GenerateUnicodeDateTimeFormat.cpp 59 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478
  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. struct CalendarPattern : public Unicode::CalendarPattern {
  31. bool contains_only_date_fields() const
  32. {
  33. return !day_period.has_value() && !hour.has_value() && !minute.has_value() && !second.has_value() && !fractional_second_digits.has_value() && !time_zone_name.has_value();
  34. }
  35. bool contains_only_time_fields() const
  36. {
  37. return !weekday.has_value() && !era.has_value() && !year.has_value() && !month.has_value() && !day.has_value();
  38. }
  39. unsigned hash() const
  40. {
  41. auto hash = pair_int_hash(pattern_index, pattern12_index);
  42. auto hash_field = [&](auto const& field) {
  43. if (field.has_value())
  44. hash = pair_int_hash(hash, static_cast<u8>(*field));
  45. else
  46. hash = pair_int_hash(hash, -1);
  47. };
  48. hash_field(era);
  49. hash_field(year);
  50. hash_field(month);
  51. hash_field(weekday);
  52. hash_field(day);
  53. hash_field(day_period);
  54. hash_field(hour);
  55. hash_field(minute);
  56. hash_field(second);
  57. hash_field(fractional_second_digits);
  58. hash_field(time_zone_name);
  59. return hash;
  60. }
  61. bool operator==(CalendarPattern const& other) const
  62. {
  63. return (pattern_index == other.pattern_index)
  64. && (pattern12_index == other.pattern12_index)
  65. && (era == other.era)
  66. && (year == other.year)
  67. && (month == other.month)
  68. && (weekday == other.weekday)
  69. && (day == other.day)
  70. && (day_period == other.day_period)
  71. && (hour == other.hour)
  72. && (minute == other.minute)
  73. && (second == other.second)
  74. && (fractional_second_digits == other.fractional_second_digits)
  75. && (time_zone_name == other.time_zone_name);
  76. }
  77. StringIndexType pattern_index { 0 };
  78. StringIndexType pattern12_index { 0 };
  79. };
  80. template<>
  81. struct AK::Formatter<CalendarPattern> : Formatter<FormatString> {
  82. ErrorOr<void> format(FormatBuilder& builder, CalendarPattern const& pattern)
  83. {
  84. auto field_to_i8 = [](auto const& field) -> i8 {
  85. if (!field.has_value())
  86. return -1;
  87. return static_cast<i8>(*field);
  88. };
  89. return Formatter<FormatString>::format(builder,
  90. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}",
  91. pattern.pattern_index,
  92. pattern.pattern12_index,
  93. field_to_i8(pattern.era),
  94. field_to_i8(pattern.year),
  95. field_to_i8(pattern.month),
  96. field_to_i8(pattern.weekday),
  97. field_to_i8(pattern.day),
  98. field_to_i8(pattern.day_period),
  99. field_to_i8(pattern.hour),
  100. field_to_i8(pattern.minute),
  101. field_to_i8(pattern.second),
  102. field_to_i8(pattern.fractional_second_digits),
  103. field_to_i8(pattern.time_zone_name));
  104. }
  105. };
  106. template<>
  107. struct AK::Traits<CalendarPattern> : public GenericTraits<CalendarPattern> {
  108. static unsigned hash(CalendarPattern const& c) { return c.hash(); }
  109. };
  110. struct CalendarFormat {
  111. CalendarPatternIndexType full_format { 0 };
  112. CalendarPatternIndexType long_format { 0 };
  113. CalendarPatternIndexType medium_format { 0 };
  114. CalendarPatternIndexType short_format { 0 };
  115. };
  116. struct CalendarSymbols {
  117. Vector<StringIndexType> narrow_symbols {};
  118. Vector<StringIndexType> short_symbols {};
  119. Vector<StringIndexType> long_symbols {};
  120. };
  121. struct Calendar {
  122. StringIndexType calendar { 0 };
  123. CalendarFormat date_formats {};
  124. CalendarFormat time_formats {};
  125. CalendarFormat date_time_formats {};
  126. Vector<CalendarPatternIndexType> available_formats {};
  127. HashMap<String, CalendarSymbols> symbols {};
  128. };
  129. struct TimeZone {
  130. StringIndexType time_zone { 0 };
  131. StringIndexType long_name { 0 };
  132. StringIndexType short_name { 0 };
  133. };
  134. struct Locale {
  135. HashMap<String, Calendar> calendars;
  136. HashMap<String, TimeZone> time_zones;
  137. };
  138. struct UnicodeLocaleData {
  139. UniqueStringStorage<StringIndexType> unique_strings;
  140. UniqueStorage<CalendarPattern, CalendarPatternIndexType> unique_patterns;
  141. HashMap<String, Locale> locales;
  142. HashMap<String, Vector<Unicode::HourCycle>> hour_cycles;
  143. Vector<String> hour_cycle_regions;
  144. HashMap<String, StringIndexType> meta_zones;
  145. Vector<String> time_zones;
  146. Vector<String> calendars;
  147. Vector<Alias> calendar_aliases {
  148. // FIXME: Aliases should come from BCP47. See: https://unicode-org.atlassian.net/browse/CLDR-15158
  149. { "gregorian"sv, "gregory"sv },
  150. };
  151. Vector<String> symbols;
  152. };
  153. static ErrorOr<void> parse_hour_cycles(String core_path, UnicodeLocaleData& locale_data)
  154. {
  155. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  156. LexicalPath time_data_path(move(core_path));
  157. time_data_path = time_data_path.append("supplemental"sv);
  158. time_data_path = time_data_path.append("timeData.json"sv);
  159. auto time_data_file = TRY(Core::File::open(time_data_path.string(), Core::OpenMode::ReadOnly));
  160. auto time_data = TRY(JsonValue::from_string(time_data_file->read_all()));
  161. auto const& supplemental_object = time_data.as_object().get("supplemental"sv);
  162. auto const& time_data_object = supplemental_object.as_object().get("timeData"sv);
  163. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Unicode::HourCycle> {
  164. if (hour_cycle == "h"sv)
  165. return Unicode::HourCycle::H12;
  166. if (hour_cycle == "H"sv)
  167. return Unicode::HourCycle::H23;
  168. if (hour_cycle == "K"sv)
  169. return Unicode::HourCycle::H11;
  170. if (hour_cycle == "k"sv)
  171. return Unicode::HourCycle::H24;
  172. return {};
  173. };
  174. time_data_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  175. auto allowed_hour_cycles_string = value.as_object().get("_allowed"sv).as_string();
  176. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  177. Vector<Unicode::HourCycle> hour_cycles;
  178. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  179. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  180. hour_cycles.append(*hour_cycle);
  181. }
  182. locale_data.hour_cycles.set(key, move(hour_cycles));
  183. if (!locale_data.hour_cycle_regions.contains_slow(key))
  184. locale_data.hour_cycle_regions.append(key);
  185. });
  186. return {};
