GenerateUnicodeDateTimeFormat.cpp 90 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278
  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_daylight_name);
  331. hash = pair_int_hash(hash, long_daylight_name);
  332. hash = pair_int_hash(hash, short_generic_name);
  333. hash = pair_int_hash(hash, long_generic_name);
  334. return hash;
  335. }
  336. bool operator==(TimeZoneNames const& other) const
  337. {
  338. return (short_standard_name == other.short_standard_name)
  339. && (long_standard_name == other.long_standard_name)
  340. && (short_daylight_name == other.short_daylight_name)
  341. && (long_daylight_name == other.long_daylight_name)
  342. && (short_generic_name == other.short_generic_name)
  343. && (long_generic_name == other.long_generic_name);
  344. }
  345. StringIndexType short_standard_name { 0 };
  346. StringIndexType long_standard_name { 0 };
  347. StringIndexType short_daylight_name { 0 };
  348. StringIndexType long_daylight_name { 0 };
  349. StringIndexType short_generic_name { 0 };
  350. StringIndexType long_generic_name { 0 };
  351. };
  352. template<>
  353. struct AK::Formatter<TimeZoneNames> : Formatter<FormatString> {
  354. ErrorOr<void> format(FormatBuilder& builder, TimeZoneNames const& time_zone)
  355. {
  356. return Formatter<FormatString>::format(builder,
  357. "{{ {}, {}, {}, {}, {}, {} }}",
  358. time_zone.short_standard_name,
  359. time_zone.long_standard_name,
  360. time_zone.short_daylight_name,
  361. time_zone.long_daylight_name,
  362. time_zone.short_generic_name,
  363. time_zone.long_generic_name);
  364. }
  365. };
  366. template<>
  367. struct AK::Traits<TimeZoneNames> : public GenericTraits<TimeZoneNames> {
  368. static unsigned hash(TimeZoneNames const& t) { return t.hash(); }
  369. };
  370. struct TimeZoneFormat {
  371. unsigned hash() const
  372. {
  373. auto hash = int_hash(symbol_ahead_sign);
  374. hash = pair_int_hash(hash, symbol_ahead_separator);
  375. hash = pair_int_hash(hash, symbol_behind_sign);
  376. hash = pair_int_hash(hash, symbol_behind_separator);
  377. hash = pair_int_hash(hash, gmt_format);
  378. hash = pair_int_hash(hash, gmt_zero_format);
  379. return hash;
  380. }
  381. bool operator==(TimeZoneFormat const& other) const
  382. {
  383. return (symbol_ahead_sign == other.symbol_ahead_sign)
  384. && (symbol_ahead_separator == other.symbol_ahead_separator)
  385. && (symbol_behind_sign == other.symbol_behind_sign)
  386. && (symbol_behind_separator == other.symbol_behind_separator)
  387. && (gmt_format == other.gmt_format)
  388. && (gmt_zero_format == other.gmt_zero_format);
  389. }
  390. StringIndexType symbol_ahead_sign { 0 };
  391. StringIndexType symbol_ahead_separator { 0 };
  392. StringIndexType symbol_behind_sign { 0 };
  393. StringIndexType symbol_behind_separator { 0 };
  394. StringIndexType gmt_format { 0 };
  395. StringIndexType gmt_zero_format { 0 };
  396. };
  397. template<>
  398. struct AK::Formatter<TimeZoneFormat> : Formatter<FormatString> {
  399. ErrorOr<void> format(FormatBuilder& builder, TimeZoneFormat const& time_zone_format)
  400. {
  401. return Formatter<FormatString>::format(builder, "{{ {}, {}, {}, {}, {}, {} }}",
  402. time_zone_format.symbol_ahead_sign,
  403. time_zone_format.symbol_ahead_separator,
  404. time_zone_format.symbol_behind_sign,
  405. time_zone_format.symbol_behind_separator,
  406. time_zone_format.gmt_format,
  407. time_zone_format.gmt_zero_format);
  408. }
  409. };
  410. template<>
  411. struct AK::Traits<TimeZoneFormat> : public GenericTraits<TimeZoneFormat> {
  412. static unsigned hash(TimeZoneFormat const& t) { return t.hash(); }
  413. };
  414. struct DayPeriod {
  415. unsigned hash() const
  416. {
  417. auto hash = int_hash(static_cast<u8>(day_period));
  418. hash = pair_int_hash(hash, begin);
  419. hash = pair_int_hash(hash, end);
  420. return hash;
  421. }
  422. bool operator==(DayPeriod const& other) const
  423. {
  424. return (day_period == other.day_period)
  425. && (begin == other.begin)
  426. && (end == other.end);
  427. }
  428. Unicode::DayPeriod day_period {};
  429. u8 begin { 0 };
  430. u8 end { 0 };
  431. };
  432. template<>
  433. struct AK::Formatter<DayPeriod> : Formatter<FormatString> {
  434. ErrorOr<void> format(FormatBuilder& builder, DayPeriod const& day_period)
  435. {
  436. return Formatter<FormatString>::format(builder,
  437. "{{ {}, {}, {} }}",
  438. static_cast<u8>(day_period.day_period),
  439. day_period.begin,
  440. day_period.end);
  441. }
  442. };
  443. template<>
  444. struct AK::Traits<DayPeriod> : public GenericTraits<DayPeriod> {
  445. static unsigned hash(DayPeriod const& d) { return d.hash(); }
  446. };
  447. using TimeZoneNamesList = Vector<TimeZoneNamesIndexType>;
  448. using DayPeriodList = Vector<DayPeriodIndexType>;
  449. using HourCycleList = Vector<Unicode::HourCycle>;
  450. template<>
  451. struct AK::Formatter<Unicode::HourCycle> : Formatter<FormatString> {
  452. ErrorOr<void> format(FormatBuilder& builder, Unicode::HourCycle hour_cycle)
  453. {
  454. return Formatter<FormatString>::format(builder, "{}", to_underlying(hour_cycle));
  455. }
  456. };
  457. struct Locale {
  458. HashMap<String, CalendarIndexType> calendars;
  459. TimeZoneNamesListIndexType time_zones { 0 };
  460. TimeZoneFormatIndexType time_zone_formats { 0 };
  461. DayPeriodListIndexType day_periods { 0 };
  462. };
  463. struct UnicodeLocaleData {
  464. UniqueStringStorage<StringIndexType> unique_strings;
  465. UniqueStorage<CalendarPattern, CalendarPatternIndexType> unique_patterns;
  466. UniqueStorage<CalendarPatternList, CalendarPatternListIndexType> unique_pattern_lists;
  467. UniqueStorage<CalendarRangePattern, CalendarRangePatternIndexType> unique_range_patterns;
  468. UniqueStorage<CalendarRangePatternList, CalendarRangePatternListIndexType> unique_range_pattern_lists;
  469. UniqueStorage<CalendarFormat, CalendarFormatIndexType> unique_formats;
  470. UniqueStorage<SymbolList, SymbolListIndexType> unique_symbol_lists;
  471. UniqueStorage<CalendarSymbols, CalendarSymbolsIndexType> unique_calendar_symbols;
  472. UniqueStorage<CalendarSymbolsList, CalendarSymbolsListIndexType> unique_calendar_symbols_lists;
  473. UniqueStorage<Calendar, CalendarIndexType> unique_calendars;
  474. UniqueStorage<TimeZoneNames, TimeZoneNamesIndexType> unique_time_zones;
  475. UniqueStorage<TimeZoneNamesList, TimeZoneNamesListIndexType> unique_time_zone_lists;
  476. UniqueStorage<TimeZoneFormat, TimeZoneFormatIndexType> unique_time_zone_formats;
  477. UniqueStorage<DayPeriod, DayPeriodIndexType> unique_day_periods;
  478. UniqueStorage<DayPeriodList, DayPeriodListIndexType> unique_day_period_lists;
  479. UniqueStorage<HourCycleList, HourCycleListIndexType> unique_hour_cycle_lists;
  480. HashMap<String, Locale> locales;
  481. HashMap<String, HourCycleListIndexType> hour_cycles;
  482. Vector<String> hour_cycle_regions;
  483. HashMap<String, Vector<TimeZone::TimeZone>> meta_zones;
  484. Vector<String> time_zones { "UTC"sv };
  485. Vector<String> calendars;
  486. Vector<Alias> calendar_aliases {
  487. // FIXME: Aliases should come from BCP47. See: https://unicode-org.atlassian.net/browse/CLDR-15158
  488. { "gregorian"sv, "gregory"sv },
  489. };
  490. };
  491. static Optional<Unicode::DayPeriod> day_period_from_string(StringView day_period)
  492. {
  493. if (day_period == "am"sv)
  494. return Unicode::DayPeriod::AM;
  495. if (day_period == "pm"sv)
  496. return Unicode::DayPeriod::PM;
  497. if (day_period == "morning1"sv)
  498. return Unicode::DayPeriod::Morning1;
  499. if (day_period == "morning2"sv)
  500. return Unicode::DayPeriod::Morning2;
  501. if (day_period == "afternoon1"sv)
  502. return Unicode::DayPeriod::Afternoon1;
  503. if (day_period == "afternoon2"sv)
  504. return Unicode::DayPeriod::Afternoon2;
  505. if (day_period == "evening1"sv)
  506. return Unicode::DayPeriod::Evening1;
  507. if (day_period == "evening2"sv)
  508. return Unicode::DayPeriod::Evening2;
  509. if (day_period == "night1"sv)
  510. return Unicode::DayPeriod::Night1;
  511. if (day_period == "night2"sv)
  512. return Unicode::DayPeriod::Night2;
  513. return {};
  514. };
  515. static ErrorOr<void> parse_hour_cycles(String core_path, UnicodeLocaleData& locale_data)
  516. {
  517. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  518. LexicalPath time_data_path(move(core_path));
  519. time_data_path = time_data_path.append("supplemental"sv);
  520. time_data_path = time_data_path.append("timeData.json"sv);
  521. auto time_data_file = TRY(Core::File::open(time_data_path.string(), Core::OpenMode::ReadOnly));
  522. auto time_data = TRY(JsonValue::from_string(time_data_file->read_all()));
  523. auto const& supplemental_object = time_data.as_object().get("supplemental"sv);
  524. auto const& time_data_object = supplemental_object.as_object().get("timeData"sv);
  525. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Unicode::HourCycle> {
