GenerateLocaleData.cpp 81 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647164816491650165116521653165416551656165716581659166016611662166316641665166616671668166916701671167216731674167516761677167816791680168116821683168416851686168716881689169016911692169316941695169616971698169917001701170217031704170517061707170817091710171117121713171417151716171717181719172017211722172317241725172617271728172917301731173217331734173517361737173817391740174117421743174417451746174717481749175017511752175317541755175617571758175917601761176217631764176517661767176817691770177117721773177417751776177717781779178017811782178317841785178617871788178917901791179217931794179517961797179817991800180118021803180418051806180718081809181018111812181318141815181618171818181918201821182218231824182518261827182818291830183118321833183418351836183718381839184018411842184318441845184618471848184918501851185218531854185518561857185818591860186118621863186418651866186718681869187018711872187318741875187618771878187918801881188218831884188518861887188818891890189118921893189418951896189718981899190019011902190319041905
  1. /*
  2. * Copyright (c) 2021-2022, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "../LibUnicode/GeneratorUtil.h" // FIXME: Move this somewhere common.
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/DeprecatedString.h>
  10. #include <AK/Error.h>
  11. #include <AK/Format.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/QuickSort.h>
  18. #include <AK/SourceGenerator.h>
  19. #include <AK/StringBuilder.h>
  20. #include <LibCore/ArgsParser.h>
  21. #include <LibCore/Directory.h>
  22. #include <LibFileSystem/FileSystem.h>
  23. static DeprecatedString format_identifier(StringView owner, DeprecatedString identifier)
  24. {
  25. identifier = identifier.replace("-"sv, "_"sv, ReplaceMode::All);
  26. if (all_of(identifier, is_ascii_digit))
  27. return DeprecatedString::formatted("{}_{}", owner[0], identifier);
  28. if (is_ascii_lower_alpha(identifier[0]))
  29. return DeprecatedString::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  30. return identifier;
  31. }
  32. struct DisplayPattern {
  33. unsigned hash() const
  34. {
  35. return pair_int_hash(locale_pattern, locale_separator);
  36. }
  37. bool operator==(DisplayPattern const& other) const
  38. {
  39. return (locale_pattern == other.locale_pattern)
  40. && (locale_separator == other.locale_separator);
  41. }
  42. size_t locale_pattern { 0 };
  43. size_t locale_separator { 0 };
  44. };
  45. template<>
  46. struct AK::Formatter<DisplayPattern> : Formatter<FormatString> {
  47. ErrorOr<void> format(FormatBuilder& builder, DisplayPattern const& patterns)
  48. {
  49. return Formatter<FormatString>::format(builder,
  50. "{{ {}, {} }}"sv,
  51. patterns.locale_pattern,
  52. patterns.locale_separator);
  53. }
  54. };
  55. template<>
  56. struct AK::Traits<DisplayPattern> : public GenericTraits<DisplayPattern> {
  57. static unsigned hash(DisplayPattern const& p) { return p.hash(); }
  58. };
  59. struct ListPatterns {
  60. unsigned hash() const
  61. {
  62. auto hash = pair_int_hash(type.hash(), style.hash());
  63. hash = pair_int_hash(hash, start);
  64. hash = pair_int_hash(hash, middle);
  65. hash = pair_int_hash(hash, end);
  66. hash = pair_int_hash(hash, pair);
  67. return hash;
  68. }
  69. bool operator==(ListPatterns const& other) const
  70. {
  71. return (type == other.type)
  72. && (style == other.style)
  73. && (start == other.start)
  74. && (middle == other.middle)
  75. && (end == other.end)
  76. && (pair == other.pair);
  77. }
  78. StringView type;
  79. StringView style;
  80. size_t start { 0 };
  81. size_t middle { 0 };
  82. size_t end { 0 };
  83. size_t pair { 0 };
  84. };
  85. template<>
  86. struct AK::Formatter<ListPatterns> : Formatter<FormatString> {
  87. ErrorOr<void> format(FormatBuilder& builder, ListPatterns const& patterns)
  88. {
  89. return Formatter<FormatString>::format(builder,
  90. "{{ ListPatternType::{}, Style::{}, {}, {}, {}, {} }}"sv,
  91. format_identifier({}, patterns.type),
  92. format_identifier({}, patterns.style),
  93. patterns.start,
  94. patterns.middle,
  95. patterns.end,
  96. patterns.pair);
  97. }
  98. };
  99. template<>
  100. struct AK::Traits<ListPatterns> : public GenericTraits<ListPatterns> {
  101. static unsigned hash(ListPatterns const& p) { return p.hash(); }
  102. };
  103. struct TextLayout {
  104. unsigned hash() const
  105. {
  106. return character_order.hash();
  107. }
  108. bool operator==(TextLayout const& other) const
  109. {
  110. return character_order == other.character_order;
  111. }
  112. StringView character_order;
  113. };
  114. template<>
  115. struct AK::Formatter<TextLayout> : Formatter<FormatString> {
  116. ErrorOr<void> format(FormatBuilder& builder, TextLayout const& patterns)
  117. {
  118. return Formatter<FormatString>::format(builder,
  119. "{{ CharacterOrder::{} }}"sv,
  120. format_identifier({}, patterns.character_order));
  121. }
  122. };
  123. template<>
  124. struct AK::Traits<TextLayout> : public GenericTraits<TextLayout> {
  125. static unsigned hash(TextLayout const& t) { return t.hash(); }
  126. };
  127. using LanguageList = Vector<size_t>;
  128. using TerritoryList = Vector<size_t>;
  129. using ScriptList = Vector<size_t>;
  130. using CurrencyList = Vector<size_t>;
  131. using CalendarList = Vector<size_t>;
  132. using DateFieldList = Vector<size_t>;
  133. using KeywordList = Vector<size_t>;
  134. using ListPatternList = Vector<size_t>;
  135. struct LocaleData {
  136. DeprecatedString language;
  137. Optional<DeprecatedString> territory;
  138. Optional<DeprecatedString> variant;
  139. size_t display_patterns { 0 };
  140. size_t languages { 0 };
  141. size_t territories { 0 };
  142. size_t scripts { 0 };
  143. size_t long_currencies { 0 };
  144. size_t short_currencies { 0 };
  145. size_t narrow_currencies { 0 };
  146. size_t numeric_currencies { 0 };
  147. size_t calendars { 0 };
  148. size_t long_date_fields { 0 };
  149. size_t short_date_fields { 0 };
  150. size_t narrow_date_fields { 0 };
  151. size_t calendar_keywords { 0 };
