GenerateUnicodeDateTimeFormat.cpp 59 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491
  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. hash = pair_int_hash(hash, skeleton_index);
  43. auto hash_field = [&](auto const& field) {
  44. if (field.has_value())
  45. hash = pair_int_hash(hash, static_cast<u8>(*field));
  46. else
  47. hash = pair_int_hash(hash, -1);
  48. };
  49. hash_field(era);
  50. hash_field(year);
  51. hash_field(month);
  52. hash_field(weekday);
  53. hash_field(day);
  54. hash_field(day_period);
  55. hash_field(hour);
  56. hash_field(minute);
  57. hash_field(second);
  58. hash_field(fractional_second_digits);
  59. hash_field(time_zone_name);
  60. return hash;
  61. }
  62. bool operator==(CalendarPattern const& other) const
  63. {
  64. return (skeleton_index == other.skeleton_index)
  65. && (pattern_index == other.pattern_index)
  66. && (pattern12_index == other.pattern12_index)
  67. && (era == other.era)
  68. && (year == other.year)
  69. && (month == other.month)
  70. && (weekday == other.weekday)
  71. && (day == other.day)
  72. && (day_period == other.day_period)
  73. && (hour == other.hour)
  74. && (minute == other.minute)
  75. && (second == other.second)
  76. && (fractional_second_digits == other.fractional_second_digits)
  77. && (time_zone_name == other.time_zone_name);
  78. }
  79. StringIndexType skeleton_index { 0 };
  80. StringIndexType pattern_index { 0 };
  81. StringIndexType pattern12_index { 0 };
  82. };
  83. template<>
  84. struct AK::Formatter<CalendarPattern> : Formatter<FormatString> {
  85. ErrorOr<void> format(FormatBuilder& builder, CalendarPattern const& pattern)
  86. {
  87. auto field_to_i8 = [](auto const& field) -> i8 {
  88. if (!field.has_value())
  89. return -1;
  90. return static_cast<i8>(*field);
  91. };
  92. return Formatter<FormatString>::format(builder,
  93. "{{ {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {}, {} }}",
  94. pattern.skeleton_index,
  95. pattern.pattern_index,
  96. pattern.pattern12_index,
  97. field_to_i8(pattern.era),
  98. field_to_i8(pattern.year),
  99. field_to_i8(pattern.month),
  100. field_to_i8(pattern.weekday),
  101. field_to_i8(pattern.day),
  102. field_to_i8(pattern.day_period),
  103. field_to_i8(pattern.hour),
  104. field_to_i8(pattern.minute),
  105. field_to_i8(pattern.second),
  106. field_to_i8(pattern.fractional_second_digits),
  107. field_to_i8(pattern.time_zone_name));
  108. }
  109. };
  110. template<>
  111. struct AK::Traits<CalendarPattern> : public GenericTraits<CalendarPattern> {
  112. static unsigned hash(CalendarPattern const& c) { return c.hash(); }
  113. };
  114. struct CalendarFormat {
  115. CalendarPatternIndexType full_format { 0 };
  116. CalendarPatternIndexType long_format { 0 };
  117. CalendarPatternIndexType medium_format { 0 };
  118. CalendarPatternIndexType short_format { 0 };
  119. };
  120. struct CalendarSymbols {
  121. Vector<StringIndexType> narrow_symbols {};
  122. Vector<StringIndexType> short_symbols {};
  123. Vector<StringIndexType> long_symbols {};
  124. };
  125. struct Calendar {
  126. StringIndexType calendar { 0 };
  127. CalendarFormat date_formats {};
  128. CalendarFormat time_formats {};
  129. CalendarFormat date_time_formats {};
  130. Vector<CalendarPatternIndexType> available_formats {};
  131. HashMap<String, CalendarSymbols> symbols {};
  132. };
  133. struct TimeZone {
  134. StringIndexType time_zone { 0 };
  135. StringIndexType long_name { 0 };
  136. StringIndexType short_name { 0 };
  137. };
  138. struct Locale {
  139. HashMap<String, Calendar> calendars;
  140. HashMap<String, TimeZone> time_zones;
  141. };
  142. struct UnicodeLocaleData {
  143. UniqueStringStorage<StringIndexType> unique_strings;
  144. UniqueStorage<CalendarPattern, CalendarPatternIndexType> unique_patterns;
  145. HashMap<String, Locale> locales;
  146. HashMap<String, Vector<Unicode::HourCycle>> hour_cycles;
  147. Vector<String> hour_cycle_regions;
  148. HashMap<String, StringIndexType> meta_zones;
  149. Vector<String> time_zones;
  150. Vector<String> calendars;
  151. Vector<Alias> calendar_aliases {
  152. // FIXME: Aliases should come from BCP47. See: https://unicode-org.atlassian.net/browse/CLDR-15158
  153. { "gregorian"sv, "gregory"sv },
  154. };
  155. Vector<String> symbols;
  156. };
  157. static ErrorOr<void> parse_hour_cycles(String core_path, UnicodeLocaleData& locale_data)
  158. {
  159. // https://unicode.org/reports/tr35/tr35-dates.html#Time_Data
  160. LexicalPath time_data_path(move(core_path));
  161. time_data_path = time_data_path.append("supplemental"sv);
  162. time_data_path = time_data_path.append("timeData.json"sv);
  163. auto time_data_file = TRY(Core::File::open(time_data_path.string(), Core::OpenMode::ReadOnly));
  164. auto time_data = TRY(JsonValue::from_string(time_data_file->read_all()));
  165. auto const& supplemental_object = time_data.as_object().get("supplemental"sv);
  166. auto const& time_data_object = supplemental_object.as_object().get("timeData"sv);
  167. auto parse_hour_cycle = [](StringView hour_cycle) -> Optional<Unicode::HourCycle> {
  168. if (hour_cycle == "h"sv)
  169. return Unicode::HourCycle::H12;
  170. if (hour_cycle == "H"sv)
  171. return Unicode::HourCycle::H23;
  172. if (hour_cycle == "K"sv)
  173. return Unicode::HourCycle::H11;
  174. if (hour_cycle == "k"sv)
  175. return Unicode::HourCycle::H24;
  176. return {};
  177. };
  178. time_data_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  179. auto allowed_hour_cycles_string = value.as_object().get("_allowed"sv).as_string();
  180. auto allowed_hour_cycles = allowed_hour_cycles_string.split_view(' ');
  181. Vector<Unicode::HourCycle> hour_cycles;
  182. for (auto allowed_hour_cycle : allowed_hour_cycles) {
  183. if (auto hour_cycle = parse_hour_cycle(allowed_hour_cycle); hour_cycle.has_value())
  184. hour_cycles.append(*hour_cycle);
  185. }
  186. locale_data.hour_cycles.set(key, move(hour_cycles));
  187. if (!locale_data.hour_cycle_regions.contains_slow(key))
  188. locale_data.hour_cycle_regions.append(key);
  189. });
  190. return {};
  191. };
  192. static ErrorOr<void> parse_meta_zones(String core_path, UnicodeLocaleData& locale_data)
  193. {
  194. // https://unicode.org/reports/tr35/tr35-dates.html#Metazones
  195. LexicalPath meta_zone_path(move(core_path));
  196. meta_zone_path = meta_zone_path.append("supplemental"sv);
  197. meta_zone_path = meta_zone_path.append("metaZones.json"sv);
  198. auto meta_zone_file = TRY(Core::File::open(meta_zone_path.string(), Core::OpenMode::ReadOnly));
  199. auto meta_zone = TRY(JsonValue::from_string(meta_zone_file->read_all()));
  200. auto const& supplemental_object = meta_zone.as_object().get("supplemental"sv);
  201. auto const& meta_zone_object = supplemental_object.as_object().get("metaZones"sv);
  202. auto const& meta_zone_array = meta_zone_object.as_object().get("metazones"sv);
  203. meta_zone_array.as_array().for_each([&](JsonValue const& value) {
  204. auto const& mapping = value.as_object().get("mapZone"sv);
  205. auto const& meta_zone = mapping.as_object().get("_other"sv);
  206. auto const& golden_zone = mapping.as_object().get("_type"sv);
  207. auto golden_zone_index = locale_data.unique_strings.ensure(golden_zone.as_string());
  208. locale_data.meta_zones.set(meta_zone.as_string(), golden_zone_index);
  209. });
  210. // UTC does not appear in metaZones.json. Define it for convenience so other parsers don't need to check for its existence.
