GenerateUnicodeDateTimeFormat.cpp 78 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975
  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 CalendarRangePatternIndexType = u16;
  31. constexpr auto s_calendar_range_pattern_index_type = "u16"sv;
  32. using CalendarFormatIndexType = u8;
  33. constexpr auto s_calendar_format_index_type = "u8"sv;
  34. using SymbolListIndexType = u16;
  35. constexpr auto s_symbol_list_index_type = "u16"sv;
  36. struct CalendarPattern : public Unicode::CalendarPattern {
  37. bool contains_only_date_fields() const
  38. {
  39. return !day_period.has_value() && !hour.has_value() && !minute.has_value() && !second.has_value() && !fractional_second_digits.has_value() && !time_zone_name.has_value();
  40. }
  41. bool contains_only_time_fields() const
  42. {
  43. return !weekday.has_value() && !era.has_value() && !year.has_value() && !month.has_value() && !day.has_value();
  44. }
  45. unsigned hash() const
  46. {
  47. auto hash = pair_int_hash(pattern_index, pattern12_index);
  48. hash = pair_int_hash(hash, skeleton_index);
  49. auto hash_field = [&](auto const& field) {
  50. if (field.has_value())
  51. hash = pair_int_hash(hash, static_cast<u8>(*field));
  52. else
  53. hash = pair_int_hash(hash, -1);
  54. };
  55. hash_field(era);
  56. hash_field(year);
  57. hash_field(month);
  58. hash_field(weekday);
  59. hash_field(day);
  60. hash_field(day_period);
  61. hash_field(hour);
  62. hash_field(minute);
  63. hash_field(second);
  64. hash_field(fractional_second_digits);
  65. hash_field(time_zone_name);
  66. return hash;
  67. }
  68. bool operator==(CalendarPattern const& other) const
  69. {
  70. return (skeleton_index == other.skeleton_index)
  71. && (pattern_index == other.pattern_index)
  72. && (pattern12_index == other.pattern12_index)
  73. && (era == other.era)
  74. && (year == other.year)
  75. && (month == other.month)
  76. && (weekday == other.weekday)
  77. && (day == other.day)
  78. && (day_period == other.day_period)
  79. && (hour == other.hour)
  80. && (minute == other.minute)
  81. && (second == other.second)
  82. && (fractional_second_digits == other.fractional_second_digits)
  83. && (time_zone_name == other.time_zone_name);
  84. }
  85. StringIndexType skeleton_index { 0 };
  86. StringIndexType pattern_index { 0 };
  87. StringIndexType pattern12_index { 0 };
  88. };
  89. template<>
  90. struct AK::Formatter<CalendarPattern> : Formatter<FormatString> {
  91. ErrorOr<void> format(FormatBuilder& builder, CalendarPattern const& pattern)
  92. {
  93. auto field_to_i8 = [](auto const& field) -> i8 {
  94. if (!field.has_value())
  95. return -1;
  96. return static_cast<i8>(*field);
  97. };
  98. return Formatter<FormatString>::format(builder,
  99. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}",
  100. pattern.skeleton_index,
  101. pattern.pattern_index,
  102. pattern.pattern12_index,
  103. field_to_i8(pattern.era),
  104. field_to_i8(pattern.year),
  105. field_to_i8(pattern.month),
  106. field_to_i8(pattern.weekday),
  107. field_to_i8(pattern.day),
  108. field_to_i8(pattern.day_period),
  109. field_to_i8(pattern.hour),
  110. field_to_i8(pattern.minute),
  111. field_to_i8(pattern.second),
  112. field_to_i8(pattern.fractional_second_digits),
  113. field_to_i8(pattern.time_zone_name));
  114. }
  115. };
  116. template<>
  117. struct AK::Traits<CalendarPattern> : public GenericTraits<CalendarPattern> {
  118. static unsigned hash(CalendarPattern const& c) { return c.hash(); }
  119. };
  120. struct CalendarRangePattern : public CalendarPattern {
  121. unsigned hash() const
  122. {
  123. auto hash = CalendarPattern::hash();
  124. if (field.has_value())
  125. hash = pair_int_hash(hash, static_cast<u8>(*field));
  126. hash = pair_int_hash(hash, start_range);
  127. hash = pair_int_hash(hash, separator);
  128. hash = pair_int_hash(hash, end_range);
  129. return hash;
  130. }
  131. bool operator==(CalendarRangePattern const& other) const
  132. {
  133. if (!CalendarPattern::operator==(other))
  134. return false;
  135. return (field == other.field)
  136. && (start_range == other.start_range)
  137. && (separator == other.separator)
  138. && (end_range == other.end_range);
  139. }
  140. Optional<Unicode::CalendarRangePattern::Field> field {};
  141. StringIndexType start_range { 0 };
  142. StringIndexType separator { 0 };
  143. StringIndexType end_range { 0 };
  144. };
  145. template<>
  146. struct AK::Formatter<CalendarRangePattern> : Formatter<FormatString> {
  147. ErrorOr<void> format(FormatBuilder& builder, CalendarRangePattern const& pattern)
  148. {
  149. auto field_to_i8 = [](auto const& field) -> i8 {
  150. if (!field.has_value())
  151. return -1;
  152. return static_cast<i8>(*field);
  153. };
  154. return Formatter<FormatString>::format(builder,
  155. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}",
  156. pattern.skeleton_index,
  157. field_to_i8(pattern.field),
  158. pattern.start_range,
  159. pattern.separator,
  160. pattern.end_range,
  161. field_to_i8(pattern.era),
  162. field_to_i8(pattern.year),
  163. field_to_i8(pattern.month),
  164. field_to_i8(pattern.weekday),
  165. field_to_i8(pattern.day),
  166. field_to_i8(pattern.day_period),
  167. field_to_i8(pattern.hour),
  168. field_to_i8(pattern.minute),
  169. field_to_i8(pattern.second),
  170. field_to_i8(pattern.fractional_second_digits),
  171. field_to_i8(pattern.time_zone_name));
  172. }
  173. };
  174. template<>
  175. struct AK::Traits<CalendarRangePattern> : public GenericTraits<CalendarRangePattern> {
  176. static unsigned hash(CalendarRangePattern const& c) { return c.hash(); }
  177. };
  178. struct CalendarFormat {
  179. unsigned hash() const
  180. {
  181. auto hash = pair_int_hash(full_format, long_format);
  182. hash = pair_int_hash(hash, medium_format);
  183. hash = pair_int_hash(hash, short_format);
  184. return hash;
  185. }
  186. bool operator==(CalendarFormat const& other) const
  187. {
  188. return (full_format == other.full_format)
  189. && (long_format == other.long_format)
  190. && (medium_format == other.medium_format)
  191. && (short_format == other.short_format);
  192. }
  193. CalendarPatternIndexType full_format { 0 };
  194. CalendarPatternIndexType long_format { 0 };
  195. CalendarPatternIndexType medium_format { 0 };
  196. CalendarPatternIndexType short_format { 0 };
  197. };
  198. template<>
  199. struct AK::Formatter<CalendarFormat> : Formatter<FormatString> {
  200. ErrorOr<void> format(FormatBuilder& builder, CalendarFormat const& pattern)
  201. {
  202. return Formatter<FormatString>::format(builder,
  203. "{{ {}, {}, {}, {} }}",
  204. pattern.full_format,
  205. pattern.long_format,
  206. pattern.medium_format,
  207. pattern.short_format);
  208. }
  209. };
  210. template<>
  211. struct AK::Traits<CalendarFormat> : public GenericTraits<CalendarFormat> {
  212. static unsigned hash(CalendarFormat const& c) { return c.hash(); }
  213. };
  214. using SymbolList = Vector<StringIndexType>;
  215. struct CalendarSymbols {
  216. SymbolListIndexType narrow_symbols { 0 };
  217. SymbolListIndexType short_symbols { 0 };
  218. SymbolListIndexType long_symbols { 0 };
  219. };
  220. struct Calendar {
  221. StringIndexType calendar { 0 };
  222. CalendarFormatIndexType date_formats { 0 };
  223. CalendarFormatIndexType time_formats { 0 };
  224. CalendarFormatIndexType date_time_formats { 0 };
  225. Vector<CalendarPatternIndexType> available_formats {};
  226. CalendarRangePatternIndexType default_range_format { 0 };
  227. Vector<CalendarRangePatternIndexType> range_formats {};
  228. Vector<CalendarRangePatternIndexType> range12_formats {};
  229. HashMap<String, CalendarSymbols> symbols {};
  230. };
  231. struct TimeZone {
  232. StringIndexType time_zone { 0 };
  233. StringIndexType long_name { 0 };
  234. StringIndexType short_name { 0 };
  235. };
  236. struct DayPeriod {
  237. Unicode::DayPeriod day_period {};
  238. u8 begin { 0 };
  239. u8 end { 0 };
  240. };
  241. struct Locale {
  242. HashMap<String, Calendar> calendars;
  243. HashMap<String, TimeZone> time_zones;
  244. Vector<DayPeriod> day_periods;
  245. };
  246. struct UnicodeLocaleData {
  247. UniqueStringStorage<StringIndexType> unique_strings;
  248. UniqueStorage<CalendarPattern, CalendarPatternIndexType> unique_patterns;
  249. UniqueStorage<CalendarRangePattern, CalendarRangePatternIndexType> unique_range_patterns;
  250. UniqueStorage<CalendarFormat, CalendarFormatIndexType> unique_formats;
  251. UniqueStorage<SymbolList, SymbolListIndexType> unique_symbol_lists;
  252. HashMap<String, Locale> locales;
  253. HashMap<String, Vector<Unicode::HourCycle>> hour_cycles;
  254. Vector<String> hour_cycle_regions;
  255. HashMap<String, StringIndexType> meta_zones;
  256. Vector<String> time_zones;
  257. Vector<String> calendars;
  258. Vector<Alias> calendar_aliases {
  259. // FIXME: Aliases should come from BCP47. See: https://unicode-org.atlassian.net/browse/CLDR-15158
  260. { "gregorian"sv, "gregory"sv },
  261. };
  262. Vector<String> symbols;
  263. };