  152. size_t collation_case_keywords { 0 };
  153. size_t collation_numeric_keywords { 0 };
  154. size_t number_system_keywords { 0 };
  155. size_t list_patterns { 0 };
  156. size_t text_layout { 0 };
  157. };
  158. struct LanguageMapping {
  159. CanonicalLanguageID key {};
  160. CanonicalLanguageID alias {};
  161. };
  162. struct CLDR {
  163. UniqueStringStorage unique_strings;
  164. UniqueStorage<DisplayPattern> unique_display_patterns;
  165. UniqueStorage<LanguageList> unique_language_lists;
  166. UniqueStorage<TerritoryList> unique_territory_lists;
  167. UniqueStorage<ScriptList> unique_script_lists;
  168. UniqueStorage<CurrencyList> unique_currency_lists;
  169. UniqueStorage<CalendarList> unique_calendar_lists;
  170. UniqueStorage<DateFieldList> unique_date_field_lists;
  171. UniqueStorage<KeywordList> unique_keyword_lists;
  172. UniqueStorage<ListPatterns> unique_list_patterns;
  173. UniqueStorage<ListPatternList> unique_list_pattern_lists;
  174. UniqueStorage<TextLayout> unique_text_layouts;
  175. HashMap<DeprecatedString, LocaleData> locales;
  176. Vector<Alias> locale_aliases;
  177. Vector<DeprecatedString> languages;
  178. HashMap<StringView, size_t> language_indices;
  179. Vector<DeprecatedString> territories;
  180. HashMap<StringView, size_t> territory_indices;
  181. Vector<DeprecatedString> scripts;
  182. HashMap<StringView, size_t> script_indices;
  183. Vector<DeprecatedString> variants;
  184. HashMap<StringView, size_t> variant_indices;
  185. Vector<DeprecatedString> currencies;
  186. HashMap<StringView, size_t> currency_indices;
  187. Vector<DeprecatedString> date_fields;
  188. HashMap<StringView, size_t> date_fields_indices;
  189. Vector<Alias> date_field_aliases {
  190. // ECMA-402 and the CLDR refer to some date fields with different names. Defining these aliases
  191. // means we can remain agnostic about the naming differences elsewhere.
  192. { "dayperiod"sv, "dayPeriod"sv },
  193. { "week"sv, "weekOfYear"sv },
  194. { "zone"sv, "timeZoneName"sv },
  195. };
  196. HashMap<DeprecatedString, Vector<DeprecatedString>> keywords;
  197. HashMap<DeprecatedString, Vector<Alias>> keyword_aliases;
  198. HashMap<DeprecatedString, DeprecatedString> keyword_names;
  199. Vector<DeprecatedString> list_pattern_types;
  200. Vector<DeprecatedString> character_orders;
  201. HashMap<DeprecatedString, size_t> language_aliases;
  202. HashMap<DeprecatedString, size_t> territory_aliases;
  203. HashMap<DeprecatedString, size_t> script_aliases;
  204. HashMap<DeprecatedString, size_t> variant_aliases;
  205. HashMap<DeprecatedString, size_t> subdivision_aliases;
  206. Vector<LanguageMapping> complex_mappings;
  207. Vector<LanguageMapping> likely_subtags;
  208. size_t max_variant_size { 0 };
  209. };
  210. // Some parsing is expected to fail. For example, the CLDR contains language mappings
  211. // with locales such as "en-GB-oed" that are canonically invalid locale IDs.
  212. #define TRY_OR_DISCARD(expression) \
  213. ({ \
  214. auto&& _temporary_result = (expression); \
  215. if (_temporary_result.is_error()) \
  216. return; \
  217. static_assert(!::AK::Detail::IsLvalueReference<decltype(_temporary_result.release_value())>, \
  218. "Do not return a reference from a fallible expression"); \
  219. _temporary_result.release_value(); \
  220. })
  221. // NOTE: We return a pointer only because ErrorOr cannot store references. You may safely assume the pointer is non-null.
  222. ErrorOr<JsonValue const*> read_json_file_with_cache(DeprecatedString const& path)
  223. {
  224. static HashMap<DeprecatedString, JsonValue> parsed_json_cache;
  225. if (auto parsed_json = parsed_json_cache.get(path); parsed_json.has_value())
  226. return &parsed_json.value();
  227. auto parsed_json = TRY(read_json_file(path));
  228. TRY(parsed_json_cache.try_set(path, move(parsed_json)));
  229. return &parsed_json_cache.get(path).value();
  230. }
  231. static ErrorOr<LanguageMapping> parse_language_mapping(CLDR& cldr, StringView key, StringView alias)
  232. {
  233. auto parsed_key = TRY(CanonicalLanguageID::parse(cldr.unique_strings, key));
  234. auto parsed_alias = TRY(CanonicalLanguageID::parse(cldr.unique_strings, alias));
  235. return LanguageMapping { move(parsed_key), move(parsed_alias) };
  236. }
  237. static ErrorOr<void> parse_core_aliases(DeprecatedString core_supplemental_path, CLDR& cldr)
  238. {
  239. LexicalPath core_aliases_path(move(core_supplemental_path));
  240. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  241. auto core_aliases = TRY(read_json_file(core_aliases_path.string()));
  242. auto const& supplemental_object = core_aliases.as_object().get_object("supplemental"sv).value();
  243. auto const& metadata_object = supplemental_object.get_object("metadata"sv).value();
  244. auto const& alias_object = metadata_object.get_object("alias"sv).value();
  245. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  246. alias_object.for_each_member([&](auto const& key, JsonValue const& value) {
  247. auto alias = value.as_object().get_deprecated_string("_replacement"sv).value();
  248. if (key.contains('-')) {
  249. auto mapping = TRY_OR_DISCARD(parse_language_mapping(cldr, key, alias));
  250. cldr.max_variant_size = max(mapping.key.variants.size(), cldr.max_variant_size);
  251. cldr.max_variant_size = max(mapping.alias.variants.size(), cldr.max_variant_size);
  252. cldr.complex_mappings.append(move(mapping));
  253. } else {
  254. alias_map.set(key, cldr.unique_strings.ensure(alias));
  255. }
  256. });
  257. };
  258. append_aliases(alias_object.get_object("languageAlias"sv).value(), cldr.language_aliases);
  259. append_aliases(alias_object.get_object("territoryAlias"sv).value(), cldr.territory_aliases);
  260. append_aliases(alias_object.get_object("scriptAlias"sv).value(), cldr.script_aliases);
  261. append_aliases(alias_object.get_object("variantAlias"sv).value(), cldr.variant_aliases);
  262. append_aliases(alias_object.get_object("subdivisionAlias"sv).value(), cldr.subdivision_aliases);
  263. return {};
  264. }
  265. static ErrorOr<void> parse_likely_subtags(DeprecatedString core_supplemental_path, CLDR& cldr)
  266. {