  526. if (hour_cycle == "h"sv)
  527. return Unicode::HourCycle::H12;
  528. if (hour_cycle == "H"sv)
  529. return Unicode::HourCycle::H23;
  530. if (hour_cycle == "K"sv)
  531. return Unicode::HourCycle::H11;
  532. if (hour_cycle == "k"sv)
  533. return Unicode::HourCycle::H24;
  534. return {};
  535. };
  536. time_data_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  537. auto allowed_hour_cycles_string = value.as_object().get("_allowed"sv).as_string();
  538. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  539. Vector<Unicode::HourCycle> hour_cycles;
  540. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  541. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  542. hour_cycles.append(*hour_cycle);
  543. }
  544. auto hour_cycles_index = locale_data.unique_hour_cycle_lists.ensure(move(hour_cycles));
  545. locale_data.hour_cycles.set(key, hour_cycles_index);
  546. if (!locale_data.hour_cycle_regions.contains_slow(key))
  547. locale_data.hour_cycle_regions.append(key);
  548. });
  549. return {};
  550. };
  551. static ErrorOr<void> parse_meta_zones(String core_path, UnicodeLocaleData& locale_data)
  552. {
  553. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  554. LexicalPath meta_zone_path(move(core_path));
  555. meta_zone_path = meta_zone_path.append("supplemental"sv);
  556. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  557. auto meta_zone_file = TRY(Core::File::open(meta_zone_path.string(), Core::OpenMode::ReadOnly));
  558. auto meta_zone = TRY(JsonValue::from_string(meta_zone_file->read_all()));
  559. auto const& supplemental_object = meta_zone.as_object().get("supplemental"sv);
  560. auto const& meta_zone_object = supplemental_object.as_object().get("metaZones"sv);
  561. auto const& meta_zone_array = meta_zone_object.as_object().get("metazones"sv);
  562. meta_zone_array.as_array().for_each([&](JsonValue const& value) {
  563. auto const& mapping = value.as_object().get("mapZone"sv);
  564. auto const& meta_zone = mapping.as_object().get("_other"sv);
  565. auto const& golden_zone = mapping.as_object().get("_type"sv);
  566. if (auto time_zone = TimeZone::time_zone_from_string(golden_zone.as_string()); time_zone.has_value()) {
  567. auto& golden_zones = locale_data.meta_zones.ensure(meta_zone.as_string());
  568. golden_zones.append(*time_zone);
  569. }
  570. });
  571. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  572. if (auto time_zone = TimeZone::time_zone_from_string("UTC"sv); time_zone.has_value())
  573. locale_data.meta_zones.set("UTC"sv, { *time_zone });
  574. return {};
  575. };
  576. static constexpr auto is_char(char ch)
  577. {
  578. return [ch](auto c) { return c == ch; };
  579. }
  580. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  581. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  582. // extra spaces around the day period. Some example expected removals:
  583. //
  584. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  585. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  586. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  587. static String remove_period_from_pattern(String pattern)
  588. {
  589. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  590. auto index = pattern.find(remove);
  591. if (!index.has_value())
  592. continue;
  593. constexpr u32 space = ' ';
  594. constexpr u32 open = '{';
  595. constexpr u32 close = '}';
  596. Utf8View utf8_pattern { pattern };
  597. Optional<u32> before_removal;
  598. Optional<u32> after_removal;
  599. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  600. before_removal = *it;
  601. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  602. after_removal = *it;
  603. if ((before_removal == space) && (after_removal != open)) {
  604. pattern = String::formatted("{}{}",
  605. pattern.substring_view(0, *index - 1),
  606. pattern.substring_view(*index + remove.length()));
  607. } else if ((after_removal == space) && (before_removal != close)) {
  608. pattern = String::formatted("{}{}",
  609. pattern.substring_view(0, *index),
  610. pattern.substring_view(*index + remove.length() + 1));
  611. } else {
  612. pattern = String::formatted("{}{}",
  613. pattern.substring_view(0, *index),
  614. pattern.substring_view(*index + remove.length()));
  615. }
  616. }
  617. return pattern;
  618. }
  619. static Optional<CalendarPattern> parse_date_time_pattern_raw(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  620. {
  621. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  622. using Unicode::CalendarPatternStyle;
  623. CalendarPattern format {};
  624. if (!skeleton.is_empty())
  625. format.skeleton_index = locale_data.unique_strings.ensure(move(skeleton));
  626. GenericLexer lexer { pattern };
  627. StringBuilder builder;
  628. bool hour12 { false };
  629. while (!lexer.is_eof()) {
  630. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  631. // processing (this just avoids conflicts with the patterns below).
  632. if (lexer.next_is(is_quote)) {
  633. builder.append(lexer.consume_quoted_string());
  634. continue;
  635. }
  636. auto starting_char = lexer.peek();
  637. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  638. // Era
  639. if (all_of(segment, is_char('G'))) {
  640. builder.append("{era}");
  641. if (segment.length() <= 3)
  642. format.era = CalendarPatternStyle::Short;
  643. else if (segment.length() == 4)
  644. format.era = CalendarPatternStyle::Long;
  645. else
  646. format.era = CalendarPatternStyle::Narrow;
  647. }
  648. // Year
  649. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  650. builder.append("{year}");
  651. if (segment.length() == 2)
  652. format.year = CalendarPatternStyle::TwoDigit;
  653. else
  654. format.year = CalendarPatternStyle::Numeric;
  655. }
  656. // Quarter
  657. else if (all_of(segment, is_any_of("qQ"sv))) {
  658. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  659. return {};
  660. }
  661. // Month
  662. else if (all_of(segment, is_any_of("ML"sv))) {
  663. builder.append("{month}");
  664. if (segment.length() == 1)
  665. format.month = CalendarPatternStyle::Numeric;
  666. else if (segment.length() == 2)
  667. format.month = CalendarPatternStyle::TwoDigit;
  668. else if (segment.length() == 3)
  669. format.month = CalendarPatternStyle::Short;
  670. else if (segment.length() == 4)
  671. format.month = CalendarPatternStyle::Long;
  672. else if (segment.length() == 5)
  673. format.month = CalendarPatternStyle::Narrow;
  674. } else if (all_of(segment, is_char('l'))) {
  675. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  676. return {};
  677. }
  678. // Week
  679. else if (all_of(segment, is_any_of("wW"sv))) {
  680. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  681. return {};
  682. }
  683. // Day
  684. else if (all_of(segment, is_char('d'))) {
  685. builder.append("{day}");
  686. if (segment.length() == 1)
  687. format.day = CalendarPatternStyle::Numeric;
  688. else
  689. format.day = CalendarPatternStyle::TwoDigit;
  690. } else if (all_of(segment, is_any_of("DFg"sv))) {
  691. builder.append("{day}");
  692. format.day = CalendarPatternStyle::Numeric;
  693. }
  694. // Weekday
  695. else if (all_of(segment, is_char('E'))) {
  696. builder.append("{weekday}");
  697. if (segment.length() == 4)
  698. format.weekday = CalendarPatternStyle::Long;
  699. else if (segment.length() == 5)
  700. format.weekday = CalendarPatternStyle::Narrow;
  701. else
  702. format.weekday = CalendarPatternStyle::Short;
  703. } else if (all_of(segment, is_any_of("ec"sv))) {
  704. builder.append("{weekday}");
  705. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  706. // pattern styles are not supported by Intl.DateTimeFormat.
  707. if (segment.length() <= 2)
  708. return {};
  709. if (segment.length() == 4)
  710. format.weekday = CalendarPatternStyle::Long;
  711. else if (segment.length() == 5)
  712. format.weekday = CalendarPatternStyle::Narrow;
  713. else
  714. format.weekday = CalendarPatternStyle::Short;
  715. }
  716. // Period
  717. else if (all_of(segment, is_any_of("ab"sv))) {
  718. builder.append("{ampm}");
  719. hour12 = true;
  720. } else if (all_of(segment, is_char('B'))) {
  721. builder.append("{dayPeriod}");
  722. hour12 = true;
  723. if (segment.length() == 4)
  724. format.day_period = CalendarPatternStyle::Long;
  725. else if (segment.length() == 5)
  726. format.day_period = CalendarPatternStyle::Narrow;
  727. else
  728. format.day_period = CalendarPatternStyle::Short;
  729. }
  730. // Hour
  731. else if (all_of(segment, is_any_of("hHKk"sv))) {
  732. builder.append("{hour}");
  733. if ((segment[0] == 'h') || (segment[0] == 'K'))
  734. hour12 = true;
  735. if (segment.length() == 1)
  736. format.hour = CalendarPatternStyle::Numeric;
  737. else
  738. format.hour = CalendarPatternStyle::TwoDigit;
  739. } else if (all_of(segment, is_any_of("jJC"sv))) {
  740. // TR-35 indicates these should not be used.
