
  52. constexpr auto s_time_zone_format_index_type = "u8"sv;
  53. using DayPeriodIndexType = u8;
  54. constexpr auto s_day_period_index_type = "u8"sv;
  55. using DayPeriodListIndexType = u8;
  56. constexpr auto s_day_period_list_index_type = "u8"sv;
  57. using HourCycleListIndexType = u8;
  58. constexpr auto s_hour_cycle_list_index_type = "u8"sv;
  59. struct CalendarPattern : public Unicode::CalendarPattern {
  60. bool contains_only_date_fields() const
  61. {
  62. return !day_period.has_value() && !hour.has_value() && !minute.has_value() && !second.has_value() && !fractional_second_digits.has_value() && !time_zone_name.has_value();
  63. }
  64. bool contains_only_time_fields() const
  65. {
  66. return !weekday.has_value() && !era.has_value() && !year.has_value() && !month.has_value() && !day.has_value();
  67. }
  68. unsigned hash() const
  69. {
  70. auto hash = pair_int_hash(pattern_index, pattern12_index);
  71. hash = pair_int_hash(hash, skeleton_index);
  72. auto hash_field = [&](auto const& field) {
  73. if (field.has_value())
  74. hash = pair_int_hash(hash, static_cast<u8>(*field));
  75. else
  76. hash = pair_int_hash(hash, -1);
  77. };
  78. hash_field(era);
  79. hash_field(year);
  80. hash_field(month);
  81. hash_field(weekday);
  82. hash_field(day);
  83. hash_field(day_period);
  84. hash_field(hour);
  85. hash_field(minute);
  86. hash_field(second);
  87. hash_field(fractional_second_digits);
  88. hash_field(time_zone_name);
  89. return hash;
  90. }
  91. bool operator==(CalendarPattern const& other) const
  92. {
  93. return (skeleton_index == other.skeleton_index)
  94. && (pattern_index == other.pattern_index)
  95. && (pattern12_index == other.pattern12_index)
  96. && (era == other.era)
  97. && (year == other.year)
  98. && (month == other.month)
  99. && (weekday == other.weekday)
  100. && (day == other.day)
  101. && (day_period == other.day_period)
  102. && (hour == other.hour)
  103. && (minute == other.minute)
  104. && (second == other.second)
  105. && (fractional_second_digits == other.fractional_second_digits)
  106. && (time_zone_name == other.time_zone_name);
  107. }
  108. StringIndexType skeleton_index { 0 };
  109. StringIndexType pattern_index { 0 };
  110. StringIndexType pattern12_index { 0 };
  111. };
  112. template<>
  113. struct AK::Formatter<CalendarPattern> : Formatter<FormatString> {
  114. ErrorOr<void> format(FormatBuilder& builder, CalendarPattern const& pattern)
  115. {
  116. auto field_to_i8 = [](auto const& field) -> i8 {
  117. if (!field.has_value())
  118. return -1;
  119. return static_cast<i8>(*field);
  120. };
  121. return Formatter<FormatString>::format(builder,
  122. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}",
  123. pattern.skeleton_index,
  124. pattern.pattern_index,
  125. pattern.pattern12_index,
  126. field_to_i8(pattern.era),
  127. field_to_i8(pattern.year),
  128. field_to_i8(pattern.month),
  129. field_to_i8(pattern.weekday),
  130. field_to_i8(pattern.day),
  131. field_to_i8(pattern.day_period),
  132. field_to_i8(pattern.hour),
  133. field_to_i8(pattern.minute),
  134. field_to_i8(pattern.second),
  135. field_to_i8(pattern.fractional_second_digits),
  136. field_to_i8(pattern.time_zone_name));
  137. }
  138. };
  139. template<>
  140. struct AK::Traits<CalendarPattern> : public GenericTraits<CalendarPattern> {
  141. static unsigned hash(CalendarPattern const& c) { return c.hash(); }
  142. };
  143. struct CalendarRangePattern : public CalendarPattern {
  144. unsigned hash() const
  145. {
  146. auto hash = CalendarPattern::hash();
  147. if (field.has_value())
  148. hash = pair_int_hash(hash, static_cast<u8>(*field));
  149. hash = pair_int_hash(hash, start_range);
  150. hash = pair_int_hash(hash, separator);
  151. hash = pair_int_hash(hash, end_range);
  152. return hash;
  153. }
  154. bool operator==(CalendarRangePattern const& other) const
  155. {
  156. if (!CalendarPattern::operator==(other))
  157. return false;
  158. return (field == other.field)
  159. && (start_range == other.start_range)
  160. && (separator == other.separator)
  161. && (end_range == other.end_range);
  162. }
  163. Optional<Unicode::CalendarRangePattern::Field> field {};
  164. StringIndexType start_range { 0 };
  165. StringIndexType separator { 0 };
  166. StringIndexType end_range { 0 };
  167. };
  168. template<>
  169. struct AK::Formatter<CalendarRangePattern> : Formatter<FormatString> {
  170. ErrorOr<void> format(FormatBuilder& builder, CalendarRangePattern const& pattern)
  171. {
  172. auto field_to_i8 = [](auto const& field) -> i8 {
  173. if (!field.has_value())
  174. return -1;
  175. return static_cast<i8>(*field);
  176. };
  177. return Formatter<FormatString>::format(builder,
  178. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}",
  179. pattern.skeleton_index,
  180. field_to_i8(pattern.field),
  181. pattern.start_range,
  182. pattern.separator,
  183. pattern.end_range,
  184. field_to_i8(pattern.era),
  185. field_to_i8(pattern.year),
  186. field_to_i8(pattern.month),
  187. field_to_i8(pattern.weekday),
  188. field_to_i8(pattern.day),
  189. field_to_i8(pattern.day_period),
  190. field_to_i8(pattern.hour),
  191. field_to_i8(pattern.minute),
  192. field_to_i8(pattern.second),
  193. field_to_i8(pattern.fractional_second_digits),
  194. field_to_i8(pattern.time_zone_name));
  195. }
  196. };
  197. template<>
  198. struct AK::Traits<CalendarRangePattern> : public GenericTraits<CalendarRangePattern> {
  199. static unsigned hash(CalendarRangePattern const& c) { return c.hash(); }
  200. };
  201. struct CalendarFormat {
  202. unsigned hash() const
  203. {
  204. auto hash = pair_int_hash(full_format, long_format);
  205. hash = pair_int_hash(hash, medium_format);
  206. hash = pair_int_hash(hash, short_format);
  207. return hash;
  208. }
  209. bool operator==(CalendarFormat const& other) const
  210. {
  211. return (full_format == other.full_format)
  212. && (long_format == other.long_format)
  213. && (medium_format == other.medium_format)
  214. && (short_format == other.short_format);
  215. }
  216. CalendarPatternIndexType full_format { 0 };
  217. CalendarPatternIndexType long_format { 0 };
  218. CalendarPatternIndexType medium_format { 0 };
  219. CalendarPatternIndexType short_format { 0 };
  220. };
  221. template<>
  222. struct AK::Formatter<CalendarFormat> : Formatter<FormatString> {
  223. ErrorOr<void> format(FormatBuilder& builder, CalendarFormat const& pattern)
  224. {
  225. return Formatter<FormatString>::format(builder,
  226. "{{ {}, {}, {}, {} }}",
  227. pattern.full_format,
  228. pattern.long_format,
  229. pattern.medium_format,
  230. pattern.short_format);
  231. }
  232. };
  233. template<>
  234. struct AK::Traits<CalendarFormat> : public GenericTraits<CalendarFormat> {
  235. static unsigned hash(CalendarFormat const& c) { return c.hash(); }
  236. };
  237. using SymbolList = Vector<StringIndexType>;
  238. struct CalendarSymbols {
  239. unsigned hash() const
  240. {
  241. auto hash = pair_int_hash(narrow_symbols, short_symbols);
  242. hash = pair_int_hash(hash, long_symbols);
  243. return hash;
  244. }
  245. bool operator==(CalendarSymbols const& other) const
  246. {
  247. return (narrow_symbols == other.narrow_symbols)
  248. && (short_symbols == other.short_symbols)
  249. && (long_symbols == other.long_symbols);
  250. }
  251. SymbolListIndexType narrow_symbols { 0 };
  252. SymbolListIndexType short_symbols { 0 };
  253. SymbolListIndexType long_symbols { 0 };
  254. };
  255. template<>
  256. struct AK::Formatter<CalendarSymbols> : Formatter<FormatString> {
  257. ErrorOr<void> format(FormatBuilder& builder, CalendarSymbols const& symbols)
  258. {
  259. return Formatter<FormatString>::format(builder,
  260. "{{ {}, {}, {} }}",
  261. symbols.narrow_symbols,
  262. symbols.short_symbols,
  263. symbols.long_symbols);
  264. }
  265. };
  266. template<>
  267. struct AK::Traits<CalendarSymbols> : public GenericTraits<CalendarSymbols> {
  268. static unsigned hash(CalendarSymbols const& c) { return c.hash(); }
  269. };
  270. using CalendarPatternList = Vector<CalendarPatternIndexType>;
  271. using CalendarRangePatternList = Vector<CalendarRangePatternIndexType>;
  272. using CalendarSymbolsList = Vector<CalendarSymbolsIndexType>;
  273. struct Calendar {
  274. unsigned hash() const
  275. {
  276. auto hash = int_hash(date_formats);
  277. hash = pair_int_hash(hash, time_formats);
  278. hash = pair_int_hash(hash, date_time_formats);
  279. hash = pair_int_hash(hash, available_formats);
  280. hash = pair_int_hash(hash, default_range_format);
  281. hash = pair_int_hash(hash, range_formats);
  282. hash = pair_int_hash(hash, range12_formats);
  283. hash = pair_int_hash(hash, symbols);
  284. return hash;
  285. }
  286. bool operator==(Calendar const& other) const
  287. {
  288. return (date_formats == other.date_formats)
  289. && (time_formats == other.time_formats)
  290. && (date_time_formats == other.date_time_formats)
  291. && (available_formats == other.available_formats)
  292. && (default_range_format == other.default_range_format)
  293. && (range_formats == other.range_formats)
  294. && (range12_formats == other.range12_formats)
  295. && (symbols == other.symbols);
  296. }
  297. CalendarFormatIndexType date_formats { 0 };