  267. LexicalPath likely_subtags_path(move(core_supplemental_path));
  268. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  269. auto likely_subtags = TRY(read_json_file(likely_subtags_path.string()));
  270. auto const& supplemental_object = likely_subtags.as_object().get_object("supplemental"sv).value();
  271. auto const& likely_subtags_object = supplemental_object.get_object("likelySubtags"sv).value();
  272. likely_subtags_object.for_each_member([&](auto const& key, JsonValue const& value) {
  273. auto mapping = TRY_OR_DISCARD(parse_language_mapping(cldr, key, value.as_string()));
  274. cldr.max_variant_size = max(mapping.key.variants.size(), cldr.max_variant_size);
  275. cldr.max_variant_size = max(mapping.alias.variants.size(), cldr.max_variant_size);
  276. cldr.likely_subtags.append(move(mapping));
  277. });
  278. return {};
  279. }
  280. static ErrorOr<void> parse_identity(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  281. {
  282. LexicalPath locale_display_names_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  283. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  284. auto const& locale_display_names = *TRY(read_json_file_with_cache(locale_display_names_path.string()));
  285. auto const& main_object = locale_display_names.as_object().get_object("main"sv).value();
  286. auto const& locale_object = main_object.get_object(locale_display_names_path.parent().basename()).value();
  287. auto const& identity_object = locale_object.get_object("identity"sv).value();
  288. auto const& language_string = identity_object.get_deprecated_string("language"sv).value();
  289. auto const& territory_string = identity_object.get_deprecated_string("territory"sv);
  290. auto const& script_string = identity_object.get_deprecated_string("script"sv);
  291. auto const& variant_string = identity_object.get_deprecated_string("variant"sv);
  292. locale.language = language_string;
  293. if (territory_string.has_value()) {
  294. locale.territory = territory_string.value();
  295. if (!cldr.territory_indices.contains(*locale.territory)) {
  296. cldr.territory_indices.set(*locale.territory, 0);
  297. cldr.territories.append(*locale.territory);
  298. }
  299. }
  300. if (script_string.has_value()) {
  301. auto const& script = script_string.value();
  302. if (!cldr.script_indices.contains(script)) {
  303. cldr.script_indices.set(script, 0);
  304. cldr.scripts.append(script);
  305. }
  306. }
  307. if (variant_string.has_value()) {
  308. locale.variant = variant_string.value();
  309. if (!cldr.variant_indices.contains(*locale.variant)) {
  310. cldr.variant_indices.set(*locale.variant, 0);
  311. cldr.variants.append(*locale.variant);
  312. }
  313. }
  314. return {};
  315. }
  316. static ErrorOr<void> parse_locale_display_patterns(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  317. {
  318. LexicalPath locale_display_names_path(move(locale_path));
  319. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  320. auto const& locale_display_names = *TRY(read_json_file_with_cache(locale_display_names_path.string()));
  321. auto const& main_object = locale_display_names.as_object().get_object("main"sv).value();
  322. auto const& locale_object = main_object.get_object(locale_display_names_path.parent().basename()).value();
  323. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  324. auto const& locale_display_patterns_object = locale_display_names_object.get_object("localeDisplayPattern"sv).value();
  325. auto const& locale_pattern = locale_display_patterns_object.get_deprecated_string("localePattern"sv).value();
  326. auto const& locale_separator = locale_display_patterns_object.get_deprecated_string("localeSeparator"sv).value();
  327. DisplayPattern patterns {};
  328. patterns.locale_pattern = cldr.unique_strings.ensure(locale_pattern);
  329. patterns.locale_separator = cldr.unique_strings.ensure(locale_separator);
  330. locale.display_patterns = cldr.unique_display_patterns.ensure(move(patterns));
  331. return {};
  332. }
  333. static ErrorOr<void> preprocess_languages(DeprecatedString locale_path, CLDR& cldr)
  334. {
  335. LexicalPath languages_path(move(locale_path));
  336. languages_path = languages_path.append("languages.json"sv);
  337. if (!FileSystem::exists(languages_path.string()))
  338. return {};
  339. auto const& locale_languages = *TRY(read_json_file_with_cache(languages_path.string()));
  340. auto const& main_object = locale_languages.as_object().get_object("main"sv).value();
  341. auto const& locale_object = main_object.get_object(languages_path.parent().basename()).value();
  342. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  343. auto const& languages_object = locale_display_names_object.get_object("languages"sv).value();
  344. languages_object.for_each_member([&](auto const& key, auto const&) {
  345. if (!key.contains("-alt-"sv) && !cldr.language_indices.contains(key)) {
  346. cldr.language_indices.set(key, 0);
  347. cldr.languages.append(key);
  348. }
  349. });
  350. return {};
  351. }
  352. static ErrorOr<void> preprocess_currencies(DeprecatedString numbers_path, CLDR& cldr)
  353. {
  354. LexicalPath currencies_path(move(numbers_path));
  355. currencies_path = currencies_path.append("currencies.json"sv);
  356. auto const& locale_currencies = *TRY(read_json_file_with_cache(currencies_path.string()));
  357. auto const& main_object = locale_currencies.as_object().get_object("main"sv).value();
  358. auto const& locale_object = main_object.get_object(currencies_path.parent().basename()).value();
  359. auto const& locale_numbers_object = locale_object.get_object("numbers"sv).value();
  360. auto const& currencies_object = locale_numbers_object.get_object("currencies"sv).value();
  361. currencies_object.for_each_member([&](auto const& key, JsonValue const&) {
  362. if (!cldr.currency_indices.contains(key)) {
  363. cldr.currency_indices.set(key, 0);
  364. cldr.currencies.append(key);
  365. }
  366. });
  367. return {};
  368. }
  369. static bool is_sanctioned_date_field(StringView field)
  370. {
  371. // This is a copy of the units sanctioned for use within ECMA-402, with names adjusted for the names used by the CLDR.