  741. return {};
  742. }
  743. // Minute
  744. else if (all_of(segment, is_char('m'))) {
  745. builder.append("{minute}");
  746. if (segment.length() == 1)
  747. format.minute = CalendarPatternStyle::Numeric;
  748. else
  749. format.minute = CalendarPatternStyle::TwoDigit;
  750. }
  751. // Second
  752. else if (all_of(segment, is_char('s'))) {
  753. builder.append("{second}");
  754. if (segment.length() == 1)
  755. format.second = CalendarPatternStyle::Numeric;
  756. else
  757. format.second = CalendarPatternStyle::TwoDigit;
  758. } else if (all_of(segment, is_char('S'))) {
  759. builder.append("{fractionalSecondDigits}");
  760. VERIFY(segment.length() <= 3);
  761. format.fractional_second_digits = static_cast<u8>(segment.length());
  762. } else if (all_of(segment, is_char('A'))) {
  763. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  764. return {};
  765. }
  766. // Zone
  767. else if (all_of(segment, is_any_of("zV"sv))) {
  768. builder.append("{timeZoneName}");
  769. if (segment.length() < 4)
  770. format.time_zone_name = CalendarPatternStyle::Short;
  771. else
  772. format.time_zone_name = CalendarPatternStyle::Long;
  773. } else if (all_of(segment, is_any_of("ZOXx"sv))) {
  774. builder.append("{timeZoneName}");
  775. if (segment.length() < 4)
  776. format.time_zone_name = CalendarPatternStyle::ShortOffset;
  777. else
  778. format.time_zone_name = CalendarPatternStyle::LongOffset;
  779. } else if (all_of(segment, is_char('v'))) {
  780. builder.append("{timeZoneName}");
  781. if (segment.length() < 4)
  782. format.time_zone_name = CalendarPatternStyle::ShortGeneric;
  783. else
  784. format.time_zone_name = CalendarPatternStyle::LongGeneric;
  785. }
  786. // Non-patterns
  787. else {
  788. builder.append(segment);
  789. }
  790. }
  791. pattern = builder.build();
  792. if (hour12) {
  793. format.pattern = remove_period_from_pattern(pattern);
  794. format.pattern12 = move(pattern);
  795. } else {
  796. format.pattern = move(pattern);
  797. }
  798. return format;
  799. }
  800. static Optional<CalendarPatternIndexType> parse_date_time_pattern(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  801. {
  802. auto format = parse_date_time_pattern_raw(move(pattern), move(skeleton), locale_data);
  803. if (!format.has_value())
  804. return {};
  805. format->pattern_index = locale_data.unique_strings.ensure(move(format->pattern));
  806. if (format->pattern12.has_value())
  807. format->pattern12_index = locale_data.unique_strings.ensure(format->pattern12.release_value());
  808. return locale_data.unique_patterns.ensure(format.release_value());
  809. }
  810. template<typename... Chars>
  811. static constexpr bool char_is_one_of(char ch, Chars&&... chars)
  812. {
  813. return ((ch == chars) || ...);
  814. }
  815. static void parse_interval_patterns(Calendar& calendar, JsonObject const& interval_formats_object, UnicodeLocaleData& locale_data)
  816. {
  817. // https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats
  818. CalendarRangePatternList range_formats {};
  819. CalendarRangePatternList range12_formats {};
  820. auto name_of_field = [&](char field) {
  821. if (char_is_one_of(field, 'G'))
  822. return Unicode::CalendarRangePattern::Field::Era;
  823. if (char_is_one_of(field, 'y', 'Y', 'u', 'U', 'r'))
  824. return Unicode::CalendarRangePattern::Field::Year;
  825. if (char_is_one_of(field, 'M', 'L'))
  826. return Unicode::CalendarRangePattern::Field::Month;
  827. if (char_is_one_of(field, 'd', 'D', 'F', 'g'))
  828. return Unicode::CalendarRangePattern::Field::Day;
  829. if (char_is_one_of(field, 'a', 'b'))
  830. return Unicode::CalendarRangePattern::Field::AmPm;
  831. if (char_is_one_of(field, 'B'))
  832. return Unicode::CalendarRangePattern::Field::DayPeriod;
  833. if (char_is_one_of(field, 'h', 'H', 'K', 'k'))
  834. return Unicode::CalendarRangePattern::Field::Hour;
  835. if (char_is_one_of(field, 'm'))
  836. return Unicode::CalendarRangePattern::Field::Minute;
  837. if (char_is_one_of(field, 's'))
  838. return Unicode::CalendarRangePattern::Field::Second;
  839. if (char_is_one_of(field, 'S'))
  840. return Unicode::CalendarRangePattern::Field::FractionalSecondDigits;
  841. VERIFY_NOT_REACHED();
  842. };
  843. auto split_default_range_pattern = [&](auto skeleton, auto const& pattern) {
  844. auto start_range_end = pattern.find('}').value() + 1;
  845. auto end_range_begin = pattern.find_last('{').value();
  846. auto start_range = pattern.substring_view(0, start_range_end);
  847. auto separator = pattern.substring_view(start_range_end, end_range_begin - start_range_end);
  848. auto end_range = pattern.substring_view(end_range_begin);
  849. CalendarRangePattern format {};
  850. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  851. format.start_range = locale_data.unique_strings.ensure(start_range);
  852. format.separator = locale_data.unique_strings.ensure(separator);
  853. format.end_range = locale_data.unique_strings.ensure(end_range);
  854. return format;
  855. };
  856. auto split_range_pattern = [&](auto skeleton, auto field, auto const& pattern, auto const& parsed_fields) {
  857. HashMap<StringView, size_t> partitions;
  858. StringView last_partition;
  859. auto begin_index = pattern.find('{');
  860. size_t end_index = 0;
  861. while (begin_index.has_value()) {
  862. end_index = pattern.find('}', *begin_index).value();
  863. auto partition = pattern.substring_view(*begin_index, end_index - *begin_index);
  864. if (partitions.contains(partition))
  865. break;
  866. partitions.set(partition, *begin_index);
  867. last_partition = partition;
  868. begin_index = pattern.find('{', end_index + 1);
  869. }
  870. VERIFY(begin_index.has_value() && !last_partition.is_empty());
  871. auto start_range_end = partitions.get(last_partition).value() + last_partition.length() + 1;
  872. auto start_range = pattern.substring_view(0, start_range_end);
  873. auto separator = pattern.substring_view(start_range_end, *begin_index - start_range_end);
  874. auto end_range = pattern.substring_view(*begin_index);
  875. CalendarRangePattern format {};
  876. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  877. format.field = field;
  878. format.start_range = locale_data.unique_strings.ensure(start_range);
  879. format.separator = locale_data.unique_strings.ensure(separator);
  880. format.end_range = locale_data.unique_strings.ensure(end_range);
  881. format.for_each_calendar_field_zipped_with(parsed_fields, [](auto& format_field, auto const& parsed_field, auto) {
  882. format_field = parsed_field;
  883. });
  884. return format;
  885. };
  886. interval_formats_object.for_each_member([&](auto const& skeleton, auto const& value) {
  887. if (skeleton == "intervalFormatFallback"sv) {
  888. auto range_format = split_default_range_pattern(skeleton, value.as_string());
  889. calendar.default_range_format = locale_data.unique_range_patterns.ensure(move(range_format));
  890. return;
  891. }
  892. value.as_object().for_each_member([&](auto const& field, auto const& pattern) {
  893. if (field.ends_with("alt-variant"sv))
  894. return;
  895. VERIFY(field.length() == 1);
  896. auto name = name_of_field(field[0]);
  897. auto format = parse_date_time_pattern_raw(pattern.as_string(), skeleton, locale_data).release_value();
  898. auto range_format = split_range_pattern(skeleton, name, format.pattern, format);
  899. range_formats.append(locale_data.unique_range_patterns.ensure(move(range_format)));
  900. if (format.pattern12.has_value()) {
  901. auto range12_pattern = split_range_pattern(skeleton, name, *format.pattern12, format);
  902. range12_formats.append(locale_data.unique_range_patterns.ensure(move(range12_pattern)));
  903. } else {
  904. range12_formats.append(range_formats.last());
  905. }
  906. });
  907. });
  908. calendar.range_formats = locale_data.unique_range_pattern_lists.ensure(move(range_formats));
  909. calendar.range12_formats = locale_data.unique_range_pattern_lists.ensure(move(range12_formats));
  910. }
  911. static void generate_missing_patterns(Calendar& calendar, CalendarPatternList& formats, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, UnicodeLocaleData& locale_data)
  912. {
  913. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  914. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  915. auto pattern = locale_data.unique_strings.get(format);
  916. auto time_pattern = locale_data.unique_strings.get(time_format);
  917. auto date_pattern = locale_data.unique_strings.get(date_format);
  918. auto new_pattern = pattern.replace("{0}", time_pattern).replace("{1}", date_pattern);
  919. return locale_data.unique_strings.ensure(move(new_pattern));
  920. };
  921. auto inject_fractional_second_digits = [&](auto format) {
  922. auto pattern = locale_data.unique_strings.get(format);
  923. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv);
  924. return locale_data.unique_strings.ensure(move(new_pattern));
  925. };
  926. auto append_if_unique = [&](auto format) {
  927. auto format_index = locale_data.unique_patterns.ensure(move(format));
  928. if (!formats.contains_slow(format_index))
  929. formats.append(format_index);
  930. };
  931. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  932. for (auto const& format : date_formats)
  933. append_if_unique(format);
  934. for (auto const& format : time_formats) {
  935. append_if_unique(format);
  936. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  937. auto new_format = format;
  938. new_format.fractional_second_digits = 2;
