GenerateUnicodeDateTimeFormat.cpp 80 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015
  1. /*
  2. * Copyright (c) 2021, 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 <LibUnicode/DateTimeFormat.h>
  26. using StringIndexType = u16;
  27. constexpr auto s_string_index_type = "u16"sv;
  28. using CalendarPatternIndexType = u16;
  29. constexpr auto s_calendar_pattern_index_type = "u16"sv;
  30. using CalendarPatternListIndexType = u8;
  31. constexpr auto s_calendar_pattern_list_index_type = "u8"sv;
  32. using CalendarRangePatternIndexType = u16;
  33. constexpr auto s_calendar_range_pattern_index_type = "u16"sv;
  34. using CalendarFormatIndexType = u8;
  35. constexpr auto s_calendar_format_index_type = "u8"sv;
  36. using SymbolListIndexType = u16;
  37. constexpr auto s_symbol_list_index_type = "u16"sv;
  38. using CalendarSymbolsIndexType = u16;
  39. constexpr auto s_calendar_symbols_index_type = "u16"sv;
  40. using CalendarSymbolsListIndexType = u8;
  41. constexpr auto s_calendar_symbols_list_index_type = "u8"sv;
  42. struct CalendarPattern : public Unicode::CalendarPattern {
  43. bool contains_only_date_fields() const
  44. {
  45. return !day_period.has_value() && !hour.has_value() && !minute.has_value() && !second.has_value() && !fractional_second_digits.has_value() && !time_zone_name.has_value();
  46. }
  47. bool contains_only_time_fields() const
  48. {
  49. return !weekday.has_value() && !era.has_value() && !year.has_value() && !month.has_value() && !day.has_value();
  50. }
  51. unsigned hash() const
  52. {
  53. auto hash = pair_int_hash(pattern_index, pattern12_index);
  54. hash = pair_int_hash(hash, skeleton_index);
  55. auto hash_field = [&](auto const& field) {
  56. if (field.has_value())
  57. hash = pair_int_hash(hash, static_cast<u8>(*field));
  58. else
  59. hash = pair_int_hash(hash, -1);
  60. };
  61. hash_field(era);
  62. hash_field(year);
  63. hash_field(month);
  64. hash_field(weekday);
  65. hash_field(day);
  66. hash_field(day_period);
  67. hash_field(hour);
  68. hash_field(minute);
  69. hash_field(second);
  70. hash_field(fractional_second_digits);
  71. hash_field(time_zone_name);
  72. return hash;
  73. }
  74. bool operator==(CalendarPattern const& other) const
  75. {
  76. return (skeleton_index == other.skeleton_index)
  77. && (pattern_index == other.pattern_index)
  78. && (pattern12_index == other.pattern12_index)
  79. && (era == other.era)
  80. && (year == other.year)
  81. && (month == other.month)
  82. && (weekday == other.weekday)
  83. && (day == other.day)
  84. && (day_period == other.day_period)
  85. && (hour == other.hour)
  86. && (minute == other.minute)
  87. && (second == other.second)
  88. && (fractional_second_digits == other.fractional_second_digits)
  89. && (time_zone_name == other.time_zone_name);
  90. }
  91. StringIndexType skeleton_index { 0 };
  92. StringIndexType pattern_index { 0 };
  93. StringIndexType pattern12_index { 0 };
  94. };
  95. template<>
  96. struct AK::Formatter<CalendarPattern> : Formatter<FormatString> {
  97. ErrorOr<void> format(FormatBuilder& builder, CalendarPattern const& pattern)
  98. {
  99. auto field_to_i8 = [](auto const& field) -> i8 {
  100. if (!field.has_value())
  101. return -1;
  102. return static_cast<i8>(*field);
  103. };
  104. return Formatter<FormatString>::format(builder,
  105. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}",
  106. pattern.skeleton_index,
  107. pattern.pattern_index,
  108. pattern.pattern12_index,
  109. field_to_i8(pattern.era),
  110. field_to_i8(pattern.year),
  111. field_to_i8(pattern.month),
  112. field_to_i8(pattern.weekday),
  113. field_to_i8(pattern.day),
  114. field_to_i8(pattern.day_period),
  115. field_to_i8(pattern.hour),
  116. field_to_i8(pattern.minute),
  117. field_to_i8(pattern.second),
  118. field_to_i8(pattern.fractional_second_digits),
  119. field_to_i8(pattern.time_zone_name));
  120. }
  121. };
  122. template<>
  123. struct AK::Traits<CalendarPattern> : public GenericTraits<CalendarPattern> {
  124. static unsigned hash(CalendarPattern const& c) { return c.hash(); }
  125. };
  126. struct CalendarRangePattern : public CalendarPattern {
  127. unsigned hash() const
  128. {
  129. auto hash = CalendarPattern::hash();
  130. if (field.has_value())
  131. hash = pair_int_hash(hash, static_cast<u8>(*field));
  132. hash = pair_int_hash(hash, start_range);
  133. hash = pair_int_hash(hash, separator);
  134. hash = pair_int_hash(hash, end_range);
  135. return hash;
  136. }
  137. bool operator==(CalendarRangePattern const& other) const
  138. {
  139. if (!CalendarPattern::operator==(other))
  140. return false;
  141. return (field == other.field)
  142. && (start_range == other.start_range)
  143. && (separator == other.separator)
  144. && (end_range == other.end_range);
  145. }
  146. Optional<Unicode::CalendarRangePattern::Field> field {};
  147. StringIndexType start_range { 0 };
  148. StringIndexType separator { 0 };
  149. StringIndexType end_range { 0 };
  150. };
  151. template<>
  152. struct AK::Formatter<CalendarRangePattern> : Formatter<FormatString> {
  153. ErrorOr<void> format(FormatBuilder& builder, CalendarRangePattern const& pattern)
  154. {
  155. auto field_to_i8 = [](auto const& field) -> i8 {
  156. if (!field.has_value())
  157. return -1;
  158. return static_cast<i8>(*field);
  159. };
  160. return Formatter<FormatString>::format(builder,
  161. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}",
  162. pattern.skeleton_index,
  163. field_to_i8(pattern.field),
  164. pattern.start_range,
  165. pattern.separator,
  166. pattern.end_range,
  167. field_to_i8(pattern.era),
  168. field_to_i8(pattern.year),
  169. field_to_i8(pattern.month),
  170. field_to_i8(pattern.weekday),
  171. field_to_i8(pattern.day),
  172. field_to_i8(pattern.day_period),
  173. field_to_i8(pattern.hour),
  174. field_to_i8(pattern.minute),
  175. field_to_i8(pattern.second),
  176. field_to_i8(pattern.fractional_second_digits),
  177. field_to_i8(pattern.time_zone_name));
  178. }
  179. };
  180. template<>
  181. struct AK::Traits<CalendarRangePattern> : public GenericTraits<CalendarRangePattern> {
  182. static unsigned hash(CalendarRangePattern const& c) { return c.hash(); }
  183. };
  184. struct CalendarFormat {
  185. unsigned hash() const
  186. {
  187. auto hash = pair_int_hash(full_format, long_format);
  188. hash = pair_int_hash(hash, medium_format);
  189. hash = pair_int_hash(hash, short_format);
  190. return hash;
  191. }
  192. bool operator==(CalendarFormat const& other) const
  193. {
  194. return (full_format == other.full_format)
  195. && (long_format == other.long_format)
  196. && (medium_format == other.medium_format)
  197. && (short_format == other.short_format);
  198. }
  199. CalendarPatternIndexType full_format { 0 };
  200. CalendarPatternIndexType long_format { 0 };
  201. CalendarPatternIndexType medium_format { 0 };
  202. CalendarPatternIndexType short_format { 0 };
  203. };
  204. template<>
  205. struct AK::Formatter<CalendarFormat> : Formatter<FormatString> {
  206. ErrorOr<void> format(FormatBuilder& builder, CalendarFormat const& pattern)
  207. {
  208. return Formatter<FormatString>::format(builder,
  209. "{{ {}, {}, {}, {} }}",
  210. pattern.full_format,
  211. pattern.long_format,
  212. pattern.medium_format,
  213. pattern.short_format);
  214. }
  215. };
  216. template<>
  217. struct AK::Traits<CalendarFormat> : public GenericTraits<CalendarFormat> {
  218. static unsigned hash(CalendarFormat const& c) { return c.hash(); }
  219. };
  220. using SymbolList = Vector<StringIndexType>;
  221. struct CalendarSymbols {
  222. unsigned hash() const
  223. {
  224. auto hash = pair_int_hash(narrow_symbols, short_symbols);
  225. hash = pair_int_hash(hash, long_symbols);
  226. return hash;
  227. }
  228. bool operator==(CalendarSymbols const& other) const
  229. {
  230. return (narrow_symbols == other.narrow_symbols)
  231. && (short_symbols == other.short_symbols)
  232. && (long_symbols == other.long_symbols);
  233. }
  234. SymbolListIndexType narrow_symbols { 0 };
  235. SymbolListIndexType short_symbols { 0 };
  236. SymbolListIndexType long_symbols { 0 };
  237. };
  238. template<>
  239. struct AK::Formatter<CalendarSymbols> : Formatter<FormatString> {
  240. ErrorOr<void> format(FormatBuilder& builder, CalendarSymbols const& symbols)
  241. {
  242. return Formatter<FormatString>::format(builder,
  243. "{{ {}, {}, {} }}",
  244. symbols.narrow_symbols,
  245. symbols.short_symbols,
  246. symbols.long_symbols);
  247. }
  248. };
  249. template<>
  250. struct AK::Traits<CalendarSymbols> : public GenericTraits<CalendarSymbols> {
  251. static unsigned hash(CalendarSymbols const& c) { return c.hash(); }