  372. // https://tc39.es/ecma402/#table-validcodeforDateField
  373. return field.is_one_of("era"sv, "year"sv, "quarter"sv, "month"sv, "week"sv, "weekday"sv, "day"sv, "dayperiod"sv, "hour"sv, "minute"sv, "second"sv, "zone"sv);
  374. }
  375. static ErrorOr<void> preprocess_date_fields(DeprecatedString dates_path, CLDR& cldr)
  376. {
  377. LexicalPath date_fields_path(move(dates_path));
  378. date_fields_path = date_fields_path.append("dateFields.json"sv);
  379. auto const& locale_date_fields = *TRY(read_json_file_with_cache(date_fields_path.string()));
  380. auto const& main_object = locale_date_fields.as_object().get_object("main"sv).value();
  381. auto const& locale_object = main_object.get_object(date_fields_path.parent().basename()).value();
  382. auto const& dates_object = locale_object.get_object("dates"sv).value();
  383. auto const& fields_object = dates_object.get_object("fields"sv).value();
  384. fields_object.for_each_member([&](auto const& key, JsonValue const&) {
  385. if (!is_sanctioned_date_field(key))
  386. return;
  387. if (!cldr.date_fields_indices.contains(key)) {
  388. cldr.date_fields_indices.set(key, 0);
  389. cldr.date_fields.append(key);
  390. }
  391. });
  392. return {};
  393. }
  394. static ErrorOr<void> parse_unicode_extension_keywords(DeprecatedString bcp47_path, CLDR& cldr)
  395. {
  396. constexpr auto desired_keywords = Array { "ca"sv, "co"sv, "hc"sv, "kf"sv, "kn"sv, "nu"sv };
  397. auto keywords = TRY(read_json_file(bcp47_path));
  398. auto const& keyword_object = keywords.as_object().get_object("keyword"sv).value();
  399. auto unicode_object = keyword_object.get_object("u"sv);
  400. if (!unicode_object.has_value())
  401. return {};
  402. unicode_object->for_each_member([&](auto const& key, auto const& value) {
  403. if (!desired_keywords.span().contains_slow(key))
  404. return;
  405. auto const& name = value.as_object().get_deprecated_string("_alias"sv).value();
  406. cldr.keyword_names.set(key, name);
  407. auto& keywords = cldr.keywords.ensure(key);
  408. // FIXME: ECMA-402 requires the list of supported collation types to include "default", but
  409. // that type does not appear in collation.json.
  410. if (key == "co" && !keywords.contains_slow("default"sv))
  411. keywords.append("default"sv);
  412. value.as_object().for_each_member([&](auto const& keyword, auto const& properties) {
  413. if (!properties.is_object())
  414. return;
  415. // Filter out values not permitted by ECMA-402.
  416. // https://tc39.es/ecma402/#sec-intl-collator-internal-slots
  417. if (key == "co"sv && keyword.is_one_of("search"sv, "standard"sv))
  418. return;
  419. // https://tc39.es/ecma402/#sec-intl.numberformat-internal-slots
  420. if (key == "nu"sv && keyword.is_one_of("finance"sv, "native"sv, "traditio"sv))
  421. return;
  422. if (auto const& preferred = properties.as_object().get_deprecated_string("_preferred"sv); preferred.has_value()) {
  423. cldr.keyword_aliases.ensure(key).append({ preferred.value(), keyword });
  424. return;
  425. }
  426. if (auto const& alias = properties.as_object().get_deprecated_string("_alias"sv); alias.has_value())
  427. cldr.keyword_aliases.ensure(key).append({ keyword, alias.value() });
  428. keywords.append(keyword);
  429. });
  430. });
  431. return {};
  432. }
  433. static Optional<DeprecatedString> find_keyword_alias(StringView key, StringView calendar, CLDR& cldr)
  434. {
  435. auto it = cldr.keyword_aliases.find(key);
  436. if (it == cldr.keyword_aliases.end())
  437. return {};
  438. auto alias = it->value.find_if([&](auto const& alias) { return calendar == alias.alias; });
  439. if (alias == it->value.end())
  440. return {};
  441. return alias->name;
  442. }
  443. static ErrorOr<void> parse_locale_languages(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  444. {
  445. LexicalPath languages_path(move(locale_path));
  446. languages_path = languages_path.append("languages.json"sv);
  447. LanguageList languages;
  448. languages.resize(cldr.languages.size());
  449. if (!FileSystem::exists(languages_path.string())) {
  450. for (size_t i = 0; i < languages.size(); ++i)
  451. languages[i] = cldr.unique_strings.ensure(cldr.languages[i]);
  452. locale.languages = cldr.unique_language_lists.ensure(move(languages));
  453. return {};
  454. }
  455. auto const& locale_languages = *TRY(read_json_file_with_cache(languages_path.string()));
  456. auto const& main_object = locale_languages.as_object().get_object("main"sv).value();
  457. auto const& locale_object = main_object.get_object(languages_path.parent().basename()).value();
  458. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  459. auto const& languages_object = locale_display_names_object.get_object("languages"sv).value();
  460. languages_object.for_each_member([&](auto const& key, JsonValue const& value) {
  461. if (key.contains("-alt-"sv))
  462. return;
  463. auto index = cldr.language_indices.get(key).value();
  464. languages[index] = cldr.unique_strings.ensure(value.as_string());
  465. });
  466. locale.languages = cldr.unique_language_lists.ensure(move(languages));
  467. return {};
  468. }
  469. static ErrorOr<void> parse_locale_territories(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  470. {
  471. LexicalPath territories_path(move(locale_path));
  472. territories_path = territories_path.append("territories.json"sv);
  473. TerritoryList territories;
  474. territories.resize(cldr.territories.size());
  475. if (!FileSystem::exists(territories_path.string())) {
  476. for (size_t i = 0; i < territories.size(); ++i)
  477. territories[i] = cldr.unique_strings.ensure(cldr.territories[i]);
  478. locale.territories = cldr.unique_territory_lists.ensure(move(territories));
  479. return {};
  480. }
  481. auto locale_territories = TRY(read_json_file(territories_path.string()));
  482. auto const& main_object = locale_territories.as_object().get_object("main"sv).value();
  483. auto const& locale_object = main_object.get_object(territories_path.parent().basename()).value();
  484. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  485. auto const& territories_object = locale_display_names_object.get_object("territories"sv).value();
  486. territories_object.for_each_member([&](auto const& key, JsonValue const& value) {
  487. if (auto index = cldr.territory_indices.get(key); index.has_value())
  488. territories[*index] = cldr.unique_strings.ensure(value.as_string());
  489. });
  490. locale.territories = cldr.unique_territory_lists.ensure(move(territories));
  491. return {};
  492. }
  493. static ErrorOr<void> parse_locale_scripts(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  494. {
  495. LexicalPath scripts_path(move(locale_path));
  496. scripts_path = scripts_path.append("scripts.json"sv);
  497. ScriptList scripts;
  498. scripts.resize(cldr.scripts.size());
  499. if (!FileSystem::exists(scripts_path.string())) {
  500. for (size_t i = 0; i < scripts.size(); ++i)
  501. scripts[i] = cldr.unique_strings.ensure(cldr.scripts[i]);
  502. locale.scripts = cldr.unique_script_lists.ensure(move(scripts));