  264. static Optional<Unicode::DayPeriod> day_period_from_string(StringView day_period)
  265. {
  266. if (day_period == "am"sv)
  267. return Unicode::DayPeriod::AM;
  268. if (day_period == "pm"sv)
  269. return Unicode::DayPeriod::PM;
  270. if (day_period == "morning1"sv)
  271. return Unicode::DayPeriod::Morning1;
  272. if (day_period == "morning2"sv)
  273. return Unicode::DayPeriod::Morning2;
  274. if (day_period == "afternoon1"sv)
  275. return Unicode::DayPeriod::Afternoon1;
  276. if (day_period == "afternoon2"sv)
  277. return Unicode::DayPeriod::Afternoon2;
  278. if (day_period == "evening1"sv)
  279. return Unicode::DayPeriod::Evening1;
  280. if (day_period == "evening2"sv)
  281. return Unicode::DayPeriod::Evening2;
  282. if (day_period == "night1"sv)
  283. return Unicode::DayPeriod::Night1;
  284. if (day_period == "night2"sv)
  285. return Unicode::DayPeriod::Night2;
  286. return {};
  287. };
  288. static ErrorOr<void> parse_hour_cycles(String core_path, UnicodeLocaleData& locale_data)
  289. {
  290. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  291. LexicalPath time_data_path(move(core_path));
  292. time_data_path = time_data_path.append("supplemental"sv);
  293. time_data_path = time_data_path.append("timeData.json"sv);
  294. auto time_data_file = TRY(Core::File::open(time_data_path.string(), Core::OpenMode::ReadOnly));
  295. auto time_data = TRY(JsonValue::from_string(time_data_file->read_all()));
  296. auto const& supplemental_object = time_data.as_object().get("supplemental"sv);
  297. auto const& time_data_object = supplemental_object.as_object().get("timeData"sv);
  298. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Unicode::HourCycle> {
  299. if (hour_cycle == "h"sv)
  300. return Unicode::HourCycle::H12;
  301. if (hour_cycle == "H"sv)
  302. return Unicode::HourCycle::H23;
  303. if (hour_cycle == "K"sv)
  304. return Unicode::HourCycle::H11;
  305. if (hour_cycle == "k"sv)
  306. return Unicode::HourCycle::H24;
  307. return {};
  308. };
  309. time_data_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  310. auto allowed_hour_cycles_string = value.as_object().get("_allowed"sv).as_string();
  311. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  312. Vector<Unicode::HourCycle> hour_cycles;
  313. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  314. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  315. hour_cycles.append(*hour_cycle);
  316. }
  317. locale_data.hour_cycles.set(key, move(hour_cycles));
  318. if (!locale_data.hour_cycle_regions.contains_slow(key))
  319. locale_data.hour_cycle_regions.append(key);
  320. });
  321. return {};
  322. };
  323. static ErrorOr<void> parse_meta_zones(String core_path, UnicodeLocaleData& locale_data)
  324. {
  325. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  326. LexicalPath meta_zone_path(move(core_path));
  327. meta_zone_path = meta_zone_path.append("supplemental"sv);
  328. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  329. auto meta_zone_file = TRY(Core::File::open(meta_zone_path.string(), Core::OpenMode::ReadOnly));
  330. auto meta_zone = TRY(JsonValue::from_string(meta_zone_file->read_all()));
  331. auto const& supplemental_object = meta_zone.as_object().get("supplemental"sv);
  332. auto const& meta_zone_object = supplemental_object.as_object().get("metaZones"sv);
  333. auto const& meta_zone_array = meta_zone_object.as_object().get("metazones"sv);
  334. meta_zone_array.as_array().for_each([&](JsonValue const& value) {
  335. auto const& mapping = value.as_object().get("mapZone"sv);
  336. auto const& meta_zone = mapping.as_object().get("_other"sv);
  337. auto const& golden_zone = mapping.as_object().get("_type"sv);
  338. auto golden_zone_index = locale_data.unique_strings.ensure(golden_zone.as_string());
  339. locale_data.meta_zones.set(meta_zone.as_string(), golden_zone_index);
  340. });
  341. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  342. locale_data.meta_zones.set("UTC"sv, locale_data.unique_strings.ensure("UTC"sv));
  343. return {};
  344. };
  345. static constexpr auto is_char(char ch)
  346. {
  347. return [ch](auto c) { return c == ch; };
  348. }
  349. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  350. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  351. // extra spaces around the day period. Some example expected removals:
  352. //
  353. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  354. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  355. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  356. static String remove_period_from_pattern(String pattern)
  357. {
  358. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  359. auto index = pattern.find(remove);
  360. if (!index.has_value())
  361. continue;
  362. constexpr u32 space = ' ';
  363. constexpr u32 open = '{';
  364. constexpr u32 close = '}';
  365. Utf8View utf8_pattern { pattern };
  366. Optional<u32> before_removal;
  367. Optional<u32> after_removal;
  368. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  369. before_removal = *it;
  370. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  371. after_removal = *it;
  372. if ((before_removal == space) && (after_removal != open)) {
  373. pattern = String::formatted("{}{}",
  374. pattern.substring_view(0, *index - 1),
  375. pattern.substring_view(*index + remove.length()));
  376. } else if ((after_removal == space) && (before_removal != close)) {
  377. pattern = String::formatted("{}{}",
  378. pattern.substring_view(0, *index),
  379. pattern.substring_view(*index + remove.length() + 1));
  380. } else {
  381. pattern = String::formatted("{}{}",
  382. pattern.substring_view(0, *index),
  383. pattern.substring_view(*index + remove.length()));
  384. }
  385. }
  386. return pattern;
  387. }
  388. static Optional<CalendarPattern> parse_date_time_pattern_raw(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  389. {
  390. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  391. using Unicode::CalendarPatternStyle;
  392. CalendarPattern format {};
  393. if (!skeleton.is_empty())
  394. format.skeleton_index = locale_data.unique_strings.ensure(move(skeleton));
  395. GenericLexer lexer { pattern };
  396. StringBuilder builder;
  397. bool hour12 { false };
  398. while (!lexer.is_eof()) {
  399. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  400. // processing (this just avoids conflicts with the patterns below).
  401. if (lexer.next_is(is_quote)) {
  402. builder.append(lexer.consume_quoted_string());
  403. continue;
  404. }
  405. auto starting_char = lexer.peek();
  406. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  407. // Era
  408. if (all_of(segment, is_char('G'))) {
  409. builder.append("{era}");
  410. if (segment.length() <= 3)
  411. format.era = CalendarPatternStyle::Short;
  412. else if (segment.length() == 4)
  413. format.era = CalendarPatternStyle::Long;
  414. else
  415. format.era = CalendarPatternStyle::Narrow;
  416. }
  417. // Year
  418. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  419. builder.append("{year}");
  420. if (segment.length() == 2)
  421. format.year = CalendarPatternStyle::TwoDigit;
  422. else
  423. format.year = CalendarPatternStyle::Numeric;
  424. }
  425. // Quarter
  426. else if (all_of(segment, is_any_of("qQ"sv))) {
  427. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  428. return {};
  429. }
  430. // Month
  431. else if (all_of(segment, is_any_of("ML"sv))) {
  432. builder.append("{month}");
  433. if (segment.length() == 1)
  434. format.month = CalendarPatternStyle::Numeric;
  435. else if (segment.length() == 2)
  436. format.month = CalendarPatternStyle::TwoDigit;
  437. else if (segment.length() == 3)
  438. format.month = CalendarPatternStyle::Short;
  439. else if (segment.length() == 4)
  440. format.month = CalendarPatternStyle::Long;
  441. else if (segment.length() == 5)
  442. format.month = CalendarPatternStyle::Narrow;
  443. } else if (all_of(segment, is_char('l'))) {
  444. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  445. return {};
  446. }
  447. // Week
  448. else if (all_of(segment, is_any_of("wW"sv))) {
  449. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  450. return {};
  451. }
  452. // Day
  453. else if (all_of(segment, is_char('d'))) {
  454. builder.append("{day}");
  455. if (segment.length() == 1)
  456. format.day = CalendarPatternStyle::Numeric;
  457. else
  458. format.day = CalendarPatternStyle::TwoDigit;
  459. } else if (all_of(segment, is_any_of("DFg"sv))) {
  460. builder.append("{day}");
  461. format.day = CalendarPatternStyle::Numeric;
  462. }
  463. // Weekday
  464. else if (all_of(segment, is_char('E'))) {
  465. builder.append("{weekday}");
  466. if (segment.length() == 4)
  467. format.weekday = CalendarPatternStyle::Long;
  468. else if (segment.length() == 5)
  469. format.weekday = CalendarPatternStyle::Narrow;
  470. else
  471. format.weekday = CalendarPatternStyle::Short;
  472. } else if (all_of(segment, is_any_of("ec"sv))) {
  473. builder.append("{weekday}");
  474. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  475. // pattern styles are not supported by Intl.DateTimeFormat.