  298. CalendarFormatIndexType time_formats { 0 };
  299. CalendarFormatIndexType date_time_formats { 0 };
  300. CalendarPatternListIndexType available_formats { 0 };
  301. CalendarRangePatternIndexType default_range_format { 0 };
  302. CalendarRangePatternListIndexType range_formats { 0 };
  303. CalendarRangePatternListIndexType range12_formats { 0 };
  304. CalendarSymbolsListIndexType symbols { 0 };
  305. };
  306. template<>
  307. struct AK::Formatter<Calendar> : Formatter<FormatString> {
  308. ErrorOr<void> format(FormatBuilder& builder, Calendar const& calendar)
  309. {
  310. return Formatter<FormatString>::format(builder,
  311. "{{ {}, {}, {}, {}, {}, {}, {}, {} }}",
  312. calendar.date_formats,
  313. calendar.time_formats,
  314. calendar.date_time_formats,
  315. calendar.available_formats,
  316. calendar.default_range_format,
  317. calendar.range_formats,
  318. calendar.range12_formats,
  319. calendar.symbols);
  320. }
  321. };
  322. template<>
  323. struct AK::Traits<Calendar> : public GenericTraits<Calendar> {
  324. static unsigned hash(Calendar const& c) { return c.hash(); }
  325. };
  326. struct TimeZoneNames {
  327. unsigned hash() const
  328. {
  329. auto hash = pair_int_hash(short_standard_name, long_standard_name);
  330. hash = pair_int_hash(hash, short_generic_name);
  331. hash = pair_int_hash(hash, long_generic_name);
  332. return hash;
  333. }
  334. bool operator==(TimeZoneNames const& other) const
  335. {
  336. return (short_standard_name == other.short_standard_name)
  337. && (long_standard_name == other.long_standard_name)
  338. && (short_generic_name == other.short_generic_name)
  339. && (long_generic_name == other.long_generic_name);
  340. }
  341. StringIndexType short_standard_name { 0 };
  342. StringIndexType long_standard_name { 0 };
  343. StringIndexType short_generic_name { 0 };
  344. StringIndexType long_generic_name { 0 };
  345. };
  346. template<>
  347. struct AK::Formatter<TimeZoneNames> : Formatter<FormatString> {
  348. ErrorOr<void> format(FormatBuilder& builder, TimeZoneNames const& time_zone)
  349. {
  350. return Formatter<FormatString>::format(builder,
  351. "{{ {}, {}, {}, {} }}",
  352. time_zone.short_standard_name,
  353. time_zone.long_standard_name,
  354. time_zone.short_generic_name,
  355. time_zone.long_generic_name);
  356. }
  357. };
  358. template<>
  359. struct AK::Traits<TimeZoneNames> : public GenericTraits<TimeZoneNames> {
  360. static unsigned hash(TimeZoneNames const& t) { return t.hash(); }
  361. };
  362. struct TimeZoneFormat {
  363. unsigned hash() const
  364. {
  365. auto hash = int_hash(symbol_ahead_sign);
  366. hash = pair_int_hash(hash, symbol_ahead_separator);
  367. hash = pair_int_hash(hash, symbol_behind_sign);
  368. hash = pair_int_hash(hash, symbol_behind_separator);
  369. hash = pair_int_hash(hash, gmt_format);
  370. hash = pair_int_hash(hash, gmt_zero_format);
  371. return hash;
  372. }
  373. bool operator==(TimeZoneFormat const& other) const
  374. {
  375. return (symbol_ahead_sign == other.symbol_ahead_sign)
  376. && (symbol_ahead_separator == other.symbol_ahead_separator)
  377. && (symbol_behind_sign == other.symbol_behind_sign)
  378. && (symbol_behind_separator == other.symbol_behind_separator)
  379. && (gmt_format == other.gmt_format)
  380. && (gmt_zero_format == other.gmt_zero_format);
  381. }
  382. StringIndexType symbol_ahead_sign { 0 };
  383. StringIndexType symbol_ahead_separator { 0 };
  384. StringIndexType symbol_behind_sign { 0 };
  385. StringIndexType symbol_behind_separator { 0 };
  386. StringIndexType gmt_format { 0 };
  387. StringIndexType gmt_zero_format { 0 };
  388. };
  389. template<>
  390. struct AK::Formatter<TimeZoneFormat> : Formatter<FormatString> {
  391. ErrorOr<void> format(FormatBuilder& builder, TimeZoneFormat const& time_zone_format)
  392. {
  393. return Formatter<FormatString>::format(builder, "{{ {}, {}, {}, {}, {}, {} }}",
  394. time_zone_format.symbol_ahead_sign,
  395. time_zone_format.symbol_ahead_separator,
  396. time_zone_format.symbol_behind_sign,
  397. time_zone_format.symbol_behind_separator,
  398. time_zone_format.gmt_format,
  399. time_zone_format.gmt_zero_format);
  400. }
  401. };
  402. template<>
  403. struct AK::Traits<TimeZoneFormat> : public GenericTraits<TimeZoneFormat> {
  404. static unsigned hash(TimeZoneFormat const& t) { return t.hash(); }
  405. };
  406. struct DayPeriod {
  407. unsigned hash() const
  408. {
  409. auto hash = int_hash(static_cast<u8>(day_period));
  410. hash = pair_int_hash(hash, begin);
  411. hash = pair_int_hash(hash, end);
  412. return hash;
  413. }
  414. bool operator==(DayPeriod const& other) const
  415. {
  416. return (day_period == other.day_period)
  417. && (begin == other.begin)
  418. && (end == other.end);
  419. }
  420. Unicode::DayPeriod day_period {};
  421. u8 begin { 0 };
  422. u8 end { 0 };
  423. };
  424. template<>
  425. struct AK::Formatter<DayPeriod> : Formatter<FormatString> {
  426. ErrorOr<void> format(FormatBuilder& builder, DayPeriod const& day_period)
  427. {
  428. return Formatter<FormatString>::format(builder,
  429. "{{ {}, {}, {} }}",
  430. static_cast<u8>(day_period.day_period),
  431. day_period.begin,
  432. day_period.end);
  433. }
  434. };
  435. template<>
  436. struct AK::Traits<DayPeriod> : public GenericTraits<DayPeriod> {
  437. static unsigned hash(DayPeriod const& d) { return d.hash(); }
  438. };
  439. using TimeZoneNamesList = Vector<TimeZoneNamesIndexType>;
  440. using DayPeriodList = Vector<DayPeriodIndexType>;
  441. using HourCycleList = Vector<Unicode::HourCycle>;
  442. template<>
  443. struct AK::Formatter<Unicode::HourCycle> : Formatter<FormatString> {
  444. ErrorOr<void> format(FormatBuilder& builder, Unicode::HourCycle hour_cycle)
  445. {
  446. return Formatter<FormatString>::format(builder, "{}", to_underlying(hour_cycle));
  447. }
  448. };
  449. struct Locale {
  450. HashMap<String, CalendarIndexType> calendars;
  451. TimeZoneNamesListIndexType time_zones { 0 };
  452. TimeZoneFormatIndexType time_zone_formats { 0 };
  453. DayPeriodListIndexType day_periods { 0 };
  454. };
  455. struct UnicodeLocaleData {
  456. UniqueStringStorage<StringIndexType> unique_strings;
  457. UniqueStorage<CalendarPattern, CalendarPatternIndexType> unique_patterns;
  458. UniqueStorage<CalendarPatternList, CalendarPatternListIndexType> unique_pattern_lists;
  459. UniqueStorage<CalendarRangePattern, CalendarRangePatternIndexType> unique_range_patterns;
  460. UniqueStorage<CalendarRangePatternList, CalendarRangePatternListIndexType> unique_range_pattern_lists;
  461. UniqueStorage<CalendarFormat, CalendarFormatIndexType> unique_formats;
  462. UniqueStorage<SymbolList, SymbolListIndexType> unique_symbol_lists;
  463. UniqueStorage<CalendarSymbols, CalendarSymbolsIndexType> unique_calendar_symbols;
  464. UniqueStorage<CalendarSymbolsList, CalendarSymbolsListIndexType> unique_calendar_symbols_lists;
  465. UniqueStorage<Calendar, CalendarIndexType> unique_calendars;
  466. UniqueStorage<TimeZoneNames, TimeZoneNamesIndexType> unique_time_zones;
  467. UniqueStorage<TimeZoneNamesList, TimeZoneNamesListIndexType> unique_time_zone_lists;
  468. UniqueStorage<TimeZoneFormat, TimeZoneFormatIndexType> unique_time_zone_formats;
  469. UniqueStorage<DayPeriod, DayPeriodIndexType> unique_day_periods;
  470. UniqueStorage<DayPeriodList, DayPeriodListIndexType> unique_day_period_lists;
  471. UniqueStorage<HourCycleList, HourCycleListIndexType> unique_hour_cycle_lists;
  472. HashMap<String, Locale> locales;
  473. HashMap<String, HourCycleListIndexType> hour_cycles;
  474. Vector<String> hour_cycle_regions;
  475. HashMap<String, Vector<TimeZone::TimeZone>> meta_zones;
  476. Vector<String> time_zones { "UTC"sv };
  477. Vector<String> calendars;
  478. Vector<Alias> calendar_aliases {
  479. // FIXME: Aliases should come from BCP47. See: https://unicode-org.atlassian.net/browse/CLDR-15158
  480. { "gregorian"sv, "gregory"sv },
  481. };
  482. };
  483. static Optional<Unicode::DayPeriod> day_period_from_string(StringView day_period)
  484. {
  485. if (day_period == "am"sv)
  486. return Unicode::DayPeriod::AM;
  487. if (day_period == "pm"sv)
  488. return Unicode::DayPeriod::PM;
  489. if (day_period == "morning1"sv)
  490. return Unicode::DayPeriod::Morning1;
  491. if (day_period == "morning2"sv)
  492. return Unicode::DayPeriod::Morning2;
  493. if (day_period == "afternoon1"sv)
  494. return Unicode::DayPeriod::Afternoon1;
  495. if (day_period == "afternoon2"sv)
  496. return Unicode::DayPeriod::Afternoon2;
  497. if (day_period == "evening1"sv)
  498. return Unicode::DayPeriod::Evening1;
  499. if (day_period == "evening2"sv)
  500. return Unicode::DayPeriod::Evening2;
  501. if (day_period == "night1"sv)
  502. return Unicode::DayPeriod::Night1;
  503. if (day_period == "night2"sv)
  504. return Unicode::DayPeriod::Night2;
  505. return {};
  506. };
  507. static ErrorOr<void> parse_hour_cycles(String core_path, UnicodeLocaleData& locale_data)
  508. {
  509. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  510. LexicalPath time_data_path(move(core_path));
  511. time_data_path = time_data_path.append("supplemental"sv);
  512. time_data_path = time_data_path.append("timeData.json"sv);