  503. return {};
  504. }
  505. auto locale_scripts = TRY(read_json_file(scripts_path.string()));
  506. auto const& main_object = locale_scripts.as_object().get_object("main"sv).value();
  507. auto const& locale_object = main_object.get_object(scripts_path.parent().basename()).value();
  508. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  509. auto const& scripts_object = locale_display_names_object.get_object("scripts"sv).value();
  510. scripts_object.for_each_member([&](auto const& key, JsonValue const& value) {
  511. if (auto index = cldr.script_indices.get(key); index.has_value())
  512. scripts[*index] = cldr.unique_strings.ensure(value.as_string());
  513. });
  514. locale.scripts = cldr.unique_script_lists.ensure(move(scripts));
  515. return {};
  516. }
  517. static ErrorOr<void> parse_locale_list_patterns(DeprecatedString misc_path, CLDR& cldr, LocaleData& locale)
  518. {
  519. LexicalPath list_patterns_path(move(misc_path));
  520. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  521. auto locale_list_patterns = TRY(read_json_file(list_patterns_path.string()));
  522. auto const& main_object = locale_list_patterns.as_object().get_object("main"sv).value();
  523. auto const& locale_object = main_object.get_object(list_patterns_path.parent().basename()).value();
  524. auto const& list_patterns_object = locale_object.get_object("listPatterns"sv).value();
  525. auto list_pattern_type = [](StringView key) {
  526. if (key.contains("type-standard"sv))
  527. return "conjunction"sv;
  528. if (key.contains("type-or"sv))
  529. return "disjunction"sv;
  530. if (key.contains("type-unit"sv))
  531. return "unit"sv;
  532. VERIFY_NOT_REACHED();
  533. };
  534. auto list_pattern_style = [](StringView key) {
  535. if (key.contains("short"sv))
  536. return "short"sv;
  537. if (key.contains("narrow"sv))
  538. return "narrow"sv;
  539. return "long"sv;
  540. };
  541. ListPatternList list_patterns;
  542. list_patterns.ensure_capacity(list_patterns_object.size());
  543. list_patterns_object.for_each_member([&](auto const& key, JsonValue const& value) {
  544. auto type = list_pattern_type(key);
  545. auto style = list_pattern_style(key);
  546. auto start = cldr.unique_strings.ensure(value.as_object().get_deprecated_string("start"sv).value());
  547. auto middle = cldr.unique_strings.ensure(value.as_object().get_deprecated_string("middle"sv).value());
  548. auto end = cldr.unique_strings.ensure(value.as_object().get_deprecated_string("end"sv).value());
  549. auto pair = cldr.unique_strings.ensure(value.as_object().get_deprecated_string("2"sv).value());
  550. if (!cldr.list_pattern_types.contains_slow(type))
  551. cldr.list_pattern_types.append(type);
  552. ListPatterns list_pattern { type, style, start, middle, end, pair };
  553. list_patterns.append(cldr.unique_list_patterns.ensure(move(list_pattern)));
  554. });
  555. locale.list_patterns = cldr.unique_list_pattern_lists.ensure(move(list_patterns));
  556. return {};
  557. }
  558. static ErrorOr<void> parse_locale_layout(DeprecatedString misc_path, CLDR& cldr, LocaleData& locale)
  559. {
  560. LexicalPath layout_path(move(misc_path));
  561. layout_path = layout_path.append("layout.json"sv);
  562. auto locale_layout = TRY(read_json_file(layout_path.string()));
  563. auto const& main_object = locale_layout.as_object().get_object("main"sv).value();
  564. auto const& locale_object = main_object.get_object(layout_path.parent().basename()).value();
  565. auto const& layout_object = locale_object.get_object("layout"sv).value();
  566. auto const& orientation_object = layout_object.get_object("orientation"sv).value();
  567. auto text_layout_character_order = [](StringView key) {
  568. if (key == "left-to-right"sv)
  569. return "ltr"sv;
  570. if (key == "right-to-left"sv)
  571. return "rtl"sv;
  572. VERIFY_NOT_REACHED();
  573. };
  574. auto character_order = orientation_object.get_deprecated_string("characterOrder"sv).value();
  575. TextLayout layout {};
  576. layout.character_order = text_layout_character_order(character_order);
  577. if (!cldr.character_orders.contains_slow(layout.character_order))
  578. cldr.character_orders.append(layout.character_order);
  579. locale.text_layout = cldr.unique_text_layouts.ensure(move(layout));
  580. return {};
  581. }
  582. static ErrorOr<void> parse_locale_currencies(DeprecatedString numbers_path, CLDR& cldr, LocaleData& locale)
  583. {
  584. LexicalPath currencies_path(move(numbers_path));
  585. currencies_path = currencies_path.append("currencies.json"sv);
  586. auto const& locale_currencies = *TRY(read_json_file_with_cache(currencies_path.string()));
  587. auto const& main_object = locale_currencies.as_object().get_object("main"sv).value();
  588. auto const& locale_object = main_object.get_object(currencies_path.parent().basename()).value();
  589. auto const& locale_numbers_object = locale_object.get_object("numbers"sv).value();
  590. auto const& currencies_object = locale_numbers_object.get_object("currencies"sv).value();
  591. CurrencyList long_currencies {};
  592. long_currencies.resize(cldr.currencies.size());
  593. CurrencyList short_currencies {};
  594. short_currencies.resize(cldr.currencies.size());
  595. CurrencyList narrow_currencies {};
  596. narrow_currencies.resize(cldr.currencies.size());
  597. CurrencyList numeric_currencies {};
  598. numeric_currencies.resize(cldr.currencies.size());
  599. currencies_object.for_each_member([&](auto const& key, JsonValue const& value) {
  600. auto long_name = value.as_object().get_deprecated_string("displayName"sv).value_or(key);
  601. auto short_name = value.as_object().get_deprecated_string("symbol"sv).value_or(key);
  602. auto narrow_name = value.as_object().get_deprecated_string("symbol-alt-narrow"sv);
  603. auto numeric_name = value.as_object().get_deprecated_string("displayName-count-other"sv);
  604. auto index = cldr.currency_indices.get(key).value();
  605. long_currencies[index] = cldr.unique_strings.ensure(move(long_name));
  606. short_currencies[index] = cldr.unique_strings.ensure(move(short_name));
  607. narrow_currencies[index] = narrow_name.has_value() ? cldr.unique_strings.ensure(narrow_name.release_value()) : 0;
  608. numeric_currencies[index] = numeric_name.has_value() ? cldr.unique_strings.ensure(numeric_name.release_value()) : long_currencies[index];
  609. });
  610. locale.long_currencies = cldr.unique_currency_lists.ensure(move(long_currencies));
  611. locale.short_currencies = cldr.unique_currency_lists.ensure(move(short_currencies));
  612. locale.narrow_currencies = cldr.unique_currency_lists.ensure(move(narrow_currencies));
  613. locale.numeric_currencies = cldr.unique_currency_lists.ensure(move(numeric_currencies));
  614. return {};
  615. }
  616. static ErrorOr<void> parse_locale_calendars(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  617. {
  618. LexicalPath locale_display_names_path(move(locale_path));
  619. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  620. auto const& locale_display_names = *TRY(read_json_file_with_cache(locale_display_names_path.string()));