  252. };
  253. using CalendarPatternList = Vector<CalendarPatternIndexType>;
  254. using CalendarSymbolsList = Vector<CalendarSymbolsIndexType>;
  255. struct Calendar {
  256. StringIndexType calendar { 0 };
  257. CalendarFormatIndexType date_formats { 0 };
  258. CalendarFormatIndexType time_formats { 0 };
  259. CalendarFormatIndexType date_time_formats { 0 };
  260. CalendarPatternListIndexType available_formats { 0 };
  261. CalendarRangePatternIndexType default_range_format { 0 };
  262. Vector<CalendarRangePatternIndexType> range_formats {};
  263. Vector<CalendarRangePatternIndexType> range12_formats {};
  264. CalendarSymbolsListIndexType symbols { 0 };
  265. };
  266. struct TimeZone {
  267. StringIndexType time_zone { 0 };
  268. StringIndexType long_name { 0 };
  269. StringIndexType short_name { 0 };
  270. };
  271. struct DayPeriod {
  272. Unicode::DayPeriod day_period {};
  273. u8 begin { 0 };
  274. u8 end { 0 };
  275. };
  276. struct Locale {
  277. HashMap<String, Calendar> calendars;
  278. HashMap<String, TimeZone> time_zones;
  279. Vector<DayPeriod> day_periods;
  280. };
  281. struct UnicodeLocaleData {
  282. UniqueStringStorage<StringIndexType> unique_strings;
  283. UniqueStorage<CalendarPattern, CalendarPatternIndexType> unique_patterns;
  284. UniqueStorage<CalendarPatternList, CalendarPatternListIndexType> unique_pattern_lists;
  285. UniqueStorage<CalendarRangePattern, CalendarRangePatternIndexType> unique_range_patterns;
  286. UniqueStorage<CalendarFormat, CalendarFormatIndexType> unique_formats;
  287. UniqueStorage<SymbolList, SymbolListIndexType> unique_symbol_lists;
  288. UniqueStorage<CalendarSymbols, CalendarSymbolsIndexType> unique_calendar_symbols;
  289. UniqueStorage<CalendarSymbolsList, CalendarSymbolsListIndexType> unique_calendar_symbols_lists;
  290. HashMap<String, Locale> locales;
  291. HashMap<String, Vector<Unicode::HourCycle>> hour_cycles;
  292. Vector<String> hour_cycle_regions;
  293. HashMap<String, StringIndexType> meta_zones;
  294. Vector<String> time_zones;
  295. Vector<String> calendars;
  296. Vector<Alias> calendar_aliases {
  297. // FIXME: Aliases should come from BCP47. See: https://unicode-org.atlassian.net/browse/CLDR-15158
  298. { "gregorian"sv, "gregory"sv },
  299. };
  300. };
  301. static Optional<Unicode::DayPeriod> day_period_from_string(StringView day_period)
  302. {
  303. if (day_period == "am"sv)
  304. return Unicode::DayPeriod::AM;
  305. if (day_period == "pm"sv)
  306. return Unicode::DayPeriod::PM;
  307. if (day_period == "morning1"sv)
  308. return Unicode::DayPeriod::Morning1;
  309. if (day_period == "morning2"sv)
  310. return Unicode::DayPeriod::Morning2;
  311. if (day_period == "afternoon1"sv)
  312. return Unicode::DayPeriod::Afternoon1;
  313. if (day_period == "afternoon2"sv)
  314. return Unicode::DayPeriod::Afternoon2;
  315. if (day_period == "evening1"sv)
  316. return Unicode::DayPeriod::Evening1;
  317. if (day_period == "evening2"sv)
  318. return Unicode::DayPeriod::Evening2;
  319. if (day_period == "night1"sv)
  320. return Unicode::DayPeriod::Night1;
  321. if (day_period == "night2"sv)
  322. return Unicode::DayPeriod::Night2;
  323. return {};
  324. };
  325. static ErrorOr<void> parse_hour_cycles(String core_path, UnicodeLocaleData& locale_data)
  326. {
  327. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  328. LexicalPath time_data_path(move(core_path));
  329. time_data_path = time_data_path.append("supplemental"sv);
  330. time_data_path = time_data_path.append("timeData.json"sv);
  331. auto time_data_file = TRY(Core::File::open(time_data_path.string(), Core::OpenMode::ReadOnly));
  332. auto time_data = TRY(JsonValue::from_string(time_data_file->read_all()));
  333. auto const& supplemental_object = time_data.as_object().get("supplemental"sv);
  334. auto const& time_data_object = supplemental_object.as_object().get("timeData"sv);
  335. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Unicode::HourCycle> {
  336. if (hour_cycle == "h"sv)
  337. return Unicode::HourCycle::H12;
  338. if (hour_cycle == "H"sv)
  339. return Unicode::HourCycle::H23;
  340. if (hour_cycle == "K"sv)
  341. return Unicode::HourCycle::H11;
  342. if (hour_cycle == "k"sv)
  343. return Unicode::HourCycle::H24;
  344. return {};
  345. };
  346. time_data_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  347. auto allowed_hour_cycles_string = value.as_object().get("_allowed"sv).as_string();
  348. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  349. Vector<Unicode::HourCycle> hour_cycles;
  350. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  351. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  352. hour_cycles.append(*hour_cycle);
  353. }
  354. locale_data.hour_cycles.set(key, move(hour_cycles));
  355. if (!locale_data.hour_cycle_regions.contains_slow(key))
  356. locale_data.hour_cycle_regions.append(key);
  357. });
  358. return {};
  359. };
  360. static ErrorOr<void> parse_meta_zones(String core_path, UnicodeLocaleData& locale_data)
  361. {
  362. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  363. LexicalPath meta_zone_path(move(core_path));
  364. meta_zone_path = meta_zone_path.append("supplemental"sv);
  365. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  366. auto meta_zone_file = TRY(Core::File::open(meta_zone_path.string(), Core::OpenMode::ReadOnly));
  367. auto meta_zone = TRY(JsonValue::from_string(meta_zone_file->read_all()));
  368. auto const& supplemental_object = meta_zone.as_object().get("supplemental"sv);
  369. auto const& meta_zone_object = supplemental_object.as_object().get("metaZones"sv);
  370. auto const& meta_zone_array = meta_zone_object.as_object().get("metazones"sv);
  371. meta_zone_array.as_array().for_each([&](JsonValue const& value) {
  372. auto const& mapping = value.as_object().get("mapZone"sv);
  373. auto const& meta_zone = mapping.as_object().get("_other"sv);
  374. auto const& golden_zone = mapping.as_object().get("_type"sv);
  375. auto golden_zone_index = locale_data.unique_strings.ensure(golden_zone.as_string());
  376. locale_data.meta_zones.set(meta_zone.as_string(), golden_zone_index);
  377. });
  378. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  379. locale_data.meta_zones.set("UTC"sv, locale_data.unique_strings.ensure("UTC"sv));
  380. return {};
  381. };
  382. static constexpr auto is_char(char ch)
  383. {
  384. return [ch](auto c) { return c == ch; };
  385. }
  386. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  387. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  388. // extra spaces around the day period. Some example expected removals:
  389. //
  390. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  391. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  392. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  393. static String remove_period_from_pattern(String pattern)
  394. {
  395. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  396. auto index = pattern.find(remove);
  397. if (!index.has_value())
  398. continue;
  399. constexpr u32 space = ' ';
  400. constexpr u32 open = '{';
  401. constexpr u32 close = '}';
  402. Utf8View utf8_pattern { pattern };
  403. Optional<u32> before_removal;
  404. Optional<u32> after_removal;
  405. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  406. before_removal = *it;
  407. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  408. after_removal = *it;
  409. if ((before_removal == space) && (after_removal != open)) {
  410. pattern = String::formatted("{}{}",
  411. pattern.substring_view(0, *index - 1),
  412. pattern.substring_view(*index + remove.length()));
  413. } else if ((after_removal == space) && (before_removal != close)) {
  414. pattern = String::formatted("{}{}",
  415. pattern.substring_view(0, *index),
  416. pattern.substring_view(*index + remove.length() + 1));
  417. } else {
  418. pattern = String::formatted("{}{}",
  419. pattern.substring_view(0, *index),
  420. pattern.substring_view(*index + remove.length()));
  421. }
  422. }
  423. return pattern;
  424. }
  425. static Optional<CalendarPattern> parse_date_time_pattern_raw(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  426. {
  427. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  428. using Unicode::CalendarPatternStyle;
  429. CalendarPattern format {};
  430. if (!skeleton.is_empty())
  431. format.skeleton_index = locale_data.unique_strings.ensure(move(skeleton));
  432. GenericLexer lexer { pattern };
  433. StringBuilder builder;
  434. bool hour12 { false };
  435. while (!lexer.is_eof()) {
  436. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  437. // processing (this just avoids conflicts with the patterns below).