  513. auto time_data_file = TRY(Core::File::open(time_data_path.string(), Core::OpenMode::ReadOnly));
  514. auto time_data = TRY(JsonValue::from_string(time_data_file->read_all()));
  515. auto const& supplemental_object = time_data.as_object().get("supplemental"sv);
  516. auto const& time_data_object = supplemental_object.as_object().get("timeData"sv);
  517. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Unicode::HourCycle> {
  518. if (hour_cycle == "h"sv)
  519. return Unicode::HourCycle::H12;
  520. if (hour_cycle == "H"sv)
  521. return Unicode::HourCycle::H23;
  522. if (hour_cycle == "K"sv)
  523. return Unicode::HourCycle::H11;
  524. if (hour_cycle == "k"sv)
  525. return Unicode::HourCycle::H24;
  526. return {};
  527. };
  528. time_data_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  529. auto allowed_hour_cycles_string = value.as_object().get("_allowed"sv).as_string();
  530. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  531. Vector<Unicode::HourCycle> hour_cycles;
  532. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  533. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  534. hour_cycles.append(*hour_cycle);
  535. }
  536. auto hour_cycles_index = locale_data.unique_hour_cycle_lists.ensure(move(hour_cycles));
  537. locale_data.hour_cycles.set(key, hour_cycles_index);
  538. if (!locale_data.hour_cycle_regions.contains_slow(key))
  539. locale_data.hour_cycle_regions.append(key);
  540. });
  541. return {};
  542. };
  543. static ErrorOr<void> parse_meta_zones(String core_path, UnicodeLocaleData& locale_data)
  544. {
  545. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  546. LexicalPath meta_zone_path(move(core_path));
  547. meta_zone_path = meta_zone_path.append("supplemental"sv);
  548. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  549. auto meta_zone_file = TRY(Core::File::open(meta_zone_path.string(), Core::OpenMode::ReadOnly));
  550. auto meta_zone = TRY(JsonValue::from_string(meta_zone_file->read_all()));
  551. auto const& supplemental_object = meta_zone.as_object().get("supplemental"sv);
  552. auto const& meta_zone_object = supplemental_object.as_object().get("metaZones"sv);
  553. auto const& meta_zone_array = meta_zone_object.as_object().get("metazones"sv);
  554. meta_zone_array.as_array().for_each([&](JsonValue const& value) {
  555. auto const& mapping = value.as_object().get("mapZone"sv);
  556. auto const& meta_zone = mapping.as_object().get("_other"sv);
  557. auto const& golden_zone = mapping.as_object().get("_type"sv);
  558. if (auto time_zone = TimeZone::time_zone_from_string(golden_zone.as_string()); time_zone.has_value()) {
  559. auto& golden_zones = locale_data.meta_zones.ensure(meta_zone.as_string());
  560. golden_zones.append(*time_zone);
  561. }
  562. });
  563. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  564. if (auto time_zone = TimeZone::time_zone_from_string("UTC"sv); time_zone.has_value())
  565. locale_data.meta_zones.set("UTC"sv, { *time_zone });
  566. return {};
  567. };
  568. static constexpr auto is_char(char ch)
  569. {
  570. return [ch](auto c) { return c == ch; };
  571. }
  572. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  573. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  574. // extra spaces around the day period. Some example expected removals:
  575. //
  576. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  577. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  578. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  579. static String remove_period_from_pattern(String pattern)
  580. {
  581. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  582. auto index = pattern.find(remove);
  583. if (!index.has_value())
  584. continue;
  585. constexpr u32 space = ' ';
  586. constexpr u32 open = '{';
  587. constexpr u32 close = '}';
  588. Utf8View utf8_pattern { pattern };
  589. Optional<u32> before_removal;
  590. Optional<u32> after_removal;
  591. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  592. before_removal = *it;
  593. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  594. after_removal = *it;
  595. if ((before_removal == space) && (after_removal != open)) {
  596. pattern = String::formatted("{}{}",
  597. pattern.substring_view(0, *index - 1),
  598. pattern.substring_view(*index + remove.length()));
  599. } else if ((after_removal == space) && (before_removal != close)) {
  600. pattern = String::formatted("{}{}",
  601. pattern.substring_view(0, *index),
  602. pattern.substring_view(*index + remove.length() + 1));
  603. } else {
  604. pattern = String::formatted("{}{}",
  605. pattern.substring_view(0, *index),
  606. pattern.substring_view(*index + remove.length()));
  607. }
  608. }
  609. return pattern;
  610. }
  611. static Optional<CalendarPattern> parse_date_time_pattern_raw(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  612. {
  613. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  614. using Unicode::CalendarPatternStyle;
  615. CalendarPattern format {};
  616. if (!skeleton.is_empty())
  617. format.skeleton_index = locale_data.unique_strings.ensure(move(skeleton));
  618. GenericLexer lexer { pattern };
  619. StringBuilder builder;
  620. bool hour12 { false };
  621. while (!lexer.is_eof()) {
  622. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  623. // processing (this just avoids conflicts with the patterns below).
  624. if (lexer.next_is(is_quote)) {
  625. builder.append(lexer.consume_quoted_string());
  626. continue;
  627. }
  628. auto starting_char = lexer.peek();
  629. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  630. // Era
  631. if (all_of(segment, is_char('G'))) {
  632. builder.append("{era}");
  633. if (segment.length() <= 3)
  634. format.era = CalendarPatternStyle::Short;
  635. else if (segment.length() == 4)
  636. format.era = CalendarPatternStyle::Long;
  637. else
  638. format.era = CalendarPatternStyle::Narrow;
  639. }
  640. // Year
  641. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  642. builder.append("{year}");
  643. if (segment.length() == 2)
  644. format.year = CalendarPatternStyle::TwoDigit;
  645. else
  646. format.year = CalendarPatternStyle::Numeric;
  647. }
  648. // Quarter
  649. else if (all_of(segment, is_any_of("qQ"sv))) {
  650. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  651. return {};
  652. }
  653. // Month
  654. else if (all_of(segment, is_any_of("ML"sv))) {
  655. builder.append("{month}");
  656. if (segment.length() == 1)
  657. format.month = CalendarPatternStyle::Numeric;
  658. else if (segment.length() == 2)
  659. format.month = CalendarPatternStyle::TwoDigit;
  660. else if (segment.length() == 3)
  661. format.month = CalendarPatternStyle::Short;
  662. else if (segment.length() == 4)
  663. format.month = CalendarPatternStyle::Long;
  664. else if (segment.length() == 5)
  665. format.month = CalendarPatternStyle::Narrow;
  666. } else if (all_of(segment, is_char('l'))) {
  667. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  668. return {};
  669. }
  670. // Week
  671. else if (all_of(segment, is_any_of("wW"sv))) {
  672. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  673. return {};
  674. }
  675. // Day
  676. else if (all_of(segment, is_char('d'))) {
  677. builder.append("{day}");
  678. if (segment.length() == 1)
  679. format.day = CalendarPatternStyle::Numeric;
  680. else
  681. format.day = CalendarPatternStyle::TwoDigit;
  682. } else if (all_of(segment, is_any_of("DFg"sv))) {
  683. builder.append("{day}");
  684. format.day = CalendarPatternStyle::Numeric;
  685. }
  686. // Weekday
  687. else if (all_of(segment, is_char('E'))) {
  688. builder.append("{weekday}");
  689. if (segment.length() == 4)
  690. format.weekday = CalendarPatternStyle::Long;
  691. else if (segment.length() == 5)
  692. format.weekday = CalendarPatternStyle::Narrow;
  693. else
  694. format.weekday = CalendarPatternStyle::Short;
  695. } else if (all_of(segment, is_any_of("ec"sv))) {
  696. builder.append("{weekday}");
  697. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  698. // pattern styles are not supported by Intl.DateTimeFormat.
  699. if (segment.length() <= 2)
  700. return {};
  701. if (segment.length() == 4)
  702. format.weekday = CalendarPatternStyle::Long;
  703. else if (segment.length() == 5)
  704. format.weekday = CalendarPatternStyle::Narrow;
  705. else
  706. format.weekday = CalendarPatternStyle::Short;
  707. }
  708. // Period
  709. else if (all_of(segment, is_any_of("ab"sv))) {
  710. builder.append("{ampm}");
  711. hour12 = true;
  712. } else if (all_of(segment, is_char('B'))) {
  713. builder.append("{dayPeriod}");
  714. hour12 = true;
  715. if (segment.length() == 4)
  716. format.day_period = CalendarPatternStyle::Long;
  717. else if (segment.length() == 5)
  718. format.day_period = CalendarPatternStyle::Narrow;
  719. else
  720. format.day_period = CalendarPatternStyle::Short;
  721. }
  722. // Hour
  723. else if (all_of(segment, is_any_of("hHKk"sv))) {
  724. builder.append("{hour}");
  725. if ((segment[0] == 'h') || (segment[0] == 'K'))
  726. hour12 = true;
  727. if (segment.length() == 1)
  728. format.hour = CalendarPatternStyle::Numeric;
  729. else
  730. format.hour = CalendarPatternStyle::TwoDigit;
  731. } else if (all_of(segment, is_any_of("jJC"sv))) {
  732. // TR-35 indicates these should not be used.