  939. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  940. if (new_format.pattern12_index != 0)
  941. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  942. time_formats_with_fractional_second_digits.append(new_format);
  943. append_if_unique(move(new_format));
  944. }
  945. }
  946. time_formats.extend(move(time_formats_with_fractional_second_digits));
  947. for (auto const& date_format : date_formats) {
  948. auto const& date_time_formats = locale_data.unique_formats.get(calendar.date_time_formats);
  949. CalendarPatternIndexType date_time_format_index = 0;
  950. if (date_format.month == Unicode::CalendarPatternStyle::Long) {
  951. if (date_format.weekday.has_value())
  952. date_time_format_index = date_time_formats.full_format;
  953. else
  954. date_time_format_index = date_time_formats.long_format;
  955. } else if (date_format.month == Unicode::CalendarPatternStyle::Short) {
  956. date_time_format_index = date_time_formats.medium_format;
  957. } else {
  958. date_time_format_index = date_time_formats.short_format;
  959. }
  960. for (auto const& time_format : time_formats) {
  961. auto format = locale_data.unique_patterns.get(date_time_format_index);
  962. if (time_format.pattern12_index != 0)
  963. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  964. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  965. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  966. if (date_field.has_value())
  967. field = date_field;
  968. });
  969. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  970. if (time_field.has_value())
  971. field = time_field;
  972. });
  973. append_if_unique(move(format));
  974. }
  975. }
  976. }
  977. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, UnicodeLocaleData& locale_data)
  978. {
  979. auto create_symbol_lists = [](size_t size) {
  980. SymbolList narrow_symbol_list;
  981. SymbolList short_symbol_list;
  982. SymbolList long_symbol_list;
  983. narrow_symbol_list.resize(size);
  984. short_symbol_list.resize(size);
  985. long_symbol_list.resize(size);
  986. return Array<SymbolList, 3> { {
  987. move(narrow_symbol_list),
  988. move(short_symbol_list),
  989. move(long_symbol_list),
  990. } };
  991. };
  992. CalendarSymbolsList symbols_list {};
  993. auto store_symbol_lists = [&](auto symbol, auto symbol_lists) {
  994. auto symbol_index = to_underlying(symbol);
  995. if (symbol_index >= symbols_list.size())
  996. symbols_list.resize(symbol_index + 1);
  997. CalendarSymbols symbols {};
  998. symbols.narrow_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[0]));
  999. symbols.short_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[1]));
  1000. symbols.long_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[2]));
  1001. auto calendar_symbols_index = locale_data.unique_calendar_symbols.ensure(move(symbols));
  1002. symbols_list[symbol_index] = calendar_symbols_index;
  1003. };
  1004. auto parse_era_symbols = [&](auto const& symbols_object) {
  1005. auto const& narrow_symbols = symbols_object.get("eraNarrow"sv).as_object();
  1006. auto const& short_symbols = symbols_object.get("eraAbbr"sv).as_object();
  1007. auto const& long_symbols = symbols_object.get("eraNames"sv).as_object();
  1008. auto symbol_lists = create_symbol_lists(2);
  1009. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1010. if (auto key_index = key.to_uint(); key_index.has_value())
  1011. symbols[*key_index] = locale_data.unique_strings.ensure(move(symbol));
  1012. };
  1013. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1014. append_symbol(symbol_lists[0], key, value.as_string());
  1015. });
  1016. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1017. append_symbol(symbol_lists[1], key, value.as_string());
  1018. });
  1019. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1020. append_symbol(symbol_lists[2], key, value.as_string());
  1021. });
  1022. store_symbol_lists(Unicode::CalendarSymbol::Era, move(symbol_lists));
  1023. };
  1024. auto parse_month_symbols = [&](auto const& symbols_object) {
  1025. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1026. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1027. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1028. auto symbol_lists = create_symbol_lists(12);
  1029. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1030. auto key_index = key.to_uint().value() - 1;
  1031. symbols[key_index] = locale_data.unique_strings.ensure(move(symbol));
  1032. };
  1033. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1034. append_symbol(symbol_lists[0], key, value.as_string());
  1035. });
  1036. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1037. append_symbol(symbol_lists[1], key, value.as_string());
  1038. });
  1039. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1040. append_symbol(symbol_lists[2], key, value.as_string());
  1041. });
  1042. store_symbol_lists(Unicode::CalendarSymbol::Month, move(symbol_lists));
  1043. };
  1044. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  1045. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1046. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1047. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1048. auto symbol_lists = create_symbol_lists(7);
  1049. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1050. if (key == "sun"sv)
  1051. symbols[to_underlying(Unicode::Weekday::Sunday)] = locale_data.unique_strings.ensure(move(symbol));
  1052. else if (key == "mon"sv)
  1053. symbols[to_underlying(Unicode::Weekday::Monday)] = locale_data.unique_strings.ensure(move(symbol));
  1054. else if (key == "tue"sv)
  1055. symbols[to_underlying(Unicode::Weekday::Tuesday)] = locale_data.unique_strings.ensure(move(symbol));
  1056. else if (key == "wed"sv)
  1057. symbols[to_underlying(Unicode::Weekday::Wednesday)] = locale_data.unique_strings.ensure(move(symbol));
  1058. else if (key == "thu"sv)
  1059. symbols[to_underlying(Unicode::Weekday::Thursday)] = locale_data.unique_strings.ensure(move(symbol));
  1060. else if (key == "fri"sv)
  1061. symbols[to_underlying(Unicode::Weekday::Friday)] = locale_data.unique_strings.ensure(move(symbol));
  1062. else if (key == "sat"sv)
  1063. symbols[to_underlying(Unicode::Weekday::Saturday)] = locale_data.unique_strings.ensure(move(symbol));
  1064. };
  1065. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1066. append_symbol(symbol_lists[0], key, value.as_string());
  1067. });
  1068. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1069. append_symbol(symbol_lists[1], key, value.as_string());
  1070. });
  1071. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1072. append_symbol(symbol_lists[2], key, value.as_string());
  1073. });
  1074. store_symbol_lists(Unicode::CalendarSymbol::Weekday, move(symbol_lists));
  1075. };
  1076. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  1077. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  1078. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  1079. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  1080. auto symbol_lists = create_symbol_lists(10);
  1081. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  1082. if (auto day_period = day_period_from_string(key); day_period.has_value())
  1083. symbols[to_underlying(*day_period)] = locale_data.unique_strings.ensure(move(symbol));
  1084. };
  1085. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1086. append_symbol(symbol_lists[0], key, value.as_string());
  1087. });
  1088. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1089. append_symbol(symbol_lists[1], key, value.as_string());
  1090. });
  1091. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  1092. append_symbol(symbol_lists[2], key, value.as_string());
  1093. });
  1094. store_symbol_lists(Unicode::CalendarSymbol::DayPeriod, move(symbol_lists));
  1095. };
  1096. parse_era_symbols(calendar_object.get("eras"sv).as_object());
  1097. parse_month_symbols(calendar_object.get("months"sv).as_object().get("format"sv).as_object());
  1098. parse_weekday_symbols(calendar_object.get("days"sv).as_object().get("format"sv).as_object());
  1099. parse_day_period_symbols(calendar_object.get("dayPeriods"sv).as_object().get("format"sv).as_object());
  1100. calendar.symbols = locale_data.unique_calendar_symbols_lists.ensure(move(symbols_list));
  1101. }
  1102. static ErrorOr<void> parse_calendars(String locale_calendars_path, UnicodeLocaleData& locale_data, Locale& locale)
  1103. {
  1104. LexicalPath calendars_path(move(locale_calendars_path));
  1105. if (!calendars_path.basename().starts_with("ca-"sv))
  1106. return {};
  1107. auto calendars_file = TRY(Core::File::open(calendars_path.string(), Core::OpenMode::ReadOnly));
  1108. auto calendars = TRY(JsonValue::from_string(calendars_file->read_all()));
  1109. auto const& main_object = calendars.as_object().get("main"sv);
  1110. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  1111. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1112. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  1113. auto parse_patterns = [&](auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) {
  1114. auto parse_pattern = [&](auto name) {
  1115. auto format = patterns_object.get(name);
  1116. auto skeleton = skeletons_object.get(name);
  1117. auto format_index = parse_date_time_pattern(format.as_string(), skeleton.as_string_or(String::empty()), locale_data).value();
  1118. if (patterns)
  1119. patterns->append(locale_data.unique_patterns.get(format_index));
  1120. return format_index;
  1121. };
  1122. CalendarFormat formats {};
  1123. formats.full_format = parse_pattern("full"sv);
  1124. formats.long_format = parse_pattern("long"sv);