  187. };
  188. static ErrorOr<void> parse_meta_zones(String core_path, UnicodeLocaleData& locale_data)
  189. {
  190. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  191. LexicalPath meta_zone_path(move(core_path));
  192. meta_zone_path = meta_zone_path.append("supplemental"sv);
  193. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  194. auto meta_zone_file = TRY(Core::File::open(meta_zone_path.string(), Core::OpenMode::ReadOnly));
  195. auto meta_zone = TRY(JsonValue::from_string(meta_zone_file->read_all()));
  196. auto const& supplemental_object = meta_zone.as_object().get("supplemental"sv);
  197. auto const& meta_zone_object = supplemental_object.as_object().get("metaZones"sv);
  198. auto const& meta_zone_array = meta_zone_object.as_object().get("metazones"sv);
  199. meta_zone_array.as_array().for_each([&](JsonValue const& value) {
  200. auto const& mapping = value.as_object().get("mapZone"sv);
  201. auto const& meta_zone = mapping.as_object().get("_other"sv);
  202. auto const& golden_zone = mapping.as_object().get("_type"sv);
  203. auto golden_zone_index = locale_data.unique_strings.ensure(golden_zone.as_string());
  204. locale_data.meta_zones.set(meta_zone.as_string(), golden_zone_index);
  205. });
  206. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  207. locale_data.meta_zones.set("UTC"sv, locale_data.unique_strings.ensure("UTC"sv));
  208. return {};
  209. };
  210. static constexpr auto is_char(char ch)
  211. {
  212. return [ch](auto c) { return c == ch; };
  213. }
  214. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  215. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  216. // extra spaces around the day period. Some example expected removals:
  217. //
  218. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  219. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  220. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  221. static String remove_period_from_pattern(String pattern)
  222. {
  223. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  224. auto index = pattern.find(remove);
  225. if (!index.has_value())
  226. continue;
  227. constexpr u32 space = ' ';
  228. constexpr u32 open = '{';
  229. constexpr u32 close = '}';
  230. Utf8View utf8_pattern { pattern };
  231. Optional<u32> before_removal;
  232. Optional<u32> after_removal;
  233. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  234. before_removal = *it;
  235. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  236. after_removal = *it;
  237. if ((before_removal == space) && (after_removal != open)) {
  238. pattern = String::formatted("{}{}",
  239. pattern.substring_view(0, *index - 1),
  240. pattern.substring_view(*index + remove.length()));
  241. } else if ((after_removal == space) && (before_removal != close)) {
  242. pattern = String::formatted("{}{}",
  243. pattern.substring_view(0, *index),
  244. pattern.substring_view(*index + remove.length() + 1));
  245. } else {
  246. pattern = String::formatted("{}{}",
  247. pattern.substring_view(0, *index),
  248. pattern.substring_view(*index + remove.length()));
  249. }
  250. }
  251. return pattern;
  252. }
  253. static Optional<CalendarPatternIndexType> parse_date_time_pattern(String pattern, UnicodeLocaleData& locale_data)
  254. {
  255. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  256. using Unicode::CalendarPatternStyle;
  257. CalendarPattern format {};
  258. GenericLexer lexer { pattern };
  259. StringBuilder builder;
  260. bool hour12 { false };
  261. while (!lexer.is_eof()) {
  262. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  263. // processing (this just avoids conflicts with the patterns below).
  264. if (lexer.next_is(is_quote)) {
  265. builder.append(lexer.consume_quoted_string());
  266. continue;
  267. }
  268. auto starting_char = lexer.peek();
  269. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  270. // Era
  271. if (all_of(segment, is_char('G'))) {
  272. builder.append("{era}");
  273. if (segment.length() <= 3)
  274. format.era = CalendarPatternStyle::Short;
  275. else if (segment.length() == 4)
  276. format.era = CalendarPatternStyle::Long;
  277. else
  278. format.era = CalendarPatternStyle::Narrow;
  279. }
  280. // Year
  281. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  282. builder.append("{year}");
  283. if (segment.length() == 2)
  284. format.year = CalendarPatternStyle::TwoDigit;
  285. else
  286. format.year = CalendarPatternStyle::Numeric;
  287. }
  288. // Quarter
  289. else if (all_of(segment, is_any_of("qQ"sv))) {
  290. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  291. return {};
  292. }
  293. // Month
  294. else if (all_of(segment, is_any_of("ML"sv))) {
  295. builder.append("{month}");
  296. if (segment.length() == 1)
  297. format.month = CalendarPatternStyle::Numeric;
  298. else if (segment.length() == 2)
  299. format.month = CalendarPatternStyle::TwoDigit;
  300. else if (segment.length() == 3)
  301. format.month = CalendarPatternStyle::Short;
  302. else if (segment.length() == 4)
  303. format.month = CalendarPatternStyle::Long;
  304. else if (segment.length() == 5)
  305. format.month = CalendarPatternStyle::Narrow;
  306. } else if (all_of(segment, is_char('l'))) {
  307. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  308. return {};
  309. }
  310. // Week
  311. else if (all_of(segment, is_any_of("wW"sv))) {
  312. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  313. return {};
  314. }
  315. // Day
  316. else if (all_of(segment, is_char('d'))) {
  317. builder.append("{day}");
  318. if (segment.length() == 1)
  319. format.day = CalendarPatternStyle::Numeric;
  320. else
  321. format.day = CalendarPatternStyle::TwoDigit;
  322. } else if (all_of(segment, is_any_of("DFg"sv))) {
  323. builder.append("{day}");
  324. format.day = CalendarPatternStyle::Numeric;
  325. }
  326. // Weekday
  327. else if (all_of(segment, is_char('E'))) {
  328. builder.append("{weekday}");
  329. if (segment.length() == 4)
  330. format.weekday = CalendarPatternStyle::Long;
  331. else if (segment.length() == 5)
  332. format.weekday = CalendarPatternStyle::Narrow;
  333. else
  334. format.weekday = CalendarPatternStyle::Short;
  335. } else if (all_of(segment, is_any_of("ec"sv))) {
  336. builder.append("{weekday}");
  337. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  338. // pattern styles are not supported by Intl.DateTimeFormat.