  733. return {};
  734. }
  735. // Minute
  736. else if (all_of(segment, is_char('m'))) {
  737. builder.append("{minute}");
  738. if (segment.length() == 1)
  739. format.minute = CalendarPatternStyle::Numeric;
  740. else
  741. format.minute = CalendarPatternStyle::TwoDigit;
  742. }
  743. // Second
  744. else if (all_of(segment, is_char('s'))) {
  745. builder.append("{second}");
  746. if (segment.length() == 1)
  747. format.second = CalendarPatternStyle::Numeric;
  748. else
  749. format.second = CalendarPatternStyle::TwoDigit;
  750. } else if (all_of(segment, is_char('S'))) {
  751. builder.append("{fractionalSecondDigits}");
  752. VERIFY(segment.length() <= 3);
  753. format.fractional_second_digits = static_cast<u8>(segment.length());
  754. } else if (all_of(segment, is_char('A'))) {
  755. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  756. return {};
  757. }
  758. // Zone
  759. else if (all_of(segment, is_any_of("zV"sv))) {
  760. builder.append("{timeZoneName}");
  761. if (segment.length() < 4)
  762. format.time_zone_name = CalendarPatternStyle::Short;
  763. else
  764. format.time_zone_name = CalendarPatternStyle::Long;
  765. } else if (all_of(segment, is_any_of("ZOXx"sv))) {
  766. builder.append("{timeZoneName}");
  767. if (segment.length() < 4)
  768. format.time_zone_name = CalendarPatternStyle::ShortOffset;
  769. else
  770. format.time_zone_name = CalendarPatternStyle::LongOffset;
  771. } else if (all_of(segment, is_char('v'))) {
  772. builder.append("{timeZoneName}");
  773. if (segment.length() < 4)
  774. format.time_zone_name = CalendarPatternStyle::ShortGeneric;
  775. else
  776. format.time_zone_name = CalendarPatternStyle::LongGeneric;
  777. }
  778. // Non-patterns
  779. else {
  780. builder.append(segment);
  781. }
  782. }
  783. pattern = builder.build();
  784. if (hour12) {
  785. format.pattern = remove_period_from_pattern(pattern);
  786. format.pattern12 = move(pattern);
  787. } else {
  788. format.pattern = move(pattern);
  789. }
  790. return format;
  791. }
  792. static Optional<CalendarPatternIndexType> parse_date_time_pattern(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  793. {
  794. auto format = parse_date_time_pattern_raw(move(pattern), move(skeleton), locale_data);
  795. if (!format.has_value())
  796. return {};
  797. format->pattern_index = locale_data.unique_strings.ensure(move(format->pattern));
  798. if (format->pattern12.has_value())
  799. format->pattern12_index = locale_data.unique_strings.ensure(format->pattern12.release_value());
  800. return locale_data.unique_patterns.ensure(format.release_value());
  801. }
  802. template<typename... Chars>
  803. static constexpr bool char_is_one_of(char ch, Chars&&... chars)
  804. {
  805. return ((ch == chars) || ...);
  806. }
  807. static void parse_interval_patterns(Calendar& calendar, JsonObject const& interval_formats_object, UnicodeLocaleData& locale_data)
  808. {
  809. // https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats
  810. CalendarRangePatternList range_formats {};
  811. CalendarRangePatternList range12_formats {};
  812. auto name_of_field = [&](char field) {
  813. if (char_is_one_of(field, 'G'))
  814. return Unicode::CalendarRangePattern::Field::Era;
  815. if (char_is_one_of(field, 'y', 'Y', 'u', 'U', 'r'))
  816. return Unicode::CalendarRangePattern::Field::Year;
  817. if (char_is_one_of(field, 'M', 'L'))
  818. return Unicode::CalendarRangePattern::Field::Month;
  819. if (char_is_one_of(field, 'd', 'D', 'F', 'g'))
  820. return Unicode::CalendarRangePattern::Field::Day;
  821. if (char_is_one_of(field, 'a', 'b'))
  822. return Unicode::CalendarRangePattern::Field::AmPm;
  823. if (char_is_one_of(field, 'B'))
  824. return Unicode::CalendarRangePattern::Field::DayPeriod;
  825. if (char_is_one_of(field, 'h', 'H', 'K', 'k'))
  826. return Unicode::CalendarRangePattern::Field::Hour;
  827. if (char_is_one_of(field, 'm'))
  828. return Unicode::CalendarRangePattern::Field::Minute;
  829. if (char_is_one_of(field, 's'))
  830. return Unicode::CalendarRangePattern::Field::Second;
  831. if (char_is_one_of(field, 'S'))
  832. return Unicode::CalendarRangePattern::Field::FractionalSecondDigits;
  833. VERIFY_NOT_REACHED();
  834. };
  835. auto split_default_range_pattern = [&](auto skeleton, auto const& pattern) {
  836. auto start_range_end = pattern.find('}').value() + 1;
  837. auto end_range_begin = pattern.find_last('{').value();
  838. auto start_range = pattern.substring_view(0, start_range_end);
  839. auto separator = pattern.substring_view(start_range_end, end_range_begin - start_range_end);
  840. auto end_range = pattern.substring_view(end_range_begin);
  841. CalendarRangePattern format {};
  842. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  843. format.start_range = locale_data.unique_strings.ensure(start_range);
  844. format.separator = locale_data.unique_strings.ensure(separator);
  845. format.end_range = locale_data.unique_strings.ensure(end_range);
  846. return format;
  847. };
  848. auto split_range_pattern = [&](auto skeleton, auto field, auto const& pattern, auto const& parsed_fields) {
  849. HashMap<StringView, size_t> partitions;
  850. StringView last_partition;
  851. auto begin_index = pattern.find('{');
  852. size_t end_index = 0;
  853. while (begin_index.has_value()) {
  854. end_index = pattern.find('}', *begin_index).value();
  855. auto partition = pattern.substring_view(*begin_index, end_index - *begin_index);
  856. if (partitions.contains(partition))
  857. break;
  858. partitions.set(partition, *begin_index);
  859. last_partition = partition;
  860. begin_index = pattern.find('{', end_index + 1);
  861. }
  862. VERIFY(begin_index.has_value() && !last_partition.is_empty());
  863. auto start_range_end = partitions.get(last_partition).value() + last_partition.length() + 1;
  864. auto start_range = pattern.substring_view(0, start_range_end);
  865. auto separator = pattern.substring_view(start_range_end, *begin_index - start_range_end);
  866. auto end_range = pattern.substring_view(*begin_index);
  867. CalendarRangePattern format {};
  868. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  869. format.field = field;
  870. format.start_range = locale_data.unique_strings.ensure(start_range);
  871. format.separator = locale_data.unique_strings.ensure(separator);
  872. format.end_range = locale_data.unique_strings.ensure(end_range);
  873. format.for_each_calendar_field_zipped_with(parsed_fields, [](auto& format_field, auto const& parsed_field, auto) {
  874. format_field = parsed_field;
  875. });
  876. return format;
  877. };
  878. interval_formats_object.for_each_member([&](auto const& skeleton, auto const& value) {
  879. if (skeleton == "intervalFormatFallback"sv) {
  880. auto range_format = split_default_range_pattern(skeleton, value.as_string());
  881. calendar.default_range_format = locale_data.unique_range_patterns.ensure(move(range_format));
  882. return;
  883. }
  884. value.as_object().for_each_member([&](auto const& field, auto const& pattern) {
  885. if (field.ends_with("alt-variant"sv))
  886. return;
  887. VERIFY(field.length() == 1);
  888. auto name = name_of_field(field[0]);
  889. auto format = parse_date_time_pattern_raw(pattern.as_string(), skeleton, locale_data).release_value();
  890. auto range_format = split_range_pattern(skeleton, name, format.pattern, format);
  891. range_formats.append(locale_data.unique_range_patterns.ensure(move(range_format)));
  892. if (format.pattern12.has_value()) {
  893. auto range12_pattern = split_range_pattern(skeleton, name, *format.pattern12, format);
  894. range12_formats.append(locale_data.unique_range_patterns.ensure(move(range12_pattern)));
  895. } else {
  896. range12_formats.append(range_formats.last());
  897. }
  898. });
  899. });
  900. calendar.range_formats = locale_data.unique_range_pattern_lists.ensure(move(range_formats));
  901. calendar.range12_formats = locale_data.unique_range_pattern_lists.ensure(move(range12_formats));
  902. }
  903. static void generate_missing_patterns(Calendar& calendar, CalendarPatternList& formats, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, UnicodeLocaleData& locale_data)
  904. {
  905. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  906. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  907. auto pattern = locale_data.unique_strings.get(format);
  908. auto time_pattern = locale_data.unique_strings.get(time_format);
  909. auto date_pattern = locale_data.unique_strings.get(date_format);
  910. auto new_pattern = pattern.replace("{0}", time_pattern).replace("{1}", date_pattern);
  911. return locale_data.unique_strings.ensure(move(new_pattern));
  912. };
  913. auto inject_fractional_second_digits = [&](auto format) {
  914. auto pattern = locale_data.unique_strings.get(format);
  915. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv);
  916. return locale_data.unique_strings.ensure(move(new_pattern));
  917. };
  918. auto append_if_unique = [&](auto format) {
  919. auto format_index = locale_data.unique_patterns.ensure(move(format));
  920. if (!formats.contains_slow(format_index))
  921. formats.append(format_index);
  922. };
  923. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  924. for (auto const& format : date_formats)
  925. append_if_unique(format);
  926. for (auto const& format : time_formats) {
  927. append_if_unique(format);
  928. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  929. auto new_format = format;
  930. new_format.fractional_second_digits = 2;
  931. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  932. if (new_format.pattern12_index != 0)
  933. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  934. time_formats_with_fractional_second_digits.append(new_format);
  935. append_if_unique(move(new_format));
  936. }
  937. }
  938. time_formats.extend(move(time_formats_with_fractional_second_digits));
  939. for (auto const& date_format : date_formats) {
  940. auto const& date_time_formats = locale_data.unique_formats.get(calendar.date_time_formats);
  941. CalendarPatternIndexType date_time_format_index = 0;
  942. if (date_format.month == Unicode::CalendarPatternStyle::Long) {
  943. if (date_format.weekday.has_value())
  944. date_time_format_index = date_time_formats.full_format;
  945. else
  946. date_time_format_index = date_time_formats.long_format;
  947. } else if (date_format.month == Unicode::CalendarPatternStyle::Short) {
  948. date_time_format_index = date_time_formats.medium_format;
  949. } else {
  950. date_time_format_index = date_time_formats.short_format;
  951. }
  952. for (auto const& time_format : time_formats) {
  953. auto format = locale_data.unique_patterns.get(date_time_format_index);
  954. if (time_format.pattern12_index != 0)
  955. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  956. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  957. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  958. if (date_field.has_value())
  959. field = date_field;
  960. });
  961. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  962. if (time_field.has_value())
  963. field = time_field;
  964. });
  965. append_if_unique(move(format));
  966. }