  211. locale_data.meta_zones.set("UTC"sv, locale_data.unique_strings.ensure("UTC"sv));
  212. return {};
  213. };
  214. static constexpr auto is_char(char ch)
  215. {
  216. return [ch](auto c) { return c == ch; };
  217. }
  218. // For patterns that are 12-hour aware, we need to generate two patterns: one with the day period
  219. // (e.g. {ampm}) in the pattern, and one without the day period. We need to take care to remove
  220. // extra spaces around the day period. Some example expected removals:
  221. //
  222. // "{hour}:{minute} {ampm}" becomes "{hour}:{minute}" (remove the space before {ampm})
  223. // "{ampm} {hour}" becomes "{hour}" (remove the space after {ampm})
  224. // "{hour}:{minute} {ampm} {timeZoneName}" becomes "{hour}:{minute} {timeZoneName}" (remove one of the spaces around {ampm})
  225. static String remove_period_from_pattern(String pattern)
  226. {
  227. for (auto remove : AK::Array { "({ampm})"sv, "{ampm}"sv, "({dayPeriod})"sv, "{dayPeriod}"sv }) {
  228. auto index = pattern.find(remove);
  229. if (!index.has_value())
  230. continue;
  231. constexpr u32 space = ' ';
  232. constexpr u32 open = '{';
  233. constexpr u32 close = '}';
  234. Utf8View utf8_pattern { pattern };
  235. Optional<u32> before_removal;
  236. Optional<u32> after_removal;
  237. for (auto it = utf8_pattern.begin(); utf8_pattern.byte_offset_of(it) < *index; ++it)
  238. before_removal = *it;
  239. if (auto it = utf8_pattern.iterator_at_byte_offset(*index + remove.length()); it != utf8_pattern.end())
  240. after_removal = *it;
  241. if ((before_removal == space) && (after_removal != open)) {
  242. pattern = String::formatted("{}{}",
  243. pattern.substring_view(0, *index - 1),
  244. pattern.substring_view(*index + remove.length()));
  245. } else if ((after_removal == space) && (before_removal != close)) {
  246. pattern = String::formatted("{}{}",
  247. pattern.substring_view(0, *index),
  248. pattern.substring_view(*index + remove.length() + 1));
  249. } else {
  250. pattern = String::formatted("{}{}",
  251. pattern.substring_view(0, *index),
  252. pattern.substring_view(*index + remove.length()));
  253. }
  254. }
  255. return pattern;
  256. }
  257. static Optional<CalendarPatternIndexType> parse_date_time_pattern(String pattern, String skeleton, UnicodeLocaleData& locale_data)
  258. {
  259. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  260. using Unicode::CalendarPatternStyle;
  261. CalendarPattern format {};
  262. if (!skeleton.is_empty())
  263. format.skeleton_index = locale_data.unique_strings.ensure(move(skeleton));
  264. GenericLexer lexer { pattern };
  265. StringBuilder builder;
  266. bool hour12 { false };
  267. while (!lexer.is_eof()) {
  268. // Literal strings enclosed by quotes are to be appended to the pattern as-is without further
  269. // processing (this just avoids conflicts with the patterns below).
  270. if (lexer.next_is(is_quote)) {
  271. builder.append(lexer.consume_quoted_string());
  272. continue;
  273. }
  274. auto starting_char = lexer.peek();
  275. auto segment = lexer.consume_while([&](char ch) { return ch == starting_char; });
  276. // Era
  277. if (all_of(segment, is_char('G'))) {
  278. builder.append("{era}");
  279. if (segment.length() <= 3)
  280. format.era = CalendarPatternStyle::Short;
  281. else if (segment.length() == 4)
  282. format.era = CalendarPatternStyle::Long;
  283. else
  284. format.era = CalendarPatternStyle::Narrow;
  285. }
  286. // Year
  287. else if (all_of(segment, is_any_of("yYuUr"sv))) {
  288. builder.append("{year}");
  289. if (segment.length() == 2)
  290. format.year = CalendarPatternStyle::TwoDigit;
  291. else
  292. format.year = CalendarPatternStyle::Numeric;
  293. }
  294. // Quarter
  295. else if (all_of(segment, is_any_of("qQ"sv))) {
  296. // Intl.DateTimeFormat does not support quarter formatting, so drop these patterns.
  297. return {};
  298. }
  299. // Month
  300. else if (all_of(segment, is_any_of("ML"sv))) {
  301. builder.append("{month}");
  302. if (segment.length() == 1)
  303. format.month = CalendarPatternStyle::Numeric;
  304. else if (segment.length() == 2)
  305. format.month = CalendarPatternStyle::TwoDigit;
  306. else if (segment.length() == 3)
  307. format.month = CalendarPatternStyle::Short;
  308. else if (segment.length() == 4)
  309. format.month = CalendarPatternStyle::Long;
  310. else if (segment.length() == 5)
  311. format.month = CalendarPatternStyle::Narrow;
  312. } else if (all_of(segment, is_char('l'))) {
  313. // Using 'l' for month formatting is deprecated by TR-35, ensure it is not used.
  314. return {};
  315. }
  316. // Week
  317. else if (all_of(segment, is_any_of("wW"sv))) {
  318. // Intl.DateTimeFormat does not support week formatting, so drop these patterns.