  339. if (segment.length() <= 2)
  340. return {};
  341. if (segment.length() == 4)
  342. format.weekday = CalendarPatternStyle::Long;
  343. else if (segment.length() == 5)
  344. format.weekday = CalendarPatternStyle::Narrow;
  345. else
  346. format.weekday = CalendarPatternStyle::Short;
  347. }
  348. // Period
  349. else if (all_of(segment, is_any_of("ab"sv))) {
  350. builder.append("{ampm}");
  351. hour12 = true;
  352. } else if (all_of(segment, is_char('B'))) {
  353. builder.append("{dayPeriod}");
  354. hour12 = true;
  355. if (segment.length() == 4)
  356. format.day_period = CalendarPatternStyle::Long;
  357. else if (segment.length() == 5)
  358. format.day_period = CalendarPatternStyle::Narrow;
  359. else
  360. format.day_period = CalendarPatternStyle::Short;
  361. }
  362. // Hour
  363. else if (all_of(segment, is_any_of("hHKk"sv))) {
  364. builder.append("{hour}");
  365. if ((segment[0] == 'h') || (segment[0] == 'K'))
  366. hour12 = true;
  367. if (segment.length() == 1)
  368. format.hour = CalendarPatternStyle::Numeric;
  369. else
  370. format.hour = CalendarPatternStyle::TwoDigit;
  371. } else if (all_of(segment, is_any_of("jJC"sv))) {
  372. // TR-35 indicates these should not be used.
  373. return {};
  374. }
  375. // Minute
  376. else if (all_of(segment, is_char('m'))) {
  377. builder.append("{minute}");
  378. if (segment.length() == 1)
  379. format.minute = CalendarPatternStyle::Numeric;
  380. else
  381. format.minute = CalendarPatternStyle::TwoDigit;
  382. }
  383. // Second
  384. else if (all_of(segment, is_char('s'))) {
  385. builder.append("{second}");
  386. if (segment.length() == 1)
  387. format.second = CalendarPatternStyle::Numeric;
  388. else
  389. format.second = CalendarPatternStyle::TwoDigit;
  390. } else if (all_of(segment, is_char('S'))) {
  391. builder.append("{fractionalSecondDigits}");
  392. VERIFY(segment.length() <= 3);
  393. format.fractional_second_digits = static_cast<u8>(segment.length());
  394. } else if (all_of(segment, is_char('A'))) {
  395. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  396. return {};
  397. }
  398. // Zone
  399. else if (all_of(segment, is_any_of("zZOvVXx"))) {
  400. builder.append("{timeZoneName}");
  401. if (segment.length() < 4)
  402. format.time_zone_name = CalendarPatternStyle::Short;
  403. else
  404. format.time_zone_name = CalendarPatternStyle::Long;
  405. }
  406. // Non-patterns
  407. else {
  408. builder.append(segment);
  409. }
  410. }
  411. pattern = builder.build();
  412. if (hour12) {
  413. auto pattern_without_period = remove_period_from_pattern(pattern);
  414. format.pattern_index = locale_data.unique_strings.ensure(move(pattern_without_period));
  415. format.pattern12_index = locale_data.unique_strings.ensure(move(pattern));
  416. } else {
  417. format.pattern_index = locale_data.unique_strings.ensure(move(pattern));
  418. }
  419. return locale_data.unique_patterns.ensure(move(format));
  420. }
  421. static void generate_missing_patterns(Calendar& calendar, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, UnicodeLocaleData& locale_data)
  422. {
  423. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  424. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  425. auto pattern = locale_data.unique_strings.get(format);
  426. auto time_pattern = locale_data.unique_strings.get(time_format);
  427. auto date_pattern = locale_data.unique_strings.get(date_format);
  428. auto new_pattern = pattern.replace("{0}", time_pattern).replace("{1}", date_pattern);
  429. return locale_data.unique_strings.ensure(move(new_pattern));
  430. };
  431. auto inject_fractional_second_digits = [&](auto format) {
  432. auto pattern = locale_data.unique_strings.get(format);
  433. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv);
  434. return locale_data.unique_strings.ensure(move(new_pattern));
  435. };
  436. auto append_if_unique = [&](auto format) {
  437. auto format_index = locale_data.unique_patterns.ensure(move(format));
  438. if (!calendar.available_formats.contains_slow(format_index))
  439. calendar.available_formats.append(format_index);
  440. };
  441. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  442. for (auto const& format : date_formats)
  443. append_if_unique(format);
  444. for (auto const& format : time_formats) {
  445. append_if_unique(format);
  446. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  447. auto new_format = format;
  448. new_format.fractional_second_digits = 2;
  449. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  450. if (new_format.pattern12_index != 0)
  451. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  452. time_formats_with_fractional_second_digits.append(new_format);
  453. append_if_unique(move(new_format));
  454. }
  455. }
  456. time_formats.extend(move(time_formats_with_fractional_second_digits));
  457. for (auto const& date_format : date_formats) {
  458. CalendarPatternIndexType date_time_format_index = 0;
  459. if (date_format.month == Unicode::CalendarPatternStyle::Long) {
  460. if (date_format.weekday.has_value())
  461. date_time_format_index = calendar.date_time_formats.full_format;
  462. else
  463. date_time_format_index = calendar.date_time_formats.long_format;
  464. } else if (date_format.month == Unicode::CalendarPatternStyle::Short) {
  465. date_time_format_index = calendar.date_time_formats.medium_format;
  466. } else {
  467. date_time_format_index = calendar.date_time_formats.short_format;
  468. }
  469. for (auto const& time_format : time_formats) {
  470. auto format = locale_data.unique_patterns.get(date_time_format_index);
  471. if (time_format.pattern12_index != 0)
  472. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  473. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  474. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  475. if (date_field.has_value())
  476. field = date_field;
  477. });
  478. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  479. if (time_field.has_value())
  480. field = time_field;
  481. });
  482. append_if_unique(move(format));
  483. }
  484. }
  485. }
  486. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, UnicodeLocaleData& locale_data)
  487. {
  488. auto ensure_symbols = [&](auto name, size_t size) -> CalendarSymbols& {
  489. if (!locale_data.symbols.contains_slow(name))
  490. locale_data.symbols.append(name);
  491. return calendar.symbols.ensure(name, [&]() {
  492. CalendarSymbols symbols {};
  493. symbols.narrow_symbols.resize(size);
  494. symbols.short_symbols.resize(size);
  495. symbols.long_symbols.resize(size);
  496. return symbols;
  497. });
  498. };
  499. auto parse_era_symbols = [&](auto const& symbols_object) {