  438. if (lexer.next_is(is_quote)) {
  439. builder.append(lexer.consume_quoted_string());
  440. continue;
  441. }
  442. auto starting_char = lexer.peek();
  443. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  444. // Era
  445. if (all_of(segment, is_char('G'))) {
  446. builder.append("{era}");
  447. if (segment.length() <= 3)
  448. format.era = CalendarPatternStyle::Short;
  449. else if (segment.length() == 4)
  450. format.era = CalendarPatternStyle::Long;
  451. else
  452. format.era = CalendarPatternStyle::Narrow;
  453. }
  454. // Year
  455. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  456. builder.append("{year}");
  457. if (segment.length() == 2)
  458. format.year = CalendarPatternStyle::TwoDigit;
  459. else
  460. format.year = CalendarPatternStyle::Numeric;
  461. }
  462. // Quarter
  463. else if (all_of(segment, is_any_of("qQ"sv))) {
  464. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  465. return {};
  466. }
  467. // Month
  468. else if (all_of(segment, is_any_of("ML"sv))) {
  469. builder.append("{month}");
  470. if (segment.length() == 1)
  471. format.month = CalendarPatternStyle::Numeric;
  472. else if (segment.length() == 2)
  473. format.month = CalendarPatternStyle::TwoDigit;
  474. else if (segment.length() == 3)
  475. format.month = CalendarPatternStyle::Short;
  476. else if (segment.length() == 4)
  477. format.month = CalendarPatternStyle::Long;
  478. else if (segment.length() == 5)
  479. format.month = CalendarPatternStyle::Narrow;
  480. } else if (all_of(segment, is_char('l'))) {
  481. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  482. return {};
  483. }
  484. // Week
  485. else if (all_of(segment, is_any_of("wW"sv))) {
  486. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  487. return {};
  488. }
  489. // Day
  490. else if (all_of(segment, is_char('d'))) {
  491. builder.append("{day}");
  492. if (segment.length() == 1)
  493. format.day = CalendarPatternStyle::Numeric;
  494. else
  495. format.day = CalendarPatternStyle::TwoDigit;
  496. } else if (all_of(segment, is_any_of("DFg"sv))) {
  497. builder.append("{day}");
  498. format.day = CalendarPatternStyle::Numeric;
  499. }
  500. // Weekday
  501. else if (all_of(segment, is_char('E'))) {
  502. builder.append("{weekday}");
  503. if (segment.length() == 4)
  504. format.weekday = CalendarPatternStyle::Long;
  505. else if (segment.length() == 5)
  506. format.weekday = CalendarPatternStyle::Narrow;
  507. else
  508. format.weekday = CalendarPatternStyle::Short;
  509. } else if (all_of(segment, is_any_of("ec"sv))) {
  510. builder.append("{weekday}");
  511. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  512. // pattern styles are not supported by Intl.DateTimeFormat.
  513. if (segment.length() <= 2)
  514. return {};
  515. if (segment.length() == 4)
  516. format.weekday = CalendarPatternStyle::Long;
  517. else if (segment.length() == 5)
  518. format.weekday = CalendarPatternStyle::Narrow;
  519. else
  520. format.weekday = CalendarPatternStyle::Short;
  521. }
  522. // Period
  523. else if (all_of(segment, is_any_of("ab"sv))) {
  524. builder.append("{ampm}");
  525. hour12 = true;
  526. } else if (all_of(segment, is_char('B'))) {
  527. builder.append("{dayPeriod}");
  528. hour12 = true;
  529. if (segment.length() == 4)
  530. format.day_period = CalendarPatternStyle::Long;
  531. else if (segment.length() == 5)
  532. format.day_period = CalendarPatternStyle::Narrow;
  533. else
  534. format.day_period = CalendarPatternStyle::Short;
  535. }
  536. // Hour
  537. else if (all_of(segment, is_any_of("hHKk"sv))) {
  538. builder.append("{hour}");
  539. if ((segment[0] == 'h') || (segment[0] == 'K'))
  540. hour12 = true;
  541. if (segment.length() == 1)
  542. format.hour = CalendarPatternStyle::Numeric;
  543. else
  544. format.hour = CalendarPatternStyle::TwoDigit;
  545. } else if (all_of(segment, is_any_of("jJC"sv))) {
  546. // TR-35 indicates these should not be used.
  547. return {};
  548. }
  549. // Minute
  550. else if (all_of(segment, is_char('m'))) {
  551. builder.append("{minute}");
  552. if (segment.length() == 1)
  553. format.minute = CalendarPatternStyle::Numeric;
  554. else
  555. format.minute = CalendarPatternStyle::TwoDigit;
  556. }
  557. // Second
  558. else if (all_of(segment, is_char('s'))) {
  559. builder.append("{second}");
  560. if (segment.length() == 1)
  561. format.second = CalendarPatternStyle::Numeric;
  562. else
  563. format.second = CalendarPatternStyle::TwoDigit;
  564. } else if (all_of(segment, is_char('S'))) {
  565. builder.append("{fractionalSecondDigits}");
  566. VERIFY(segment.length() <= 3);
  567. format.fractional_second_digits = static_cast<u8>(segment.length());
  568. } else if (all_of(segment, is_char('A'))) {
  569. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  570. return {};
  571. }
  572. // Zone
  573. else if (all_of(segment, is_any_of("zZOvVXx"))) {
  574. builder.append("{timeZoneName}");
  575. if (segment.length() < 4)
  576. format.time_zone_name = CalendarPatternStyle::Short;
  577. else
  578. format.time_zone_name = CalendarPatternStyle::Long;
  579. }
  580. // Non-patterns
  581. else {
  582. builder.append(segment);
  583. }
  584. }
  585. pattern = builder.build();
  586. if (hour12) {
  587. format.pattern = remove_period_from_pattern(pattern);
  588. format.pattern12 = move(pattern);
  589. } else {
  590. format.pattern = move(pattern);
  591. }
  592. return format;
  593. }
  594. static Optional<CalendarPatternIndexType> parse_date_time_pattern(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  595. {
  596. auto format = parse_date_time_pattern_raw(move(pattern), move(skeleton), locale_data);
  597. if (!format.has_value())
  598. return {};
  599. format->pattern_index = locale_data.unique_strings.ensure(move(format->pattern));
  600. if (format->pattern12.has_value())
  601. format->pattern12_index = locale_data.unique_strings.ensure(format->pattern12.release_value());
  602. return locale_data.unique_patterns.ensure(format.release_value());
  603. }
  604. template<typename... Chars>
  605. static constexpr bool char_is_one_of(char ch, Chars&&... chars)
  606. {
  607. return ((ch == chars) || ...);
  608. }
  609. static void parse_interval_patterns(Calendar& calendar, JsonObject const& interval_formats_object, UnicodeLocaleData& locale_data)
  610. {
  611. // https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats
  612. auto name_of_field = [&](char field) {
  613. if (char_is_one_of(field, 'G'))
  614. return Unicode::CalendarRangePattern::Field::Era;
  615. if (char_is_one_of(field, 'y', 'Y', 'u', 'U', 'r'))
  616. return Unicode::CalendarRangePattern::Field::Year;
  617. if (char_is_one_of(field, 'M', 'L'))
  618. return Unicode::CalendarRangePattern::Field::Month;
  619. if (char_is_one_of(field, 'd', 'D', 'F', 'g'))
  620. return Unicode::CalendarRangePattern::Field::Day;
  621. if (char_is_one_of(field, 'a', 'b'))
  622. return Unicode::CalendarRangePattern::Field::AmPm;
  623. if (char_is_one_of(field, 'B'))
  624. return Unicode::CalendarRangePattern::Field::DayPeriod;
  625. if (char_is_one_of(field, 'h', 'H', 'K', 'k'))
  626. return Unicode::CalendarRangePattern::Field::Hour;
  627. if (char_is_one_of(field, 'm'))
  628. return Unicode::CalendarRangePattern::Field::Minute;
  629. if (char_is_one_of(field, 's'))
  630. return Unicode::CalendarRangePattern::Field::Second;
  631. if (char_is_one_of(field, 'S'))
  632. return Unicode::CalendarRangePattern::Field::FractionalSecondDigits;
  633. VERIFY_NOT_REACHED();
  634. };
  635. auto split_default_range_pattern = [&](auto skeleton, auto const& pattern) {
  636. auto start_range_end = pattern.find('}').value() + 1;
  637. auto end_range_begin = pattern.find_last('{').value();
  638. auto start_range = pattern.substring_view(0, start_range_end);
  639. auto separator = pattern.substring_view(start_range_end, end_range_begin - start_range_end);
  640. auto end_range = pattern.substring_view(end_range_begin);
  641. CalendarRangePattern format {};
  642. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  643. format.start_range = locale_data.unique_strings.ensure(start_range);
  644. format.separator = locale_data.unique_strings.ensure(separator);
  645. format.end_range = locale_data.unique_strings.ensure(end_range);
  646. return format;
  647. };
  648. auto split_range_pattern = [&](auto skeleton, auto field, auto const& pattern, auto const& parsed_fields) {
  649. HashMap<StringView, size_t> partitions;
  650. StringView last_partition;
  651. auto begin_index = pattern.find('{');
  652. size_t end_index = 0;
  653. while (begin_index.has_value()) {
  654. end_index = pattern.find('}', *begin_index).value();
  655. auto partition = pattern.substring_view(*begin_index, end_index - *begin_index);
  656. if (partitions.contains(partition))
  657. break;
  658. partitions.set(partition, *begin_index);
  659. last_partition = partition;
  660. begin_index = pattern.find('{', end_index + 1);
  661. }
  662. VERIFY(begin_index.has_value() && !last_partition.is_empty());
  663. auto start_range_end = partitions.get(last_partition).value() + last_partition.length() + 1;
  664. auto start_range = pattern.substring_view(0, start_range_end);
  665. auto separator = pattern.substring_view(start_range_end, *begin_index - start_range_end);
  666. auto end_range = pattern.substring_view(*begin_index);
  667. CalendarRangePattern format {};
  668. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  669. format.field = field;
  670. format.start_range = locale_data.unique_strings.ensure(start_range);
  671. format.separator = locale_data.unique_strings.ensure(separator);