  621. auto const& main_object = locale_display_names.as_object().get_object("main"sv).value();
  622. auto const& locale_object = main_object.get_object(locale_display_names_path.parent().basename()).value();
  623. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  624. if (!locale_display_names_object.has_object("types"sv))
  625. return {};
  626. auto const& types_object = locale_display_names_object.get_object("types"sv).value();
  627. auto const& calendar_object = types_object.get_object("calendar"sv).value();
  628. auto const& supported_calendars = cldr.keywords.find("ca"sv)->value;
  629. CalendarList calendars;
  630. calendars.resize(supported_calendars.size());
  631. calendar_object.for_each_member([&](auto const& key, auto const& calendar) {
  632. auto index = supported_calendars.find_first_index(key);
  633. if (!index.has_value()) {
  634. auto alias = find_keyword_alias("ca"sv, key, cldr);
  635. index = supported_calendars.find_first_index(*alias);
  636. }
  637. calendars[*index] = cldr.unique_strings.ensure(calendar.as_string());
  638. });
  639. locale.calendars = cldr.unique_calendar_lists.ensure(move(calendars));
  640. return {};
  641. }
  642. static ErrorOr<void> parse_locale_date_fields(DeprecatedString dates_path, CLDR& cldr, LocaleData& locale)
  643. {
  644. LexicalPath date_fields_path(move(dates_path));
  645. date_fields_path = date_fields_path.append("dateFields.json"sv);
  646. auto const& locale_date_fields = *TRY(read_json_file_with_cache(date_fields_path.string()));
  647. auto const& main_object = locale_date_fields.as_object().get_object("main"sv).value();
  648. auto const& locale_object = main_object.get_object(date_fields_path.parent().basename()).value();
  649. auto const& dates_object = locale_object.get_object("dates"sv).value();
  650. auto const& fields_object = dates_object.get_object("fields"sv).value();
  651. DateFieldList long_date_fields {};
  652. long_date_fields.resize(cldr.date_fields.size());
  653. DateFieldList short_date_fields {};
  654. short_date_fields.resize(cldr.date_fields.size());
  655. DateFieldList narrow_date_fields {};
  656. narrow_date_fields.resize(cldr.date_fields.size());
  657. fields_object.for_each_member([&](auto const& key, JsonValue const& value) {
  658. if (!is_sanctioned_date_field(key))
  659. return;
  660. auto const& long_name = value.as_object().get_deprecated_string("displayName"sv).value();
  661. auto const& short_name = fields_object.get_object(DeprecatedString::formatted("{}-short", key))->get_deprecated_string("displayName"sv).value();
  662. auto const& narrow_name = fields_object.get_object(DeprecatedString::formatted("{}-narrow", key))->get_deprecated_string("displayName"sv).value();
  663. auto index = cldr.date_fields_indices.get(key).value();
  664. long_date_fields[index] = cldr.unique_strings.ensure(long_name);
  665. short_date_fields[index] = cldr.unique_strings.ensure(short_name);
  666. narrow_date_fields[index] = cldr.unique_strings.ensure(narrow_name);
  667. });
  668. locale.long_date_fields = cldr.unique_date_field_lists.ensure(move(long_date_fields));
  669. locale.short_date_fields = cldr.unique_date_field_lists.ensure(move(short_date_fields));
  670. locale.narrow_date_fields = cldr.unique_date_field_lists.ensure(move(narrow_date_fields));
  671. return {};
  672. }
  673. static ErrorOr<void> parse_number_system_keywords(DeprecatedString locale_numbers_path, CLDR& cldr, LocaleData& locale)
  674. {
  675. LexicalPath numbers_path(move(locale_numbers_path));
  676. numbers_path = numbers_path.append("numbers.json"sv);
  677. auto numbers = TRY(read_json_file(numbers_path.string()));
  678. auto const& main_object = numbers.as_object().get_object("main"sv).value();
  679. auto const& locale_object = main_object.get_object(numbers_path.parent().basename()).value();
  680. auto const& locale_numbers_object = locale_object.get_object("numbers"sv).value();
  681. auto const& default_numbering_system_object = locale_numbers_object.get_deprecated_string("defaultNumberingSystem"sv).value();
  682. auto const& other_numbering_systems_object = locale_numbers_object.get_object("otherNumberingSystems"sv).value();
  683. KeywordList keywords {};
  684. auto append_numbering_system = [&](DeprecatedString system_name) {
  685. if (auto system_alias = find_keyword_alias("nu"sv, system_name, cldr); system_alias.has_value())
  686. system_name = system_alias.release_value();
  687. auto index = cldr.unique_strings.ensure(move(system_name));
  688. if (!keywords.contains_slow(index))
  689. keywords.append(move(index));
  690. };
  691. append_numbering_system(default_numbering_system_object);
  692. other_numbering_systems_object.for_each_member([&](auto const&, JsonValue const& value) {
  693. append_numbering_system(value.as_string());
  694. });
  695. locale_numbers_object.for_each_member([&](auto const& key, JsonValue const& value) {
  696. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  697. return;
  698. append_numbering_system(value.as_string());
  699. });
  700. locale.number_system_keywords = cldr.unique_keyword_lists.ensure(move(keywords));
  701. return {};
  702. }
  703. static ErrorOr<void> parse_calendar_keywords(DeprecatedString locale_dates_path, CLDR& cldr, LocaleData& locale)
  704. {
  705. KeywordList keywords {};
  706. TRY(Core::Directory::for_each_entry(locale_dates_path, Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  707. if (!entry.name.starts_with("ca-"sv))
  708. return IterationDecision::Continue;
  709. // The generic calendar is not a supported Unicode calendar key, so skip it:
  710. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  711. if (entry.name == "ca-generic.json"sv)
  712. return IterationDecision::Continue;
  713. auto locale_calendars_path = LexicalPath::join(directory.path().string(), entry.name).string();
  714. LexicalPath calendars_path(move(locale_calendars_path));
  715. auto calendars = TRY(read_json_file(calendars_path.string()));
  716. auto const& main_object = calendars.as_object().get_object("main"sv).value();
  717. auto const& locale_object = main_object.get_object(calendars_path.parent().basename()).value();
  718. auto const& dates_object = locale_object.get_object("dates"sv).value();
  719. auto const& calendars_object = dates_object.get_object("calendars"sv).value();
  720. calendars_object.for_each_member([&](auto calendar_name, JsonValue const&) {
  721. if (auto calendar_alias = find_keyword_alias("ca"sv, calendar_name, cldr); calendar_alias.has_value())
  722. calendar_name = calendar_alias.release_value();
  723. keywords.append(cldr.unique_strings.ensure(calendar_name));
  724. });
  725. return IterationDecision::Continue;
  726. }));
  727. locale.calendar_keywords = cldr.unique_keyword_lists.ensure(move(keywords));
  728. return {};
  729. }
  730. static void fill_in_collation_keywords(CLDR& cldr, LocaleData& locale)
  731. {
  732. // FIXME: If collation data becomes available in the CLDR, parse per-locale ordering from there.