  500. auto const& narrow_symbols = symbols_object.get("eraNarrow"sv).as_object();
  501. auto const& short_symbols = symbols_object.get("eraAbbr"sv).as_object();
  502. auto const& long_symbols = symbols_object.get("eraNames"sv).as_object();
  503. auto& symbols = ensure_symbols("era"sv, 2);
  504. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  505. if (auto key_index = key.to_uint(); key_index.has_value())
  506. symbols[*key_index] = locale_data.unique_strings.ensure(move(symbol));
  507. };
  508. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  509. append_symbol(symbols.narrow_symbols, key, value.as_string());
  510. });
  511. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  512. append_symbol(symbols.short_symbols, key, value.as_string());
  513. });
  514. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  515. append_symbol(symbols.long_symbols, key, value.as_string());
  516. });
  517. };
  518. auto parse_month_symbols = [&](auto const& symbols_object) {
  519. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  520. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  521. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  522. auto& month_symbols = ensure_symbols("month"sv, 12);
  523. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  524. auto key_index = key.to_uint().value() - 1;
  525. symbols[key_index] = locale_data.unique_strings.ensure(move(symbol));
  526. };
  527. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  528. append_symbol(month_symbols.narrow_symbols, key, value.as_string());
  529. });
  530. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  531. append_symbol(month_symbols.short_symbols, key, value.as_string());
  532. });
  533. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  534. append_symbol(month_symbols.long_symbols, key, value.as_string());
  535. });
  536. };
  537. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  538. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  539. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  540. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  541. auto& weekday_symbols = ensure_symbols("weekday"sv, 7);
  542. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  543. if (key == "sun"sv)
  544. symbols[to_underlying(Unicode::Weekday::Sunday)] = locale_data.unique_strings.ensure(move(symbol));
  545. else if (key == "mon"sv)
  546. symbols[to_underlying(Unicode::Weekday::Monday)] = locale_data.unique_strings.ensure(move(symbol));
  547. else if (key == "tue"sv)
  548. symbols[to_underlying(Unicode::Weekday::Tuesday)] = locale_data.unique_strings.ensure(move(symbol));
  549. else if (key == "wed"sv)
  550. symbols[to_underlying(Unicode::Weekday::Wednesday)] = locale_data.unique_strings.ensure(move(symbol));
  551. else if (key == "thu"sv)
  552. symbols[to_underlying(Unicode::Weekday::Thursday)] = locale_data.unique_strings.ensure(move(symbol));
  553. else if (key == "fri"sv)
  554. symbols[to_underlying(Unicode::Weekday::Friday)] = locale_data.unique_strings.ensure(move(symbol));
  555. else if (key == "sat"sv)
  556. symbols[to_underlying(Unicode::Weekday::Saturday)] = locale_data.unique_strings.ensure(move(symbol));
  557. };
  558. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  559. append_symbol(weekday_symbols.narrow_symbols, key, value.as_string());
  560. });
  561. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  562. append_symbol(weekday_symbols.short_symbols, key, value.as_string());
  563. });
  564. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  565. append_symbol(weekday_symbols.long_symbols, key, value.as_string());
  566. });
  567. };
  568. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  569. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  570. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  571. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  572. auto& day_period_symbols = ensure_symbols("dayPeriod"sv, 6);
  573. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  574. if (key == "am"sv)
  575. symbols[to_underlying(Unicode::DayPeriod::AM)] = locale_data.unique_strings.ensure(move(symbol));
  576. else if (key == "pm"sv)
  577. symbols[to_underlying(Unicode::DayPeriod::PM)] = locale_data.unique_strings.ensure(move(symbol));
  578. else if (key == "morning1"sv)
  579. symbols[to_underlying(Unicode::DayPeriod::Morning)] = locale_data.unique_strings.ensure(move(symbol));
  580. else if (key == "afternoon1"sv)
  581. symbols[to_underlying(Unicode::DayPeriod::Afternoon)] = locale_data.unique_strings.ensure(move(symbol));
  582. else if (key == "evening1"sv)
  583. symbols[to_underlying(Unicode::DayPeriod::Evening)] = locale_data.unique_strings.ensure(move(symbol));
  584. else if (key == "night1"sv)
  585. symbols[to_underlying(Unicode::DayPeriod::Night)] = locale_data.unique_strings.ensure(move(symbol));
  586. };
  587. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  588. append_symbol(day_period_symbols.narrow_symbols, key, value.as_string());
  589. });
  590. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  591. append_symbol(day_period_symbols.short_symbols, key, value.as_string());
  592. });
  593. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  594. append_symbol(day_period_symbols.long_symbols, key, value.as_string());
  595. });
  596. };
  597. parse_era_symbols(calendar_object.get("eras"sv).as_object());
  598. parse_month_symbols(calendar_object.get("months"sv).as_object().get("format"sv).as_object());
  599. parse_weekday_symbols(calendar_object.get("days"sv).as_object().get("format"sv).as_object());
  600. parse_day_period_symbols(calendar_object.get("dayPeriods"sv).as_object().get("format"sv).as_object());
  601. }
  602. static ErrorOr<void> parse_calendars(String locale_calendars_path, UnicodeLocaleData& locale_data, Locale& locale)
  603. {
  604. LexicalPath calendars_path(move(locale_calendars_path));
  605. if (!calendars_path.basename().starts_with("ca-"sv))
  606. return {};
  607. auto calendars_file = TRY(Core::File::open(calendars_path.string(), Core::OpenMode::ReadOnly));
  608. auto calendars = TRY(JsonValue::from_string(calendars_file->read_all()));
  609. auto const& main_object = calendars.as_object().get("main"sv);
  610. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  611. auto const& dates_object = locale_object.as_object().get("dates"sv);
  612. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  613. auto ensure_calendar = [&](auto const& calendar) -> Calendar& {
  614. return locale.calendars.ensure(calendar, [&]() {
  615. auto calendar_index = locale_data.unique_strings.ensure(calendar);
  616. return Calendar { .calendar = calendar_index };
  617. });