  476. if (segment.length() <= 2)
  477. return {};
  478. if (segment.length() == 4)
  479. format.weekday = CalendarPatternStyle::Long;
  480. else if (segment.length() == 5)
  481. format.weekday = CalendarPatternStyle::Narrow;
  482. else
  483. format.weekday = CalendarPatternStyle::Short;
  484. }
  485. // Period
  486. else if (all_of(segment, is_any_of("ab"sv))) {
  487. builder.append("{ampm}");
  488. hour12 = true;
  489. } else if (all_of(segment, is_char('B'))) {
  490. builder.append("{dayPeriod}");
  491. hour12 = true;
  492. if (segment.length() == 4)
  493. format.day_period = CalendarPatternStyle::Long;
  494. else if (segment.length() == 5)
  495. format.day_period = CalendarPatternStyle::Narrow;
  496. else
  497. format.day_period = CalendarPatternStyle::Short;
  498. }
  499. // Hour
  500. else if (all_of(segment, is_any_of("hHKk"sv))) {
  501. builder.append("{hour}");
  502. if ((segment[0] == 'h') || (segment[0] == 'K'))
  503. hour12 = true;
  504. if (segment.length() == 1)
  505. format.hour = CalendarPatternStyle::Numeric;
  506. else
  507. format.hour = CalendarPatternStyle::TwoDigit;
  508. } else if (all_of(segment, is_any_of("jJC"sv))) {
  509. // TR-35 indicates these should not be used.
  510. return {};
  511. }
  512. // Minute
  513. else if (all_of(segment, is_char('m'))) {
  514. builder.append("{minute}");
  515. if (segment.length() == 1)
  516. format.minute = CalendarPatternStyle::Numeric;
  517. else
  518. format.minute = CalendarPatternStyle::TwoDigit;
  519. }
  520. // Second
  521. else if (all_of(segment, is_char('s'))) {
  522. builder.append("{second}");
  523. if (segment.length() == 1)
  524. format.second = CalendarPatternStyle::Numeric;
  525. else
  526. format.second = CalendarPatternStyle::TwoDigit;
  527. } else if (all_of(segment, is_char('S'))) {
  528. builder.append("{fractionalSecondDigits}");
  529. VERIFY(segment.length() <= 3);
  530. format.fractional_second_digits = static_cast<u8>(segment.length());
  531. } else if (all_of(segment, is_char('A'))) {
  532. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  533. return {};
  534. }
  535. // Zone
  536. else if (all_of(segment, is_any_of("zZOvVXx"))) {
  537. builder.append("{timeZoneName}");
  538. if (segment.length() < 4)
  539. format.time_zone_name = CalendarPatternStyle::Short;
  540. else
  541. format.time_zone_name = CalendarPatternStyle::Long;
  542. }
  543. // Non-patterns
  544. else {
  545. builder.append(segment);
  546. }
  547. }
  548. pattern = builder.build();
  549. if (hour12) {
  550. format.pattern = remove_period_from_pattern(pattern);
  551. format.pattern12 = move(pattern);
  552. } else {
  553. format.pattern = move(pattern);
  554. }
  555. return format;
  556. }
  557. static Optional<CalendarPatternIndexType> parse_date_time_pattern(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  558. {
  559. auto format = parse_date_time_pattern_raw(move(pattern), move(skeleton), locale_data);
  560. if (!format.has_value())
  561. return {};
  562. format->pattern_index = locale_data.unique_strings.ensure(move(format->pattern));
  563. if (format->pattern12.has_value())
  564. format->pattern12_index = locale_data.unique_strings.ensure(format->pattern12.release_value());
  565. return locale_data.unique_patterns.ensure(format.release_value());
  566. }
  567. template<typename... Chars>
  568. static constexpr bool char_is_one_of(char ch, Chars&&... chars)
  569. {
  570. return ((ch == chars) || ...);
  571. }
  572. static void parse_interval_patterns(Calendar& calendar, JsonObject const& interval_formats_object, UnicodeLocaleData& locale_data)
  573. {
  574. // https://unicode.org/reports/tr35/tr35-dates.html#intervalFormats
  575. auto name_of_field = [&](char field) {
  576. if (char_is_one_of(field, 'G'))
  577. return Unicode::CalendarRangePattern::Field::Era;
  578. if (char_is_one_of(field, 'y', 'Y', 'u', 'U', 'r'))
  579. return Unicode::CalendarRangePattern::Field::Year;
  580. if (char_is_one_of(field, 'M', 'L'))
  581. return Unicode::CalendarRangePattern::Field::Month;
  582. if (char_is_one_of(field, 'd', 'D', 'F', 'g'))
  583. return Unicode::CalendarRangePattern::Field::Day;
  584. if (char_is_one_of(field, 'a', 'b'))
  585. return Unicode::CalendarRangePattern::Field::AmPm;
  586. if (char_is_one_of(field, 'B'))
  587. return Unicode::CalendarRangePattern::Field::DayPeriod;
  588. if (char_is_one_of(field, 'h', 'H', 'K', 'k'))
  589. return Unicode::CalendarRangePattern::Field::Hour;
  590. if (char_is_one_of(field, 'm'))
  591. return Unicode::CalendarRangePattern::Field::Minute;
  592. if (char_is_one_of(field, 's'))
  593. return Unicode::CalendarRangePattern::Field::Second;
  594. if (char_is_one_of(field, 'S'))
  595. return Unicode::CalendarRangePattern::Field::FractionalSecondDigits;
  596. VERIFY_NOT_REACHED();
  597. };
  598. auto split_default_range_pattern = [&](auto skeleton, auto const& pattern) {
  599. auto start_range_end = pattern.find('}').value() + 1;
  600. auto end_range_begin = pattern.find_last('{').value();
  601. auto start_range = pattern.substring_view(0, start_range_end);
  602. auto separator = pattern.substring_view(start_range_end, end_range_begin - start_range_end);
  603. auto end_range = pattern.substring_view(end_range_begin);
  604. CalendarRangePattern format {};
  605. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  606. format.start_range = locale_data.unique_strings.ensure(start_range);
  607. format.separator = locale_data.unique_strings.ensure(separator);
  608. format.end_range = locale_data.unique_strings.ensure(end_range);
  609. return format;
  610. };
  611. auto split_range_pattern = [&](auto skeleton, auto field, auto const& pattern, auto const& parsed_fields) {
  612. HashMap<StringView, size_t> partitions;
  613. StringView last_partition;
  614. auto begin_index = pattern.find('{');
  615. size_t end_index = 0;
  616. while (begin_index.has_value()) {
  617. end_index = pattern.find('}', *begin_index).value();
  618. auto partition = pattern.substring_view(*begin_index, end_index - *begin_index);
  619. if (partitions.contains(partition))
  620. break;
  621. partitions.set(partition, *begin_index);
  622. last_partition = partition;
  623. begin_index = pattern.find('{', end_index + 1);
  624. }
  625. VERIFY(begin_index.has_value() && !last_partition.is_empty());
  626. auto start_range_end = partitions.get(last_partition).value() + last_partition.length() + 1;
  627. auto start_range = pattern.substring_view(0, start_range_end);
  628. auto separator = pattern.substring_view(start_range_end, *begin_index - start_range_end);
  629. auto end_range = pattern.substring_view(*begin_index);
  630. CalendarRangePattern format {};
  631. format.skeleton_index = locale_data.unique_strings.ensure(skeleton);
  632. format.field = field;
  633. format.start_range = locale_data.unique_strings.ensure(start_range);
  634. format.separator = locale_data.unique_strings.ensure(separator);
  635. format.end_range = locale_data.unique_strings.ensure(end_range);
  636. format.for_each_calendar_field_zipped_with(parsed_fields, [](auto& format_field, auto const& parsed_field, auto) {
  637. format_field = parsed_field;
  638. });
  639. return format;
  640. };
  641. interval_formats_object.for_each_member([&](auto const& skeleton, auto const& value) {
  642. if (skeleton == "intervalFormatFallback"sv) {
  643. auto range_format = split_default_range_pattern(skeleton, value.as_string());
  644. calendar.default_range_format = locale_data.unique_range_patterns.ensure(move(range_format));
  645. return;
  646. }
  647. value.as_object().for_each_member([&](auto const& field, auto const& pattern) {
  648. if (field.ends_with("alt-variant"sv))
  649. return;
  650. VERIFY(field.length() == 1);
  651. auto name = name_of_field(field[0]);
  652. auto format = parse_date_time_pattern_raw(pattern.as_string(), skeleton, locale_data).release_value();
  653. auto range_format = split_range_pattern(skeleton, name, format.pattern, format);
  654. calendar.range_formats.append(locale_data.unique_range_patterns.ensure(move(range_format)));
  655. if (format.pattern12.has_value()) {
  656. auto range12_pattern = split_range_pattern(skeleton, name, *format.pattern12, format);
  657. calendar.range12_formats.append(locale_data.unique_range_patterns.ensure(move(range12_pattern)));
  658. } else {
  659. calendar.range12_formats.append(calendar.range_formats.last());
  660. }
  661. });
  662. });
  663. }
  664. static void generate_missing_patterns(Calendar& calendar, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, UnicodeLocaleData& locale_data)
  665. {