  319. return {};
  320. }
  321. // Day
  322. else if (all_of(segment, is_char('d'))) {
  323. builder.append("{day}");
  324. if (segment.length() == 1)
  325. format.day = CalendarPatternStyle::Numeric;
  326. else
  327. format.day = CalendarPatternStyle::TwoDigit;
  328. } else if (all_of(segment, is_any_of("DFg"sv))) {
  329. builder.append("{day}");
  330. format.day = CalendarPatternStyle::Numeric;
  331. }
  332. // Weekday
  333. else if (all_of(segment, is_char('E'))) {
  334. builder.append("{weekday}");
  335. if (segment.length() == 4)
  336. format.weekday = CalendarPatternStyle::Long;
  337. else if (segment.length() == 5)
  338. format.weekday = CalendarPatternStyle::Narrow;
  339. else
  340. format.weekday = CalendarPatternStyle::Short;
  341. } else if (all_of(segment, is_any_of("ec"sv))) {
  342. builder.append("{weekday}");
  343. // TR-35 defines "e", "c", and "cc" as as numeric, and "ee" as 2-digit, but those
  344. // pattern styles are not supported by Intl.DateTimeFormat.
  345. if (segment.length() <= 2)
  346. return {};
  347. if (segment.length() == 4)
  348. format.weekday = CalendarPatternStyle::Long;
  349. else if (segment.length() == 5)
  350. format.weekday = CalendarPatternStyle::Narrow;
  351. else
  352. format.weekday = CalendarPatternStyle::Short;
  353. }
  354. // Period
  355. else if (all_of(segment, is_any_of("ab"sv))) {
  356. builder.append("{ampm}");
  357. hour12 = true;
  358. } else if (all_of(segment, is_char('B'))) {
  359. builder.append("{dayPeriod}");
  360. hour12 = true;
  361. if (segment.length() == 4)
  362. format.day_period = CalendarPatternStyle::Long;
  363. else if (segment.length() == 5)
  364. format.day_period = CalendarPatternStyle::Narrow;
  365. else
  366. format.day_period = CalendarPatternStyle::Short;
  367. }
  368. // Hour
  369. else if (all_of(segment, is_any_of("hHKk"sv))) {
  370. builder.append("{hour}");
  371. if ((segment[0] == 'h') || (segment[0] == 'K'))
  372. hour12 = true;
  373. if (segment.length() == 1)
  374. format.hour = CalendarPatternStyle::Numeric;
  375. else
  376. format.hour = CalendarPatternStyle::TwoDigit;
  377. } else if (all_of(segment, is_any_of("jJC"sv))) {
  378. // TR-35 indicates these should not be used.
  379. return {};
  380. }
  381. // Minute
  382. else if (all_of(segment, is_char('m'))) {
  383. builder.append("{minute}");
  384. if (segment.length() == 1)
  385. format.minute = CalendarPatternStyle::Numeric;
  386. else
  387. format.minute = CalendarPatternStyle::TwoDigit;
  388. }
  389. // Second
  390. else if (all_of(segment, is_char('s'))) {
  391. builder.append("{second}");
  392. if (segment.length() == 1)
  393. format.second = CalendarPatternStyle::Numeric;
  394. else
  395. format.second = CalendarPatternStyle::TwoDigit;
  396. } else if (all_of(segment, is_char('S'))) {
  397. builder.append("{fractionalSecondDigits}");
  398. VERIFY(segment.length() <= 3);
  399. format.fractional_second_digits = static_cast<u8>(segment.length());
  400. } else if (all_of(segment, is_char('A'))) {
  401. // Intl.DateTimeFormat does not support millisecond formatting, so drop these patterns.
  402. return {};
  403. }
  404. // Zone
  405. else if (all_of(segment, is_any_of("zZOvVXx"))) {
  406. builder.append("{timeZoneName}");
  407. if (segment.length() < 4)
  408. format.time_zone_name = CalendarPatternStyle::Short;
  409. else
  410. format.time_zone_name = CalendarPatternStyle::Long;
  411. }
  412. // Non-patterns
  413. else {
  414. builder.append(segment);
  415. }
  416. }
  417. pattern = builder.build();
  418. if (hour12) {
  419. auto pattern_without_period = remove_period_from_pattern(pattern);
  420. format.pattern_index = locale_data.unique_strings.ensure(move(pattern_without_period));
  421. format.pattern12_index = locale_data.unique_strings.ensure(move(pattern));
  422. } else {
  423. format.pattern_index = locale_data.unique_strings.ensure(move(pattern));
  424. }
  425. return locale_data.unique_patterns.ensure(move(format));
  426. }
  427. static void generate_missing_patterns(Calendar& calendar, Vector<CalendarPattern> date_formats, Vector<CalendarPattern> time_formats, UnicodeLocaleData& locale_data)
  428. {
  429. // https://unicode.org/reports/tr35/tr35-dates.html#Missing_Skeleton_Fields
  430. auto replace_pattern = [&](auto format, auto time_format, auto date_format) {
  431. auto pattern = locale_data.unique_strings.get(format);
  432. auto time_pattern = locale_data.unique_strings.get(time_format);
  433. auto date_pattern = locale_data.unique_strings.get(date_format);
  434. auto new_pattern = pattern.replace("{0}", time_pattern).replace("{1}", date_pattern);
  435. return locale_data.unique_strings.ensure(move(new_pattern));
  436. };
  437. auto inject_fractional_second_digits = [&](auto format) {
  438. auto pattern = locale_data.unique_strings.get(format);
  439. auto new_pattern = pattern.replace("{second}"sv, "{second}{decimal}{fractionalSecondDigits}"sv);
  440. return locale_data.unique_strings.ensure(move(new_pattern));
  441. };
  442. auto append_if_unique = [&](auto format) {
  443. auto format_index = locale_data.unique_patterns.ensure(move(format));
  444. if (!calendar.available_formats.contains_slow(format_index))
  445. calendar.available_formats.append(format_index);
  446. };
  447. Vector<CalendarPattern> time_formats_with_fractional_second_digits;
  448. for (auto const& format : date_formats)
  449. append_if_unique(format);
  450. for (auto const& format : time_formats) {
  451. append_if_unique(format);
  452. if (format.second.has_value() && !format.fractional_second_digits.has_value()) {
  453. auto new_format = format;
  454. new_format.fractional_second_digits = 2;
  455. new_format.pattern_index = inject_fractional_second_digits(new_format.pattern_index);
  456. if (new_format.pattern12_index != 0)
  457. new_format.pattern12_index = inject_fractional_second_digits(new_format.pattern12_index);
  458. time_formats_with_fractional_second_digits.append(new_format);
  459. append_if_unique(move(new_format));
  460. }
  461. }
  462. time_formats.extend(move(time_formats_with_fractional_second_digits));
  463. for (auto const& date_format : date_formats) {
  464. CalendarPatternIndexType date_time_format_index = 0;
  465. if (date_format.month == Unicode::CalendarPatternStyle::Long) {
  466. if (date_format.weekday.has_value())
  467. date_time_format_index = calendar.date_time_formats.full_format;
  468. else
  469. date_time_format_index = calendar.date_time_formats.long_format;
  470. } else if (date_format.month == Unicode::CalendarPatternStyle::Short) {
  471. date_time_format_index = calendar.date_time_formats.medium_format;
  472. } else {
  473. date_time_format_index = calendar.date_time_formats.short_format;
  474. }
  475. for (auto const& time_format : time_formats) {
  476. auto format = locale_data.unique_patterns.get(date_time_format_index);