  733. auto create_list_with_default_first = [&](auto key, auto default_value) {
  734. auto& values = cldr.keywords.find(key)->value;
  735. quick_sort(values, [&](auto const& lhs, auto const& rhs) {
  736. if (lhs == default_value)
  737. return true;
  738. if (rhs == default_value)
  739. return false;
  740. return lhs < rhs;
  741. });
  742. KeywordList keywords;
  743. keywords.ensure_capacity(values.size());
  744. for (auto const& value : values)
  745. keywords.append(cldr.unique_strings.ensure(value));
  746. return cldr.unique_keyword_lists.ensure(move(keywords));
  747. };
  748. static auto kf_index = create_list_with_default_first("kf"sv, "upper"sv);
  749. static auto kn_index = create_list_with_default_first("kn"sv, "true"sv);
  750. locale.collation_case_keywords = kf_index;
  751. locale.collation_numeric_keywords = kn_index;
  752. }
  753. static ErrorOr<void> parse_default_content_locales(DeprecatedString core_path, CLDR& cldr)
  754. {
  755. LexicalPath default_content_path(move(core_path));
  756. default_content_path = default_content_path.append("defaultContent.json"sv);
  757. auto default_content = TRY(read_json_file(default_content_path.string()));
  758. auto const& default_content_array = default_content.as_object().get_array("defaultContent"sv).value();
  759. default_content_array.for_each([&](JsonValue const& value) {
  760. auto locale = value.as_string();
  761. StringView default_locale = locale;
  762. while (true) {
  763. if (cldr.locales.contains(default_locale))
  764. break;
  765. auto pos = default_locale.find_last('-');
  766. if (!pos.has_value())
  767. return;
  768. default_locale = default_locale.substring_view(0, *pos);
  769. }
  770. if (default_locale != locale)
  771. cldr.locale_aliases.append({ default_locale, move(locale) });
  772. });
  773. return {};
  774. }
  775. static ErrorOr<void> define_aliases_without_scripts(CLDR& cldr)
  776. {
  777. // From ECMA-402: https://tc39.es/ecma402/#sec-internal-slots
  778. //
  779. // For locales that include a script subtag in addition to language and region, the
  780. // corresponding locale without a script subtag must also be supported.
  781. //
  782. // So we define aliases for locales that contain all three subtags, but we must also take
  783. // care to handle when the locale itself or the locale without a script subtag are an alias
  784. // by way of default-content locales.
  785. auto find_alias = [&](auto const& locale) {
  786. return cldr.locale_aliases.find_if([&](auto const& alias) { return locale == alias.alias; });
  787. };
  788. auto append_alias_without_script = [&](auto const& locale) -> ErrorOr<void> {
  789. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, locale));
  790. if ((parsed_locale.language == 0) || (parsed_locale.script == 0) || (parsed_locale.region == 0))
  791. return {};
  792. auto locale_without_script = DeprecatedString::formatted("{}-{}",
  793. cldr.unique_strings.get(parsed_locale.language),
  794. cldr.unique_strings.get(parsed_locale.region));
  795. if (cldr.locales.contains(locale_without_script))
  796. return {};
  797. if (find_alias(locale_without_script) != cldr.locale_aliases.end())
  798. return {};
  799. if (auto it = find_alias(locale); it != cldr.locale_aliases.end())
  800. cldr.locale_aliases.append({ it->name, locale_without_script });
  801. else
  802. cldr.locale_aliases.append({ locale, locale_without_script });
  803. return {};
  804. };
  805. for (auto const& locale : cldr.locales)
  806. TRY(append_alias_without_script(locale.key));
  807. for (auto const& locale : cldr.locale_aliases)
  808. TRY(append_alias_without_script(locale.alias));
  809. return {};
  810. }
  811. static ErrorOr<void> parse_all_locales(DeprecatedString bcp47_path, DeprecatedString core_path, DeprecatedString locale_names_path, DeprecatedString misc_path, DeprecatedString numbers_path, DeprecatedString dates_path, CLDR& cldr)
  812. {
  813. LexicalPath core_supplemental_path(core_path);
  814. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  815. VERIFY(FileSystem::is_directory(core_supplemental_path.string()));
  816. TRY(parse_core_aliases(core_supplemental_path.string(), cldr));
  817. TRY(parse_likely_subtags(core_supplemental_path.string(), cldr));
  818. auto remove_variants_from_path = [&](DeprecatedString path) -> ErrorOr<DeprecatedString> {
  819. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, LexicalPath::basename(path)));
  820. StringBuilder builder;
  821. builder.append(cldr.unique_strings.get(parsed_locale.language));
  822. if (auto script = cldr.unique_strings.get(parsed_locale.script); !script.is_empty())
  823. builder.appendff("-{}", script);
  824. if (auto region = cldr.unique_strings.get(parsed_locale.region); !region.is_empty())
  825. builder.appendff("-{}", region);
  826. return builder.to_deprecated_string();
  827. };
  828. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", locale_names_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  829. auto locale_path = LexicalPath::join(directory.path().string(), entry.name).string();
  830. auto language = TRY(remove_variants_from_path(locale_path));
  831. auto& locale = cldr.locales.ensure(language);
  832. TRY(parse_identity(locale_path, cldr, locale));
  833. return IterationDecision::Continue;
  834. }));
  835. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", locale_names_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  836. auto locale_path = LexicalPath::join(directory.path().string(), entry.name).string();
  837. TRY(preprocess_languages(locale_path, cldr));
  838. return IterationDecision::Continue;
  839. }));
  840. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", numbers_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  841. auto numbers_path = LexicalPath::join(directory.path().string(), entry.name).string();
  842. TRY(preprocess_currencies(numbers_path, cldr));
  843. return IterationDecision::Continue;
  844. }));
  845. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", dates_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  846. auto dates_path = LexicalPath::join(directory.path().string(), entry.name).string();
  847. TRY(preprocess_date_fields(dates_path, cldr));