  666. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  667. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  668. auto pattern = locale_data.unique_strings.get(format);
  669. auto time_pattern = locale_data.unique_strings.get(time_format);
  670. auto date_pattern = locale_data.unique_strings.get(date_format);
  671. auto new_pattern = pattern.replace("{0}", time_pattern).replace("{1}", date_pattern);
  672. return locale_data.unique_strings.ensure(move(new_pattern));
  673. };
  674. auto inject_fractional_second_digits = [&](auto format) {
  675. auto pattern = locale_data.unique_strings.get(format);
  676. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv);
  677. return locale_data.unique_strings.ensure(move(new_pattern));
  678. };
  679. auto append_if_unique = [&](auto format) {
  680. auto format_index = locale_data.unique_patterns.ensure(move(format));
  681. if (!calendar.available_formats.contains_slow(format_index))
  682. calendar.available_formats.append(format_index);
  683. };
  684. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  685. for (auto const& format : date_formats)
  686. append_if_unique(format);
  687. for (auto const& format : time_formats) {
  688. append_if_unique(format);
  689. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  690. auto new_format = format;
  691. new_format.fractional_second_digits = 2;
  692. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  693. if (new_format.pattern12_index != 0)
  694. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  695. time_formats_with_fractional_second_digits.append(new_format);
  696. append_if_unique(move(new_format));
  697. }
  698. }
  699. time_formats.extend(move(time_formats_with_fractional_second_digits));
  700. for (auto const& date_format : date_formats) {
  701. auto const& date_time_formats = locale_data.unique_formats.get(calendar.date_time_formats);
  702. CalendarPatternIndexType date_time_format_index = 0;
  703. if (date_format.month == Unicode::CalendarPatternStyle::Long) {
  704. if (date_format.weekday.has_value())
  705. date_time_format_index = date_time_formats.full_format;
  706. else
  707. date_time_format_index = date_time_formats.long_format;
  708. } else if (date_format.month == Unicode::CalendarPatternStyle::Short) {
  709. date_time_format_index = date_time_formats.medium_format;
  710. } else {
  711. date_time_format_index = date_time_formats.short_format;
  712. }
  713. for (auto const& time_format : time_formats) {
  714. auto format = locale_data.unique_patterns.get(date_time_format_index);
  715. if (time_format.pattern12_index != 0)
  716. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  717. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  718. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  719. if (date_field.has_value())
  720. field = date_field;
  721. });
  722. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  723. if (time_field.has_value())
  724. field = time_field;
  725. });
  726. append_if_unique(move(format));
  727. }
  728. }
  729. }
  730. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, UnicodeLocaleData& locale_data)
  731. {
  732. auto ensure_symbols = [&](auto name) -> CalendarSymbols& {
  733. if (!locale_data.symbols.contains_slow(name))
  734. locale_data.symbols.append(name);
  735. return calendar.symbols.ensure(name);
  736. };
  737. auto create_symbol_lists = [](size_t size) {
  738. SymbolList narrow_symbol_list;
  739. SymbolList short_symbol_list;
  740. SymbolList long_symbol_list;
  741. narrow_symbol_list.resize(size);
  742. short_symbol_list.resize(size);
  743. long_symbol_list.resize(size);
  744. return Array<SymbolList, 3> { {
  745. move(narrow_symbol_list),
  746. move(short_symbol_list),
  747. move(long_symbol_list),
  748. } };
  749. };
  750. auto store_symbol_lists = [&](auto name, auto symbol_lists) {
  751. auto& symbols = ensure_symbols(name);
  752. symbols.narrow_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[0]));
  753. symbols.short_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[1]));
  754. symbols.long_symbols = locale_data.unique_symbol_lists.ensure(move(symbol_lists[2]));
  755. };
  756. auto parse_era_symbols = [&](auto const& symbols_object) {
  757. auto const& narrow_symbols = symbols_object.get("eraNarrow"sv).as_object();
  758. auto const& short_symbols = symbols_object.get("eraAbbr"sv).as_object();
  759. auto const& long_symbols = symbols_object.get("eraNames"sv).as_object();
  760. auto symbol_lists = create_symbol_lists(2);
  761. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  762. if (auto key_index = key.to_uint(); key_index.has_value())
  763. symbols[*key_index] = locale_data.unique_strings.ensure(move(symbol));
  764. };
  765. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  766. append_symbol(symbol_lists[0], key, value.as_string());
  767. });
  768. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  769. append_symbol(symbol_lists[1], key, value.as_string());
  770. });
  771. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  772. append_symbol(symbol_lists[2], key, value.as_string());
  773. });
  774. store_symbol_lists("era"sv, move(symbol_lists));
  775. };
  776. auto parse_month_symbols = [&](auto const& symbols_object) {
  777. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  778. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  779. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  780. auto symbol_lists = create_symbol_lists(12);
  781. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  782. auto key_index = key.to_uint().value() - 1;
  783. symbols[key_index] = locale_data.unique_strings.ensure(move(symbol));
  784. };
  785. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  786. append_symbol(symbol_lists[0], key, value.as_string());
  787. });
  788. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  789. append_symbol(symbol_lists[1], key, value.as_string());
  790. });
  791. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  792. append_symbol(symbol_lists[2], key, value.as_string());
  793. });
  794. store_symbol_lists("month"sv, move(symbol_lists));
  795. };
  796. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  797. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  798. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  799. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  800. auto symbol_lists = create_symbol_lists(7);
  801. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  802. if (key == "sun"sv)
  803. symbols[to_underlying(Unicode::Weekday::Sunday)] = locale_data.unique_strings.ensure(move(symbol));
  804. else if (key == "mon"sv)
  805. symbols[to_underlying(Unicode::Weekday::Monday)] = locale_data.unique_strings.ensure(move(symbol));
  806. else if (key == "tue"sv)
  807. symbols[to_underlying(Unicode::Weekday::Tuesday)] = locale_data.unique_strings.ensure(move(symbol));
  808. else if (key == "wed"sv)
  809. symbols[to_underlying(Unicode::Weekday::Wednesday)] = locale_data.unique_strings.ensure(move(symbol));
  810. else if (key == "thu"sv)
  811. symbols[to_underlying(Unicode::Weekday::Thursday)] = locale_data.unique_strings.ensure(move(symbol));
  812. else if (key == "fri"sv)
  813. symbols[to_underlying(Unicode::Weekday::Friday)] = locale_data.unique_strings.ensure(move(symbol));
  814. else if (key == "sat"sv)
  815. symbols[to_underlying(Unicode::Weekday::Saturday)] = locale_data.unique_strings.ensure(move(symbol));
  816. };
  817. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  818. append_symbol(symbol_lists[0], key, value.as_string());
  819. });
  820. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  821. append_symbol(symbol_lists[1], key, value.as_string());
  822. });
  823. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  824. append_symbol(symbol_lists[2], key, value.as_string());
  825. });
  826. store_symbol_lists("weekday"sv, move(symbol_lists));
  827. };
  828. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  829. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  830. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  831. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  832. auto symbol_lists = create_symbol_lists(10);
  833. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  834. if (auto day_period = day_period_from_string(key); day_period.has_value())
  835. symbols[to_underlying(*day_period)] = locale_data.unique_strings.ensure(move(symbol));
  836. };
  837. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  838. append_symbol(symbol_lists[0], key, value.as_string());
  839. });