  477. if (time_format.pattern12_index != 0)
  478. format.pattern12_index = replace_pattern(format.pattern_index, time_format.pattern12_index, date_format.pattern_index);
  479. format.pattern_index = replace_pattern(format.pattern_index, time_format.pattern_index, date_format.pattern_index);
  480. format.for_each_calendar_field_zipped_with(date_format, [](auto& field, auto const& date_field, auto) {
  481. if (date_field.has_value())
  482. field = date_field;
  483. });
  484. format.for_each_calendar_field_zipped_with(time_format, [](auto& field, auto const& time_field, auto) {
  485. if (time_field.has_value())
  486. field = time_field;
  487. });
  488. append_if_unique(move(format));
  489. }
  490. }
  491. }
  492. static void parse_calendar_symbols(Calendar& calendar, JsonObject const& calendar_object, UnicodeLocaleData& locale_data)
  493. {
  494. auto ensure_symbols = [&](auto name, size_t size) -> CalendarSymbols& {
  495. if (!locale_data.symbols.contains_slow(name))
  496. locale_data.symbols.append(name);
  497. return calendar.symbols.ensure(name, [&]() {
  498. CalendarSymbols symbols {};
  499. symbols.narrow_symbols.resize(size);
  500. symbols.short_symbols.resize(size);
  501. symbols.long_symbols.resize(size);
  502. return symbols;
  503. });
  504. };
  505. auto parse_era_symbols = [&](auto const& symbols_object) {
  506. auto const& narrow_symbols = symbols_object.get("eraNarrow"sv).as_object();
  507. auto const& short_symbols = symbols_object.get("eraAbbr"sv).as_object();
  508. auto const& long_symbols = symbols_object.get("eraNames"sv).as_object();
  509. auto& symbols = ensure_symbols("era"sv, 2);
  510. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  511. if (auto key_index = key.to_uint(); key_index.has_value())
  512. symbols[*key_index] = locale_data.unique_strings.ensure(move(symbol));
  513. };
  514. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  515. append_symbol(symbols.narrow_symbols, key, value.as_string());
  516. });
  517. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  518. append_symbol(symbols.short_symbols, key, value.as_string());
  519. });
  520. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  521. append_symbol(symbols.long_symbols, key, value.as_string());
  522. });
  523. };
  524. auto parse_month_symbols = [&](auto const& symbols_object) {
  525. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  526. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  527. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  528. auto& month_symbols = ensure_symbols("month"sv, 12);
  529. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  530. auto key_index = key.to_uint().value() - 1;
  531. symbols[key_index] = locale_data.unique_strings.ensure(move(symbol));
  532. };
  533. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  534. append_symbol(month_symbols.narrow_symbols, key, value.as_string());
  535. });
  536. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  537. append_symbol(month_symbols.short_symbols, key, value.as_string());
  538. });
  539. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  540. append_symbol(month_symbols.long_symbols, key, value.as_string());
  541. });
  542. };
  543. auto parse_weekday_symbols = [&](auto const& symbols_object) {
  544. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  545. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  546. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  547. auto& weekday_symbols = ensure_symbols("weekday"sv, 7);
  548. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  549. if (key == "sun"sv)
  550. symbols[to_underlying(Unicode::Weekday::Sunday)] = locale_data.unique_strings.ensure(move(symbol));
  551. else if (key == "mon"sv)
  552. symbols[to_underlying(Unicode::Weekday::Monday)] = locale_data.unique_strings.ensure(move(symbol));
  553. else if (key == "tue"sv)
  554. symbols[to_underlying(Unicode::Weekday::Tuesday)] = locale_data.unique_strings.ensure(move(symbol));
  555. else if (key == "wed"sv)
  556. symbols[to_underlying(Unicode::Weekday::Wednesday)] = locale_data.unique_strings.ensure(move(symbol));
  557. else if (key == "thu"sv)
  558. symbols[to_underlying(Unicode::Weekday::Thursday)] = locale_data.unique_strings.ensure(move(symbol));
  559. else if (key == "fri"sv)
  560. symbols[to_underlying(Unicode::Weekday::Friday)] = locale_data.unique_strings.ensure(move(symbol));
  561. else if (key == "sat"sv)
  562. symbols[to_underlying(Unicode::Weekday::Saturday)] = locale_data.unique_strings.ensure(move(symbol));
  563. };
  564. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  565. append_symbol(weekday_symbols.narrow_symbols, key, value.as_string());
  566. });
  567. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  568. append_symbol(weekday_symbols.short_symbols, key, value.as_string());
  569. });
  570. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  571. append_symbol(weekday_symbols.long_symbols, key, value.as_string());
  572. });
  573. };
  574. auto parse_day_period_symbols = [&](auto const& symbols_object) {
  575. auto const& narrow_symbols = symbols_object.get("narrow"sv).as_object();
  576. auto const& short_symbols = symbols_object.get("abbreviated"sv).as_object();
  577. auto const& long_symbols = symbols_object.get("wide"sv).as_object();
  578. auto& day_period_symbols = ensure_symbols("dayPeriod"sv, 6);
  579. auto append_symbol = [&](auto& symbols, auto const& key, auto symbol) {
  580. if (key == "am"sv)
  581. symbols[to_underlying(Unicode::DayPeriod::AM)] = locale_data.unique_strings.ensure(move(symbol));
  582. else if (key == "pm"sv)
  583. symbols[to_underlying(Unicode::DayPeriod::PM)] = locale_data.unique_strings.ensure(move(symbol));
  584. else if (key == "morning1"sv)
  585. symbols[to_underlying(Unicode::DayPeriod::Morning)] = locale_data.unique_strings.ensure(move(symbol));
  586. else if (key == "afternoon1"sv)
  587. symbols[to_underlying(Unicode::DayPeriod::Afternoon)] = locale_data.unique_strings.ensure(move(symbol));
  588. else if (key == "evening1"sv)
  589. symbols[to_underlying(Unicode::DayPeriod::Evening)] = locale_data.unique_strings.ensure(move(symbol));
  590. else if (key == "night1"sv)
  591. symbols[to_underlying(Unicode::DayPeriod::Night)] = locale_data.unique_strings.ensure(move(symbol));
  592. };
  593. narrow_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  594. append_symbol(day_period_symbols.narrow_symbols, key, value.as_string());
  595. });
  596. short_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  597. append_symbol(day_period_symbols.short_symbols, key, value.as_string());
  598. });
  599. long_symbols.for_each_member([&](auto const& key, JsonValue const& value) {
  600. append_symbol(day_period_symbols.long_symbols, key, value.as_string());
  601. });
  602. };
  603. parse_era_symbols(calendar_object.get("eras"sv).as_object());