  672. format.end_range = locale_data.unique_strings.ensure(end_range);
  673. format.for_each_calendar_field_zipped_with(parsed_fields, [](auto& format_field, auto const& parsed_field, auto) {
  674. format_field = parsed_field;
  675. });
  676. return format;
  677. };
  678. interval_formats_object.for_each_member([&](auto const& skeleton, auto const& value) {
  679. if (skeleton == "intervalFormatFallback"sv) {
  680. auto range_format = split_default_range_pattern(skeleton, value.as_string());
  681. calendar.default_range_format = locale_data.unique_range_patterns.ensure(move(range_format));
  682. return;
  683. }
  684. value.as_object().for_each_member([&](auto const& field, auto const& pattern) {
  685. if (field.ends_with("alt-variant"sv))
  686. return;
  687. VERIFY(field.length() == 1);
  688. auto name = name_of_field(field[0]);
  689. auto format = parse_date_time_pattern_raw(pattern.as_string(), skeleton, locale_data).release_value();
  690. auto range_format = split_range_pattern(skeleton, name, format.pattern, format);
  691. calendar.range_formats.append(locale_data.unique_range_patterns.ensure(move(range_format)));
  692. if (format.pattern12.has_value()) {
  693. auto range12_pattern = split_range_pattern(skeleton, name, *format.pattern12, format);
  694. calendar.range12_formats.append(locale_data.unique_range_patterns.ensure(move(range12_pattern)));
  695. } else {
  696. calendar.range12_formats.append(calendar.range_formats.last());
  697. }
  698. });
  699. });
  700. }
  701. static void generate_missing_patterns(Calendar& calendar, CalendarPatternList& formats, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, UnicodeLocaleData& locale_data)
  702. {
  703. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  704. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  705. auto pattern = locale_data.unique_strings.get(format);
  706. auto time_pattern = locale_data.unique_strings.get(time_format);
  707. auto date_pattern = locale_data.unique_strings.get(date_format);
  708. auto new_pattern = pattern.replace("{0}", time_pattern).replace("{1}", date_pattern);
  709. return locale_data.unique_strings.ensure(move(new_pattern));
  710. };
  711. auto inject_fractional_second_digits = [&](auto format) {
  712. auto pattern = locale_data.unique_strings.get(format);
  713. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv);
  714. return locale_data.unique_strings.ensure(move(new_pattern));
  715. };
  716. auto append_if_unique = [&](auto format) {
  717. auto format_index = locale_data.unique_patterns.ensure(move(format));
  718. if (!formats.contains_slow(format_index))
  719. formats.append(format_index);
  720. };
  721. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  722. for (auto const& format : date_formats)
  723. append_if_unique(format);
  724. for (auto const& format : time_formats) {
  725. append_if_unique(format);
  726. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  727. auto new_format = format;
  728. new_format.fractional_second_digits = 2;
  729. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  730. if (new_format.pattern12_index != 0)
  731. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  732. time_formats_with_fractional_second_digits.append(new_format);
  733. append_if_unique(move(new_format));
  734. }
  735. }
  736. time_formats.extend(move(time_formats_with_fractional_second_digits));
  737. for (auto const& date_format : date_formats) {
  738. auto const& date_time_formats = locale_data.unique_formats.get(calendar.date_time_formats);
  739. CalendarPatternIndexType date_time_format_index = 0;
  740. if (date_format.month == Unicode::CalendarPatternStyle::Long) {
  741. if (date_format.weekday.has_value())
  742. date_time_format_index = date_time_formats.full_format;
  743. else
  744. date_time_format_index = date_time_formats.long_format;
  745. } else if (date_format.month == Unicode::CalendarPatternStyle::Short) {
  746. date_time_format_index = date_time_formats.medium_format;
  747. } else {
  748. date_time_format_index = date_time_formats.short_format;
  749. }
  750. for (auto const& time_format : time_formats) {
  751. auto format = locale_data.unique_patterns.get(date_time_format_index);
  752. if (time_format.pattern12_index != 0)
  753. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  754. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  755. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  756. if (date_field.has_value())
  757. field = date_field;
  758. });
  759. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  760. if (time_field.has_value())
  761. field = time_field;
  762. });
  763. append_if_unique(move(format));
  764. }
  765. }
  766. }
  767. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, UnicodeLocaleData& locale_data)
  768. {
  769. auto create_symbol_lists = [](size_t size) {
  770. SymbolList narrow_symbol_list;
  771. SymbolList short_symbol_list;
  772. SymbolList long_symbol_list;
  773. narrow_symbol_list.resize(size);
  774. short_symbol_list.resize(size);
  775. long_symbol_list.resize(size);
  776. return Array<SymbolList, 3> { {
  777. move(narrow_symbol_list),
  778. move(short_symbol_list),
  779. move(long_symbol_list),
  780. } };
  781. };
  782. CalendarSymbolsList symbols_list {};
  783. auto store_symbol_lists = [&](auto symbol, auto symbol_lists) {
  784. auto symbol_index = to_underlying(symbol);
  785. if (symbol_index >= symbols_list.size())
  786. symbols_list.resize(symbol_index + 1);
  787. CalendarSymbols symbols {};
  788. symbols.narrow_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[0]));
  789. symbols.short_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[1]));
  790. symbols.long_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[2]));
  791. auto calendar_symbols_index = locale_data.unique_calendar_symbols.ensure(move(symbols));
  792. symbols_list[symbol_index] = calendar_symbols_index;
  793. };
  794. auto parse_era_symbols = [&](auto const& symbols_object) {
  795. auto const& narrow_symbols = symbols_object.get("eraNarrow"sv).as_object();
  796. auto const& short_symbols = symbols_object.get("eraAbbr"sv).as_object();
  797. auto const& long_symbols = symbols_object.get("eraNames"sv).as_object();
  798. auto symbol_lists = create_symbol_lists(2);
  799. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  800. if (auto key_index = key.to_uint(); key_index.has_value())
  801. symbols[*key_index] = locale_data.unique_strings.ensure(move(symbol));
  802. };
  803. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  804. append_symbol(symbol_lists[0], key, value.as_string());
  805. });
  806. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  807. append_symbol(symbol_lists[1], key, value.as_string());
  808. });
  809. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  810. append_symbol(symbol_lists[2], key, value.as_string());
  811. });
  812. store_symbol_lists(Unicode::CalendarSymbol::Era, move(symbol_lists));
  813. };
  814. auto parse_month_symbols = [&](auto const& symbols_object) {
  815. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  816. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  817. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  818. auto symbol_lists = create_symbol_lists(12);
  819. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  820. auto key_index = key.to_uint().value() - 1;
  821. symbols[key_index] = locale_data.unique_strings.ensure(move(symbol));
  822. };
  823. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  824. append_symbol(symbol_lists[0], key, value.as_string());
  825. });
  826. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  827. append_symbol(symbol_lists[1], key, value.as_string());
  828. });
  829. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  830. append_symbol(symbol_lists[2], key, value.as_string());
  831. });
  832. store_symbol_lists(Unicode::CalendarSymbol::Month, move(symbol_lists));
  833. };
  834. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  835. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  836. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  837. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  838. auto symbol_lists = create_symbol_lists(7);
  839. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  840. if (key == "sun"sv)
  841. symbols[to_underlying(Unicode::Weekday::Sunday)] = locale_data.unique_strings.ensure(move(symbol));
  842. else if (key == "mon"sv)
  843. symbols[to_underlying(Unicode::Weekday::Monday)] = locale_data.unique_strings.ensure(move(symbol));
  844. else if (key == "tue"sv)
  845. symbols[to_underlying(Unicode::Weekday::Tuesday)] = locale_data.unique_strings.ensure(move(symbol));
  846. else if (key == "wed"sv)
  847. symbols[to_underlying(Unicode::Weekday::Wednesday)] = locale_data.unique_strings.ensure(move(symbol));
  848. else if (key == "thu"sv)
  849. symbols[to_underlying(Unicode::Weekday::Thursday)] = locale_data.unique_strings.ensure(move(symbol));
  850. else if (key == "fri"sv)
  851. symbols[to_underlying(Unicode::Weekday::Friday)] = locale_data.unique_strings.ensure(move(symbol));
  852. else if (key == "sat"sv)
  853. symbols[to_underlying(Unicode::Weekday::Saturday)] = locale_data.unique_strings.ensure(move(symbol));
  854. };
  855. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  856. append_symbol(symbol_lists[0], key, value.as_string());
  857. });
  858. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  859. append_symbol(symbol_lists[1], key, value.as_string());
  860. });
  861. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  862. append_symbol(symbol_lists[2], key, value.as_string());
  863. });
  864. store_symbol_lists(Unicode::CalendarSymbol::Weekday, move(symbol_lists));