  618. };
  619. auto parse_patterns = [&](auto& formats, auto const& patterns_object, Vector<CalendarPattern>* patterns) {
  620. auto parse_pattern = [&](auto name) {
  621. auto format = patterns_object.get(name);
  622. auto format_index = parse_date_time_pattern(format.as_string(), locale_data).value();
  623. if (patterns)
  624. patterns->append(locale_data.unique_patterns.get(format_index));
  625. return format_index;
  626. };
  627. formats.full_format = parse_pattern("full"sv);
  628. formats.long_format = parse_pattern("long"sv);
  629. formats.medium_format = parse_pattern("medium"sv);
  630. formats.short_format = parse_pattern("short"sv);
  631. };
  632. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  633. // The generic calendar is not a supported Unicode calendar key, so skip it:
  634. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  635. if (calendar_name == "generic"sv)
  636. return;
  637. auto& calendar = ensure_calendar(calendar_name);
  638. if (!locale_data.calendars.contains_slow(calendar_name))
  639. locale_data.calendars.append(calendar_name);
  640. Vector<CalendarPattern> date_formats;
  641. Vector<CalendarPattern> time_formats;
  642. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  643. parse_patterns(calendar.date_formats, date_formats_object.as_object(), &date_formats);
  644. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  645. parse_patterns(calendar.time_formats, time_formats_object.as_object(), &time_formats);
  646. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  647. parse_patterns(calendar.date_time_formats, date_time_formats_object.as_object(), nullptr);
  648. auto const& available_formats = date_time_formats_object.as_object().get("availableFormats"sv);
  649. available_formats.as_object().for_each_member([&](auto const&, JsonValue const& pattern) {
  650. auto pattern_index = parse_date_time_pattern(pattern.as_string(), locale_data);
  651. if (!pattern_index.has_value())
  652. return;
  653. auto const& format = locale_data.unique_patterns.get(*pattern_index);
  654. if (format.contains_only_date_fields())
  655. date_formats.append(format);
  656. else if (format.contains_only_time_fields())
  657. time_formats.append(format);
  658. if (!calendar.available_formats.contains_slow(*pattern_index))
  659. calendar.available_formats.append(*pattern_index);
  660. });
  661. generate_missing_patterns(calendar, move(date_formats), move(time_formats), locale_data);
  662. parse_calendar_symbols(calendar, value.as_object(), locale_data);
  663. });
  664. return {};
  665. }
  666. static ErrorOr<void> parse_time_zone_names(String locale_time_zone_names_path, UnicodeLocaleData& locale_data, Locale& locale)
  667. {
  668. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  669. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  670. auto time_zone_names_file = TRY(Core::File::open(time_zone_names_path.string(), Core::OpenMode::ReadOnly));
  671. auto time_zone_names = TRY(JsonValue::from_string(time_zone_names_file->read_all()));
  672. auto const& main_object = time_zone_names.as_object().get("main"sv);
  673. auto const& locale_object = main_object.as_object().get(time_zone_names_path.parent().basename());
  674. auto const& dates_object = locale_object.as_object().get("dates"sv);
  675. auto const& time_zone_names_object = dates_object.as_object().get("timeZoneNames"sv);
  676. auto const& meta_zone_object = time_zone_names_object.as_object().get("metazone"sv);
  677. if (meta_zone_object.is_null())
  678. return {};
  679. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object) -> Optional<StringIndexType> {
  680. auto const& names = meta_zone_object.get(type);
  681. if (!names.is_object())
  682. return {};
  683. auto const& daylight = names.as_object().get("daylight"sv);
  684. if (daylight.is_string())
  685. return locale_data.unique_strings.ensure(daylight.as_string());
  686. auto const& standard = names.as_object().get("standard"sv);
  687. if (standard.is_string())
  688. return locale_data.unique_strings.ensure(standard.as_string());
  689. return {};
  690. };
  691. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  692. auto golden_zone = locale_data.meta_zones.get(meta_zone).value();
  693. TimeZone time_zone { .time_zone = golden_zone };
  694. if (auto long_name = parse_name("long"sv, meta_zone_object); long_name.has_value())
  695. time_zone.long_name = long_name.value();
  696. if (auto short_name = parse_name("short"sv, meta_zone_object); short_name.has_value())
  697. time_zone.short_name = short_name.value();
  698. auto const& time_zone_name = locale_data.unique_strings.get(golden_zone);
  699. if (!locale_data.time_zones.contains_slow(time_zone_name))
  700. locale_data.time_zones.append(time_zone_name);
  701. locale.time_zones.set(time_zone_name, move(time_zone));
  702. };
  703. meta_zone_object.as_object().for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  704. parse_time_zone(meta_zone, value.as_object());
  705. });
  706. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  707. auto const& zone_object = time_zone_names_object.as_object().get("zone"sv);
  708. auto const& etc_object = zone_object.as_object().get("Etc"sv);
  709. auto const& utc_object = etc_object.as_object().get("UTC"sv);
  710. parse_time_zone("UTC"sv, utc_object.as_object());
  711. return {};
  712. }
  713. static ErrorOr<void> parse_all_locales(String core_path, String dates_path, UnicodeLocaleData& locale_data)
  714. {
  715. TRY(parse_hour_cycles(core_path, locale_data));
  716. TRY(parse_meta_zones(move(core_path), locale_data));
  717. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  718. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  719. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  720. StringBuilder builder;
  721. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  722. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  723. builder.appendff("-{}", script);
  724. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  725. builder.appendff("-{}", region);
  726. return builder.build();
  727. };
  728. while (dates_iterator.has_next()) {
  729. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  730. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  731. auto language = TRY(remove_variants_from_path(dates_path));
  732. auto& locale = locale_data.locales.ensure(language);
  733. while (calendars_iterator.has_next()) {
  734. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  735. TRY(parse_calendars(move(calendars_path), locale_data, locale));
  736. }
  737. TRY(parse_time_zone_names(move(dates_path), locale_data, locale));
  738. }
  739. return {};
  740. }