  604. parse_month_symbols(calendar_object.get("months"sv).as_object().get("format"sv).as_object());
  605. parse_weekday_symbols(calendar_object.get("days"sv).as_object().get("format"sv).as_object());
  606. parse_day_period_symbols(calendar_object.get("dayPeriods"sv).as_object().get("format"sv).as_object());
  607. }
  608. static ErrorOr<void> parse_calendars(String locale_calendars_path, UnicodeLocaleData& locale_data, Locale& locale)
  609. {
  610. LexicalPath calendars_path(move(locale_calendars_path));
  611. if (!calendars_path.basename().starts_with("ca-"sv))
  612. return {};
  613. auto calendars_file = TRY(Core::File::open(calendars_path.string(), Core::OpenMode::ReadOnly));
  614. auto calendars = TRY(JsonValue::from_string(calendars_file->read_all()));
  615. auto const& main_object = calendars.as_object().get("main"sv);
  616. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  617. auto const& dates_object = locale_object.as_object().get("dates"sv);
  618. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  619. auto ensure_calendar = [&](auto const& calendar) -> Calendar& {
  620. return locale.calendars.ensure(calendar, [&]() {
  621. auto calendar_index = locale_data.unique_strings.ensure(calendar);
  622. return Calendar { .calendar = calendar_index };
  623. });
  624. };
  625. auto parse_patterns = [&](auto& formats, auto const& patterns_object, auto const& skeletons_object, Vector<CalendarPattern>* patterns) {
  626. auto parse_pattern = [&](auto name) {
  627. auto format = patterns_object.get(name);
  628. auto skeleton = skeletons_object.get(name);
  629. auto format_index = parse_date_time_pattern(format.as_string(), skeleton.as_string_or(String::empty()), locale_data).value();
  630. if (patterns)
  631. patterns->append(locale_data.unique_patterns.get(format_index));
  632. return format_index;
  633. };
  634. formats.full_format = parse_pattern("full"sv);
  635. formats.long_format = parse_pattern("long"sv);
  636. formats.medium_format = parse_pattern("medium"sv);
  637. formats.short_format = parse_pattern("short"sv);
  638. };
  639. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  640. // The generic calendar is not a supported Unicode calendar key, so skip it:
  641. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  642. if (calendar_name == "generic"sv)
  643. return;
  644. auto& calendar = ensure_calendar(calendar_name);
  645. if (!locale_data.calendars.contains_slow(calendar_name))
  646. locale_data.calendars.append(calendar_name);
  647. Vector<CalendarPattern> date_formats;
  648. Vector<CalendarPattern> time_formats;
  649. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  650. auto const& date_skeletons_object = value.as_object().get("dateSkeletons"sv);
  651. parse_patterns(calendar.date_formats, date_formats_object.as_object(), date_skeletons_object.as_object(), &date_formats);
  652. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  653. auto const& time_skeletons_object = value.as_object().get("timeSkeletons"sv);
  654. parse_patterns(calendar.time_formats, time_formats_object.as_object(), time_skeletons_object.as_object(), &time_formats);
  655. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  656. parse_patterns(calendar.date_time_formats, date_time_formats_object.as_object(), JsonObject {}, nullptr);
  657. auto const& available_formats = date_time_formats_object.as_object().get("availableFormats"sv);
  658. available_formats.as_object().for_each_member([&](auto const& skeleton, JsonValue const& pattern) {
  659. auto pattern_index = parse_date_time_pattern(pattern.as_string(), skeleton, locale_data);
  660. if (!pattern_index.has_value())
  661. return;
  662. auto const& format = locale_data.unique_patterns.get(*pattern_index);
  663. if (format.contains_only_date_fields())
  664. date_formats.append(format);
  665. else if (format.contains_only_time_fields())
  666. time_formats.append(format);
  667. if (!calendar.available_formats.contains_slow(*pattern_index))
  668. calendar.available_formats.append(*pattern_index);
  669. });
  670. generate_missing_patterns(calendar, move(date_formats), move(time_formats), locale_data);
  671. parse_calendar_symbols(calendar, value.as_object(), locale_data);
  672. });
  673. return {};
  674. }
  675. static ErrorOr<void> parse_time_zone_names(String locale_time_zone_names_path, UnicodeLocaleData& locale_data, Locale& locale)
  676. {
  677. LexicalPath time_zone_names_path(move(locale_time_zone_names_path));
  678. time_zone_names_path = time_zone_names_path.append("timeZoneNames.json"sv);
  679. auto time_zone_names_file = TRY(Core::File::open(time_zone_names_path.string(), Core::OpenMode::ReadOnly));
  680. auto time_zone_names = TRY(JsonValue::from_string(time_zone_names_file->read_all()));
  681. auto const& main_object = time_zone_names.as_object().get("main"sv);
  682. auto const& locale_object = main_object.as_object().get(time_zone_names_path.parent().basename());
  683. auto const& dates_object = locale_object.as_object().get("dates"sv);
  684. auto const& time_zone_names_object = dates_object.as_object().get("timeZoneNames"sv);
  685. auto const& meta_zone_object = time_zone_names_object.as_object().get("metazone"sv);
  686. if (meta_zone_object.is_null())
  687. return {};
  688. auto parse_name = [&](StringView type, JsonObject const& meta_zone_object) -> Optional<StringIndexType> {
  689. auto const& names = meta_zone_object.get(type);
  690. if (!names.is_object())
  691. return {};
  692. auto const& daylight = names.as_object().get("daylight"sv);
  693. if (daylight.is_string())
  694. return locale_data.unique_strings.ensure(daylight.as_string());
  695. auto const& standard = names.as_object().get("standard"sv);
  696. if (standard.is_string())
  697. return locale_data.unique_strings.ensure(standard.as_string());
  698. return {};
  699. };
  700. auto parse_time_zone = [&](StringView meta_zone, JsonObject const& meta_zone_object) {
  701. auto golden_zone = locale_data.meta_zones.get(meta_zone).value();
  702. TimeZone time_zone { .time_zone = golden_zone };
  703. if (auto long_name = parse_name("long"sv, meta_zone_object); long_name.has_value())
  704. time_zone.long_name = long_name.value();
  705. if (auto short_name = parse_name("short"sv, meta_zone_object); short_name.has_value())
  706. time_zone.short_name = short_name.value();
  707. auto const& time_zone_name = locale_data.unique_strings.get(golden_zone);
  708. if (!locale_data.time_zones.contains_slow(time_zone_name))
  709. locale_data.time_zones.append(time_zone_name);
  710. locale.time_zones.set(time_zone_name, move(time_zone));
  711. };
  712. meta_zone_object.as_object().for_each_member([&](auto const& meta_zone, JsonValue const& value) {
  713. parse_time_zone(meta_zone, value.as_object());
  714. });
  715. // The long and short names for UTC are not under the "timeZoneNames/metazone" object, but are under "timeZoneNames/zone/Etc".