  1125. formats.medium_format = parse_pattern("medium"sv);
  1126. formats.short_format = parse_pattern("short"sv);
  1127. return locale_data.unique_formats.ensure(move(formats));
  1128. };
  1129. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  1130. // The generic calendar is not a supported Unicode calendar key, so skip it:
  1131. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  1132. if (calendar_name == "generic"sv)
  1133. return;
  1134. Calendar calendar {};
  1135. CalendarPatternList available_formats {};
  1136. if (!locale_data.calendars.contains_slow(calendar_name))
  1137. locale_data.calendars.append(calendar_name);
  1138. Vector<CalendarPattern> date_formats;
  1139. Vector<CalendarPattern> time_formats;
  1140. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  1141. auto const& date_skeletons_object = value.as_object().get("dateSkeletons"sv);
  1142. calendar.date_formats = parse_patterns(date_formats_object.as_object(), date_skeletons_object.as_object(), &date_formats);
  1143. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  1144. auto const& time_skeletons_object = value.as_object().get("timeSkeletons"sv);
  1145. calendar.time_formats = parse_patterns(time_formats_object.as_object(), time_skeletons_object.as_object(), &time_formats);
  1146. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  1147. calendar.date_time_formats = parse_patterns(date_time_formats_object.as_object(), JsonObject {}, nullptr);
  1148. auto const& available_formats_object = date_time_formats_object.as_object().get("availableFormats"sv);
  1149. available_formats_object.as_object().for_each_member([&](auto const& skeleton, JsonValue const& pattern) {
  1150. auto pattern_index = parse_date_time_pattern(pattern.as_string(), skeleton, locale_data);
  1151. if (!pattern_index.has_value())
  1152. return;
  1153. auto const& format = locale_data.unique_patterns.get(*pattern_index);
  1154. if (format.contains_only_date_fields())
  1155. date_formats.append(format);
  1156. else if (format.contains_only_time_fields())
  1157. time_formats.append(format);
  1158. if (!available_formats.contains_slow(*pattern_index))
  1159. available_formats.append(*pattern_index);
  1160. });
  1161. auto const& interval_formats_object = date_time_formats_object.as_object().get("intervalFormats"sv);
  1162. parse_interval_patterns(calendar, interval_formats_object.as_object(), locale_data);
  1163. generate_missing_patterns(calendar, available_formats, move(date_formats), move(time_formats), locale_data);
  1164. parse_calendar_symbols(calendar, value.as_object(), locale_data);
  1165. calendar.available_formats = locale_data.unique_pattern_lists.ensure(move(available_formats));
  1166. locale.calendars.set(calendar_name, locale_data.unique_calendars.ensure(move(calendar)));
  1167. });
  1168. return {};
  1169. }
  1170. static ErrorOr<void> parse_time_zone_names(String locale_time_zone_names_path, UnicodeLocaleData& locale_data, Locale& locale)
  1171. {
  1172. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  1173. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  1174. auto time_zone_names_file = TRY(Core::File::open(time_zone_names_path.string(), Core::OpenMode::ReadOnly));
  1175. auto time_zone_names = TRY(JsonValue::from_string(time_zone_names_file->read_all()));
  1176. auto const& main_object = time_zone_names.as_object().get("main"sv);
  1177. auto const& locale_object = main_object.as_object().get(time_zone_names_path.parent().basename());
  1178. auto const& dates_object = locale_object.as_object().get("dates"sv);
  1179. auto const& time_zone_names_object = dates_object.as_object().get("timeZoneNames"sv);
  1180. auto const& meta_zone_object = time_zone_names_object.as_object().get("metazone"sv);
  1181. auto const& hour_format_string = time_zone_names_object.as_object().get("hourFormat"sv);
  1182. auto const& gmt_format_string = time_zone_names_object.as_object().get("gmtFormat"sv);
  1183. auto const& gmt_zero_format_string = time_zone_names_object.as_object().get("gmtZeroFormat"sv);
  1184. if (meta_zone_object.is_null())
  1185. return {};
  1186. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object, StringView key) -> Optional<StringIndexType> {
  1187. auto const& names = meta_zone_object.get(type);
  1188. if (!names.is_object())
  1189. return {};
  1190. auto const& name = names.as_object().get(key);
  1191. if (name.is_string())
  1192. return locale_data.unique_strings.ensure(name.as_string());
  1193. return {};
  1194. };
  1195. auto parse_hour_format = [&](auto const& format, auto& time_zone_formats) {
  1196. auto hour_formats = format.split_view(';');
  1197. auto hour_format_ahead_start = hour_formats[0].find('H').value();
  1198. auto separator_ahead_start = hour_formats[0].find_last('H').value() + 1;
  1199. auto separator_ahead_end = hour_formats[0].find('m').value();
  1200. auto hour_format_behind_start = hour_formats[1].find('H').value();
  1201. auto separator_behind_start = hour_formats[1].find_last('H').value() + 1;
  1202. auto separator_behind_end = hour_formats[1].find('m').value();
  1203. auto symbol_ahead_sign = hour_formats[0].substring_view(0, hour_format_ahead_start);
  1204. auto symbol_ahead_separator = hour_formats[0].substring_view(separator_ahead_start, separator_ahead_end - separator_ahead_start);
  1205. auto symbol_behind_sign = hour_formats[1].substring_view(0, hour_format_behind_start);
  1206. auto symbol_behind_separator = hour_formats[1].substring_view(separator_behind_start, separator_behind_end - separator_behind_start);
  1207. time_zone_formats.symbol_ahead_sign = locale_data.unique_strings.ensure(symbol_ahead_sign);
  1208. time_zone_formats.symbol_ahead_separator = locale_data.unique_strings.ensure(symbol_ahead_separator);
  1209. time_zone_formats.symbol_behind_sign = locale_data.unique_strings.ensure(symbol_behind_sign);
  1210. time_zone_formats.symbol_behind_separator = locale_data.unique_strings.ensure(symbol_behind_separator);
  1211. };
  1212. TimeZoneNamesList time_zones;
  1213. TimeZoneFormat time_zone_formats {};
  1214. parse_hour_format(hour_format_string.as_string(), time_zone_formats);
  1215. time_zone_formats.gmt_format = locale_data.unique_strings.ensure(gmt_format_string.as_string());
  1216. time_zone_formats.gmt_zero_format = locale_data.unique_strings.ensure(gmt_zero_format_string.as_string());
  1217. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  1218. auto golden_zones = locale_data.meta_zones.find(meta_zone);
  1219. if (golden_zones == locale_data.meta_zones.end())
  1220. return;
  1221. TimeZoneNames time_zone_names {};
  1222. if (auto name = parse_name("long"sv, meta_zone_object, "standard"sv); name.has_value())
  1223. time_zone_names.long_standard_name = name.value();
  1224. if (auto name = parse_name("short"sv, meta_zone_object, "standard"sv); name.has_value())
  1225. time_zone_names.short_standard_name = name.value();
  1226. if (auto name = parse_name("long"sv, meta_zone_object, "daylight"sv); name.has_value())
  1227. time_zone_names.long_daylight_name = name.value();
  1228. if (auto name = parse_name("short"sv, meta_zone_object, "daylight"sv); name.has_value())
  1229. time_zone_names.short_daylight_name = name.value();
  1230. if (auto name = parse_name("long"sv, meta_zone_object, "generic"sv); name.has_value())
  1231. time_zone_names.long_generic_name = name.value();
  1232. if (auto name = parse_name("short"sv, meta_zone_object, "generic"sv); name.has_value())
  1233. time_zone_names.short_generic_name = name.value();
  1234. auto time_zone_index = locale_data.unique_time_zones.ensure(move(time_zone_names));
  1235. for (auto golden_zone : golden_zones->value) {
  1236. auto time_zone = to_underlying(golden_zone);
  1237. if (time_zone >= time_zones.size())
  1238. time_zones.resize(time_zone + 1);
  1239. time_zones[time_zone] = time_zone_index;
  1240. }
  1241. };
  1242. meta_zone_object.as_object().for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  1243. parse_time_zone(meta_zone, value.as_object());
  1244. });
  1245. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  1246. auto const& zone_object = time_zone_names_object.as_object().get("zone"sv);
  1247. auto const& etc_object = zone_object.as_object().get("Etc"sv);
  1248. auto const& utc_object = etc_object.as_object().get("UTC"sv);
  1249. parse_time_zone("UTC"sv, utc_object.as_object());
  1250. locale.time_zones = locale_data.unique_time_zone_lists.ensure(move(time_zones));
  1251. locale.time_zone_formats = locale_data.unique_time_zone_formats.ensure(move(time_zone_formats));
  1252. return {};
  1253. }
  1254. static ErrorOr<void> parse_day_periods(String core_path, UnicodeLocaleData& locale_data)
  1255. {
  1256. // https://unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rule_Sets
  1257. LexicalPath day_periods_path(move(core_path));
  1258. day_periods_path = day_periods_path.append("supplemental"sv);
  1259. day_periods_path = day_periods_path.append("dayPeriods.json"sv);
  1260. auto day_periods_file = TRY(Core::File::open(day_periods_path.string(), Core::OpenMode::ReadOnly));
  1261. auto locale_day_periods = TRY(JsonValue::from_string(day_periods_file->read_all()));
  1262. auto const& supplemental_object = locale_day_periods.as_object().get("supplemental"sv);
  1263. auto const& day_periods_object = supplemental_object.as_object().get("dayPeriodRuleSet"sv);
  1264. auto parse_hour = [](auto const& time) {
  1265. auto hour_end_index = time.find(':').value();
  1266. // The times are in the form "hours:minutes", but we only need the hour segment.