  865. };
  866. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  867. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  868. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  869. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  870. auto symbol_lists = create_symbol_lists(10);
  871. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  872. if (auto day_period = day_period_from_string(key); day_period.has_value())
  873. symbols[to_underlying(*day_period)] = locale_data.unique_strings.ensure(move(symbol));
  874. };
  875. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  876. append_symbol(symbol_lists[0], key, value.as_string());
  877. });
  878. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  879. append_symbol(symbol_lists[1], key, value.as_string());
  880. });
  881. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  882. append_symbol(symbol_lists[2], key, value.as_string());
  883. });
  884. store_symbol_lists(Unicode::CalendarSymbol::DayPeriod, move(symbol_lists));
  885. };
  886. parse_era_symbols(calendar_object.get("eras"sv).as_object());
  887. parse_month_symbols(calendar_object.get("months"sv).as_object().get("format"sv).as_object());
  888. parse_weekday_symbols(calendar_object.get("days"sv).as_object().get("format"sv).as_object());
  889. parse_day_period_symbols(calendar_object.get("dayPeriods"sv).as_object().get("format"sv).as_object());
  890. calendar.symbols = locale_data.unique_calendar_symbols_lists.ensure(move(symbols_list));
  891. }
  892. static ErrorOr<void> parse_calendars(String locale_calendars_path, UnicodeLocaleData& locale_data, Locale& locale)
  893. {
  894. LexicalPath calendars_path(move(locale_calendars_path));
  895. if (!calendars_path.basename().starts_with("ca-"sv))
  896. return {};
  897. auto calendars_file = TRY(Core::File::open(calendars_path.string(), Core::OpenMode::ReadOnly));
  898. auto calendars = TRY(JsonValue::from_string(calendars_file->read_all()));
  899. auto const& main_object = calendars.as_object().get("main"sv);
  900. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  901. auto const& dates_object = locale_object.as_object().get("dates"sv);
  902. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  903. auto ensure_calendar = [&](auto const& calendar) -> Calendar& {
  904. return locale.calendars.ensure(calendar, [&]() {
  905. auto calendar_index = locale_data.unique_strings.ensure(calendar);
  906. return Calendar { .calendar = calendar_index };
  907. });
  908. };
  909. auto parse_patterns = [&](auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) {
  910. auto parse_pattern = [&](auto name) {
  911. auto format = patterns_object.get(name);
  912. auto skeleton = skeletons_object.get(name);
  913. auto format_index = parse_date_time_pattern(format.as_string(), skeleton.as_string_or(String::empty()), locale_data).value();
  914. if (patterns)
  915. patterns->append(locale_data.unique_patterns.get(format_index));
  916. return format_index;
  917. };
  918. CalendarFormat formats {};
  919. formats.full_format = parse_pattern("full"sv);
  920. formats.long_format = parse_pattern("long"sv);
  921. formats.medium_format = parse_pattern("medium"sv);
  922. formats.short_format = parse_pattern("short"sv);
  923. return locale_data.unique_formats.ensure(move(formats));
  924. };
  925. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  926. // The generic calendar is not a supported Unicode calendar key, so skip it:
  927. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  928. if (calendar_name == "generic"sv)
  929. return;
  930. auto& calendar = ensure_calendar(calendar_name);
  931. CalendarPatternList available_formats {};
  932. if (!locale_data.calendars.contains_slow(calendar_name))
  933. locale_data.calendars.append(calendar_name);
  934. Vector<CalendarPattern> date_formats;
  935. Vector<CalendarPattern> time_formats;
  936. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  937. auto const& date_skeletons_object = value.as_object().get("dateSkeletons"sv);
  938. calendar.date_formats = parse_patterns(date_formats_object.as_object(), date_skeletons_object.as_object(), &date_formats);
  939. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  940. auto const& time_skeletons_object = value.as_object().get("timeSkeletons"sv);
  941. calendar.time_formats = parse_patterns(time_formats_object.as_object(), time_skeletons_object.as_object(), &time_formats);
  942. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  943. calendar.date_time_formats = parse_patterns(date_time_formats_object.as_object(), JsonObject {}, nullptr);
  944. auto const& available_formats_object = date_time_formats_object.as_object().get("availableFormats"sv);
  945. available_formats_object.as_object().for_each_member([&](auto const& skeleton, JsonValue const& pattern) {
  946. auto pattern_index = parse_date_time_pattern(pattern.as_string(), skeleton, locale_data);
  947. if (!pattern_index.has_value())
  948. return;
  949. auto const& format = locale_data.unique_patterns.get(*pattern_index);
  950. if (format.contains_only_date_fields())
  951. date_formats.append(format);
  952. else if (format.contains_only_time_fields())
  953. time_formats.append(format);
  954. if (!available_formats.contains_slow(*pattern_index))
  955. available_formats.append(*pattern_index);
  956. });
  957. auto const& interval_formats_object = date_time_formats_object.as_object().get("intervalFormats"sv);
  958. parse_interval_patterns(calendar, interval_formats_object.as_object(), locale_data);
  959. generate_missing_patterns(calendar, available_formats, move(date_formats), move(time_formats), locale_data);
  960. parse_calendar_symbols(calendar, value.as_object(), locale_data);
  961. calendar.available_formats = locale_data.unique_pattern_lists.ensure(move(available_formats));
  962. });
  963. return {};
  964. }
  965. static ErrorOr<void> parse_time_zone_names(String locale_time_zone_names_path, UnicodeLocaleData& locale_data, Locale& locale)
  966. {
  967. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  968. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  969. auto time_zone_names_file = TRY(Core::File::open(time_zone_names_path.string(), Core::OpenMode::ReadOnly));
  970. auto time_zone_names = TRY(JsonValue::from_string(time_zone_names_file->read_all()));
  971. auto const& main_object = time_zone_names.as_object().get("main"sv);
  972. auto const& locale_object = main_object.as_object().get(time_zone_names_path.parent().basename());
  973. auto const& dates_object = locale_object.as_object().get("dates"sv);
  974. auto const& time_zone_names_object = dates_object.as_object().get("timeZoneNames"sv);
  975. auto const& meta_zone_object = time_zone_names_object.as_object().get("metazone"sv);
  976. if (meta_zone_object.is_null())
  977. return {};
  978. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object) -> Optional<StringIndexType> {
  979. auto const& names = meta_zone_object.get(type);
  980. if (!names.is_object())
  981. return {};
  982. auto const& daylight = names.as_object().get("daylight"sv);
  983. if (daylight.is_string())
  984. return locale_data.unique_strings.ensure(daylight.as_string());
  985. auto const& standard = names.as_object().get("standard"sv);
  986. if (standard.is_string())
  987. return locale_data.unique_strings.ensure(standard.as_string());
  988. return {};
  989. };
  990. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  991. auto golden_zone = locale_data.meta_zones.get(meta_zone).value();
  992. TimeZone time_zone { .time_zone = golden_zone };
  993. if (auto long_name = parse_name("long"sv, meta_zone_object); long_name.has_value())
  994. time_zone.long_name = long_name.value();
  995. if (auto short_name = parse_name("short"sv, meta_zone_object); short_name.has_value())
  996. time_zone.short_name = short_name.value();
  997. auto const& time_zone_name = locale_data.unique_strings.get(golden_zone);
  998. if (!locale_data.time_zones.contains_slow(time_zone_name))
  999. locale_data.time_zones.append(time_zone_name);
  1000. locale.time_zones.set(time_zone_name, move(time_zone));
  1001. };
  1002. meta_zone_object.as_object().for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  1003. parse_time_zone(meta_zone, value.as_object());
  1004. });
  1005. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  1006. auto const& zone_object = time_zone_names_object.as_object().get("zone"sv);
  1007. auto const& etc_object = zone_object.as_object().get("Etc"sv);
  1008. auto const& utc_object = etc_object.as_object().get("UTC"sv);
  1009. parse_time_zone("UTC"sv, utc_object.as_object());
  1010. return {};
  1011. }
  1012. static ErrorOr<void> parse_day_periods(String core_path, UnicodeLocaleData& locale_data)
  1013. {
  1014. // https://unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rule_Sets
  1015. LexicalPath day_periods_path(move(core_path));
  1016. day_periods_path = day_periods_path.append("supplemental"sv);
  1017. day_periods_path = day_periods_path.append("dayPeriods.json"sv);
  1018. auto day_periods_file = TRY(Core::File::open(day_periods_path.string(), Core::OpenMode::ReadOnly));
  1019. auto day_periods = TRY(JsonValue::from_string(day_periods_file->read_all()));
  1020. auto const& supplemental_object = day_periods.as_object().get("supplemental"sv);
  1021. auto const& day_periods_object = supplemental_object.as_object().get("dayPeriodRuleSet"sv);
  1022. auto parse_hour = [](auto const& time) {
  1023. auto hour_end_index = time.find(':').value();
  1024. // The times are in the form "hours:minutes", but we only need the hour segment.
  1025. // TR-35 explicitly states that minutes other than :00 must not be used.