  840. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  841. append_symbol(symbol_lists[1], key, value.as_string());
  842. });
  843. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  844. append_symbol(symbol_lists[2], key, value.as_string());
  845. });
  846. store_symbol_lists("dayPeriod"sv, move(symbol_lists));
  847. };
  848. parse_era_symbols(calendar_object.get("eras"sv).as_object());
  849. parse_month_symbols(calendar_object.get("months"sv).as_object().get("format"sv).as_object());
  850. parse_weekday_symbols(calendar_object.get("days"sv).as_object().get("format"sv).as_object());
  851. parse_day_period_symbols(calendar_object.get("dayPeriods"sv).as_object().get("format"sv).as_object());
  852. }
  853. static ErrorOr<void> parse_calendars(String locale_calendars_path, UnicodeLocaleData& locale_data, Locale& locale)
  854. {
  855. LexicalPath calendars_path(move(locale_calendars_path));
  856. if (!calendars_path.basename().starts_with("ca-"sv))
  857. return {};
  858. auto calendars_file = TRY(Core::File::open(calendars_path.string(), Core::OpenMode::ReadOnly));
  859. auto calendars = TRY(JsonValue::from_string(calendars_file->read_all()));
  860. auto const& main_object = calendars.as_object().get("main"sv);
  861. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  862. auto const& dates_object = locale_object.as_object().get("dates"sv);
  863. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  864. auto ensure_calendar = [&](auto const& calendar) -> Calendar& {
  865. return locale.calendars.ensure(calendar, [&]() {
  866. auto calendar_index = locale_data.unique_strings.ensure(calendar);
  867. return Calendar { .calendar = calendar_index };
  868. });
  869. };
  870. auto parse_patterns = [&](auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) {
  871. auto parse_pattern = [&](auto name) {
  872. auto format = patterns_object.get(name);
  873. auto skeleton = skeletons_object.get(name);
  874. auto format_index = parse_date_time_pattern(format.as_string(), skeleton.as_string_or(String::empty()), locale_data).value();
  875. if (patterns)
  876. patterns->append(locale_data.unique_patterns.get(format_index));
  877. return format_index;
  878. };
  879. CalendarFormat formats {};
  880. formats.full_format = parse_pattern("full"sv);
  881. formats.long_format = parse_pattern("long"sv);
  882. formats.medium_format = parse_pattern("medium"sv);
  883. formats.short_format = parse_pattern("short"sv);
  884. return locale_data.unique_formats.ensure(move(formats));
  885. };
  886. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  887. // The generic calendar is not a supported Unicode calendar key, so skip it:
  888. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  889. if (calendar_name == "generic"sv)
  890. return;
  891. auto& calendar = ensure_calendar(calendar_name);
  892. if (!locale_data.calendars.contains_slow(calendar_name))
  893. locale_data.calendars.append(calendar_name);
  894. Vector<CalendarPattern> date_formats;
  895. Vector<CalendarPattern> time_formats;
  896. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  897. auto const& date_skeletons_object = value.as_object().get("dateSkeletons"sv);
  898. calendar.date_formats = parse_patterns(date_formats_object.as_object(), date_skeletons_object.as_object(), &date_formats);
  899. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  900. auto const& time_skeletons_object = value.as_object().get("timeSkeletons"sv);
  901. calendar.time_formats = parse_patterns(time_formats_object.as_object(), time_skeletons_object.as_object(), &time_formats);
  902. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  903. calendar.date_time_formats = parse_patterns(date_time_formats_object.as_object(), JsonObject {}, nullptr);
  904. auto const& available_formats = date_time_formats_object.as_object().get("availableFormats"sv);
  905. available_formats.as_object().for_each_member([&](auto const& skeleton, JsonValue const& pattern) {
  906. auto pattern_index = parse_date_time_pattern(pattern.as_string(), skeleton, locale_data);
  907. if (!pattern_index.has_value())
  908. return;
  909. auto const& format = locale_data.unique_patterns.get(*pattern_index);
  910. if (format.contains_only_date_fields())
  911. date_formats.append(format);
  912. else if (format.contains_only_time_fields())
  913. time_formats.append(format);
  914. if (!calendar.available_formats.contains_slow(*pattern_index))
  915. calendar.available_formats.append(*pattern_index);
  916. });
  917. auto const& interval_formats_object = date_time_formats_object.as_object().get("intervalFormats"sv);
  918. parse_interval_patterns(calendar, interval_formats_object.as_object(), locale_data);
  919. generate_missing_patterns(calendar, move(date_formats), move(time_formats), locale_data);
  920. parse_calendar_symbols(calendar, value.as_object(), locale_data);
  921. });
  922. return {};
  923. }
  924. static ErrorOr<void> parse_time_zone_names(String locale_time_zone_names_path, UnicodeLocaleData& locale_data, Locale& locale)
  925. {
  926. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  927. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  928. auto time_zone_names_file = TRY(Core::File::open(time_zone_names_path.string(), Core::OpenMode::ReadOnly));
  929. auto time_zone_names = TRY(JsonValue::from_string(time_zone_names_file->read_all()));
  930. auto const& main_object = time_zone_names.as_object().get("main"sv);
  931. auto const& locale_object = main_object.as_object().get(time_zone_names_path.parent().basename());
  932. auto const& dates_object = locale_object.as_object().get("dates"sv);
  933. auto const& time_zone_names_object = dates_object.as_object().get("timeZoneNames"sv);
  934. auto const& meta_zone_object = time_zone_names_object.as_object().get("metazone"sv);
  935. if (meta_zone_object.is_null())
  936. return {};
  937. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object) -> Optional<StringIndexType> {
  938. auto const& names = meta_zone_object.get(type);
  939. if (!names.is_object())
  940. return {};
  941. auto const& daylight = names.as_object().get("daylight"sv);
  942. if (daylight.is_string())
  943. return locale_data.unique_strings.ensure(daylight.as_string());
  944. auto const& standard = names.as_object().get("standard"sv);
  945. if (standard.is_string())
  946. return locale_data.unique_strings.ensure(standard.as_string());
  947. return {};
  948. };
  949. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  950. auto golden_zone = locale_data.meta_zones.get(meta_zone).value();
  951. TimeZone time_zone { .time_zone = golden_zone };
  952. if (auto long_name = parse_name("long"sv, meta_zone_object); long_name.has_value())
  953. time_zone.long_name = long_name.value();
  954. if (auto short_name = parse_name("short"sv, meta_zone_object); short_name.has_value())
  955. time_zone.short_name = short_name.value();
  956. auto const& time_zone_name = locale_data.unique_strings.get(golden_zone);
  957. if (!locale_data.time_zones.contains_slow(time_zone_name))
  958. locale_data.time_zones.append(time_zone_name);
  959. locale.time_zones.set(time_zone_name, move(time_zone));
  960. };
  961. meta_zone_object.as_object().for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  962. parse_time_zone(meta_zone, value.as_object());
  963. });
  964. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  965. auto const& zone_object = time_zone_names_object.as_object().get("zone"sv);
  966. auto const& etc_object = zone_object.as_object().get("Etc"sv);
  967. auto const& utc_object = etc_object.as_object().get("UTC"sv);
  968. parse_time_zone("UTC"sv, utc_object.as_object());
  969. return {};
  970. }
  971. static ErrorOr<void> parse_day_periods(String core_path, UnicodeLocaleData& locale_data)
  972. {
  973. // https://unicode.org/reports/tr35/tr35-dates.html#Day_Period_Rule_Sets
  974. LexicalPath day_periods_path(move(core_path));
  975. day_periods_path = day_periods_path.append("supplemental"sv);
  976. day_periods_path = day_periods_path.append("dayPeriods.json"sv);
  977. auto day_periods_file = TRY(Core::File::open(day_periods_path.string(), Core::OpenMode::ReadOnly));
  978. auto day_periods = TRY(JsonValue::from_string(day_periods_file->read_all()));
  979. auto const& supplemental_object = day_periods.as_object().get("supplemental"sv);
  980. auto const& day_periods_object = supplemental_object.as_object().get("dayPeriodRuleSet"sv);
  981. auto parse_hour = [](auto const& time) {
  982. auto hour_end_index = time.find(':').value();
  983. // The times are in the form "hours:minutes", but we only need the hour segment.
  984. // TR-35 explicitly states that minutes other than :00 must not be used.