  1267. // TR-35 explicitly states that minutes other than :00 must not be used.
  1268. VERIFY(time.substring_view(hour_end_index) == ":00"sv);
  1269. auto hour = time.substring_view(0, hour_end_index);
  1270. return hour.template to_uint<u8>().value();
  1271. };
  1272. auto parse_day_period = [&](auto const& symbol, auto const& ranges) -> Optional<DayPeriod> {
  1273. auto day_period = day_period_from_string(symbol);
  1274. if (!day_period.has_value())
  1275. return {};
  1276. auto begin = parse_hour(ranges.get("_from"sv).as_string());
  1277. auto end = parse_hour(ranges.get("_before"sv).as_string());
  1278. return DayPeriod { *day_period, begin, end };
  1279. };
  1280. day_periods_object.as_object().for_each_member([&](auto const& language, JsonValue const& value) {
  1281. auto locale = locale_data.locales.find(language);
  1282. if (locale == locale_data.locales.end())
  1283. return;
  1284. DayPeriodList day_periods;
  1285. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& ranges) {
  1286. if (auto day_period = parse_day_period(symbol, ranges.as_object()); day_period.has_value()) {
  1287. auto day_period_index = locale_data.unique_day_periods.ensure(day_period.release_value());
  1288. day_periods.append(day_period_index);
  1289. }
  1290. });
  1291. locale->value.day_periods = locale_data.unique_day_period_lists.ensure(move(day_periods));
  1292. });
  1293. return {};
  1294. };
  1295. static ErrorOr<void> parse_all_locales(String core_path, String dates_path, UnicodeLocaleData& locale_data)
  1296. {
  1297. TRY(parse_hour_cycles(core_path, locale_data));
  1298. TRY(parse_meta_zones(core_path, locale_data));
  1299. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  1300. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  1301. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  1302. StringBuilder builder;
  1303. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  1304. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  1305. builder.appendff("-{}", script);
  1306. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  1307. builder.appendff("-{}", region);
  1308. return builder.build();
  1309. };
  1310. while (dates_iterator.has_next()) {
  1311. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  1312. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  1313. auto language = TRY(remove_variants_from_path(dates_path));
  1314. auto& locale = locale_data.locales.ensure(language);
  1315. while (calendars_iterator.has_next()) {
  1316. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  1317. TRY(parse_calendars(move(calendars_path), locale_data, locale));
  1318. }
  1319. TRY(parse_time_zone_names(move(dates_path), locale_data, locale));
  1320. }
  1321. TRY(parse_day_periods(move(core_path), locale_data));
  1322. return {};
  1323. }
  1324. static String format_identifier(StringView owner, String identifier)
  1325. {
  1326. identifier = identifier.replace("-"sv, "_"sv, true);
  1327. identifier = identifier.replace("/"sv, "_"sv, true);
  1328. if (all_of(identifier, is_ascii_digit))
  1329. return String::formatted("{}_{}", owner[0], identifier);
  1330. if (is_ascii_lower_alpha(identifier[0]))
  1331. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  1332. return identifier;
  1333. }
  1334. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  1335. {
  1336. StringBuilder builder;
  1337. SourceGenerator generator { builder };
  1338. generator.append(R"~~~(
  1339. #pragma once
  1340. #include <AK/Types.h>
  1341. namespace Unicode {
  1342. )~~~");
  1343. generate_enum(generator, format_identifier, "Calendar"sv, {}, locale_data.calendars, locale_data.calendar_aliases);
  1344. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, locale_data.hour_cycle_regions);
  1345. generator.append(R"~~~(
  1346. }
  1347. )~~~");
  1348. VERIFY(file.write(generator.as_string_view()));
  1349. }
  1350. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  1351. {
  1352. StringBuilder builder;
  1353. SourceGenerator generator { builder };
  1354. generator.set("string_index_type"sv, s_string_index_type);
  1355. generator.set("calendar_pattern_index_type"sv, s_calendar_pattern_index_type);
  1356. generator.set("calendar_pattern_list_index_type"sv, s_calendar_pattern_list_index_type);
  1357. generator.set("calendar_range_pattern_index_type"sv, s_calendar_range_pattern_index_type);
  1358. generator.set("calendar_range_pattern_list_index_type"sv, s_calendar_range_pattern_list_index_type);
  1359. generator.set("calendar_format_index_type"sv, s_calendar_format_index_type);
  1360. generator.set("symbol_list_index_type"sv, s_symbol_list_index_type);
  1361. generator.set("calendar_symbols_index_type"sv, s_calendar_symbols_index_type);
  1362. generator.set("calendar_symbols_list_index_type"sv, s_calendar_symbols_list_index_type);
  1363. generator.set("calendar_index_type"sv, s_calendar_index_type);
  1364. generator.set("time_zone_index_type"sv, s_time_zone_index_type);
  1365. generator.set("time_zone_list_index_type"sv, s_time_zone_list_index_type);
  1366. generator.set("day_period_index_type"sv, s_day_period_index_type);
  1367. generator.set("day_period_list_index_type"sv, s_day_period_list_index_type);
  1368. generator.append(R"~~~(
  1369. #include <AK/Array.h>
  1370. #include <AK/BinarySearch.h>
  1371. #include <AK/Optional.h>
  1372. #include <AK/StringView.h>
  1373. #include <LibTimeZone/TimeZone.h>
  1374. #include <LibUnicode/DateTimeFormat.h>
  1375. #include <LibUnicode/Locale.h>
  1376. #include <LibUnicode/UnicodeDateTimeFormat.h>
  1377. namespace Unicode {
  1378. )~~~");
  1379. locale_data.unique_strings.generate(generator);
  1380. generator.append(R"~~~(
  1381. template <typename SourceType, typename TargetType>
  1382. static void convert_calendar_fields(SourceType const& source, TargetType& target)
  1383. {
  1384. if (source.era != -1)
  1385. target.era = static_cast<CalendarPatternStyle>(source.era);
  1386. if (source.year != -1)
  1387. target.year = static_cast<CalendarPatternStyle>(source.year);
  1388. if (source.month != -1)
  1389. target.month = static_cast<CalendarPatternStyle>(source.month);
  1390. if (source.weekday != -1)
  1391. target.weekday = static_cast<CalendarPatternStyle>(source.weekday);
  1392. if (source.day != -1)
  1393. target.day = static_cast<CalendarPatternStyle>(source.day);
  1394. if (source.day_period != -1)
  1395. target.day_period = static_cast<CalendarPatternStyle>(source.day_period);
  1396. if (source.hour != -1)
  1397. target.hour = static_cast<CalendarPatternStyle>(source.hour);
  1398. if (source.minute != -1)
  1399. target.minute = static_cast<CalendarPatternStyle>(source.minute);
  1400. if (source.second != -1)
  1401. target.second = static_cast<CalendarPatternStyle>(source.second);
  1402. if (source.fractional_second_digits != -1)
  1403. target.fractional_second_digits = static_cast<u8>(source.fractional_second_digits);
  1404. if (source.time_zone_name != -1)
  1405. target.time_zone_name = static_cast<CalendarPatternStyle>(source.time_zone_name);
  1406. }
  1407. struct CalendarPatternImpl {
  1408. CalendarPattern to_unicode_calendar_pattern() const {
  1409. CalendarPattern calendar_pattern {};
  1410. calendar_pattern.skeleton = s_string_list[skeleton];
  1411. calendar_pattern.pattern = s_string_list[pattern];
  1412. if (pattern12 != 0)
  1413. calendar_pattern.pattern12 = s_string_list[pattern12];
  1414. convert_calendar_fields(*this, calendar_pattern);
  1415. return calendar_pattern;
  1416. }
  1417. @string_index_type@ skeleton { 0 };
  1418. @string_index_type@ pattern { 0 };
  1419. @string_index_type@ pattern12 { 0 };
  1420. i8 era { -1 };
  1421. i8 year { -1 };
  1422. i8 month { -1 };
  1423. i8 weekday { -1 };
  1424. i8 day { -1 };
  1425. i8 day_period { -1 };
  1426. i8 hour { -1 };
  1427. i8 minute { -1 };
  1428. i8 second { -1 };
  1429. i8 fractional_second_digits { -1 };
  1430. i8 time_zone_name { -1 };
  1431. };
  1432. struct CalendarRangePatternImpl {
  1433. CalendarRangePattern to_unicode_calendar_range_pattern() const {
  1434. CalendarRangePattern calendar_range_pattern {};
  1435. if (field != -1)
  1436. calendar_range_pattern.field = static_cast<CalendarRangePattern::Field>(field);
  1437. calendar_range_pattern.start_range = s_string_list[start_range];
  1438. calendar_range_pattern.separator = s_string_list[separator];
  1439. calendar_range_pattern.end_range = s_string_list[end_range];
  1440. convert_calendar_fields(*this, calendar_range_pattern);
  1441. return calendar_range_pattern;
  1442. }
  1443. @string_index_type@ skeleton { 0 };
  1444. i8 field { -1 };
  1445. @string_index_type@ start_range { 0 };
  1446. @string_index_type@ separator { 0 };
  1447. @string_index_type@ end_range { 0 };
  1448. i8 era { -1 };
  1449. i8 year { -1 };
  1450. i8 month { -1 };
  1451. i8 weekday { -1 };
  1452. i8 day { -1 };
  1453. i8 day_period { -1 };
  1454. i8 hour { -1 };
  1455. i8 minute { -1 };
  1456. i8 second { -1 };
  1457. i8 fractional_second_digits { -1 };
  1458. i8 time_zone_name { -1 };
  1459. };
  1460. )~~~");