  967. }
  968. }
  969. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, UnicodeLocaleData& locale_data)
  970. {
  971. auto create_symbol_lists = [](size_t size) {
  972. SymbolList narrow_symbol_list;
  973. SymbolList short_symbol_list;
  974. SymbolList long_symbol_list;
  975. narrow_symbol_list.resize(size);
  976. short_symbol_list.resize(size);
  977. long_symbol_list.resize(size);
  978. return Array<SymbolList, 3> { {
  979. move(narrow_symbol_list),
  980. move(short_symbol_list),
  981. move(long_symbol_list),
  982. } };
  983. };
  984. CalendarSymbolsList symbols_list {};
  985. auto store_symbol_lists = [&](auto symbol, auto symbol_lists) {
  986. auto symbol_index = to_underlying(symbol);
  987. if (symbol_index >= symbols_list.size())
  988. symbols_list.resize(symbol_index + 1);
  989. CalendarSymbols symbols {};
  990. symbols.narrow_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[0]));
  991. symbols.short_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[1]));
  992. symbols.long_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[2]));
  993. auto calendar_symbols_index = locale_data.unique_calendar_symbols.ensure(move(symbols));
  994. symbols_list[symbol_index] = calendar_symbols_index;
  995. };
  996. auto parse_era_symbols = [&](auto const& symbols_object) {
  997. auto const& narrow_symbols = symbols_object.get("eraNarrow"sv).as_object();
  998. auto const& short_symbols = symbols_object.get("eraAbbr"sv).as_object();
  999. auto const& long_symbols = symbols_object.get("eraNames"sv).as_object();
  1000. auto symbol_lists = create_symbol_lists(2);
  1001. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1002. if (auto key_index = key.to_uint(); key_index.has_value())
  1003. symbols[*key_index] = locale_data.unique_strings.ensure(move(symbol));
  1004. };
  1005. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1006. append_symbol(symbol_lists[0], key, value.as_string());
  1007. });
  1008. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1009. append_symbol(symbol_lists[1], key, value.as_string());
  1010. });
  1011. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1012. append_symbol(symbol_lists[2], key, value.as_string());
  1013. });
  1014. store_symbol_lists(Unicode::CalendarSymbol::Era, move(symbol_lists));
  1015. };
  1016. auto parse_month_symbols = [&](auto const& symbols_object) {
  1017. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1018. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1019. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1020. auto symbol_lists = create_symbol_lists(12);
  1021. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1022. auto key_index = key.to_uint().value() - 1;
  1023. symbols[key_index] = locale_data.unique_strings.ensure(move(symbol));
  1024. };
  1025. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1026. append_symbol(symbol_lists[0], key, value.as_string());
  1027. });
  1028. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1029. append_symbol(symbol_lists[1], key, value.as_string());
  1030. });
  1031. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1032. append_symbol(symbol_lists[2], key, value.as_string());
  1033. });
  1034. store_symbol_lists(Unicode::CalendarSymbol::Month, move(symbol_lists));
  1035. };
  1036. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  1037. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1038. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1039. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1040. auto symbol_lists = create_symbol_lists(7);
  1041. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1042. if (key == "sun"sv)
  1043. symbols[to_underlying(Unicode::Weekday::Sunday)] = locale_data.unique_strings.ensure(move(symbol));
  1044. else if (key == "mon"sv)
  1045. symbols[to_underlying(Unicode::Weekday::Monday)] = locale_data.unique_strings.ensure(move(symbol));
  1046. else if (key == "tue"sv)
  1047. symbols[to_underlying(Unicode::Weekday::Tuesday)] = locale_data.unique_strings.ensure(move(symbol));
  1048. else if (key == "wed"sv)
  1049. symbols[to_underlying(Unicode::Weekday::Wednesday)] = locale_data.unique_strings.ensure(move(symbol));
  1050. else if (key == "thu"sv)
  1051. symbols[to_underlying(Unicode::Weekday::Thursday)] = locale_data.unique_strings.ensure(move(symbol));
  1052. else if (key == "fri"sv)
  1053. symbols[to_underlying(Unicode::Weekday::Friday)] = locale_data.unique_strings.ensure(move(symbol));
  1054. else if (key == "sat"sv)
  1055. symbols[to_underlying(Unicode::Weekday::Saturday)] = locale_data.unique_strings.ensure(move(symbol));
  1056. };
  1057. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1058. append_symbol(symbol_lists[0], key, value.as_string());
  1059. });
  1060. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1061. append_symbol(symbol_lists[1], key, value.as_string());
  1062. });
  1063. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1064. append_symbol(symbol_lists[2], key, value.as_string());
  1065. });
  1066. store_symbol_lists(Unicode::CalendarSymbol::Weekday, move(symbol_lists));
  1067. };
  1068. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  1069. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1070. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1071. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1072. auto symbol_lists = create_symbol_lists(10);
  1073. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1074. if (auto day_period = day_period_from_string(key); day_period.has_value())
  1075. symbols[to_underlying(*day_period)] = locale_data.unique_strings.ensure(move(symbol));
  1076. };
  1077. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1078. append_symbol(symbol_lists[0], key, value.as_string());
  1079. });
  1080. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1081. append_symbol(symbol_lists[1], key, value.as_string());
  1082. });
  1083. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1084. append_symbol(symbol_lists[2], key, value.as_string());
  1085. });
  1086. store_symbol_lists(Unicode::CalendarSymbol::DayPeriod, move(symbol_lists));
  1087. };
  1088. parse_era_symbols(calendar_object.get("eras"sv).as_object());
  1089. parse_month_symbols(calendar_object.get("months"sv).as_object().get("format"sv).as_object());
  1090. parse_weekday_symbols(calendar_object.get("days"sv).as_object().get("format"sv).as_object());
  1091. parse_day_period_symbols(calendar_object.get("dayPeriods"sv).as_object().get("format"sv).as_object());
  1092. calendar.symbols = locale_data.unique_calendar_symbols_lists.ensure(move(symbols_list));
  1093. }
  1094. static ErrorOr<void> parse_calendars(String locale_calendars_path, UnicodeLocaleData& locale_data, Locale& locale)
  1095. {
  1096. LexicalPath calendars_path(move(locale_calendars_path));
  1097. if (!calendars_path.basename().starts_with("ca-"sv))
  1098. return {};
  1099. auto calendars_file = TRY(Core::File::open(calendars_path.string(), Core::OpenMode::ReadOnly));
  1100. auto calendars = TRY(JsonValue::from_string(calendars_file->read_all()));
  1101. auto const& main_object = calendars.as_object().get("main"sv);
  1102. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  1103. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1104. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  1105. auto parse_patterns = [&](auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) {
  1106. auto parse_pattern = [&](auto name) {
  1107. auto format = patterns_object.get(name);
  1108. auto skeleton = skeletons_object.get(name);
  1109. auto format_index = parse_date_time_pattern(format.as_string(), skeleton.as_string_or(String::empty()), locale_data).value();
  1110. if (patterns)
  1111. patterns->append(locale_data.unique_patterns.get(format_index));
  1112. return format_index;
  1113. };
  1114. CalendarFormat formats {};
  1115. formats.full_format = parse_pattern("full"sv);
  1116. formats.long_format = parse_pattern("long"sv);
  1117. formats.medium_format = parse_pattern("medium"sv);
  1118. formats.short_format = parse_pattern("short"sv);
  1119. return locale_data.unique_formats.ensure(move(formats));
  1120. };
  1121. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  1122. // The generic calendar is not a supported Unicode calendar key, so skip it:
  1123. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  1124. if (calendar_name == "generic"sv)
  1125. return;
  1126. Calendar calendar {};
  1127. CalendarPatternList available_formats {};
  1128. if (!locale_data.calendars.contains_slow(calendar_name))
  1129. locale_data.calendars.append(calendar_name);
  1130. Vector<CalendarPattern> date_formats;
  1131. Vector<CalendarPattern> time_formats;
  1132. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  1133. auto const& date_skeletons_object = value.as_object().get("dateSkeletons"sv);
  1134. calendar.date_formats = parse_patterns(date_formats_object.as_object(), date_skeletons_object.as_object(), &date_formats);
  1135. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  1136. auto const& time_skeletons_object = value.as_object().get("timeSkeletons"sv);
  1137. calendar.time_formats = parse_patterns(time_formats_object.as_object(), time_skeletons_object.as_object(), &time_formats);
  1138. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  1139. calendar.date_time_formats = parse_patterns(date_time_formats_object.as_object(), JsonObject {}, nullptr);
  1140. auto const& available_formats_object = date_time_formats_object.as_object().get("availableFormats"sv);
  1141. available_formats_object.as_object().for_each_member([&](auto const& skeleton, JsonValue const& pattern) {
  1142. auto pattern_index = parse_date_time_pattern(pattern.as_string(), skeleton, locale_data);
  1143. if (!pattern_index.has_value())
  1144. return;
  1145. auto const& format = locale_data.unique_patterns.get(*pattern_index);
  1146. if (format.contains_only_date_fields())
  1147. date_formats.append(format);
  1148. else if (format.contains_only_time_fields())
  1149. time_formats.append(format);
  1150. if (!available_formats.contains_slow(*pattern_index))
  1151. available_formats.append(*pattern_index);
  1152. });
  1153. auto const& interval_formats_object = date_time_formats_object.as_object().get("intervalFormats"sv);
  1154. parse_interval_patterns(calendar, interval_formats_object.as_object(), locale_data);
  1155. generate_missing_patterns(calendar, available_formats, move(date_formats), move(time_formats), locale_data);
  1156. parse_calendar_symbols(calendar, value.as_object(), locale_data);
  1157. calendar.available_formats = locale_data.unique_pattern_lists.ensure(move(available_formats));
  1158. locale.calendars.set(calendar_name, locale_data.unique_calendars.ensure(move(calendar)));
  1159. });
  1160. return {};
  1161. }
  1162. static ErrorOr<void> parse_time_zone_names(String locale_time_zone_names_path, UnicodeLocaleData& locale_data, Locale& locale)
  1163. {
  1164. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  1165. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  1166. auto time_zone_names_file = TRY(Core::File::open(time_zone_names_path.string(), Core::OpenMode::ReadOnly));