  985. VERIFY(time.substring_view(hour_end_index) == ":00"sv);
  986. auto hour = time.substring_view(0, hour_end_index);
  987. return hour.template to_uint<u8>().value();
  988. };
  989. auto parse_day_period = [&](auto const& symbol, auto const& ranges) -> Optional<DayPeriod> {
  990. auto day_period = day_period_from_string(symbol);
  991. if (!day_period.has_value())
  992. return {};
  993. auto begin = parse_hour(ranges.get("_from"sv).as_string());
  994. auto end = parse_hour(ranges.get("_before"sv).as_string());
  995. return DayPeriod { *day_period, begin, end };
  996. };
  997. day_periods_object.as_object().for_each_member([&](auto const& language, JsonValue const& value) {
  998. auto locale = locale_data.locales.find(language);
  999. if (locale == locale_data.locales.end())
  1000. return;
  1001. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& ranges) {
  1002. if (auto day_period = parse_day_period(symbol, ranges.as_object()); day_period.has_value())
  1003. locale->value.day_periods.append(day_period.release_value());
  1004. });
  1005. });
  1006. return {};
  1007. };
  1008. static ErrorOr<void> parse_all_locales(String core_path, String dates_path, UnicodeLocaleData& locale_data)
  1009. {
  1010. TRY(parse_hour_cycles(core_path, locale_data));
  1011. TRY(parse_meta_zones(core_path, locale_data));
  1012. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  1013. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  1014. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  1015. StringBuilder builder;
  1016. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  1017. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  1018. builder.appendff("-{}", script);
  1019. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  1020. builder.appendff("-{}", region);
  1021. return builder.build();
  1022. };
  1023. while (dates_iterator.has_next()) {
  1024. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  1025. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  1026. auto language = TRY(remove_variants_from_path(dates_path));
  1027. auto& locale = locale_data.locales.ensure(language);
  1028. while (calendars_iterator.has_next()) {
  1029. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  1030. TRY(parse_calendars(move(calendars_path), locale_data, locale));
  1031. }
  1032. TRY(parse_time_zone_names(move(dates_path), locale_data, locale));
  1033. }
  1034. TRY(parse_day_periods(move(core_path), locale_data));
  1035. return {};
  1036. }
  1037. static String format_identifier(StringView owner, String identifier)
  1038. {
  1039. identifier = identifier.replace("-"sv, "_"sv, true);
  1040. identifier = identifier.replace("/"sv, "_"sv, true);
  1041. if (all_of(identifier, is_ascii_digit))
  1042. return String::formatted("{}_{}", owner[0], identifier);
  1043. if (is_ascii_lower_alpha(identifier[0]))
  1044. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  1045. return identifier;
  1046. }
  1047. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  1048. {
  1049. StringBuilder builder;
  1050. SourceGenerator generator { builder };
  1051. generator.append(R"~~~(
  1052. #pragma once
  1053. #include <AK/Optional.h>
  1054. #include <AK/StringView.h>
  1055. #include <LibUnicode/Forward.h>
  1056. namespace Unicode {
  1057. )~~~");
  1058. generate_enum(generator, format_identifier, "Calendar"sv, {}, locale_data.calendars, locale_data.calendar_aliases);
  1059. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, locale_data.hour_cycle_regions);
  1060. generate_enum(generator, format_identifier, "CalendarSymbol"sv, {}, locale_data.symbols);
  1061. generate_enum(generator, format_identifier, "TimeZone"sv, {}, locale_data.time_zones);
  1062. generator.append(R"~~~(
  1063. namespace Detail {
  1064. Optional<Calendar> calendar_from_string(StringView calendar);
  1065. Optional<HourCycleRegion> hour_cycle_region_from_string(StringView hour_cycle_region);
  1066. Vector<Unicode::HourCycle> get_regional_hour_cycles(StringView region);
  1067. Optional<Unicode::CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar);
  1068. Optional<Unicode::CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar);
  1069. Optional<Unicode::CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar);
  1070. Vector<Unicode::CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar);
  1071. Optional<Unicode::CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar);
  1072. Vector<Unicode::CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton);
  1073. Vector<Unicode::CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton);
  1074. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Era value);
  1075. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Month value);
  1076. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Weekday value);
  1077. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::DayPeriod value);
  1078. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour);
  1079. Optional<TimeZone> time_zone_from_string(StringView time_zone);
  1080. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style);
  1081. }
  1082. }
  1083. )~~~");
  1084. VERIFY(file.write(generator.as_string_view()));
  1085. }
  1086. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  1087. {
  1088. StringBuilder builder;
  1089. SourceGenerator generator { builder };
  1090. generator.set("string_index_type"sv, s_string_index_type);
  1091. generator.set("calendar_pattern_index_type"sv, s_calendar_pattern_index_type);
  1092. generator.set("calendar_range_pattern_index_type"sv, s_calendar_range_pattern_index_type);
  1093. generator.set("calendar_format_index_type"sv, s_calendar_format_index_type);
  1094. generator.set("symbol_list_index_type"sv, s_symbol_list_index_type);
  1095. generator.set("calendar_symbols_size"sv, String::number(locale_data.symbols.size()));
  1096. generator.append(R"~~~(
  1097. #include <AK/Array.h>
  1098. #include <AK/BinarySearch.h>
  1099. #include <LibUnicode/DateTimeFormat.h>
  1100. #include <LibUnicode/Locale.h>
  1101. #include <LibUnicode/UnicodeDateTimeFormat.h>
  1102. namespace Unicode::Detail {
  1103. )~~~");
  1104. locale_data.unique_strings.generate(generator);
  1105. generator.append(R"~~~(
  1106. template <typename SourceType, typename TargetType>
  1107. void convert_calendar_fields(SourceType const& source, TargetType& target)
  1108. {
  1109. if (source.era != -1)
  1110. target.era = static_cast<Unicode::CalendarPatternStyle>(source.era);
  1111. if (source.year != -1)
  1112. target.year = static_cast<Unicode::CalendarPatternStyle>(source.year);
  1113. if (source.month != -1)
  1114. target.month = static_cast<Unicode::CalendarPatternStyle>(source.month);
  1115. if (source.weekday != -1)
  1116. target.weekday = static_cast<Unicode::CalendarPatternStyle>(source.weekday);
  1117. if (source.day != -1)
  1118. target.day = static_cast<Unicode::CalendarPatternStyle>(source.day);
  1119. if (source.day_period != -1)
  1120. target.day_period = static_cast<Unicode::CalendarPatternStyle>(source.day_period);
  1121. if (source.hour != -1)
  1122. target.hour = static_cast<Unicode::CalendarPatternStyle>(source.hour);
  1123. if (source.minute != -1)
  1124. target.minute = static_cast<Unicode::CalendarPatternStyle>(source.minute);
  1125. if (source.second != -1)
  1126. target.second = static_cast<Unicode::CalendarPatternStyle>(source.second);
  1127. if (source.fractional_second_digits != -1)
  1128. target.fractional_second_digits = static_cast<u8>(source.fractional_second_digits);
  1129. if (source.time_zone_name != -1)
  1130. target.time_zone_name = static_cast<Unicode::CalendarPatternStyle>(source.time_zone_name);
  1131. }
  1132. struct CalendarPattern {
  1133. Unicode::CalendarPattern to_unicode_calendar_pattern() const {
  1134. Unicode::CalendarPattern calendar_pattern {};
  1135. calendar_pattern.skeleton = s_string_list[skeleton];
  1136. calendar_pattern.pattern = s_string_list[pattern];
  1137. if (pattern12 != 0)
  1138. calendar_pattern.pattern12 = s_string_list[pattern12];
  1139. convert_calendar_fields(*this, calendar_pattern);
  1140. return calendar_pattern;
  1141. }
  1142. @string_index_type@ skeleton { 0 };
  1143. @string_index_type@ pattern { 0 };
  1144. @string_index_type@ pattern12 { 0 };
  1145. i8 era { -1 };
  1146. i8 year { -1 };
  1147. i8 month { -1 };
  1148. i8 weekday { -1 };
  1149. i8 day { -1 };
  1150. i8 day_period { -1 };
  1151. i8 hour { -1 };
  1152. i8 minute { -1 };
  1153. i8 second { -1 };
  1154. i8 fractional_second_digits { -1 };
  1155. i8 time_zone_name { -1 };
  1156. };
  1157. struct CalendarRangePattern {
  1158. Unicode::CalendarRangePattern to_unicode_calendar_range_pattern() const {
  1159. Unicode::CalendarRangePattern calendar_range_pattern {};
  1160. if (field != -1)
  1161. calendar_range_pattern.field = static_cast<Unicode::CalendarRangePattern::Field>(field);
  1162. calendar_range_pattern.start_range = s_string_list[start_range];
  1163. calendar_range_pattern.separator = s_string_list[separator];
  1164. calendar_range_pattern.end_range = s_string_list[end_range];
  1165. convert_calendar_fields(*this, calendar_range_pattern);
  1166. return calendar_range_pattern;
  1167. }
  1168. @string_index_type@ skeleton { 0 };
  1169. i8 field { -1 };
  1170. @string_index_type@ start_range { 0 };
  1171. @string_index_type@ separator { 0 };
  1172. @string_index_type@ end_range { 0 };
  1173. i8 era { -1 };
  1174. i8 year { -1 };
  1175. i8 month { -1 };
  1176. i8 weekday { -1 };
  1177. i8 day { -1 };
  1178. i8 day_period { -1 };