  1461. locale_data.unique_patterns.generate(generator, "CalendarPatternImpl"sv, "s_calendar_patterns"sv, 10);
  1462. locale_data.unique_pattern_lists.generate(generator, s_calendar_pattern_index_type, "s_calendar_pattern_lists"sv);
  1463. locale_data.unique_range_patterns.generate(generator, "CalendarRangePatternImpl"sv, "s_calendar_range_patterns"sv, 10);
  1464. locale_data.unique_range_pattern_lists.generate(generator, s_calendar_range_pattern_index_type, "s_calendar_range_pattern_lists"sv);
  1465. generator.append(R"~~~(
  1466. struct CalendarFormatImpl {
  1467. CalendarFormat to_unicode_calendar_format() const {
  1468. CalendarFormat calendar_format {};
  1469. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  1470. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  1471. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  1472. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  1473. return calendar_format;
  1474. }
  1475. @calendar_pattern_index_type@ full_format { 0 };
  1476. @calendar_pattern_index_type@ long_format { 0 };
  1477. @calendar_pattern_index_type@ medium_format { 0 };
  1478. @calendar_pattern_index_type@ short_format { 0 };
  1479. };
  1480. struct CalendarSymbols {
  1481. @symbol_list_index_type@ narrow_symbols { 0 };
  1482. @symbol_list_index_type@ short_symbols { 0 };
  1483. @symbol_list_index_type@ long_symbols { 0 };
  1484. };
  1485. struct CalendarData {
  1486. @calendar_format_index_type@ date_formats { 0 };
  1487. @calendar_format_index_type@ time_formats { 0 };
  1488. @calendar_format_index_type@ date_time_formats { 0 };
  1489. @calendar_pattern_list_index_type@ available_formats { 0 };
  1490. @calendar_range_pattern_index_type@ default_range_format { 0 };
  1491. @calendar_range_pattern_list_index_type@ range_formats { 0 };
  1492. @calendar_range_pattern_list_index_type@ range12_formats { 0 };
  1493. @calendar_symbols_list_index_type@ symbols { 0 };
  1494. };
  1495. struct TimeZoneNames {
  1496. @string_index_type@ short_standard_name { 0 };
  1497. @string_index_type@ long_standard_name { 0 };
  1498. @string_index_type@ short_daylight_name { 0 };
  1499. @string_index_type@ long_daylight_name { 0 };
  1500. @string_index_type@ short_generic_name { 0 };
  1501. @string_index_type@ long_generic_name { 0 };
  1502. };
  1503. struct TimeZoneFormatImpl {
  1504. TimeZoneFormat to_time_zone_format() const {
  1505. TimeZoneFormat time_zone_format {};
  1506. time_zone_format.symbol_ahead_sign = s_string_list[symbol_ahead_sign];
  1507. time_zone_format.symbol_ahead_separator = s_string_list[symbol_ahead_separator];
  1508. time_zone_format.symbol_behind_sign = s_string_list[symbol_behind_sign];
  1509. time_zone_format.symbol_behind_separator = s_string_list[symbol_behind_separator];
  1510. time_zone_format.gmt_format = s_string_list[gmt_format];
  1511. time_zone_format.gmt_zero_format = s_string_list[gmt_zero_format];
  1512. return time_zone_format;
  1513. }
  1514. @string_index_type@ symbol_ahead_sign { 0 };
  1515. @string_index_type@ symbol_ahead_separator { 0 };
  1516. @string_index_type@ symbol_behind_sign { 0 };
  1517. @string_index_type@ symbol_behind_separator { 0 };
  1518. @string_index_type@ gmt_format { 0 };
  1519. @string_index_type@ gmt_zero_format { 0 };
  1520. };
  1521. struct DayPeriodData {
  1522. u8 day_period { 0 };
  1523. u8 begin { 0 };
  1524. u8 end { 0 };
  1525. };
  1526. )~~~");
  1527. locale_data.unique_formats.generate(generator, "CalendarFormatImpl"sv, "s_calendar_formats"sv, 10);
  1528. locale_data.unique_symbol_lists.generate(generator, s_string_index_type, "s_symbol_lists"sv);
  1529. locale_data.unique_calendar_symbols.generate(generator, "CalendarSymbols"sv, "s_calendar_symbols"sv, 10);
  1530. locale_data.unique_calendar_symbols_lists.generate(generator, s_calendar_symbols_index_type, "s_calendar_symbol_lists"sv);
  1531. locale_data.unique_calendars.generate(generator, "CalendarData"sv, "s_calendars"sv, 10);
  1532. locale_data.unique_time_zones.generate(generator, "TimeZoneNames"sv, "s_time_zones"sv, 30);
  1533. locale_data.unique_time_zone_lists.generate(generator, s_time_zone_index_type, "s_time_zone_lists"sv);
  1534. locale_data.unique_time_zone_formats.generate(generator, "TimeZoneFormatImpl"sv, "s_time_zone_formats"sv, 30);
  1535. locale_data.unique_day_periods.generate(generator, "DayPeriodData"sv, "s_day_periods"sv, 30);
  1536. locale_data.unique_day_period_lists.generate(generator, s_day_period_index_type, "s_day_period_lists"sv);
  1537. locale_data.unique_hour_cycle_lists.generate(generator, "u8"sv, "s_hour_cycle_lists"sv);
  1538. auto append_calendars = [&](String name, auto const& calendars) {
  1539. generator.set("name", name);
  1540. generator.set("size", String::number(calendars.size()));
  1541. generator.append(R"~~~(
  1542. static constexpr Array<@calendar_index_type@, @size@> @name@ { {)~~~");
  1543. bool first = true;
  1544. for (auto const& calendar_key : locale_data.calendars) {
  1545. auto calendar = calendars.find(calendar_key)->value;
  1546. generator.append(first ? " " : ", ");
  1547. generator.append(String::number(calendar));
  1548. first = false;
  1549. }
  1550. generator.append(" } };");
  1551. };
  1552. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  1553. generator.set("type", type);
  1554. generator.set("name", name);
  1555. generator.set("size", String::number(keys.size()));
  1556. generator.append(R"~~~(
  1557. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1558. bool first = true;
  1559. for (auto const& key : keys) {
  1560. auto const& value = map.find(key)->value;
  1561. auto mapping = mapping_getter(value);
  1562. generator.append(first ? " " : ", ");
  1563. generator.append(String::number(mapping));
  1564. first = false;
  1565. }
  1566. generator.append(" } };");
  1567. };
  1568. auto locales = locale_data.locales.keys();
  1569. quick_sort(locales);
  1570. 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); });
  1571. append_mapping(locales, locale_data.locales, s_time_zone_index_type, "s_locale_time_zones"sv, [](auto const& locale) { return locale.time_zones; });
  1572. 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; });
  1573. append_mapping(locales, locale_data.locales, s_day_period_index_type, "s_locale_day_periods"sv, [](auto const& locale) { return locale.day_periods; });
  1574. 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; });
  1575. generator.append("\n");
  1576. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1577. HashValueMap<String> hashes;
  1578. hashes.ensure_capacity(values.size());
  1579. for (auto const& value : values)
  1580. hashes.set(value.hash(), format_identifier(enum_title, value));
  1581. for (auto const& alias : aliases)
  1582. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1583. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1584. };
  1585. append_from_string("Calendar"sv, "calendar"sv, locale_data.calendars, locale_data.calendar_aliases);
  1586. append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, locale_data.hour_cycle_regions);
  1587. generator.append(R"~~~(
  1588. Vector<HourCycle> get_regional_hour_cycles(StringView region)
  1589. {
  1590. auto region_value = hour_cycle_region_from_string(region);
  1591. if (!region_value.has_value())
  1592. return {};
  1593. auto region_index = to_underlying(*region_value);
  1594. auto regional_hour_cycles_index = s_hour_cycles.at(region_index);
  1595. auto const& regional_hour_cycles = s_hour_cycle_lists.at(regional_hour_cycles_index);
  1596. Vector<HourCycle> hour_cycles;
  1597. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1598. for (auto hour_cycle : regional_hour_cycles)
  1599. hour_cycles.unchecked_append(static_cast<HourCycle>(hour_cycle));
  1600. return hour_cycles;
  1601. }
  1602. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1603. {
  1604. auto locale_value = locale_from_string(locale);
  1605. if (!locale_value.has_value())
  1606. return nullptr;
  1607. auto calendar_value = calendar_from_string(calendar);
  1608. if (!calendar_value.has_value())
  1609. return nullptr;
  1610. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1611. size_t calendar_index = to_underlying(*calendar_value);
  1612. auto const& calendar_indices = s_locale_calendars.at(locale_index);
  1613. calendar_index = calendar_indices[calendar_index];
  1614. return &s_calendars[calendar_index];
  1615. }
  1616. Optional<CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1617. {
  1618. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1619. auto const& formats = s_calendar_formats.at(data->date_formats);
  1620. return formats.to_unicode_calendar_format();
  1621. }
  1622. return {};
  1623. }
  1624. Optional<CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1625. {
  1626. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1627. auto const& formats = s_calendar_formats.at(data->time_formats);
  1628. return formats.to_unicode_calendar_format();
  1629. }
  1630. return {};
  1631. }
  1632. Optional<CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1633. {