  1167. auto time_zone_names = TRY(JsonValue::from_string(time_zone_names_file->read_all()));
  1168. auto const& main_object = time_zone_names.as_object().get("main"sv);
  1169. auto const& locale_object = main_object.as_object().get(time_zone_names_path.parent().basename());
  1170. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1171. auto const& time_zone_names_object = dates_object.as_object().get("timeZoneNames"sv);
  1172. auto const& meta_zone_object = time_zone_names_object.as_object().get("metazone"sv);
  1173. auto const& hour_format_string = time_zone_names_object.as_object().get("hourFormat"sv);
  1174. auto const& gmt_format_string = time_zone_names_object.as_object().get("gmtFormat"sv);
  1175. auto const& gmt_zero_format_string = time_zone_names_object.as_object().get("gmtZeroFormat"sv);
  1176. if (meta_zone_object.is_null())
  1177. return {};
  1178. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object, Span<StringView const> keys) -> Optional<StringIndexType> {
  1179. auto const& names = meta_zone_object.get(type);
  1180. if (!names.is_object())
  1181. return {};
  1182. for (auto key : keys) {
  1183. auto const& name = names.as_object().get(key);
  1184. if (name.is_string())
  1185. return locale_data.unique_strings.ensure(name.as_string());
  1186. }
  1187. return {};
  1188. };
  1189. auto parse_hour_format = [&](auto const& format, auto& time_zone_formats) {
  1190. auto hour_formats = format.split_view(';');
  1191. auto hour_format_ahead_start = hour_formats[0].find('H').value();
  1192. auto separator_ahead_start = hour_formats[0].find_last('H').value() + 1;
  1193. auto separator_ahead_end = hour_formats[0].find('m').value();
  1194. auto hour_format_behind_start = hour_formats[1].find('H').value();
  1195. auto separator_behind_start = hour_formats[1].find_last('H').value() + 1;
  1196. auto separator_behind_end = hour_formats[1].find('m').value();
  1197. auto symbol_ahead_sign = hour_formats[0].substring_view(0, hour_format_ahead_start);
  1198. auto symbol_ahead_separator = hour_formats[0].substring_view(separator_ahead_start, separator_ahead_end - separator_ahead_start);
  1199. auto symbol_behind_sign = hour_formats[1].substring_view(0, hour_format_behind_start);
  1200. auto symbol_behind_separator = hour_formats[1].substring_view(separator_behind_start, separator_behind_end - separator_behind_start);
  1201. time_zone_formats.symbol_ahead_sign = locale_data.unique_strings.ensure(symbol_ahead_sign);
  1202. time_zone_formats.symbol_ahead_separator = locale_data.unique_strings.ensure(symbol_ahead_separator);
  1203. time_zone_formats.symbol_behind_sign = locale_data.unique_strings.ensure(symbol_behind_sign);
  1204. time_zone_formats.symbol_behind_separator = locale_data.unique_strings.ensure(symbol_behind_separator);
  1205. };
  1206. TimeZoneNamesList time_zones;
  1207. TimeZoneFormat time_zone_formats {};
  1208. parse_hour_format(hour_format_string.as_string(), time_zone_formats);
  1209. time_zone_formats.gmt_format = locale_data.unique_strings.ensure(gmt_format_string.as_string());
  1210. time_zone_formats.gmt_zero_format = locale_data.unique_strings.ensure(gmt_zero_format_string.as_string());
  1211. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  1212. constexpr auto standard_keys = Array { "daylight"sv, "standard"sv };
  1213. constexpr auto generic_keys = Array { "generic"sv };
  1214. auto golden_zones = locale_data.meta_zones.find(meta_zone);
  1215. if (golden_zones == locale_data.meta_zones.end())
  1216. return;
  1217. TimeZoneNames time_zone_names {};
  1218. if (auto name = parse_name("long"sv, meta_zone_object, standard_keys); name.has_value())
  1219. time_zone_names.long_standard_name = name.value();
  1220. if (auto name = parse_name("short"sv, meta_zone_object, standard_keys); name.has_value())
  1221. time_zone_names.short_standard_name = name.value();
  1222. if (auto name = parse_name("long"sv, meta_zone_object, generic_keys); name.has_value())
  1223. time_zone_names.long_generic_name = name.value();
  1224. if (auto name = parse_name("short"sv, meta_zone_object, generic_keys); name.has_value())
  1225. time_zone_names.short_generic_name = name.value();
  1226. auto time_zone_index = locale_data.unique_time_zones.ensure(move(time_zone_names));
  1227. for (auto golden_zone : golden_zones->value) {
  1228. auto time_zone = to_underlying(golden_zone);
  1229. if (time_zone >= time_zones.size())
  1230. time_zones.resize(time_zone + 1);
  1231. time_zones[time_zone] = time_zone_index;
  1232. }
  1233. };
  1234. meta_zone_object.as_object().for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  1235. parse_time_zone(meta_zone, value.as_object());
  1236. });
  1237. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  1238. auto const& zone_object = time_zone_names_object.as_object().get("zone"sv);
  1239. auto const& etc_object = zone_object.as_object().get("Etc"sv);
  1240. auto const& utc_object = etc_object.as_object().get("UTC"sv);
  1241. parse_time_zone("UTC"sv, utc_object.as_object());
  1242. locale.time_zones = locale_data.unique_time_zone_lists.ensure(move(time_zones));
  1243. locale.time_zone_formats = locale_data.unique_time_zone_formats.ensure(move(time_zone_formats));
  1244. return {};
  1245. }
  1246. static ErrorOr<void> parse_day_periods(String core_path, UnicodeLocaleData& locale_data)
  1247. {
  1248. // https://unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rule_Sets
  1249. LexicalPath day_periods_path(move(core_path));
  1250. day_periods_path = day_periods_path.append("supplemental"sv);
  1251. day_periods_path = day_periods_path.append("dayPeriods.json"sv);
  1252. auto day_periods_file = TRY(Core::File::open(day_periods_path.string(), Core::OpenMode::ReadOnly));
  1253. auto locale_day_periods = TRY(JsonValue::from_string(day_periods_file->read_all()));
  1254. auto const& supplemental_object = locale_day_periods.as_object().get("supplemental"sv);
  1255. auto const& day_periods_object = supplemental_object.as_object().get("dayPeriodRuleSet"sv);
  1256. auto parse_hour = [](auto const& time) {
  1257. auto hour_end_index = time.find(':').value();
  1258. // The times are in the form "hours:minutes", but we only need the hour segment.
  1259. // TR-35 explicitly states that minutes other than :00 must not be used.
  1260. VERIFY(time.substring_view(hour_end_index) == ":00"sv);
  1261. auto hour = time.substring_view(0, hour_end_index);
  1262. return hour.template to_uint<u8>().value();
  1263. };
  1264. auto parse_day_period = [&](auto const& symbol, auto const& ranges) -> Optional<DayPeriod> {
  1265. auto day_period = day_period_from_string(symbol);
  1266. if (!day_period.has_value())
  1267. return {};
  1268. auto begin = parse_hour(ranges.get("_from"sv).as_string());
  1269. auto end = parse_hour(ranges.get("_before"sv).as_string());
  1270. return DayPeriod { *day_period, begin, end };
  1271. };
  1272. day_periods_object.as_object().for_each_member([&](auto const& language, JsonValue const& value) {
  1273. auto locale = locale_data.locales.find(language);
  1274. if (locale == locale_data.locales.end())
  1275. return;
  1276. DayPeriodList day_periods;
  1277. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& ranges) {
  1278. if (auto day_period = parse_day_period(symbol, ranges.as_object()); day_period.has_value()) {
  1279. auto day_period_index = locale_data.unique_day_periods.ensure(day_period.release_value());
  1280. day_periods.append(day_period_index);
  1281. }
  1282. });
  1283. locale->value.day_periods = locale_data.unique_day_period_lists.ensure(move(day_periods));
  1284. });
  1285. return {};
  1286. };
  1287. static ErrorOr<void> parse_all_locales(String core_path, String dates_path, UnicodeLocaleData& locale_data)
  1288. {
  1289. TRY(parse_hour_cycles(core_path, locale_data));
  1290. TRY(parse_meta_zones(core_path, locale_data));
  1291. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  1292. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  1293. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  1294. StringBuilder builder;
  1295. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  1296. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  1297. builder.appendff("-{}", script);
  1298. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  1299. builder.appendff("-{}", region);
  1300. return builder.build();
  1301. };
  1302. while (dates_iterator.has_next()) {
  1303. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  1304. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  1305. auto language = TRY(remove_variants_from_path(dates_path));
  1306. auto& locale = locale_data.locales.ensure(language);
  1307. while (calendars_iterator.has_next()) {
  1308. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  1309. TRY(parse_calendars(move(calendars_path), locale_data, locale));
  1310. }
  1311. TRY(parse_time_zone_names(move(dates_path), locale_data, locale));
  1312. }
  1313. TRY(parse_day_periods(move(core_path), locale_data));
  1314. return {};
  1315. }
  1316. static String format_identifier(StringView owner, String identifier)
  1317. {
  1318. identifier = identifier.replace("-"sv, "_"sv, true);
  1319. identifier = identifier.replace("/"sv, "_"sv, true);
  1320. if (all_of(identifier, is_ascii_digit))
  1321. return String::formatted("{}_{}", owner[0], identifier);
  1322. if (is_ascii_lower_alpha(identifier[0]))
  1323. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  1324. return identifier;
  1325. }
  1326. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  1327. {
  1328. StringBuilder builder;
  1329. SourceGenerator generator { builder };
  1330. generator.append(R"~~~(
  1331. #pragma once
  1332. #include <AK/Types.h>
  1333. namespace Unicode {
  1334. )~~~");
  1335. generate_enum(generator, format_identifier, "Calendar"sv, {}, locale_data.calendars, locale_data.calendar_aliases);
  1336. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, locale_data.hour_cycle_regions);
  1337. generator.append(R"~~~(
  1338. }
  1339. )~~~");
  1340. VERIFY(file.write(generator.as_string_view()));
  1341. }
  1342. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  1343. {
  1344. StringBuilder builder;
  1345. SourceGenerator generator { builder };
  1346. generator.set("string_index_type"sv, s_string_index_type);
  1347. generator.set("calendar_pattern_index_type"sv, s_calendar_pattern_index_type);
  1348. generator.set("calendar_pattern_list_index_type"sv, s_calendar_pattern_list_index_type);
  1349. generator.set("calendar_range_pattern_index_type"sv, s_calendar_range_pattern_index_type);
  1350. generator.set("calendar_range_pattern_list_index_type"sv, s_calendar_range_pattern_list_index_type);
  1351. generator.set("calendar_format_index_type"sv, s_calendar_format_index_type);
  1352. generator.set("symbol_list_index_type"sv, s_symbol_list_index_type);
  1353. generator.set("calendar_symbols_index_type"sv, s_calendar_symbols_index_type);
  1354. generator.set("calendar_symbols_list_index_type"sv, s_calendar_symbols_list_index_type);
  1355. generator.set("calendar_index_type"sv, s_calendar_index_type);
  1356. generator.set("time_zone_index_type"sv, s_time_zone_index_type);
  1357. generator.set("time_zone_list_index_type"sv, s_time_zone_list_index_type);