  1179. i8 hour { -1 };
  1180. i8 minute { -1 };
  1181. i8 second { -1 };
  1182. i8 fractional_second_digits { -1 };
  1183. i8 time_zone_name { -1 };
  1184. };
  1185. )~~~");
  1186. locale_data.unique_patterns.generate(generator, "CalendarPattern"sv, "s_calendar_patterns"sv, 10);
  1187. locale_data.unique_range_patterns.generate(generator, "CalendarRangePattern"sv, "s_calendar_range_patterns"sv, 10);
  1188. generator.append(R"~~~(
  1189. struct CalendarFormat {
  1190. Unicode::CalendarFormat to_unicode_calendar_format() const {
  1191. Unicode::CalendarFormat calendar_format {};
  1192. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  1193. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  1194. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  1195. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  1196. return calendar_format;
  1197. }
  1198. @calendar_pattern_index_type@ full_format { 0 };
  1199. @calendar_pattern_index_type@ long_format { 0 };
  1200. @calendar_pattern_index_type@ medium_format { 0 };
  1201. @calendar_pattern_index_type@ short_format { 0 };
  1202. };
  1203. using CalendarSymbols = Span<@string_index_type@ const>;
  1204. struct CalendarData {
  1205. @string_index_type@ calendar { 0 };
  1206. @calendar_format_index_type@ date_formats {};
  1207. @calendar_format_index_type@ time_formats {};
  1208. @calendar_format_index_type@ date_time_formats {};
  1209. Span<@calendar_pattern_index_type@ const> available_formats {};
  1210. @calendar_range_pattern_index_type@ default_range_format { 0 };
  1211. Span<@calendar_range_pattern_index_type@ const> range_formats {};
  1212. Span<@calendar_range_pattern_index_type@ const> range12_formats {};
  1213. Array<CalendarSymbols, @calendar_symbols_size@> symbols {};
  1214. };
  1215. struct TimeZoneData {
  1216. @string_index_type@ time_zone { 0 };
  1217. @string_index_type@ long_name { 0 };
  1218. @string_index_type@ short_name { 0 };
  1219. };
  1220. struct DayPeriodData {
  1221. u8 day_period { 0 };
  1222. u8 begin { 0 };
  1223. u8 end { 0 };
  1224. };
  1225. )~~~");
  1226. locale_data.unique_formats.generate(generator, "CalendarFormat"sv, "s_calendar_formats"sv, 10);
  1227. locale_data.unique_symbol_lists.generate(generator, s_string_index_type, "s_symbol_lists"sv);
  1228. auto append_pattern_list = [&](auto name, auto type, auto const& formats) {
  1229. generator.set("name", move(name));
  1230. generator.set("type", move(type));
  1231. generator.set("size", String::number(formats.size()));
  1232. generator.append(R"~~~(
  1233. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1234. bool first = true;
  1235. for (auto format : formats) {
  1236. generator.append(first ? " " : ", ");
  1237. generator.append(String::number(format));
  1238. first = false;
  1239. }
  1240. generator.append(" } };");
  1241. };
  1242. auto append_calendar_symbols = [&](auto name, auto symbol, auto const& symbols) {
  1243. name = String::formatted("{}_{}", name, symbol.to_lowercase());
  1244. generator.set("narrow_symbols", String::number(symbols.narrow_symbols));
  1245. generator.set("short_symbols", String::number(symbols.short_symbols));
  1246. generator.set("long_symbols", String::number(symbols.long_symbols));
  1247. generator.set("name", name);
  1248. generator.append(R"~~~(
  1249. static constexpr Array<@symbol_list_index_type@, 3> @name@ { { @narrow_symbols@, @short_symbols@, @long_symbols@ } };)~~~");
  1250. return name;
  1251. };
  1252. auto append_calendars = [&](String name, auto const& calendars) {
  1253. auto format_name = [&](StringView calendar_key, StringView type) {
  1254. return String::formatted("{}_{}_{}", name, calendar_key, type);
  1255. };
  1256. Vector<String> symbols_names;
  1257. for (auto const& calendar_key : locale_data.calendars) {
  1258. auto const& calendar = calendars.find(calendar_key)->value;
  1259. append_pattern_list(format_name(calendar_key, "formats"sv), s_calendar_pattern_index_type, calendar.available_formats);
  1260. append_pattern_list(format_name(calendar_key, "range_formats"sv), s_calendar_range_pattern_index_type, calendar.range_formats);
  1261. append_pattern_list(format_name(calendar_key, "range12_formats"sv), s_calendar_range_pattern_index_type, calendar.range12_formats);
  1262. generator.append("\n");
  1263. auto symbols_name = format_name(calendar_key, "symbols"sv);
  1264. for (auto const& symbol_key : locale_data.symbols) {
  1265. auto const& symbols = calendar.symbols.find(symbol_key)->value;
  1266. auto name = append_calendar_symbols(symbols_name, symbol_key, symbols);
  1267. symbols_names.append(name);
  1268. }
  1269. generator.append("\n");
  1270. }
  1271. generator.set("name", name);
  1272. generator.set("size", String::number(calendars.size()));
  1273. generator.append(R"~~~(
  1274. static constexpr Array<CalendarData, @size@> @name@ { {)~~~");
  1275. for (auto const& calendar_key : locale_data.calendars) {
  1276. auto const& calendar = calendars.find(calendar_key)->value;
  1277. generator.set("calendar"sv, String::number(calendar.calendar));
  1278. generator.set("date_formats"sv, String::number(calendar.date_formats));
  1279. generator.set("time_formats"sv, String::number(calendar.time_formats));
  1280. generator.set("date_time_formats"sv, String::number(calendar.date_time_formats));
  1281. generator.set("formats", format_name(calendar_key, "formats"sv));
  1282. generator.set("default_range_format", String::number(calendar.default_range_format));
  1283. generator.set("range_formats", format_name(calendar_key, "range_formats"sv));
  1284. generator.set("range12_formats", format_name(calendar_key, "range12_formats"sv));
  1285. generator.append(R"~~~(
  1286. { @calendar@, )~~~");
  1287. generator.append("@date_formats@, @time_formats@, @date_time_formats@, @formats@.span(), ");
  1288. generator.append("@default_range_format@, @range_formats@.span(), @range12_formats@.span(), {");
  1289. bool first = true;
  1290. for (auto const& symbols_name : symbols_names) {
  1291. generator.append(first ? " " : ", ");
  1292. generator.append(String::formatted("{}.span()", symbols_name));
  1293. first = false;
  1294. }
  1295. generator.append(" } },");
  1296. }
  1297. generator.append(R"~~~(
  1298. } };
  1299. )~~~");
  1300. };
  1301. auto append_time_zones = [&](String name, auto const& time_zones) {
  1302. generator.set("name", name);
  1303. generator.set("size", String::number(locale_data.time_zones.size()));
  1304. generator.append(R"~~~(
  1305. static constexpr Array<TimeZoneData, @size@> @name@ { {)~~~");
  1306. constexpr size_t max_values_per_row = 20;
  1307. size_t values_in_current_row = 0;
  1308. for (auto const& time_zone_key : locale_data.time_zones) {
  1309. auto time_zone = time_zones.find(time_zone_key);
  1310. if (values_in_current_row++ > 0)
  1311. generator.append(" ");
  1312. if (time_zone == time_zones.end()) {
  1313. generator.append("{},");
  1314. } else {
  1315. generator.set("time_zone", String::number(time_zone->value.time_zone));
  1316. generator.set("long_name", String::number(time_zone->value.long_name));
  1317. generator.set("short_name", String::number(time_zone->value.short_name));
  1318. generator.append("{ @time_zone@, @long_name@, @short_name@ },");
  1319. }
  1320. if (values_in_current_row == max_values_per_row) {
  1321. values_in_current_row = 0;
  1322. generator.append("\n ");
  1323. }
  1324. }
  1325. generator.append(R"~~~(
  1326. } };
  1327. )~~~");
  1328. };
  1329. auto append_day_periods = [&](String name, auto const& day_periods) {
  1330. generator.set("name", move(name));
  1331. generator.set("size", String::number(day_periods.size()));
  1332. generator.append(R"~~~(
  1333. static constexpr Array<DayPeriodData, @size@> @name@ { {)~~~");
  1334. for (auto const& day_period : day_periods) {
  1335. generator.set("day_period", String::number(static_cast<u8>(day_period.day_period)));
  1336. generator.set("begin", String::number(day_period.begin));
  1337. generator.set("end", String::number(day_period.end));
  1338. generator.append(" { @day_period@, @begin@, @end@ },");
  1339. }
  1340. generator.append(" } };");
  1341. };
  1342. auto append_hour_cycles = [&](String name, auto const& hour_cycle_region) {
  1343. auto const& hour_cycles = locale_data.hour_cycles.find(hour_cycle_region)->value;
  1344. generator.set("name", name);
  1345. generator.set("size", String::number(hour_cycles.size()));
  1346. generator.append(R"~~~(
  1347. static constexpr Array<u8, @size@> @name@ { { )~~~");
  1348. for (auto hour_cycle : hour_cycles) {
  1349. generator.set("hour_cycle", String::number(static_cast<u8>(hour_cycle)));
  1350. generator.append("@hour_cycle@, ");
  1351. }
  1352. generator.append(" } };");
  1353. };
  1354. generate_mapping(generator, locale_data.locales, "CalendarData"sv, "s_calendars"sv, "s_calendars_{}", [&](auto const& name, auto const& value) { append_calendars(name, value.calendars); });
  1355. 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); });
  1356. 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); });
  1357. 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); });
  1358. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1359. HashValueMap<String> hashes;
  1360. hashes.ensure_capacity(values.size());
  1361. for (auto const& value : values)
  1362. hashes.set(value.hash(), format_identifier(enum_title, value));
  1363. for (auto const& alias : aliases)
  1364. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1365. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1366. };