  716. auto const& zone_object = time_zone_names_object.as_object().get("zone"sv);
  717. auto const& etc_object = zone_object.as_object().get("Etc"sv);
  718. auto const& utc_object = etc_object.as_object().get("UTC"sv);
  719. parse_time_zone("UTC"sv, utc_object.as_object());
  720. return {};
  721. }
  722. static ErrorOr<void> parse_all_locales(String core_path, String dates_path, UnicodeLocaleData& locale_data)
  723. {
  724. TRY(parse_hour_cycles(core_path, locale_data));
  725. TRY(parse_meta_zones(move(core_path), locale_data));
  726. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  727. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  728. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  729. StringBuilder builder;
  730. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  731. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  732. builder.appendff("-{}", script);
  733. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  734. builder.appendff("-{}", region);
  735. return builder.build();
  736. };
  737. while (dates_iterator.has_next()) {
  738. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  739. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  740. auto language = TRY(remove_variants_from_path(dates_path));
  741. auto& locale = locale_data.locales.ensure(language);
  742. while (calendars_iterator.has_next()) {
  743. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  744. TRY(parse_calendars(move(calendars_path), locale_data, locale));
  745. }
  746. TRY(parse_time_zone_names(move(dates_path), locale_data, locale));
  747. }
  748. return {};
  749. }
  750. static String format_identifier(StringView owner, String identifier)
  751. {
  752. identifier = identifier.replace("-"sv, "_"sv, true);
  753. identifier = identifier.replace("/"sv, "_"sv, true);
  754. if (all_of(identifier, is_ascii_digit))
  755. return String::formatted("{}_{}", owner[0], identifier);
  756. if (is_ascii_lower_alpha(identifier[0]))
  757. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  758. return identifier;
  759. }
  760. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  761. {
  762. StringBuilder builder;
  763. SourceGenerator generator { builder };
  764. generator.append(R"~~~(
  765. #pragma once
  766. #include <AK/Optional.h>
  767. #include <AK/StringView.h>
  768. #include <LibUnicode/Forward.h>
  769. namespace Unicode {
  770. )~~~");
  771. generate_enum(generator, format_identifier, "Calendar"sv, {}, locale_data.calendars, locale_data.calendar_aliases);
  772. generate_enum(generator, format_identifier, "HourCycleRegion"sv, {}, locale_data.hour_cycle_regions);
  773. generate_enum(generator, format_identifier, "CalendarSymbol"sv, {}, locale_data.symbols);
  774. generate_enum(generator, format_identifier, "TimeZone"sv, {}, locale_data.time_zones);
  775. generator.append(R"~~~(
  776. namespace Detail {
  777. Optional<Calendar> calendar_from_string(StringView calendar);
  778. Optional<HourCycleRegion> hour_cycle_region_from_string(StringView hour_cycle_region);
  779. Vector<Unicode::HourCycle> get_regional_hour_cycles(StringView region);
  780. Optional<Unicode::CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar);
  781. Optional<Unicode::CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar);
  782. Optional<Unicode::CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar);
  783. Vector<Unicode::CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar);
  784. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Era value);
  785. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Month value);
  786. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Weekday value);
  787. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::DayPeriod value);
  788. Optional<TimeZone> time_zone_from_string(StringView time_zone);
  789. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style);
  790. }
  791. }
  792. )~~~");
  793. VERIFY(file.write(generator.as_string_view()));
  794. }
  795. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  796. {
  797. StringBuilder builder;
  798. SourceGenerator generator { builder };
  799. generator.set("string_index_type"sv, s_string_index_type);
  800. generator.set("calendar_pattern_index_type"sv, s_calendar_pattern_index_type);
  801. generator.set("calendar_symbols_size"sv, String::number(locale_data.symbols.size()));
  802. generator.append(R"~~~(
  803. #include <AK/Array.h>
  804. #include <AK/BinarySearch.h>
  805. #include <LibUnicode/DateTimeFormat.h>
  806. #include <LibUnicode/Locale.h>
  807. #include <LibUnicode/UnicodeDateTimeFormat.h>
  808. namespace Unicode::Detail {
  809. )~~~");
  810. locale_data.unique_strings.generate(generator);
  811. generator.append(R"~~~(
  812. struct CalendarPattern {
  813. Unicode::CalendarPattern to_unicode_calendar_pattern() const {
  814. Unicode::CalendarPattern calendar_pattern {};
  815. calendar_pattern.skeleton = s_string_list[skeleton];
  816. calendar_pattern.pattern = s_string_list[pattern];
  817. if (pattern12 != 0)
  818. calendar_pattern.pattern12 = s_string_list[pattern12];
  819. if (era != -1)
  820. calendar_pattern.era = static_cast<Unicode::CalendarPatternStyle>(era);
  821. if (year != -1)
  822. calendar_pattern.year = static_cast<Unicode::CalendarPatternStyle>(year);
  823. if (month != -1)
  824. calendar_pattern.month = static_cast<Unicode::CalendarPatternStyle>(month);
  825. if (weekday != -1)
  826. calendar_pattern.weekday = static_cast<Unicode::CalendarPatternStyle>(weekday);
  827. if (day != -1)
  828. calendar_pattern.day = static_cast<Unicode::CalendarPatternStyle>(day);
  829. if (day_period != -1)
  830. calendar_pattern.day_period = static_cast<Unicode::CalendarPatternStyle>(day_period);
  831. if (hour != -1)
  832. calendar_pattern.hour = static_cast<Unicode::CalendarPatternStyle>(hour);
  833. if (minute != -1)
  834. calendar_pattern.minute = static_cast<Unicode::CalendarPatternStyle>(minute);
  835. if (second != -1)
  836. calendar_pattern.second = static_cast<Unicode::CalendarPatternStyle>(second);
  837. if (fractional_second_digits != -1)
  838. calendar_pattern.fractional_second_digits = static_cast<u8>(fractional_second_digits);
  839. if (time_zone_name != -1)
  840. calendar_pattern.time_zone_name = static_cast<Unicode::CalendarPatternStyle>(time_zone_name);
  841. return calendar_pattern;
  842. }
  843. @string_index_type@ skeleton { 0 };
  844. @string_index_type@ pattern { 0 };
  845. @string_index_type@ pattern12 { 0 };
  846. i8 era { -1 };
  847. i8 year { -1 };
  848. i8 month { -1 };
  849. i8 weekday { -1 };
  850. i8 day { -1 };
  851. i8 day_period { -1 };
  852. i8 hour { -1 };
  853. i8 minute { -1 };
  854. i8 second { -1 };
  855. i8 fractional_second_digits { -1 };
  856. i8 time_zone_name { -1 };
  857. };
  858. )~~~");
  859. locale_data.unique_patterns.generate(generator, "CalendarPattern"sv, "s_calendar_patterns"sv, 10);
  860. generator.append(R"~~~(
  861. struct CalendarFormat {
  862. Unicode::CalendarFormat to_unicode_calendar_format() const {
  863. Unicode::CalendarFormat calendar_format {};
  864. calendar_format.full_format = s_calendar_patterns[full_format].to_unicode_calendar_pattern();
  865. calendar_format.long_format = s_calendar_patterns[long_format].to_unicode_calendar_pattern();
  866. calendar_format.medium_format = s_calendar_patterns[medium_format].to_unicode_calendar_pattern();
  867. calendar_format.short_format = s_calendar_patterns[short_format].to_unicode_calendar_pattern();
  868. return calendar_format;