  741. static String format_identifier(StringView owner, String identifier)
  742. {
  743. identifier = identifier.replace("-"sv, "_"sv, true);
  744. identifier = identifier.replace("/"sv, "_"sv, true);
  745. if (all_of(identifier, is_ascii_digit))
  746. return String::formatted("{}_{}", owner[0], identifier);
  747. if (is_ascii_lower_alpha(identifier[0]))
  748. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  749. return identifier;
  750. }
  751. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  752. {
  753. StringBuilder builder;
  754. SourceGenerator generator { builder };
  755. generator.append(R"~~~(
  756. #pragma once
  757. #include <AK/Optional.h>
  758. #include <AK/StringView.h>
  759. #include <LibUnicode/Forward.h>
  760. namespace Unicode {
  761. )~~~");
  762. generate_enum(generator, format_identifier, "Calendar"sv, {}, locale_data.calendars, locale_data.calendar_aliases);
  763. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, locale_data.hour_cycle_regions);
  764. generate_enum(generator, format_identifier, "CalendarSymbol"sv, {}, locale_data.symbols);
  765. generate_enum(generator, format_identifier, "TimeZone"sv, {}, locale_data.time_zones);
  766. generator.append(R"~~~(
  767. namespace Detail {
  768. Optional<Calendar> calendar_from_string(StringView calendar);
  769. Optional<HourCycleRegion> hour_cycle_region_from_string(StringView hour_cycle_region);
  770. Vector<Unicode::HourCycle> get_regional_hour_cycles(StringView region);
  771. Optional<Unicode::CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar);
  772. Optional<Unicode::CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar);
  773. Optional<Unicode::CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar);
  774. Vector<Unicode::CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar);
  775. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Era value);
  776. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Month value);
  777. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Weekday value);
  778. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::DayPeriod value);
  779. Optional<TimeZone> time_zone_from_string(StringView time_zone);
  780. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style);
  781. }
  782. }
  783. )~~~");
  784. VERIFY(file.write(generator.as_string_view()));
  785. }
  786. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  787. {
  788. StringBuilder builder;
  789. SourceGenerator generator { builder };
  790. generator.set("string_index_type"sv, s_string_index_type);
  791. generator.set("calendar_pattern_index_type"sv, s_calendar_pattern_index_type);
  792. generator.set("calendar_symbols_size"sv, String::number(locale_data.symbols.size()));
  793. generator.append(R"~~~(
  794. #include <AK/Array.h>
  795. #include <AK/BinarySearch.h>
  796. #include <LibUnicode/DateTimeFormat.h>
  797. #include <LibUnicode/Locale.h>
  798. #include <LibUnicode/UnicodeDateTimeFormat.h>
  799. namespace Unicode::Detail {
  800. )~~~");
  801. locale_data.unique_strings.generate(generator);
  802. generator.append(R"~~~(
  803. struct CalendarPattern {
  804. Unicode::CalendarPattern to_unicode_calendar_pattern() const {
  805. Unicode::CalendarPattern calendar_pattern {};
  806. calendar_pattern.pattern = s_string_list[pattern];
  807. if (pattern12 != 0)
  808. calendar_pattern.pattern12 = s_string_list[pattern12];
  809. if (era != -1)
  810. calendar_pattern.era = static_cast<Unicode::CalendarPatternStyle>(era);
  811. if (year != -1)
  812. calendar_pattern.year = static_cast<Unicode::CalendarPatternStyle>(year);
  813. if (month != -1)
  814. calendar_pattern.month = static_cast<Unicode::CalendarPatternStyle>(month);
  815. if (weekday != -1)
  816. calendar_pattern.weekday = static_cast<Unicode::CalendarPatternStyle>(weekday);
  817. if (day != -1)
  818. calendar_pattern.day = static_cast<Unicode::CalendarPatternStyle>(day);
  819. if (day_period != -1)
  820. calendar_pattern.day_period = static_cast<Unicode::CalendarPatternStyle>(day_period);
  821. if (hour != -1)
  822. calendar_pattern.hour = static_cast<Unicode::CalendarPatternStyle>(hour);
  823. if (minute != -1)
  824. calendar_pattern.minute = static_cast<Unicode::CalendarPatternStyle>(minute);
  825. if (second != -1)
  826. calendar_pattern.second = static_cast<Unicode::CalendarPatternStyle>(second);
  827. if (fractional_second_digits != -1)
  828. calendar_pattern.fractional_second_digits = static_cast<u8>(fractional_second_digits);
  829. if (time_zone_name != -1)
  830. calendar_pattern.time_zone_name = static_cast<Unicode::CalendarPatternStyle>(time_zone_name);
  831. return calendar_pattern;
  832. }
  833. @string_index_type@ pattern { 0 };
  834. @string_index_type@ pattern12 { 0 };
  835. i8 era { -1 };
  836. i8 year { -1 };
  837. i8 month { -1 };
  838. i8 weekday { -1 };
  839. i8 day { -1 };
  840. i8 day_period { -1 };
  841. i8 hour { -1 };
  842. i8 minute { -1 };
  843. i8 second { -1 };
  844. i8 fractional_second_digits { -1 };
  845. i8 time_zone_name { -1 };
  846. };
  847. )~~~");
  848. locale_data.unique_patterns.generate(generator, "CalendarPattern"sv, "s_calendar_patterns"sv, 10);
  849. generator.append(R"~~~(
  850. struct CalendarFormat {
  851. Unicode::CalendarFormat to_unicode_calendar_format() const {
  852. Unicode::CalendarFormat calendar_format {};
  853. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  854. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  855. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  856. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  857. return calendar_format;
  858. }
  859. @calendar_pattern_index_type@ full_format { 0 };
  860. @calendar_pattern_index_type@ long_format { 0 };
  861. @calendar_pattern_index_type@ medium_format { 0 };
  862. @calendar_pattern_index_type@ short_format { 0 };
  863. };
  864. using CalendarSymbols = Span<@string_index_type@ const>;
  865. struct CalendarData {
  866. @string_index_type@ calendar { 0 };
  867. CalendarFormat date_formats {};
  868. CalendarFormat time_formats {};
  869. CalendarFormat date_time_formats {};
  870. Span<@calendar_pattern_index_type@ const> available_formats {};
  871. Array<Span<CalendarSymbols const>, @calendar_symbols_size@> symbols {};
  872. };
  873. struct TimeZoneData {
  874. @string_index_type@ time_zone { 0 };
  875. @string_index_type@ long_name { 0 };
  876. @string_index_type@ short_name { 0 };
  877. };
  878. )~~~");
  879. auto append_calendar_format = [&](auto const& calendar_format) {
  880. generator.set("full_format", String::number(calendar_format.full_format));
  881. generator.set("long_format", String::number(calendar_format.long_format));