  1026. VERIFY(time.substring_view(hour_end_index) == ":00"sv);
  1027. auto hour = time.substring_view(0, hour_end_index);
  1028. return hour.template to_uint<u8>().value();
  1029. };
  1030. auto parse_day_period = [&](auto const& symbol, auto const& ranges) -> Optional<DayPeriod> {
  1031. auto day_period = day_period_from_string(symbol);
  1032. if (!day_period.has_value())
  1033. return {};
  1034. auto begin = parse_hour(ranges.get("_from"sv).as_string());
  1035. auto end = parse_hour(ranges.get("_before"sv).as_string());
  1036. return DayPeriod { *day_period, begin, end };
  1037. };
  1038. day_periods_object.as_object().for_each_member([&](auto const& language, JsonValue const& value) {
  1039. auto locale = locale_data.locales.find(language);
  1040. if (locale == locale_data.locales.end())
  1041. return;
  1042. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& ranges) {
  1043. if (auto day_period = parse_day_period(symbol, ranges.as_object()); day_period.has_value())
  1044. locale->value.day_periods.append(day_period.release_value());
  1045. });
  1046. });
  1047. return {};
  1048. };
  1049. static ErrorOr<void> parse_all_locales(String core_path, String dates_path, UnicodeLocaleData& locale_data)
  1050. {
  1051. TRY(parse_hour_cycles(core_path, locale_data));
  1052. TRY(parse_meta_zones(core_path, locale_data));
  1053. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  1054. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  1055. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  1056. StringBuilder builder;
  1057. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  1058. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  1059. builder.appendff("-{}", script);
  1060. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  1061. builder.appendff("-{}", region);
  1062. return builder.build();
  1063. };
  1064. while (dates_iterator.has_next()) {
  1065. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  1066. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  1067. auto language = TRY(remove_variants_from_path(dates_path));
  1068. auto& locale = locale_data.locales.ensure(language);
  1069. while (calendars_iterator.has_next()) {
  1070. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  1071. TRY(parse_calendars(move(calendars_path), locale_data, locale));
  1072. }
  1073. TRY(parse_time_zone_names(move(dates_path), locale_data, locale));
  1074. }
  1075. TRY(parse_day_periods(move(core_path), locale_data));
  1076. return {};
  1077. }
  1078. static String format_identifier(StringView owner, String identifier)
  1079. {
  1080. identifier = identifier.replace("-"sv, "_"sv, true);
  1081. identifier = identifier.replace("/"sv, "_"sv, true);
  1082. if (all_of(identifier, is_ascii_digit))
  1083. return String::formatted("{}_{}", owner[0], identifier);
  1084. if (is_ascii_lower_alpha(identifier[0]))
  1085. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  1086. return identifier;
  1087. }
  1088. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  1089. {
  1090. StringBuilder builder;
  1091. SourceGenerator generator { builder };
  1092. generator.append(R"~~~(
  1093. #pragma once
  1094. #include <AK/Optional.h>
  1095. #include <AK/StringView.h>
  1096. #include <LibUnicode/Forward.h>
  1097. namespace Unicode {
  1098. )~~~");
  1099. generate_enum(generator, format_identifier, "Calendar"sv, {}, locale_data.calendars, locale_data.calendar_aliases);
  1100. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, locale_data.hour_cycle_regions);
  1101. generate_enum(generator, format_identifier, "TimeZone"sv, {}, locale_data.time_zones);
  1102. generator.append(R"~~~(
  1103. namespace Detail {
  1104. Optional<Calendar> calendar_from_string(StringView calendar);
  1105. Optional<HourCycleRegion> hour_cycle_region_from_string(StringView hour_cycle_region);
  1106. Vector<Unicode::HourCycle> get_regional_hour_cycles(StringView region);
  1107. Optional<Unicode::CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar);
  1108. Optional<Unicode::CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar);
  1109. Optional<Unicode::CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar);
  1110. Vector<Unicode::CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar);
  1111. Optional<Unicode::CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar);
  1112. Vector<Unicode::CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton);
  1113. Vector<Unicode::CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton);
  1114. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Era value);
  1115. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Month value);
  1116. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Weekday value);
  1117. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::DayPeriod value);
  1118. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour);
  1119. Optional<TimeZone> time_zone_from_string(StringView time_zone);
  1120. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style);
  1121. }
  1122. }
  1123. )~~~");
  1124. VERIFY(file.write(generator.as_string_view()));
  1125. }
  1126. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  1127. {
  1128. StringBuilder builder;
  1129. SourceGenerator generator { builder };
  1130. generator.set("string_index_type"sv, s_string_index_type);
  1131. generator.set("calendar_pattern_index_type"sv, s_calendar_pattern_index_type);
  1132. generator.set("calendar_pattern_list_index_type"sv, s_calendar_pattern_list_index_type);
  1133. generator.set("calendar_range_pattern_index_type"sv, s_calendar_range_pattern_index_type);
  1134. generator.set("calendar_format_index_type"sv, s_calendar_format_index_type);
  1135. generator.set("symbol_list_index_type"sv, s_symbol_list_index_type);
  1136. generator.set("calendar_symbols_index_type"sv, s_calendar_symbols_index_type);
  1137. generator.set("calendar_symbols_list_index_type"sv, s_calendar_symbols_list_index_type);
  1138. generator.append(R"~~~(
  1139. #include <AK/Array.h>
  1140. #include <AK/BinarySearch.h>
  1141. #include <LibUnicode/DateTimeFormat.h>
  1142. #include <LibUnicode/Locale.h>
  1143. #include <LibUnicode/UnicodeDateTimeFormat.h>
  1144. namespace Unicode::Detail {
  1145. )~~~");
  1146. locale_data.unique_strings.generate(generator);
  1147. generator.append(R"~~~(
  1148. template <typename SourceType, typename TargetType>
  1149. void convert_calendar_fields(SourceType const& source, TargetType& target)
  1150. {
  1151. if (source.era != -1)
  1152. target.era = static_cast<Unicode::CalendarPatternStyle>(source.era);
  1153. if (source.year != -1)
  1154. target.year = static_cast<Unicode::CalendarPatternStyle>(source.year);
  1155. if (source.month != -1)
  1156. target.month = static_cast<Unicode::CalendarPatternStyle>(source.month);
  1157. if (source.weekday != -1)
  1158. target.weekday = static_cast<Unicode::CalendarPatternStyle>(source.weekday);
  1159. if (source.day != -1)
  1160. target.day = static_cast<Unicode::CalendarPatternStyle>(source.day);
  1161. if (source.day_period != -1)
  1162. target.day_period = static_cast<Unicode::CalendarPatternStyle>(source.day_period);
  1163. if (source.hour != -1)
  1164. target.hour = static_cast<Unicode::CalendarPatternStyle>(source.hour);
  1165. if (source.minute != -1)
  1166. target.minute = static_cast<Unicode::CalendarPatternStyle>(source.minute);
  1167. if (source.second != -1)
  1168. target.second = static_cast<Unicode::CalendarPatternStyle>(source.second);
  1169. if (source.fractional_second_digits != -1)
  1170. target.fractional_second_digits = static_cast<u8>(source.fractional_second_digits);
  1171. if (source.time_zone_name != -1)
  1172. target.time_zone_name = static_cast<Unicode::CalendarPatternStyle>(source.time_zone_name);
  1173. }
  1174. struct CalendarPattern {
  1175. Unicode::CalendarPattern to_unicode_calendar_pattern() const {
  1176. Unicode::CalendarPattern calendar_pattern {};
  1177. calendar_pattern.skeleton = s_string_list[skeleton];
  1178. calendar_pattern.pattern = s_string_list[pattern];
  1179. if (pattern12 != 0)
  1180. calendar_pattern.pattern12 = s_string_list[pattern12];
  1181. convert_calendar_fields(*this, calendar_pattern);
  1182. return calendar_pattern;
  1183. }
  1184. @string_index_type@ skeleton { 0 };
  1185. @string_index_type@ pattern { 0 };
  1186. @string_index_type@ pattern12 { 0 };
  1187. i8 era { -1 };
  1188. i8 year { -1 };
  1189. i8 month { -1 };
  1190. i8 weekday { -1 };
  1191. i8 day { -1 };
  1192. i8 day_period { -1 };
  1193. i8 hour { -1 };
  1194. i8 minute { -1 };
  1195. i8 second { -1 };
  1196. i8 fractional_second_digits { -1 };
  1197. i8 time_zone_name { -1 };
  1198. };
  1199. struct CalendarRangePattern {
  1200. Unicode::CalendarRangePattern to_unicode_calendar_range_pattern() const {
  1201. Unicode::CalendarRangePattern calendar_range_pattern {};
  1202. if (field != -1)
  1203. calendar_range_pattern.field = static_cast<Unicode::CalendarRangePattern::Field>(field);
  1204. calendar_range_pattern.start_range = s_string_list[start_range];
  1205. calendar_range_pattern.separator = s_string_list[separator];
  1206. calendar_range_pattern.end_range = s_string_list[end_range];
  1207. convert_calendar_fields(*this, calendar_range_pattern);
  1208. return calendar_range_pattern;
  1209. }
  1210. @string_index_type@ skeleton { 0 };
  1211. i8 field { -1 };
  1212. @string_index_type@ start_range { 0 };
  1213. @string_index_type@ separator { 0 };
  1214. @string_index_type@ end_range { 0 };
  1215. i8 era { -1 };
  1216. i8 year { -1 };
  1217. i8 month { -1 };