  869. }
  870. @calendar_pattern_index_type@ full_format { 0 };
  871. @calendar_pattern_index_type@ long_format { 0 };
  872. @calendar_pattern_index_type@ medium_format { 0 };
  873. @calendar_pattern_index_type@ short_format { 0 };
  874. };
  875. using CalendarSymbols = Span<@string_index_type@ const>;
  876. struct CalendarData {
  877. @string_index_type@ calendar { 0 };
  878. CalendarFormat date_formats {};
  879. CalendarFormat time_formats {};
  880. CalendarFormat date_time_formats {};
  881. Span<@calendar_pattern_index_type@ const> available_formats {};
  882. Array<Span<CalendarSymbols const>, @calendar_symbols_size@> symbols {};
  883. };
  884. struct TimeZoneData {
  885. @string_index_type@ time_zone { 0 };
  886. @string_index_type@ long_name { 0 };
  887. @string_index_type@ short_name { 0 };
  888. };
  889. )~~~");
  890. auto append_calendar_format = [&](auto const& calendar_format) {
  891. generator.set("full_format", String::number(calendar_format.full_format));
  892. generator.set("long_format", String::number(calendar_format.long_format));
  893. generator.set("medium_format", String::number(calendar_format.medium_format));
  894. generator.set("short_format", String::number(calendar_format.short_format));
  895. generator.append("{ @full_format@, @long_format@, @medium_format@, @short_format@ },");
  896. };
  897. auto append_calendar_symbols_for_style = [&](auto name, auto style, auto const& symbols) {
  898. name = String::formatted("{}_{}", name, style);
  899. generator.set("name", name);
  900. generator.set("size", String::number(symbols.size()));
  901. generator.append(R"~~~(
  902. static constexpr Array<@string_index_type@, @size@> @name@ { {)~~~");
  903. bool first = true;
  904. for (auto symbol : symbols) {
  905. generator.append(first ? " " : ", ");
  906. generator.append(String::number(symbol));
  907. first = false;
  908. }
  909. generator.append(" } };");
  910. return name;
  911. };
  912. auto append_calendar_symbols = [&](auto name, auto symbol, auto const& symbols) {
  913. name = String::formatted("{}_{}", name, symbol.to_lowercase());
  914. auto narrow_symbols = append_calendar_symbols_for_style(name, "narrow"sv, symbols.narrow_symbols);
  915. auto short_symbols = append_calendar_symbols_for_style(name, "short"sv, symbols.short_symbols);
  916. auto long_symbols = append_calendar_symbols_for_style(name, "long"sv, symbols.long_symbols);
  917. generator.set("narrow_symbols", move(narrow_symbols));
  918. generator.set("short_symbols", move(short_symbols));
  919. generator.set("long_symbols", move(long_symbols));
  920. generator.set("name", name);
  921. generator.append(R"~~~(
  922. static constexpr Array<CalendarSymbols, 3> @name@ { @narrow_symbols@.span(), @short_symbols@.span(), @long_symbols@.span() };
  923. )~~~");
  924. return name;
  925. };
  926. auto append_calendars = [&](String name, auto const& calendars) {
  927. auto format_name = [&](StringView calendar_key, StringView type) {
  928. return String::formatted("{}_{}_{}", name, calendar_key, type);
  929. };
  930. Vector<String> symbols_names;
  931. for (auto const& calendar_key : locale_data.calendars) {
  932. auto const& calendar = calendars.find(calendar_key)->value;
  933. generator.set("name", format_name(calendar_key, "formats"sv));
  934. generator.set("size", String::number(calendar.available_formats.size()));
  935. generator.append(R"~~~(
  936. static constexpr Array<@calendar_pattern_index_type@, @size@> @name@ { {)~~~");
  937. bool first = true;
  938. for (auto format : calendar.available_formats) {
  939. generator.append(first ? " " : ", ");
  940. generator.append(String::number(format));
  941. first = false;
  942. }
  943. generator.append(" } };\n");
  944. auto symbols_name = format_name(calendar_key, "symbols"sv);
  945. for (auto const& symbol_key : locale_data.symbols) {
  946. auto const& symbols = calendar.symbols.find(symbol_key)->value;
  947. auto name = append_calendar_symbols(symbols_name, symbol_key, symbols);
  948. symbols_names.append(name);
  949. }
  950. }
  951. generator.set("name", name);
  952. generator.set("size", String::number(calendars.size()));
  953. generator.append(R"~~~(
  954. static constexpr Array<CalendarData, @size@> @name@ { {)~~~");
  955. for (auto const& calendar_key : locale_data.calendars) {
  956. auto const& calendar = calendars.find(calendar_key)->value;
  957. generator.set("name", format_name(calendar_key, "formats"sv));
  958. generator.set("calendar"sv, String::number(calendar.calendar));
  959. generator.append(R"~~~(
  960. { @calendar@, )~~~");
  961. append_calendar_format(calendar.date_formats);
  962. generator.append(" ");
  963. append_calendar_format(calendar.time_formats);
  964. generator.append(" ");
  965. append_calendar_format(calendar.date_time_formats);
  966. generator.append(" @name@.span(), {");
  967. bool first = true;
  968. for (auto const& symbols_name : symbols_names) {
  969. generator.append(first ? " " : ", ");
  970. generator.append(String::formatted("{}.span()", symbols_name));
  971. first = false;
  972. }
  973. generator.append(" } },");
  974. }
  975. generator.append(R"~~~(
  976. } };
  977. )~~~");
  978. };
  979. auto append_time_zones = [&](String name, auto const& time_zones) {
  980. generator.set("name", name);
  981. generator.set("size", String::number(locale_data.time_zones.size()));
  982. generator.append(R"~~~(
  983. static constexpr Array<TimeZoneData, @size@> @name@ { {)~~~");
  984. constexpr size_t max_values_per_row = 20;
  985. size_t values_in_current_row = 0;
  986. for (auto const& time_zone_key : locale_data.time_zones) {
  987. auto time_zone = time_zones.find(time_zone_key);
  988. if (values_in_current_row++ > 0)
  989. generator.append(" ");
  990. if (time_zone == time_zones.end()) {
  991. generator.append("{},");
  992. } else {
  993. generator.set("time_zone", String::number(time_zone->value.time_zone));
  994. generator.set("long_name", String::number(time_zone->value.long_name));
  995. generator.set("short_name", String::number(time_zone->value.short_name));
  996. generator.append("{ @time_zone@, @long_name@, @short_name@ },");
  997. }
  998. if (values_in_current_row == max_values_per_row) {
  999. values_in_current_row = 0;
  1000. generator.append("\n ");
  1001. }
  1002. }
  1003. generator.append(R"~~~(
  1004. } };
  1005. )~~~");
  1006. };
  1007. auto append_hour_cycles = [&](String name, auto const& hour_cycle_region) {
  1008. auto const& hour_cycles = locale_data.hour_cycles.find(hour_cycle_region)->value;
  1009. generator.set("name", name);
  1010. generator.set("size", String::number(hour_cycles.size()));
  1011. generator.append(R"~~~(
  1012. static constexpr Array<u8, @size@> @name@ { { )~~~");
  1013. for (auto hour_cycle : hour_cycles) {
  1014. generator.set("hour_cycle", String::number(static_cast<u8>(hour_cycle)));
  1015. generator.append("@hour_cycle@, ");
  1016. }
  1017. generator.append("} };");
  1018. };
  1019. generate_mapping(generator, locale_data.locales, "CalendarData"sv, "s_calendars"sv, "s_calendars_{}", [&](auto const& name, auto const& value) { append_calendars(name, value.calendars); });
  1020. 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); });
  1021. 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); });
  1022. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1023. HashValueMap<String> hashes;
  1024. hashes.ensure_capacity(values.size());
  1025. for (auto const& value : values)