  882. generator.set("medium_format", String::number(calendar_format.medium_format));
  883. generator.set("short_format", String::number(calendar_format.short_format));
  884. generator.append("{ @full_format@, @long_format@, @medium_format@, @short_format@ },");
  885. };
  886. auto append_calendar_symbols_for_style = [&](auto name, auto style, auto const& symbols) {
  887. name = String::formatted("{}_{}", name, style);
  888. generator.set("name", name);
  889. generator.set("size", String::number(symbols.size()));
  890. generator.append(R"~~~(
  891. static constexpr Array<@string_index_type@, @size@> @name@ { {)~~~");
  892. bool first = true;
  893. for (auto symbol : symbols) {
  894. generator.append(first ? " " : ", ");
  895. generator.append(String::number(symbol));
  896. first = false;
  897. }
  898. generator.append(" } };");
  899. return name;
  900. };
  901. auto append_calendar_symbols = [&](auto name, auto symbol, auto const& symbols) {
  902. name = String::formatted("{}_{}", name, symbol.to_lowercase());
  903. auto narrow_symbols = append_calendar_symbols_for_style(name, "narrow"sv, symbols.narrow_symbols);
  904. auto short_symbols = append_calendar_symbols_for_style(name, "short"sv, symbols.short_symbols);
  905. auto long_symbols = append_calendar_symbols_for_style(name, "long"sv, symbols.long_symbols);
  906. generator.set("narrow_symbols", move(narrow_symbols));
  907. generator.set("short_symbols", move(short_symbols));
  908. generator.set("long_symbols", move(long_symbols));
  909. generator.set("name", name);
  910. generator.append(R"~~~(
  911. static constexpr Array<CalendarSymbols, 3> @name@ { @narrow_symbols@.span(), @short_symbols@.span(), @long_symbols@.span() };
  912. )~~~");
  913. return name;
  914. };
  915. auto append_calendars = [&](String name, auto const& calendars) {
  916. auto format_name = [&](StringView calendar_key, StringView type) {
  917. return String::formatted("{}_{}_{}", name, calendar_key, type);
  918. };
  919. Vector<String> symbols_names;
  920. for (auto const& calendar_key : locale_data.calendars) {
  921. auto const& calendar = calendars.find(calendar_key)->value;
  922. generator.set("name", format_name(calendar_key, "formats"sv));
  923. generator.set("size", String::number(calendar.available_formats.size()));
  924. generator.append(R"~~~(
  925. static constexpr Array<@calendar_pattern_index_type@, @size@> @name@ { {)~~~");
  926. bool first = true;
  927. for (auto format : calendar.available_formats) {
  928. generator.append(first ? " " : ", ");
  929. generator.append(String::number(format));
  930. first = false;
  931. }
  932. generator.append(" } };\n");
  933. auto symbols_name = format_name(calendar_key, "symbols"sv);
  934. for (auto const& symbol_key : locale_data.symbols) {
  935. auto const& symbols = calendar.symbols.find(symbol_key)->value;
  936. auto name = append_calendar_symbols(symbols_name, symbol_key, symbols);
  937. symbols_names.append(name);
  938. }
  939. }
  940. generator.set("name", name);
  941. generator.set("size", String::number(calendars.size()));
  942. generator.append(R"~~~(
  943. static constexpr Array<CalendarData, @size@> @name@ { {)~~~");
  944. for (auto const& calendar_key : locale_data.calendars) {
  945. auto const& calendar = calendars.find(calendar_key)->value;
  946. generator.set("name", format_name(calendar_key, "formats"sv));
  947. generator.set("calendar"sv, String::number(calendar.calendar));
  948. generator.append(R"~~~(
  949. { @calendar@, )~~~");
  950. append_calendar_format(calendar.date_formats);
  951. generator.append(" ");
  952. append_calendar_format(calendar.time_formats);
  953. generator.append(" ");
  954. append_calendar_format(calendar.date_time_formats);
  955. generator.append(" @name@.span(), {");
  956. bool first = true;
  957. for (auto const& symbols_name : symbols_names) {
  958. generator.append(first ? " " : ", ");
  959. generator.append(String::formatted("{}.span()", symbols_name));
  960. first = false;
  961. }
  962. generator.append(" } },");
  963. }
  964. generator.append(R"~~~(
  965. } };
  966. )~~~");
  967. };
  968. auto append_time_zones = [&](String name, auto const& time_zones) {
  969. generator.set("name", name);
  970. generator.set("size", String::number(locale_data.time_zones.size()));
  971. generator.append(R"~~~(
  972. static constexpr Array<TimeZoneData, @size@> @name@ { {)~~~");
  973. constexpr size_t max_values_per_row = 20;
  974. size_t values_in_current_row = 0;
  975. for (auto const& time_zone_key : locale_data.time_zones) {
  976. auto time_zone = time_zones.find(time_zone_key);
  977. if (values_in_current_row++ > 0)
  978. generator.append(" ");
  979. if (time_zone == time_zones.end()) {
  980. generator.append("{},");
  981. } else {
  982. generator.set("time_zone", String::number(time_zone->value.time_zone));
  983. generator.set("long_name", String::number(time_zone->value.long_name));
  984. generator.set("short_name", String::number(time_zone->value.short_name));
  985. generator.append("{ @time_zone@, @long_name@, @short_name@ },");
  986. }
  987. if (values_in_current_row == max_values_per_row) {
  988. values_in_current_row = 0;
  989. generator.append("\n ");
  990. }
  991. }
  992. generator.append(R"~~~(
  993. } };
  994. )~~~");
  995. };
  996. auto append_hour_cycles = [&](String name, auto const& hour_cycle_region) {
  997. auto const& hour_cycles = locale_data.hour_cycles.find(hour_cycle_region)->value;
  998. generator.set("name", name);
  999. generator.set("size", String::number(hour_cycles.size()));
  1000. generator.append(R"~~~(
  1001. static constexpr Array<u8, @size@> @name@ { { )~~~");
  1002. for (auto hour_cycle : hour_cycles) {
  1003. generator.set("hour_cycle", String::number(static_cast<u8>(hour_cycle)));
  1004. generator.append("@hour_cycle@, ");
  1005. }
  1006. generator.append("} };");
  1007. };
  1008. generate_mapping(generator, locale_data.locales, "CalendarData"sv, "s_calendars"sv, "s_calendars_{}", [&](auto const& name, auto const& value) { append_calendars(name, value.calendars); });
  1009. 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); });
  1010. 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); });
  1011. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1012. HashValueMap<String> hashes;
  1013. hashes.ensure_capacity(values.size());
  1014. for (auto const& value : values)
  1015. hashes.set(value.hash(), format_identifier(enum_title, value));
  1016. for (auto const& alias : aliases)
  1017. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1018. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1019. };
  1020. append_from_string("Calendar"sv, "calendar"sv, locale_data.calendars, locale_data.calendar_aliases);