  1218. i8 weekday { -1 };
  1219. i8 day { -1 };
  1220. i8 day_period { -1 };
  1221. i8 hour { -1 };
  1222. i8 minute { -1 };
  1223. i8 second { -1 };
  1224. i8 fractional_second_digits { -1 };
  1225. i8 time_zone_name { -1 };
  1226. };
  1227. )~~~");
  1228. locale_data.unique_patterns.generate(generator, "CalendarPattern"sv, "s_calendar_patterns"sv, 10);
  1229. locale_data.unique_pattern_lists.generate(generator, s_calendar_pattern_index_type, "s_calendar_pattern_lists"sv);
  1230. locale_data.unique_range_patterns.generate(generator, "CalendarRangePattern"sv, "s_calendar_range_patterns"sv, 10);
  1231. generator.append(R"~~~(
  1232. struct CalendarFormat {
  1233. Unicode::CalendarFormat to_unicode_calendar_format() const {
  1234. Unicode::CalendarFormat calendar_format {};
  1235. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  1236. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  1237. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  1238. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  1239. return calendar_format;
  1240. }
  1241. @calendar_pattern_index_type@ full_format { 0 };
  1242. @calendar_pattern_index_type@ long_format { 0 };
  1243. @calendar_pattern_index_type@ medium_format { 0 };
  1244. @calendar_pattern_index_type@ short_format { 0 };
  1245. };
  1246. struct CalendarSymbols {
  1247. @symbol_list_index_type@ narrow_symbols { 0 };
  1248. @symbol_list_index_type@ short_symbols { 0 };
  1249. @symbol_list_index_type@ long_symbols { 0 };
  1250. };
  1251. struct CalendarData {
  1252. @string_index_type@ calendar { 0 };
  1253. @calendar_format_index_type@ date_formats { 0 };
  1254. @calendar_format_index_type@ time_formats { 0 };
  1255. @calendar_format_index_type@ date_time_formats { 0 };
  1256. @calendar_pattern_list_index_type@ available_formats { 0 };
  1257. @calendar_range_pattern_index_type@ default_range_format { 0 };
  1258. Span<@calendar_range_pattern_index_type@ const> range_formats {};
  1259. Span<@calendar_range_pattern_index_type@ const> range12_formats {};
  1260. @calendar_symbols_list_index_type@ symbols { 0 };
  1261. };
  1262. struct TimeZoneData {
  1263. @string_index_type@ time_zone { 0 };
  1264. @string_index_type@ long_name { 0 };
  1265. @string_index_type@ short_name { 0 };
  1266. };
  1267. struct DayPeriodData {
  1268. u8 day_period { 0 };
  1269. u8 begin { 0 };
  1270. u8 end { 0 };
  1271. };
  1272. )~~~");
  1273. locale_data.unique_formats.generate(generator, "CalendarFormat"sv, "s_calendar_formats"sv, 10);
  1274. locale_data.unique_symbol_lists.generate(generator, s_string_index_type, "s_symbol_lists"sv);
  1275. locale_data.unique_calendar_symbols.generate(generator, "CalendarSymbols"sv, "s_calendar_symbols"sv, 10);
  1276. locale_data.unique_calendar_symbols_lists.generate(generator, s_calendar_symbols_index_type, "s_calendar_symbol_lists"sv);
  1277. auto append_pattern_list = [&](auto name, auto type, auto const& formats) {
  1278. generator.set("name", move(name));
  1279. generator.set("type", move(type));
  1280. generator.set("size", String::number(formats.size()));
  1281. generator.append(R"~~~(
  1282. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1283. bool first = true;
  1284. for (auto format : formats) {
  1285. generator.append(first ? " " : ", ");
  1286. generator.append(String::number(format));
  1287. first = false;
  1288. }
  1289. generator.append(" } };");
  1290. };
  1291. auto append_calendars = [&](String name, auto const& calendars) {
  1292. auto format_name = [&](StringView calendar_key, StringView type) {
  1293. return String::formatted("{}_{}_{}", name, calendar_key, type);
  1294. };
  1295. for (auto const& calendar_key : locale_data.calendars) {
  1296. auto const& calendar = calendars.find(calendar_key)->value;
  1297. append_pattern_list(format_name(calendar_key, "range_formats"sv), s_calendar_range_pattern_index_type, calendar.range_formats);
  1298. append_pattern_list(format_name(calendar_key, "range12_formats"sv), s_calendar_range_pattern_index_type, calendar.range12_formats);
  1299. generator.append("\n");
  1300. }
  1301. generator.set("name", name);
  1302. generator.set("size", String::number(calendars.size()));
  1303. generator.append(R"~~~(
  1304. static constexpr Array<CalendarData, @size@> @name@ { {)~~~");
  1305. for (auto const& calendar_key : locale_data.calendars) {
  1306. auto const& calendar = calendars.find(calendar_key)->value;
  1307. generator.set("calendar"sv, String::number(calendar.calendar));
  1308. generator.set("date_formats"sv, String::number(calendar.date_formats));
  1309. generator.set("time_formats"sv, String::number(calendar.time_formats));
  1310. generator.set("date_time_formats"sv, String::number(calendar.date_time_formats));
  1311. generator.set("formats", String::number(calendar.available_formats));
  1312. generator.set("default_range_format", String::number(calendar.default_range_format));
  1313. generator.set("range_formats", format_name(calendar_key, "range_formats"sv));
  1314. generator.set("range12_formats", format_name(calendar_key, "range12_formats"sv));
  1315. generator.set("symbols"sv, String::number(calendar.symbols));
  1316. generator.append(R"~~~(
  1317. { @calendar@, )~~~");
  1318. generator.append("@date_formats@, @time_formats@, @date_time_formats@, @formats@, ");
  1319. generator.append("@default_range_format@, @range_formats@.span(), @range12_formats@.span(), ");
  1320. generator.append("@symbols@ }, ");
  1321. }
  1322. generator.append(R"~~~(
  1323. } };
  1324. )~~~");
  1325. };
  1326. auto append_time_zones = [&](String name, auto const& time_zones) {
  1327. generator.set("name", name);
  1328. generator.set("size", String::number(locale_data.time_zones.size()));
  1329. generator.append(R"~~~(
  1330. static constexpr Array<TimeZoneData, @size@> @name@ { {)~~~");
  1331. constexpr size_t max_values_per_row = 20;
  1332. size_t values_in_current_row = 0;
  1333. for (auto const& time_zone_key : locale_data.time_zones) {
  1334. auto time_zone = time_zones.find(time_zone_key);
  1335. if (values_in_current_row++ > 0)
  1336. generator.append(" ");
  1337. if (time_zone == time_zones.end()) {
  1338. generator.append("{},");
  1339. } else {
  1340. generator.set("time_zone", String::number(time_zone->value.time_zone));
  1341. generator.set("long_name", String::number(time_zone->value.long_name));
  1342. generator.set("short_name", String::number(time_zone->value.short_name));
  1343. generator.append("{ @time_zone@, @long_name@, @short_name@ },");
  1344. }
  1345. if (values_in_current_row == max_values_per_row) {
  1346. values_in_current_row = 0;
  1347. generator.append("\n ");
  1348. }
  1349. }
  1350. generator.append(R"~~~(
  1351. } };
  1352. )~~~");
  1353. };
  1354. auto append_day_periods = [&](String name, auto const& day_periods) {
  1355. generator.set("name", move(name));
  1356. generator.set("size", String::number(day_periods.size()));
  1357. generator.append(R"~~~(
  1358. static constexpr Array<DayPeriodData, @size@> @name@ { {)~~~");
  1359. for (auto const& day_period : day_periods) {
  1360. generator.set("day_period", String::number(static_cast<u8>(day_period.day_period)));
  1361. generator.set("begin", String::number(day_period.begin));
  1362. generator.set("end", String::number(day_period.end));
  1363. generator.append(" { @day_period@, @begin@, @end@ },");
  1364. }
  1365. generator.append(" } };");
  1366. };
  1367. auto append_hour_cycles = [&](String name, auto const& hour_cycle_region) {
  1368. auto const& hour_cycles = locale_data.hour_cycles.find(hour_cycle_region)->value;
  1369. generator.set("name", name);
  1370. generator.set("size", String::number(hour_cycles.size()));
  1371. generator.append(R"~~~(
  1372. static constexpr Array<u8, @size@> @name@ { { )~~~");
  1373. for (auto hour_cycle : hour_cycles) {
  1374. generator.set("hour_cycle", String::number(static_cast<u8>(hour_cycle)));
  1375. generator.append("@hour_cycle@, ");
  1376. }
  1377. generator.append(" } };");
  1378. };
  1379. generate_mapping(generator, locale_data.locales, "CalendarData"sv, "s_calendars"sv, "s_calendars_{}", [&](auto const& name, auto const& value) { append_calendars(name, value.calendars); });
  1380. generate_mapping(generator, locale_data.locales, "TimeZoneData"sv, "s_time_zones"sv, "s_time_zones_{}", [&](auto const& name, auto const& value) { append_time_zones(name, value.time_zones); });
  1381. generate_mapping(generator, locale_data.locales, "DayPeriodData"sv, "s_day_periods"sv, "s_day_periods_{}", [&](auto const& name, auto const& value) { append_day_periods(name, value.day_periods); });
  1382. generate_mapping(generator, locale_data.hour_cycle_regions, "u8"sv, "s_hour_cycles"sv, "s_hour_cycles_{}", [&](auto const& name, auto const& value) { append_hour_cycles(name, value); });
  1383. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1384. HashValueMap<String> hashes;
  1385. hashes.ensure_capacity(values.size());
  1386. for (auto const& value : values)
  1387. hashes.set(value.hash(), format_identifier(enum_title, value));
  1388. for (auto const& alias : aliases)
  1389. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1390. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1391. };
  1392. append_from_string("Calendar"sv, "calendar"sv, locale_data.calendars, locale_data.calendar_aliases);
  1393. append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, locale_data.hour_cycle_regions);
  1394. append_from_string("TimeZone"sv, "time_zone"sv, locale_data.time_zones);
  1395. generator.append(R"~~~(