  1026. hashes.set(value.hash(), format_identifier(enum_title, value));
  1027. for (auto const& alias : aliases)
  1028. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1029. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1030. };
  1031. append_from_string("Calendar"sv, "calendar"sv, locale_data.calendars, locale_data.calendar_aliases);
  1032. append_from_string("HourCycleRegion"sv, "hour_cycle_region"sv, locale_data.hour_cycle_regions);
  1033. append_from_string("TimeZone"sv, "time_zone"sv, locale_data.time_zones);
  1034. generator.append(R"~~~(
  1035. Vector<Unicode::HourCycle> get_regional_hour_cycles(StringView region)
  1036. {
  1037. auto region_value = hour_cycle_region_from_string(region);
  1038. if (!region_value.has_value())
  1039. return {};
  1040. auto region_index = to_underlying(*region_value);
  1041. auto const& regional_hour_cycles = s_hour_cycles.at(region_index);
  1042. Vector<Unicode::HourCycle> hour_cycles;
  1043. hour_cycles.ensure_capacity(regional_hour_cycles.size());
  1044. for (auto hour_cycle : regional_hour_cycles)
  1045. hour_cycles.unchecked_append(static_cast<Unicode::HourCycle>(hour_cycle));
  1046. return hour_cycles;
  1047. }
  1048. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  1049. {
  1050. auto locale_value = locale_from_string(locale);
  1051. if (!locale_value.has_value())
  1052. return nullptr;
  1053. auto calendar_value = calendar_from_string(calendar);
  1054. if (!calendar_value.has_value())
  1055. return nullptr;
  1056. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1057. auto calendar_index = to_underlying(*calendar_value);
  1058. auto const& calendars = s_calendars.at(locale_index);
  1059. return &calendars[calendar_index];
  1060. }
  1061. Optional<Unicode::CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  1062. {
  1063. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr)
  1064. return data->date_formats.to_unicode_calendar_format();
  1065. return {};
  1066. }
  1067. Optional<Unicode::CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  1068. {
  1069. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr)
  1070. return data->time_formats.to_unicode_calendar_format();
  1071. return {};
  1072. }
  1073. Optional<Unicode::CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  1074. {
  1075. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr)
  1076. return data->date_time_formats.to_unicode_calendar_format();
  1077. return {};
  1078. }
  1079. Vector<Unicode::CalendarPattern> get_calendar_available_formats(StringView locale, StringView calendar)
  1080. {
  1081. Vector<Unicode::CalendarPattern> result {};
  1082. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1083. result.ensure_capacity(data->available_formats.size());
  1084. for (auto const& format : data->available_formats)
  1085. result.unchecked_append(s_calendar_patterns[format].to_unicode_calendar_pattern());
  1086. }
  1087. return result;
  1088. }
  1089. static CalendarSymbols find_calendar_symbols(StringView locale, StringView calendar, CalendarSymbol symbol, CalendarPatternStyle style)
  1090. {
  1091. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr) {
  1092. auto symbol_index = to_underlying(symbol);
  1093. auto style_index = to_underlying(style);
  1094. auto symbols = data->symbols.at(symbol_index);
  1095. VERIFY(style_index < symbols.size());
  1096. return symbols.at(style_index);
  1097. }
  1098. return {};
  1099. }
  1100. Optional<StringView> get_calendar_era_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Era value)
  1101. {
  1102. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Era, style);
  1103. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1104. return s_string_list[symbols.at(value_index)];
  1105. return {};
  1106. }
  1107. Optional<StringView> get_calendar_month_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Month value)
  1108. {
  1109. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Month, style);
  1110. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1111. return s_string_list[symbols.at(value_index)];
  1112. return {};
  1113. }
  1114. Optional<StringView> get_calendar_weekday_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::Weekday value)
  1115. {
  1116. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::Weekday, style);
  1117. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1118. return s_string_list[symbols.at(value_index)];
  1119. return {};
  1120. }
  1121. Optional<StringView> get_calendar_day_period_symbol(StringView locale, StringView calendar, CalendarPatternStyle style, Unicode::DayPeriod value)
  1122. {
  1123. auto symbols = find_calendar_symbols(locale, calendar, CalendarSymbol::DayPeriod, style);
  1124. if (auto value_index = to_underlying(value); value_index < symbols.size())
  1125. return s_string_list[symbols.at(value_index)];
  1126. return {};
  1127. }
  1128. static TimeZoneData const* find_time_zone_data(StringView locale, StringView time_zone)
  1129. {
  1130. auto locale_value = locale_from_string(locale);
  1131. if (!locale_value.has_value())
  1132. return nullptr;
  1133. auto time_zone_value = time_zone_from_string(time_zone);
  1134. if (!time_zone_value.has_value())
  1135. return nullptr;
  1136. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1137. auto time_zone_index = to_underlying(*time_zone_value);
  1138. auto const& time_zones = s_time_zones.at(locale_index);
  1139. return &time_zones[time_zone_index];
  1140. }
  1141. Optional<StringView> get_time_zone_name(StringView locale, StringView time_zone, CalendarPatternStyle style)
  1142. {
  1143. if (auto const* data = find_time_zone_data(locale, time_zone); data != nullptr) {
  1144. @string_index_type@ time_zone_index = 0;
  1145. switch (style) {
  1146. case CalendarPatternStyle::Long:
  1147. time_zone_index = data->long_name;
  1148. break;
  1149. case CalendarPatternStyle::Short:
  1150. time_zone_index = data->short_name;
  1151. break;
  1152. default:
  1153. VERIFY_NOT_REACHED();
  1154. }
  1155. if (time_zone_index != 0)
  1156. return s_string_list[time_zone_index];
  1157. }
  1158. return {};
  1159. }
  1160. }
  1161. )~~~");
  1162. VERIFY(file.write(generator.as_string_view()));
  1163. }
  1164. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1165. {
  1166. StringView generated_header_path;
  1167. StringView generated_implementation_path;
  1168. StringView core_path;
  1169. StringView dates_path;
  1170. Core::ArgsParser args_parser;
  1171. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1172. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1173. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1174. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1175. args_parser.parse(arguments);
  1176. auto open_file = [&](StringView path) -> ErrorOr<NonnullRefPtr<Core::File>> {
  1177. if (path.is_empty()) {
  1178. args_parser.print_usage(stderr, arguments.argv[0]);
  1179. return Error::from_string_literal("Must provide all command line options"sv);
  1180. }
  1181. return Core::File::open(path, Core::OpenMode::ReadWrite);
  1182. };
  1183. auto generated_header_file = TRY(open_file(generated_header_path));
  1184. auto generated_implementation_file = TRY(open_file(generated_implementation_path));
  1185. UnicodeLocaleData locale_data;
  1186. TRY(parse_all_locales(core_path, dates_path, locale_data));
  1187. generate_unicode_locale_header(generated_header_file, locale_data);
  1188. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  1189. return 0;
  1190. }