  1021. append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, locale_data.hour_cycle_regions);
  1022. append_from_string("TimeZone"sv, "time_zone"sv, locale_data.time_zones);
  1023. generator.append(R"~~~(
  1024. Vector<Unicode::HourCycle> get_regional_hour_cycles(StringView region)
  1025. {
  1026. auto region_value = hour_cycle_region_from_string(region);
  1027. if (!region_value.has_value())
  1028. return {};
  1029. auto region_index = to_underlying(*region_value);
  1030. auto const& regional_hour_cycles = s_hour_cycles.at(region_index);
  1031. Vector<Unicode::HourCycle> hour_cycles;
  1032. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1033. for (auto hour_cycle : regional_hour_cycles)
  1034. hour_cycles.unchecked_append(static_cast<Unicode::HourCycle>(hour_cycle));
  1035. return hour_cycles;
  1036. }
  1037. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1038. {
  1039. auto locale_value = locale_from_string(locale);
  1040. if (!locale_value.has_value())
  1041. return nullptr;
  1042. auto calendar_value = calendar_from_string(calendar);
  1043. if (!calendar_value.has_value())
  1044. return nullptr;
  1045. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1046. auto calendar_index = to_underlying(*calendar_value);
  1047. auto const& calendars = s_calendars.at(locale_index);
  1048. return &calendars[calendar_index];
  1049. }
  1050. Optional<Unicode::CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1051. {
  1052. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr)
  1053. return data->date_formats.to_unicode_calendar_format();
  1054. return {};
  1055. }
  1056. Optional<Unicode::CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1057. {
  1058. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr)
  1059. return data->time_formats.to_unicode_calendar_format();
  1060. return {};
  1061. }
  1062. Optional<Unicode::CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1063. {
  1064. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr)
  1065. return data->date_time_formats.to_unicode_calendar_format();
  1066. return {};
  1067. }
  1068. Vector<Unicode::CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1069. {
  1070. Vector<Unicode::CalendarPattern> result {};
  1071. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1072. result.ensure_capacity(data->available_formats.size());
  1073. for (auto const& format : data->available_formats)
  1074. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1075. }
  1076. return result;
  1077. }
  1078. static CalendarSymbols find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1079. {
  1080. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1081. auto symbol_index = to_underlying(symbol);
  1082. auto style_index = to_underlying(style);
  1083. auto symbols = data->symbols.at(symbol_index);
  1084. VERIFY(style_index < symbols.size());
  1085. return symbols.at(style_index);
  1086. }
  1087. return {};
  1088. }
  1089. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Era value)
  1090. {
  1091. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1092. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1093. return s_string_list[symbols.at(value_index)];
  1094. return {};
  1095. }
  1096. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Month value)
  1097. {
  1098. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1099. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1100. return s_string_list[symbols.at(value_index)];
  1101. return {};
  1102. }
  1103. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Weekday value)
  1104. {
  1105. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1106. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1107. return s_string_list[symbols.at(value_index)];
  1108. return {};
  1109. }
  1110. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::DayPeriod value)
  1111. {
  1112. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1113. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1114. return s_string_list[symbols.at(value_index)];
  1115. return {};
  1116. }
  1117. static TimeZoneData const* find_time_zone_data(StringView locale, StringView time_zone)
  1118. {
  1119. auto locale_value = locale_from_string(locale);
  1120. if (!locale_value.has_value())
  1121. return nullptr;
  1122. auto time_zone_value = time_zone_from_string(time_zone);
  1123. if (!time_zone_value.has_value())
  1124. return nullptr;
  1125. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1126. auto time_zone_index = to_underlying(*time_zone_value);
  1127. auto const& time_zones = s_time_zones.at(locale_index);
  1128. return &time_zones[time_zone_index];
  1129. }
  1130. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style)
  1131. {
  1132. if (auto const* data = find_time_zone_data(locale, time_zone); data != nullptr) {
  1133. @string_index_type@ time_zone_index = 0;
  1134. switch (style) {
  1135. case CalendarPatternStyle::Long:
  1136. time_zone_index = data->long_name;
  1137. break;
  1138. case CalendarPatternStyle::Short:
  1139. time_zone_index = data->short_name;
  1140. break;
  1141. default:
  1142. VERIFY_NOT_REACHED();
  1143. }
  1144. if (time_zone_index != 0)
  1145. return s_string_list[time_zone_index];
  1146. }
  1147. return {};
  1148. }
  1149. }
  1150. )~~~");
  1151. VERIFY(file.write(generator.as_string_view()));
  1152. }
  1153. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1154. {
  1155. StringView generated_header_path;
  1156. StringView generated_implementation_path;
  1157. StringView core_path;
  1158. StringView dates_path;
  1159. Core::ArgsParser args_parser;
  1160. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1161. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1162. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1163. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1164. args_parser.parse(arguments);
  1165. auto open_file = [&](StringView path) -> ErrorOr<NonnullRefPtr<Core::File>> {
  1166. if (path.is_empty()) {
  1167. args_parser.print_usage(stderr, arguments.argv[0]);
  1168. return Error::from_string_literal("Must provide all command line options"sv);
  1169. }
  1170. return Core::File::open(path, Core::OpenMode::ReadWrite);
  1171. };
  1172. auto generated_header_file = TRY(open_file(generated_header_path));
  1173. auto generated_implementation_file = TRY(open_file(generated_implementation_path));
  1174. UnicodeLocaleData locale_data;
  1175. TRY(parse_all_locales(core_path, dates_path, locale_data));
  1176. generate_unicode_locale_header(generated_header_file, locale_data);
  1177. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  1178. return 0;
  1179. }