  1367. append_from_string("Calendar"sv, "calendar"sv, locale_data.calendars, locale_data.calendar_aliases);
  1368. append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, locale_data.hour_cycle_regions);
  1369. append_from_string("TimeZone"sv, "time_zone"sv, locale_data.time_zones);
  1370. generator.append(R"~~~(
  1371. Vector<Unicode::HourCycle> get_regional_hour_cycles(StringView region)
  1372. {
  1373. auto region_value = hour_cycle_region_from_string(region);
  1374. if (!region_value.has_value())
  1375. return {};
  1376. auto region_index = to_underlying(*region_value);
  1377. auto const& regional_hour_cycles = s_hour_cycles.at(region_index);
  1378. Vector<Unicode::HourCycle> hour_cycles;
  1379. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1380. for (auto hour_cycle : regional_hour_cycles)
  1381. hour_cycles.unchecked_append(static_cast<Unicode::HourCycle>(hour_cycle));
  1382. return hour_cycles;
  1383. }
  1384. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1385. {
  1386. auto locale_value = locale_from_string(locale);
  1387. if (!locale_value.has_value())
  1388. return nullptr;
  1389. auto calendar_value = calendar_from_string(calendar);
  1390. if (!calendar_value.has_value())
  1391. return nullptr;
  1392. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1393. auto calendar_index = to_underlying(*calendar_value);
  1394. auto const& calendars = s_calendars.at(locale_index);
  1395. return &calendars[calendar_index];
  1396. }
  1397. Optional<Unicode::CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1398. {
  1399. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1400. auto const& formats = s_calendar_formats.at(data->date_formats);
  1401. return formats.to_unicode_calendar_format();
  1402. }
  1403. return {};
  1404. }
  1405. Optional<Unicode::CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1406. {
  1407. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1408. auto const& formats = s_calendar_formats.at(data->time_formats);
  1409. return formats.to_unicode_calendar_format();
  1410. }
  1411. return {};
  1412. }
  1413. Optional<Unicode::CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1414. {
  1415. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1416. auto const& formats = s_calendar_formats.at(data->date_time_formats);
  1417. return formats.to_unicode_calendar_format();
  1418. }
  1419. return {};
  1420. }
  1421. Vector<Unicode::CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1422. {
  1423. Vector<Unicode::CalendarPattern> result {};
  1424. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1425. result.ensure_capacity(data->available_formats.size());
  1426. for (auto const& format : data->available_formats)
  1427. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1428. }
  1429. return result;
  1430. }
  1431. Optional<Unicode::CalendarRangePattern> get_calendar_default_range_format(StringView locale, StringView calendar)
  1432. {
  1433. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1434. auto const& pattern = s_calendar_range_patterns[data->default_range_format];
  1435. return pattern.to_unicode_calendar_range_pattern();
  1436. }
  1437. return {};
  1438. }
  1439. Vector<Unicode::CalendarRangePattern> get_calendar_range_formats(StringView locale, StringView calendar, StringView skeleton) {
  1440. Vector<Unicode::CalendarRangePattern> result {};
  1441. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1442. for (auto format : data->range_formats) {
  1443. auto const& pattern = s_calendar_range_patterns[format];
  1444. if (skeleton == s_string_list[pattern.skeleton])
  1445. result.append(pattern.to_unicode_calendar_range_pattern());
  1446. }
  1447. }
  1448. return result;
  1449. }
  1450. Vector<Unicode::CalendarRangePattern> get_calendar_range12_formats(StringView locale, StringView calendar, StringView skeleton) {
  1451. Vector<Unicode::CalendarRangePattern> result {};
  1452. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1453. for (auto format : data->range12_formats) {
  1454. auto const& pattern = s_calendar_range_patterns[format];
  1455. if (skeleton == s_string_list[pattern.skeleton])
  1456. result.append(pattern.to_unicode_calendar_range_pattern());
  1457. }
  1458. }
  1459. return result;
  1460. }
  1461. static CalendarSymbols find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1462. {
  1463. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1464. auto symbol_index = to_underlying(symbol);
  1465. auto style_index = to_underlying(style);
  1466. auto symbols = data->symbols.at(symbol_index);
  1467. VERIFY(style_index < symbols.size());
  1468. auto symbol_list_index = symbols.at(style_index);
  1469. return s_symbol_lists.at(symbol_list_index);
  1470. }
  1471. return {};
  1472. }
  1473. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Era value)
  1474. {
  1475. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1476. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1477. return s_string_list[symbols.at(value_index)];
  1478. return {};
  1479. }
  1480. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Month value)
  1481. {
  1482. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1483. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1484. return s_string_list[symbols.at(value_index)];
  1485. return {};
  1486. }
  1487. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Weekday value)
  1488. {
  1489. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1490. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1491. return s_string_list[symbols.at(value_index)];
  1492. return {};
  1493. }
  1494. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::DayPeriod value)
  1495. {
  1496. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1497. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1498. return s_string_list[symbols.at(value_index)];
  1499. return {};
  1500. }
  1501. Optional<StringView> get_calendar_day_period_symbol_for_hour(StringView locale, StringView calendar, CalendarPatternStyle style, u8 hour)
  1502. {
  1503. auto locale_value = locale_from_string(locale);
  1504. if (!locale_value.has_value())
  1505. return {};
  1506. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1507. auto day_periods = s_day_periods[locale_index];
  1508. for (auto const& day_period : day_periods) {
  1509. if ((day_period.begin <= hour) && (hour < day_period.end)) {
  1510. auto period = static_cast<Unicode::DayPeriod>(day_period.day_period);
  1511. return Detail::get_calendar_day_period_symbol(locale, calendar, style, period);
  1512. }
  1513. }
  1514. // Fallback to fixed periods if the locale does not have flexible day periods.
  1515. // TR-35 states that the meaning of AM and PM does not change with locale.
  1516. if (hour < 12)
  1517. return Detail::get_calendar_day_period_symbol(locale, calendar, style, Unicode::DayPeriod::AM);
  1518. return Detail::get_calendar_day_period_symbol(locale, calendar, style, Unicode::DayPeriod::PM);
  1519. }
  1520. static TimeZoneData const* find_time_zone_data(StringView locale, StringView time_zone)
  1521. {
  1522. auto locale_value = locale_from_string(locale);
  1523. if (!locale_value.has_value())
  1524. return nullptr;
  1525. auto time_zone_value = time_zone_from_string(time_zone);
  1526. if (!time_zone_value.has_value())
  1527. return nullptr;
  1528. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1529. auto time_zone_index = to_underlying(*time_zone_value);
  1530. auto const& time_zones = s_time_zones.at(locale_index);
  1531. return &time_zones[time_zone_index];
  1532. }
  1533. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style)
  1534. {
  1535. if (auto const* data = find_time_zone_data(locale, time_zone); data != nullptr) {
  1536. @string_index_type@ time_zone_index = 0;
  1537. switch (style) {
  1538. case CalendarPatternStyle::Long:
  1539. time_zone_index = data->long_name;
  1540. break;
  1541. case CalendarPatternStyle::Short:
  1542. time_zone_index = data->short_name;
  1543. break;
  1544. default:
  1545. VERIFY_NOT_REACHED();
  1546. }
  1547. if (time_zone_index != 0)
  1548. return s_string_list[time_zone_index];
  1549. }
  1550. return {};
  1551. }
  1552. }
  1553. )~~~");
  1554. VERIFY(file.write(generator.as_string_view()));
  1555. }
  1556. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1557. {
  1558. StringView generated_header_path;
  1559. StringView generated_implementation_path;
  1560. StringView core_path;
  1561. StringView dates_path;
  1562. Core::ArgsParser args_parser;
  1563. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1564. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1565. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1566. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1567. args_parser.parse(arguments);
  1568. auto open_file = [&](StringView path) -> ErrorOr<NonnullRefPtr<Core::File>> {
  1569. if (path.is_empty()) {
  1570. args_parser.print_usage(stderr, arguments.argv[0]);
  1571. return Error::from_string_literal("Must provide all command line options"sv);
  1572. }
  1573. return Core::File::open(path, Core::OpenMode::ReadWrite);
  1574. };
  1575. auto generated_header_file = TRY(open_file(generated_header_path));
  1576. auto generated_implementation_file = TRY(open_file(generated_implementation_path));
  1577. UnicodeLocaleData locale_data;
  1578. TRY(parse_all_locales(core_path, dates_path, locale_data));
  1579. generate_unicode_locale_header(generated_header_file, locale_data);
  1580. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  1581. return 0;
  1582. }