  1634. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1635. auto const& formats = s_calendar_formats.at(data->date_time_formats);
  1636. return formats.to_unicode_calendar_format();
  1637. }
  1638. return {};
  1639. }
  1640. Vector<CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1641. {
  1642. Vector<CalendarPattern> result {};
  1643. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1644. auto const& available_formats = s_calendar_pattern_lists.at(data->available_formats);
  1645. result.ensure_capacity(available_formats.size());
  1646. for (auto const& format : available_formats)
  1647. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1648. }
  1649. return result;
  1650. }
  1651. Optional<CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar)
  1652. {
  1653. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1654. auto const& pattern = s_calendar_range_patterns[data->default_range_format];
  1655. return pattern.to_unicode_calendar_range_pattern();
  1656. }
  1657. return {};
  1658. }
  1659. Vector<CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton)
  1660. {
  1661. Vector<CalendarRangePattern> result {};
  1662. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1663. auto const& range_formats = s_calendar_range_pattern_lists.at(data->range_formats);
  1664. for (auto format : range_formats) {
  1665. auto const& pattern = s_calendar_range_patterns[format];
  1666. if (skeleton == s_string_list[pattern.skeleton])
  1667. result.append(pattern.to_unicode_calendar_range_pattern());
  1668. }
  1669. }
  1670. return result;
  1671. }
  1672. Vector<CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton)
  1673. {
  1674. Vector<CalendarRangePattern> result {};
  1675. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1676. auto const& range12_formats = s_calendar_range_pattern_lists.at(data->range12_formats);
  1677. for (auto format : range12_formats) {
  1678. auto const& pattern = s_calendar_range_patterns[format];
  1679. if (skeleton == s_string_list[pattern.skeleton])
  1680. result.append(pattern.to_unicode_calendar_range_pattern());
  1681. }
  1682. }
  1683. return result;
  1684. }
  1685. static Span<@string_index_type@ const> find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1686. {
  1687. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1688. auto const& symbols_list = s_calendar_symbol_lists[data->symbols];
  1689. auto symbol_index = to_underlying(symbol);
  1690. auto calendar_symbols_index = symbols_list.at(symbol_index);
  1691. auto const& symbols = s_calendar_symbols.at(calendar_symbols_index);
  1692. @symbol_list_index_type@ symbol_list_index = 0;
  1693. switch (style) {
  1694. case CalendarPatternStyle::Narrow:
  1695. symbol_list_index = symbols.narrow_symbols;
  1696. break;
  1697. case CalendarPatternStyle::Short:
  1698. symbol_list_index = symbols.short_symbols;
  1699. break;
  1700. case CalendarPatternStyle::Long:
  1701. symbol_list_index = symbols.long_symbols;
  1702. break;
  1703. default:
  1704. VERIFY_NOT_REACHED();
  1705. }
  1706. return s_symbol_lists.at(symbol_list_index);
  1707. }
  1708. return {};
  1709. }
  1710. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Era value)
  1711. {
  1712. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1713. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1714. return s_string_list[symbols.at(value_index)];
  1715. return {};
  1716. }
  1717. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Month value)
  1718. {
  1719. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1720. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1721. return s_string_list[symbols.at(value_index)];
  1722. return {};
  1723. }
  1724. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Weekday value)
  1725. {
  1726. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1727. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1728. return s_string_list[symbols.at(value_index)];
  1729. return {};
  1730. }
  1731. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, DayPeriod value)
  1732. {
  1733. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1734. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1735. return s_string_list[symbols.at(value_index)];
  1736. return {};
  1737. }
  1738. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour)
  1739. {
  1740. auto locale_value = locale_from_string(locale);
  1741. if (!locale_value.has_value())
  1742. return {};
  1743. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1744. auto day_periods_index = s_locale_day_periods[locale_index];
  1745. auto day_periods = s_day_period_lists[day_periods_index];
  1746. for (auto day_period_index : day_periods) {
  1747. auto day_period = s_day_periods[day_period_index];
  1748. auto h = hour;
  1749. if (day_period.begin > day_period.end) {
  1750. day_period.end += 24;
  1751. h += 24;
  1752. }
  1753. if ((day_period.begin <= h) && (h < day_period.end)) {
  1754. auto period = static_cast<DayPeriod>(day_period.day_period);
  1755. return get_calendar_day_period_symbol(locale, calendar, style, period);
  1756. }
  1757. }
  1758. // Fallback to fixed periods if the locale does not have flexible day periods.
  1759. // TR-35 states that the meaning of AM and PM does not change with locale.
  1760. if (hour < 12)
  1761. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::AM);
  1762. return get_calendar_day_period_symbol(locale, calendar, style, DayPeriod::PM);
  1763. }
  1764. Optional<TimeZoneFormat> get_time_zone_format(StringView locale)
  1765. {
  1766. auto locale_value = locale_from_string(locale);
  1767. if (!locale_value.has_value())
  1768. return {};
  1769. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1770. auto time_zone_format_index = s_locale_time_zone_formats.at(locale_index);
  1771. auto const& time_zone_format = s_time_zone_formats.at(time_zone_format_index);
  1772. return time_zone_format.to_time_zone_format();
  1773. }
  1774. static TimeZoneNames const* find_time_zone_names(StringView locale, StringView time_zone)
  1775. {
  1776. auto locale_value = locale_from_string(locale);
  1777. if (!locale_value.has_value())
  1778. return nullptr;
  1779. auto time_zone_value = ::TimeZone::time_zone_from_string(time_zone);
  1780. if (!time_zone_value.has_value())
  1781. return nullptr;
  1782. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1783. size_t time_zone_index = to_underlying(*time_zone_value);
  1784. auto time_zone_list_index = s_locale_time_zones.at(locale_index);
  1785. auto const& time_zone_list = s_time_zone_lists.at(time_zone_list_index);
  1786. if (time_zone_list.size() <= time_zone_index)
  1787. return nullptr;
  1788. time_zone_index = time_zone_list.at(time_zone_index);
  1789. return &s_time_zones[time_zone_index];
  1790. }
  1791. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style, TimeZone::InDST in_dst)
  1792. {
  1793. if (auto const* data = find_time_zone_names(locale, time_zone); data != nullptr) {
  1794. size_t name_index = 0;
  1795. switch (style) {
  1796. case CalendarPatternStyle::Short:
  1797. name_index = (in_dst == TimeZone::InDST::No) ? data->short_standard_name : data->short_daylight_name;
  1798. break;
  1799. case CalendarPatternStyle::Long:
  1800. name_index = (in_dst == TimeZone::InDST::No) ? data->long_standard_name : data->long_daylight_name;
  1801. break;
  1802. case CalendarPatternStyle::ShortGeneric:
  1803. name_index = data->short_generic_name;
  1804. break;
  1805. case CalendarPatternStyle::LongGeneric:
  1806. name_index = data->long_generic_name;
  1807. break;
  1808. default:
  1809. VERIFY_NOT_REACHED();
  1810. }
  1811. if (name_index != 0)
  1812. return s_string_list[name_index];
  1813. }
  1814. return {};
  1815. }
  1816. }
  1817. )~~~");
  1818. VERIFY(file.write(generator.as_string_view()));
  1819. }
  1820. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1821. {
  1822. StringView generated_header_path;
  1823. StringView generated_implementation_path;
  1824. StringView core_path;
  1825. StringView dates_path;
  1826. Core::ArgsParser args_parser;
  1827. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1828. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1829. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1830. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1831. args_parser.parse(arguments);
  1832. auto open_file = [&](StringView path) -> ErrorOr<NonnullRefPtr<Core::File>> {
  1833. if (path.is_empty()) {
  1834. args_parser.print_usage(stderr, arguments.argv[0]);
  1835. return Error::from_string_literal("Must provide all command line options"sv);
  1836. }
  1837. return Core::File::open(path, Core::OpenMode::ReadWrite);
  1838. };
  1839. auto generated_header_file = TRY(open_file(generated_header_path));
  1840. auto generated_implementation_file = TRY(open_file(generated_implementation_path));
  1841. UnicodeLocaleData locale_data;
  1842. TRY(parse_all_locales(core_path, dates_path, locale_data));
  1843. generate_unicode_locale_header(generated_header_file, locale_data);
  1844. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  1845. return 0;
  1846. }