  1358. generator.set("day_period_index_type"sv, s_day_period_index_type);
  1359. generator.set("day_period_list_index_type"sv, s_day_period_list_index_type);
  1360. generator.append(R"~~~(
  1361. #include <AK/Array.h>
  1362. #include <AK/BinarySearch.h>
  1363. #include <AK/Optional.h>
  1364. #include <AK/StringView.h>
  1365. #include <LibTimeZone/TimeZone.h>
  1366. #include <LibUnicode/DateTimeFormat.h>
  1367. #include <LibUnicode/Locale.h>
  1368. #include <LibUnicode/UnicodeDateTimeFormat.h>
  1369. namespace Unicode {
  1370. )~~~");
  1371. locale_data.unique_strings.generate(generator);
  1372. generator.append(R"~~~(
  1373. template <typename SourceType, typename TargetType>
  1374. static void convert_calendar_fields(SourceType const& source, TargetType& target)
  1375. {
  1376. if (source.era != -1)
  1377. target.era = static_cast<CalendarPatternStyle>(source.era);
  1378. if (source.year != -1)
  1379. target.year = static_cast<CalendarPatternStyle>(source.year);
  1380. if (source.month != -1)
  1381. target.month = static_cast<CalendarPatternStyle>(source.month);
  1382. if (source.weekday != -1)
  1383. target.weekday = static_cast<CalendarPatternStyle>(source.weekday);
  1384. if (source.day != -1)
  1385. target.day = static_cast<CalendarPatternStyle>(source.day);
  1386. if (source.day_period != -1)
  1387. target.day_period = static_cast<CalendarPatternStyle>(source.day_period);
  1388. if (source.hour != -1)
  1389. target.hour = static_cast<CalendarPatternStyle>(source.hour);
  1390. if (source.minute != -1)
  1391. target.minute = static_cast<CalendarPatternStyle>(source.minute);
  1392. if (source.second != -1)
  1393. target.second = static_cast<CalendarPatternStyle>(source.second);
  1394. if (source.fractional_second_digits != -1)
  1395. target.fractional_second_digits = static_cast<u8>(source.fractional_second_digits);
  1396. if (source.time_zone_name != -1)
  1397. target.time_zone_name = static_cast<CalendarPatternStyle>(source.time_zone_name);
  1398. }
  1399. struct CalendarPatternImpl {
  1400. CalendarPattern to_unicode_calendar_pattern() const {
  1401. CalendarPattern calendar_pattern {};
  1402. calendar_pattern.skeleton = s_string_list[skeleton];
  1403. calendar_pattern.pattern = s_string_list[pattern];
  1404. if (pattern12 != 0)
  1405. calendar_pattern.pattern12 = s_string_list[pattern12];
  1406. convert_calendar_fields(*this, calendar_pattern);
  1407. return calendar_pattern;
  1408. }
  1409. @string_index_type@ skeleton { 0 };
  1410. @string_index_type@ pattern { 0 };
  1411. @string_index_type@ pattern12 { 0 };
  1412. i8 era { -1 };
  1413. i8 year { -1 };
  1414. i8 month { -1 };
  1415. i8 weekday { -1 };
  1416. i8 day { -1 };
  1417. i8 day_period { -1 };
  1418. i8 hour { -1 };
  1419. i8 minute { -1 };
  1420. i8 second { -1 };
  1421. i8 fractional_second_digits { -1 };
  1422. i8 time_zone_name { -1 };
  1423. };
  1424. struct CalendarRangePatternImpl {
  1425. CalendarRangePattern to_unicode_calendar_range_pattern() const {
  1426. CalendarRangePattern calendar_range_pattern {};
  1427. if (field != -1)
  1428. calendar_range_pattern.field = static_cast<CalendarRangePattern::Field>(field);
  1429. calendar_range_pattern.start_range = s_string_list[start_range];
  1430. calendar_range_pattern.separator = s_string_list[separator];
  1431. calendar_range_pattern.end_range = s_string_list[end_range];
  1432. convert_calendar_fields(*this, calendar_range_pattern);
  1433. return calendar_range_pattern;
  1434. }
  1435. @string_index_type@ skeleton { 0 };
  1436. i8 field { -1 };
  1437. @string_index_type@ start_range { 0 };
  1438. @string_index_type@ separator { 0 };
  1439. @string_index_type@ end_range { 0 };
  1440. i8 era { -1 };
  1441. i8 year { -1 };
  1442. i8 month { -1 };
  1443. i8 weekday { -1 };
  1444. i8 day { -1 };
  1445. i8 day_period { -1 };
  1446. i8 hour { -1 };
  1447. i8 minute { -1 };
  1448. i8 second { -1 };
  1449. i8 fractional_second_digits { -1 };
  1450. i8 time_zone_name { -1 };
  1451. };
  1452. )~~~");
  1453. locale_data.unique_patterns.generate(generator, "CalendarPatternImpl"sv, "s_calendar_patterns"sv, 10);
  1454. locale_data.unique_pattern_lists.generate(generator, s_calendar_pattern_index_type, "s_calendar_pattern_lists"sv);
  1455. locale_data.unique_range_patterns.generate(generator, "CalendarRangePatternImpl"sv, "s_calendar_range_patterns"sv, 10);
  1456. locale_data.unique_range_pattern_lists.generate(generator, s_calendar_range_pattern_index_type, "s_calendar_range_pattern_lists"sv);
  1457. generator.append(R"~~~(
  1458. struct CalendarFormatImpl {
  1459. CalendarFormat to_unicode_calendar_format() const {
  1460. CalendarFormat calendar_format {};
  1461. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  1462. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  1463. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  1464. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  1465. return calendar_format;
  1466. }
  1467. @calendar_pattern_index_type@ full_format { 0 };
  1468. @calendar_pattern_index_type@ long_format { 0 };
  1469. @calendar_pattern_index_type@ medium_format { 0 };
  1470. @calendar_pattern_index_type@ short_format { 0 };
  1471. };
  1472. struct CalendarSymbols {
  1473. @symbol_list_index_type@ narrow_symbols { 0 };
  1474. @symbol_list_index_type@ short_symbols { 0 };
  1475. @symbol_list_index_type@ long_symbols { 0 };
  1476. };
  1477. struct CalendarData {
  1478. @calendar_format_index_type@ date_formats { 0 };
  1479. @calendar_format_index_type@ time_formats { 0 };
  1480. @calendar_format_index_type@ date_time_formats { 0 };
  1481. @calendar_pattern_list_index_type@ available_formats { 0 };
  1482. @calendar_range_pattern_index_type@ default_range_format { 0 };
  1483. @calendar_range_pattern_list_index_type@ range_formats { 0 };
  1484. @calendar_range_pattern_list_index_type@ range12_formats { 0 };
  1485. @calendar_symbols_list_index_type@ symbols { 0 };
  1486. };
  1487. struct TimeZoneNames {
  1488. @string_index_type@ short_standard_name { 0 };
  1489. @string_index_type@ long_standard_name { 0 };
  1490. @string_index_type@ short_generic_name { 0 };
  1491. @string_index_type@ long_generic_name { 0 };
  1492. };
  1493. struct TimeZoneFormatImpl {
  1494. TimeZoneFormat to_time_zone_format() const {
  1495. TimeZoneFormat time_zone_format {};
  1496. time_zone_format.symbol_ahead_sign = s_string_list[symbol_ahead_sign];
  1497. time_zone_format.symbol_ahead_separator = s_string_list[symbol_ahead_separator];
  1498. time_zone_format.symbol_behind_sign = s_string_list[symbol_behind_sign];
  1499. time_zone_format.symbol_behind_separator = s_string_list[symbol_behind_separator];
  1500. time_zone_format.gmt_format = s_string_list[gmt_format];
  1501. time_zone_format.gmt_zero_format = s_string_list[gmt_zero_format];
  1502. return time_zone_format;
  1503. }
  1504. @string_index_type@ symbol_ahead_sign { 0 };
  1505. @string_index_type@ symbol_ahead_separator { 0 };
  1506. @string_index_type@ symbol_behind_sign { 0 };
  1507. @string_index_type@ symbol_behind_separator { 0 };
  1508. @string_index_type@ gmt_format { 0 };
  1509. @string_index_type@ gmt_zero_format { 0 };
  1510. };
  1511. struct DayPeriodData {
  1512. u8 day_period { 0 };
  1513. u8 begin { 0 };
  1514. u8 end { 0 };
  1515. };
  1516. )~~~");
  1517. locale_data.unique_formats.generate(generator, "CalendarFormatImpl"sv, "s_calendar_formats"sv, 10);
  1518. locale_data.unique_symbol_lists.generate(generator, s_string_index_type, "s_symbol_lists"sv);
  1519. locale_data.unique_calendar_symbols.generate(generator, "CalendarSymbols"sv, "s_calendar_symbols"sv, 10);
  1520. locale_data.unique_calendar_symbols_lists.generate(generator, s_calendar_symbols_index_type, "s_calendar_symbol_lists"sv);
  1521. locale_data.unique_calendars.generate(generator, "CalendarData"sv, "s_calendars"sv, 10);
  1522. locale_data.unique_time_zones.generate(generator, "TimeZoneNames"sv, "s_time_zones"sv, 30);
  1523. locale_data.unique_time_zone_lists.generate(generator, s_time_zone_index_type, "s_time_zone_lists"sv);
  1524. locale_data.unique_time_zone_formats.generate(generator, "TimeZoneFormatImpl"sv, "s_time_zone_formats"sv, 30);
  1525. locale_data.unique_day_periods.generate(generator, "DayPeriodData"sv, "s_day_periods"sv, 30);
  1526. locale_data.unique_day_period_lists.generate(generator, s_day_period_index_type, "s_day_period_lists"sv);
  1527. locale_data.unique_hour_cycle_lists.generate(generator, "u8"sv, "s_hour_cycle_lists"sv);
  1528. auto append_calendars = [&](String name, auto const& calendars) {
  1529. generator.set("name", name);
  1530. generator.set("size", String::number(calendars.size()));
  1531. generator.append(R"~~~(
  1532. static constexpr Array<@calendar_index_type@, @size@> @name@ { {)~~~");
  1533. bool first = true;
  1534. for (auto const& calendar_key : locale_data.calendars) {
  1535. auto calendar = calendars.find(calendar_key)->value;
  1536. generator.append(first ? " " : ", ");
  1537. generator.append(String::number(calendar));
  1538. first = false;
  1539. }
  1540. generator.append(" } };");
  1541. };
  1542. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  1543. generator.set("type", type);
  1544. generator.set("name", name);
  1545. generator.set("size", String::number(keys.size()));
  1546. generator.append(R"~~~(
  1547. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1548. bool first = true;
  1549. for (auto const& key : keys) {
  1550. auto const& value = map.find(key)->value;
  1551. auto mapping = mapping_getter(value);
  1552. generator.append(first ? " " : ", ");
  1553. generator.append(String::number(mapping));
  1554. first = false;
  1555. }
  1556. generator.append(" } };");
  1557. };
  1558. auto locales = locale_data.locales.keys();
  1559. quick_sort(locales);
  1560. generate_mapping(generator, locale_data.locales, s_calendar_index_type, "s_locale_calendars"sv, "s_calendars_{}", format_identifier, [&](auto const& name, auto const& value) { append_calendars(name, value.calendars); });
  1561. append_mapping(locales, locale_data.locales, s_time_zone_index_type, "s_locale_time_zones"sv, [](auto const& locale) { return locale.time_zones; });
  1562. append_mapping(locales, locale_data.locales, s_time_zone_format_index_type, "s_locale_time_zone_formats"sv, [](auto const& locale) { return locale.time_zone_formats; });
  1563. append_mapping(locales, locale_data.locales, s_day_period_index_type, "s_locale_day_periods"sv, [](auto const& locale) { return locale.day_periods; });
  1564. append_mapping(locale_data.hour_cycle_regions, locale_data.hour_cycles, s_hour_cycle_list_index_type, "s_hour_cycles"sv, [](auto const& hour_cycles) { return hour_cycles; });
  1565. generator.append("\n");
  1566. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1567. HashValueMap<String> hashes;
  1568. hashes.ensure_capacity(values.size());
  1569. for (auto const& value : values)
  1570. hashes.set(value.hash(), format_identifier(enum_title, value));
  1571. for (auto const& alias : aliases)
  1572. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1573. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1574. };