  1396. Vector<Unicode::HourCycle> get_regional_hour_cycles(StringView region)
  1397. {
  1398. auto region_value = hour_cycle_region_from_string(region);
  1399. if (!region_value.has_value())
  1400. return {};
  1401. auto region_index = to_underlying(*region_value);
  1402. auto const& regional_hour_cycles = s_hour_cycles.at(region_index);
  1403. Vector<Unicode::HourCycle> hour_cycles;
  1404. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1405. for (auto hour_cycle : regional_hour_cycles)
  1406. hour_cycles.unchecked_append(static_cast<Unicode::HourCycle>(hour_cycle));
  1407. return hour_cycles;
  1408. }
  1409. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1410. {
  1411. auto locale_value = locale_from_string(locale);
  1412. if (!locale_value.has_value())
  1413. return nullptr;
  1414. auto calendar_value = calendar_from_string(calendar);
  1415. if (!calendar_value.has_value())
  1416. return nullptr;
  1417. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1418. auto calendar_index = to_underlying(*calendar_value);
  1419. auto const& calendars = s_calendars.at(locale_index);
  1420. return &calendars[calendar_index];
  1421. }
  1422. Optional<Unicode::CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1423. {
  1424. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1425. auto const& formats = s_calendar_formats.at(data->date_formats);
  1426. return formats.to_unicode_calendar_format();
  1427. }
  1428. return {};
  1429. }
  1430. Optional<Unicode::CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1431. {
  1432. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1433. auto const& formats = s_calendar_formats.at(data->time_formats);
  1434. return formats.to_unicode_calendar_format();
  1435. }
  1436. return {};
  1437. }
  1438. Optional<Unicode::CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1439. {
  1440. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1441. auto const& formats = s_calendar_formats.at(data->date_time_formats);
  1442. return formats.to_unicode_calendar_format();
  1443. }
  1444. return {};
  1445. }
  1446. Vector<Unicode::CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1447. {
  1448. Vector<Unicode::CalendarPattern> result {};
  1449. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1450. auto const& available_formats = s_calendar_pattern_lists.at(data->available_formats);
  1451. result.ensure_capacity(available_formats.size());
  1452. for (auto const& format : available_formats)
  1453. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1454. }
  1455. return result;
  1456. }
  1457. Optional<Unicode::CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar)
  1458. {
  1459. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1460. auto const& pattern = s_calendar_range_patterns[data->default_range_format];
  1461. return pattern.to_unicode_calendar_range_pattern();
  1462. }
  1463. return {};
  1464. }
  1465. Vector<Unicode::CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton) {
  1466. Vector<Unicode::CalendarRangePattern> result {};
  1467. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1468. for (auto format : data->range_formats) {
  1469. auto const& pattern = s_calendar_range_patterns[format];
  1470. if (skeleton == s_string_list[pattern.skeleton])
  1471. result.append(pattern.to_unicode_calendar_range_pattern());
  1472. }
  1473. }
  1474. return result;
  1475. }
  1476. Vector<Unicode::CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton) {
  1477. Vector<Unicode::CalendarRangePattern> result {};
  1478. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1479. for (auto format : data->range12_formats) {
  1480. auto const& pattern = s_calendar_range_patterns[format];
  1481. if (skeleton == s_string_list[pattern.skeleton])
  1482. result.append(pattern.to_unicode_calendar_range_pattern());
  1483. }
  1484. }
  1485. return result;
  1486. }
  1487. static Span<@string_index_type@ const> find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1488. {
  1489. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1490. auto const& symbols_list = s_calendar_symbol_lists[data->symbols];
  1491. auto symbol_index = to_underlying(symbol);
  1492. auto calendar_symbols_index = symbols_list.at(symbol_index);
  1493. auto const& symbols = s_calendar_symbols.at(calendar_symbols_index);
  1494. @symbol_list_index_type@ symbol_list_index = 0;
  1495. switch (style) {
  1496. case CalendarPatternStyle::Narrow:
  1497. symbol_list_index = symbols.narrow_symbols;
  1498. break;
  1499. case CalendarPatternStyle::Short:
  1500. symbol_list_index = symbols.short_symbols;
  1501. break;
  1502. case CalendarPatternStyle::Long:
  1503. symbol_list_index = symbols.long_symbols;
  1504. break;
  1505. default:
  1506. VERIFY_NOT_REACHED();
  1507. }
  1508. return s_symbol_lists.at(symbol_list_index);
  1509. }
  1510. return {};
  1511. }
  1512. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Era value)
  1513. {
  1514. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1515. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1516. return s_string_list[symbols.at(value_index)];
  1517. return {};
  1518. }
  1519. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Month value)
  1520. {
  1521. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1522. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1523. return s_string_list[symbols.at(value_index)];
  1524. return {};
  1525. }
  1526. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Weekday value)
  1527. {
  1528. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1529. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1530. return s_string_list[symbols.at(value_index)];
  1531. return {};
  1532. }
  1533. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::DayPeriod value)
  1534. {
  1535. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1536. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1537. return s_string_list[symbols.at(value_index)];
  1538. return {};
  1539. }
  1540. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour)
  1541. {
  1542. auto locale_value = locale_from_string(locale);
  1543. if (!locale_value.has_value())
  1544. return {};
  1545. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1546. auto day_periods = s_day_periods[locale_index];
  1547. for (auto const& day_period : day_periods) {
  1548. if ((day_period.begin <= hour) && (hour < day_period.end)) {
  1549. auto period = static_cast<Unicode::DayPeriod>(day_period.day_period);
  1550. return Detail::get_calendar_day_period_symbol(locale, calendar, style, period);
  1551. }
  1552. }
  1553. // Fallback to fixed periods if the locale does not have flexible day periods.
  1554. // TR-35 states that the meaning of AM and PM does not change with locale.
  1555. if (hour < 12)
  1556. return Detail::get_calendar_day_period_symbol(locale, calendar, style, Unicode::DayPeriod::AM);
  1557. return Detail::get_calendar_day_period_symbol(locale, calendar, style, Unicode::DayPeriod::PM);
  1558. }
  1559. static TimeZoneData const* find_time_zone_data(StringView locale, StringView time_zone)
  1560. {
  1561. auto locale_value = locale_from_string(locale);
  1562. if (!locale_value.has_value())
  1563. return nullptr;
  1564. auto time_zone_value = time_zone_from_string(time_zone);
  1565. if (!time_zone_value.has_value())
  1566. return nullptr;
  1567. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1568. auto time_zone_index = to_underlying(*time_zone_value);
  1569. auto const& time_zones = s_time_zones.at(locale_index);
  1570. return &time_zones[time_zone_index];
  1571. }
  1572. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style)
  1573. {
  1574. if (auto const* data = find_time_zone_data(locale, time_zone); data != nullptr) {
  1575. @string_index_type@ time_zone_index = 0;
  1576. switch (style) {
  1577. case CalendarPatternStyle::Long:
  1578. time_zone_index = data->long_name;
  1579. break;
  1580. case CalendarPatternStyle::Short:
  1581. time_zone_index = data->short_name;
  1582. break;
  1583. default:
  1584. VERIFY_NOT_REACHED();
  1585. }
  1586. if (time_zone_index != 0)
  1587. return s_string_list[time_zone_index];
  1588. }
  1589. return {};
  1590. }
  1591. }
  1592. )~~~");
  1593. VERIFY(file.write(generator.as_string_view()));
  1594. }
  1595. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1596. {
  1597. StringView generated_header_path;
  1598. StringView generated_implementation_path;
  1599. StringView core_path;
  1600. StringView dates_path;
  1601. Core::ArgsParser args_parser;
  1602. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1603. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1604. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1605. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1606. args_parser.parse(arguments);
  1607. auto open_file = [&](StringView path) -> ErrorOr<NonnullRefPtr<Core::File>> {
  1608. if (path.is_empty()) {
  1609. args_parser.print_usage(stderr, arguments.argv[0]);
  1610. return Error::from_string_literal("Must provide all command line options"sv);
  1611. }
  1612. return Core::File::open(path, Core::OpenMode::ReadWrite);
  1613. };
  1614. auto generated_header_file = TRY(open_file(generated_header_path));
  1615. auto generated_implementation_file = TRY(open_file(generated_implementation_path));
  1616. UnicodeLocaleData locale_data;
  1617. TRY(parse_all_locales(core_path, dates_path, locale_data));
  1618. generate_unicode_locale_header(generated_header_file, locale_data);
  1619. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  1620. return 0;
  1621. }