  1575. append_from_string("Calendar"sv, "calendar"sv, locale_data.calendars, locale_data.calendar_aliases);
  1576. append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, locale_data.hour_cycle_regions);
  1577. generator.append(R"~~~(
  1578. Vector<HourCycle> get_regional_hour_cycles(StringView region)
  1579. {
  1580. auto region_value = hour_cycle_region_from_string(region);
  1581. if (!region_value.has_value())
  1582. return {};
  1583. auto region_index = to_underlying(*region_value);
  1584. auto regional_hour_cycles_index = s_hour_cycles.at(region_index);
  1585. auto const& regional_hour_cycles = s_hour_cycle_lists.at(regional_hour_cycles_index);
  1586. Vector<HourCycle> hour_cycles;
  1587. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1588. for (auto hour_cycle : regional_hour_cycles)
  1589. hour_cycles.unchecked_append(static_cast<HourCycle>(hour_cycle));
  1590. return hour_cycles;
  1591. }
  1592. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1593. {
  1594. auto locale_value = locale_from_string(locale);
  1595. if (!locale_value.has_value())
  1596. return nullptr;
  1597. auto calendar_value = calendar_from_string(calendar);
  1598. if (!calendar_value.has_value())
  1599. return nullptr;
  1600. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1601. size_t calendar_index = to_underlying(*calendar_value);
  1602. auto const& calendar_indices = s_locale_calendars.at(locale_index);
  1603. calendar_index = calendar_indices[calendar_index];
  1604. return &s_calendars[calendar_index];
  1605. }
  1606. Optional<CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1607. {
  1608. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1609. auto const& formats = s_calendar_formats.at(data->date_formats);
  1610. return formats.to_unicode_calendar_format();
  1611. }
  1612. return {};
  1613. }
  1614. Optional<CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1615. {
  1616. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1617. auto const& formats = s_calendar_formats.at(data->time_formats);
  1618. return formats.to_unicode_calendar_format();
  1619. }
  1620. return {};
  1621. }
  1622. Optional<CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1623. {
  1624. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1625. auto const& formats = s_calendar_formats.at(data->date_time_formats);
  1626. return formats.to_unicode_calendar_format();
  1627. }
  1628. return {};
  1629. }
  1630. Vector<CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1631. {
  1632. Vector<CalendarPattern> result {};
  1633. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1634. auto const& available_formats = s_calendar_pattern_lists.at(data->available_formats);
  1635. result.ensure_capacity(available_formats.size());
  1636. for (auto const& format : available_formats)
  1637. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1638. }
  1639. return result;
  1640. }
  1641. Optional<CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar)
  1642. {
  1643. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1644. auto const& pattern = s_calendar_range_patterns[data->default_range_format];
  1645. return pattern.to_unicode_calendar_range_pattern();
  1646. }
  1647. return {};
  1648. }
  1649. Vector<CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton)
  1650. {
  1651. Vector<CalendarRangePattern> result {};
  1652. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1653. auto const& range_formats = s_calendar_range_pattern_lists.at(data->range_formats);
  1654. for (auto format : range_formats) {
  1655. auto const& pattern = s_calendar_range_patterns[format];
  1656. if (skeleton == s_string_list[pattern.skeleton])
  1657. result.append(pattern.to_unicode_calendar_range_pattern());
  1658. }
  1659. }
  1660. return result;
  1661. }
  1662. Vector<CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton)
  1663. {
  1664. Vector<CalendarRangePattern> result {};
  1665. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1666. auto const& range12_formats = s_calendar_range_pattern_lists.at(data->range12_formats);
  1667. for (auto format : range12_formats) {
  1668. auto const& pattern = s_calendar_range_patterns[format];
  1669. if (skeleton == s_string_list[pattern.skeleton])
  1670. result.append(pattern.to_unicode_calendar_range_pattern());
  1671. }
  1672. }
  1673. return result;
  1674. }
  1675. static Span<@string_index_type@ const> find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1676. {
  1677. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1678. auto const& symbols_list = s_calendar_symbol_lists[data->symbols];
  1679. auto symbol_index = to_underlying(symbol);
  1680. auto calendar_symbols_index = symbols_list.at(symbol_index);
  1681. auto const& symbols = s_calendar_symbols.at(calendar_symbols_index);
  1682. @symbol_list_index_type@ symbol_list_index = 0;
  1683. switch (style) {
  1684. case CalendarPatternStyle::Narrow:
  1685. symbol_list_index = symbols.narrow_symbols;
  1686. break;
  1687. case CalendarPatternStyle::Short:
  1688. symbol_list_index = symbols.short_symbols;
  1689. break;
  1690. case CalendarPatternStyle::Long:
  1691. symbol_list_index = symbols.long_symbols;
  1692. break;
  1693. default:
  1694. VERIFY_NOT_REACHED();
  1695. }
  1696. return s_symbol_lists.at(symbol_list_index);
  1697. }
  1698. return {};
  1699. }
  1700. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Era value)
  1701. {
  1702. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1703. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1704. return s_string_list[symbols.at(value_index)];
  1705. return {};
  1706. }
  1707. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Month value)
  1708. {
  1709. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1710. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1711. return s_string_list[symbols.at(value_index)];
  1712. return {};
  1713. }
  1714. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Weekday value)
  1715. {
  1716. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1717. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1718. return s_string_list[symbols.at(value_index)];
  1719. return {};
  1720. }
  1721. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, DayPeriod value)
  1722. {
  1723. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1724. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1725. return s_string_list[symbols.at(value_index)];
  1726. return {};
  1727. }
  1728. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour)
  1729. {
  1730. auto locale_value = locale_from_string(locale);
  1731. if (!locale_value.has_value())
  1732. return {};
  1733. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1734. auto day_periods_index = s_locale_day_periods[locale_index];
  1735. auto day_periods = s_day_period_lists[day_periods_index];
  1736. for (auto day_period_index : day_periods) {
  1737. auto day_period = s_day_periods[day_period_index];
  1738. auto h = hour;
  1739. if (day_period.begin > day_period.end) {
  1740. day_period.end += 24;
  1741. h += 24;
  1742. }
  1743. if ((day_period.begin <= h) && (h < day_period.end)) {
  1744. auto period = static_cast<DayPeriod>(day_period.day_period);
  1745. return get_calendar_day_period_symbol(locale, calendar, style, period);
  1746. }
  1747. }
  1748. // Fallback to fixed periods if the locale does not have flexible day periods.
  1749. // TR-35 states that the meaning of AM and PM does not change with locale.
  1750. if (hour < 12)
  1751. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::AM);
  1752. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::PM);
  1753. }
  1754. Optional<TimeZoneFormat> get_time_zone_format(StringView locale)
  1755. {
  1756. auto locale_value = locale_from_string(locale);
  1757. if (!locale_value.has_value())
  1758. return {};
  1759. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1760. auto time_zone_format_index = s_locale_time_zone_formats.at(locale_index);
  1761. auto const& time_zone_format = s_time_zone_formats.at(time_zone_format_index);
  1762. return time_zone_format.to_time_zone_format();
  1763. }
  1764. static TimeZoneNames const* find_time_zone_names(StringView locale, StringView time_zone)
  1765. {
  1766. auto locale_value = locale_from_string(locale);
  1767. if (!locale_value.has_value())
  1768. return nullptr;
  1769. auto time_zone_value = ::TimeZone::time_zone_from_string(time_zone);
  1770. if (!time_zone_value.has_value())
  1771. return nullptr;
  1772. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1773. size_t time_zone_index = to_underlying(*time_zone_value);
  1774. auto time_zone_list_index = s_locale_time_zones.at(locale_index);
  1775. auto const& time_zone_list = s_time_zone_lists.at(time_zone_list_index);
  1776. if (time_zone_list.size() <= time_zone_index)
  1777. return nullptr;
  1778. time_zone_index = time_zone_list.at(time_zone_index);
  1779. return &s_time_zones[time_zone_index];
  1780. }
  1781. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style)
  1782. {
  1783. if (auto const* data = find_time_zone_names(locale, time_zone); data != nullptr) {
  1784. size_t name_index = 0;
  1785. switch (style) {
  1786. case CalendarPatternStyle::Short:
  1787. name_index = data->short_standard_name;
  1788. break;
  1789. case CalendarPatternStyle::Long:
  1790. name_index = data->long_standard_name;
  1791. break;
  1792. case CalendarPatternStyle::ShortGeneric:
  1793. name_index = data->short_generic_name;
  1794. break;
  1795. case CalendarPatternStyle::LongGeneric:
  1796. name_index = data->long_generic_name;
  1797. break;
  1798. default:
  1799. VERIFY_NOT_REACHED();
  1800. }
  1801. if (name_index != 0)
  1802. return s_string_list[name_index];
  1803. }
  1804. return {};
  1805. }
  1806. }
  1807. )~~~");
  1808. VERIFY(file.write(generator.as_string_view()));
  1809. }
  1810. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1811. {
  1812. StringView generated_header_path;
  1813. StringView generated_implementation_path;
  1814. StringView core_path;
  1815. StringView dates_path;
  1816. Core::ArgsParser args_parser;
  1817. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1818. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1819. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1820. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1821. args_parser.parse(arguments);
  1822. auto open_file = [&](StringView path) -> ErrorOr<NonnullRefPtr<Core::File>> {
  1823. if (path.is_empty()) {
  1824. args_parser.print_usage(stderr, arguments.argv[0]);
  1825. return Error::from_string_literal("Must provide all command line options"sv);
  1826. }
  1827. return Core::File::open(path, Core::OpenMode::ReadWrite);
  1828. };
  1829. auto generated_header_file = TRY(open_file(generated_header_path));
  1830. auto generated_implementation_file = TRY(open_file(generated_implementation_path));
  1831. UnicodeLocaleData locale_data;
  1832. TRY(parse_all_locales(core_path, dates_path, locale_data));
  1833. generate_unicode_locale_header(generated_header_file, locale_data);
  1834. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  1835. return 0;
  1836. }