  848. return IterationDecision::Continue;
  849. }));
  850. auto update_indices = [](auto& keys, auto& indices) {
  851. quick_sort(keys);
  852. for (size_t i = 0; i < keys.size(); ++i)
  853. indices.set(keys[i], i);
  854. };
  855. update_indices(cldr.languages, cldr.language_indices);
  856. update_indices(cldr.territories, cldr.territory_indices);
  857. update_indices(cldr.scripts, cldr.script_indices);
  858. update_indices(cldr.variants, cldr.variant_indices);
  859. update_indices(cldr.currencies, cldr.currency_indices);
  860. update_indices(cldr.date_fields, cldr.date_fields_indices);
  861. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/bcp47", bcp47_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  862. auto bcp47_path = LexicalPath::join(directory.path().string(), entry.name).string();
  863. TRY(parse_unicode_extension_keywords(move(bcp47_path), cldr));
  864. return IterationDecision::Continue;
  865. }));
  866. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", locale_names_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  867. auto locale_path = LexicalPath::join(directory.path().string(), entry.name).string();
  868. auto language = TRY(remove_variants_from_path(locale_path));
  869. auto& locale = cldr.locales.ensure(language);
  870. TRY(parse_locale_display_patterns(locale_path, cldr, locale));
  871. TRY(parse_locale_languages(locale_path, cldr, locale));
  872. TRY(parse_locale_territories(locale_path, cldr, locale));
  873. TRY(parse_locale_scripts(locale_path, cldr, locale));
  874. TRY(parse_locale_calendars(locale_path, cldr, locale));
  875. return IterationDecision::Continue;
  876. }));
  877. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", misc_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  878. auto misc_path = LexicalPath::join(directory.path().string(), entry.name).string();
  879. auto language = TRY(remove_variants_from_path(misc_path));
  880. auto& locale = cldr.locales.ensure(language);
  881. TRY(parse_locale_list_patterns(misc_path, cldr, locale));
  882. TRY(parse_locale_layout(misc_path, cldr, locale));
  883. return IterationDecision::Continue;
  884. }));
  885. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", numbers_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  886. auto numbers_path = LexicalPath::join(directory.path().string(), entry.name).string();
  887. auto language = TRY(remove_variants_from_path(numbers_path));
  888. auto& locale = cldr.locales.ensure(language);
  889. TRY(parse_locale_currencies(numbers_path, cldr, locale));
  890. TRY(parse_number_system_keywords(numbers_path, cldr, locale));
  891. fill_in_collation_keywords(cldr, locale);
  892. return IterationDecision::Continue;
  893. }));
  894. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", dates_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  895. auto dates_path = LexicalPath::join(directory.path().string(), entry.name).string();
  896. auto language = TRY(remove_variants_from_path(dates_path));
  897. auto& locale = cldr.locales.ensure(language);
  898. TRY(parse_locale_date_fields(dates_path, cldr, locale));
  899. TRY(parse_calendar_keywords(dates_path, cldr, locale));
  900. return IterationDecision::Continue;
  901. }));
  902. TRY(parse_default_content_locales(move(core_path), cldr));
  903. TRY(define_aliases_without_scripts(cldr));
  904. return {};
  905. }
  906. static ErrorOr<void> generate_unicode_locale_header(Core::InputBufferedFile& file, CLDR& cldr)
  907. {
  908. StringBuilder builder;
  909. SourceGenerator generator { builder };
  910. generator.append(R"~~~(
  911. #pragma once
  912. #include <AK/Types.h>
  913. namespace Locale {
  914. )~~~");
  915. auto locales = cldr.locales.keys();
  916. auto keywords = cldr.keywords.keys();
  917. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, cldr.locale_aliases);
  918. generate_enum(generator, format_identifier, "Language"sv, {}, cldr.languages);
  919. generate_enum(generator, format_identifier, "Territory"sv, {}, cldr.territories);
  920. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, cldr.scripts);
  921. generate_enum(generator, format_identifier, "Currency"sv, {}, cldr.currencies);
  922. generate_enum(generator, format_identifier, "DateField"sv, {}, cldr.date_fields, cldr.date_field_aliases);
  923. generate_enum(generator, format_identifier, "Variant"sv, {}, cldr.variants);
  924. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, cldr.list_pattern_types);
  925. generate_enum(generator, format_identifier, "CharacterOrder"sv, {}, cldr.character_orders);
  926. generate_enum(generator, format_identifier, "Key"sv, {}, keywords);
  927. for (auto& keyword : cldr.keywords) {
  928. auto const& keyword_name = cldr.keyword_names.find(keyword.key)->value;
  929. auto enum_name = DeprecatedString::formatted("Keyword{}", format_identifier({}, keyword_name));
  930. if (auto aliases = cldr.keyword_aliases.find(keyword.key); aliases != cldr.keyword_aliases.end())
  931. generate_enum(generator, format_identifier, enum_name, {}, keyword.value, aliases->value);
  932. else
  933. generate_enum(generator, format_identifier, enum_name, {}, keyword.value);
  934. }
  935. generator.append(R"~~~(
  936. }
  937. )~~~");
  938. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  939. return {};
  940. }
  941. static ErrorOr<void> generate_unicode_locale_implementation(Core::InputBufferedFile& file, CLDR& cldr)
  942. {
  943. auto string_index_type = cldr.unique_strings.type_that_fits();
  944. StringBuilder builder;
  945. SourceGenerator generator { builder };
  946. generator.set("string_index_type"sv, string_index_type);
  947. generator.set("locales_size"sv, DeprecatedString::number(cldr.locales.size()));
  948. generator.set("territories_size", DeprecatedString::number(cldr.territories.size()));
  949. generator.set("variants_size", DeprecatedString::number(cldr.max_variant_size));
  950. generator.append(R"~~~(
  951. #include <AK/Array.h>
  952. #include <AK/BinarySearch.h>
  953. #include <AK/Optional.h>
  954. #include <AK/Span.h>
  955. #include <AK/String.h>
  956. #include <AK/StringView.h>
  957. #include <AK/Vector.h>
  958. #include <LibLocale/DateTimeFormat.h>
  959. #include <LibLocale/Locale.h>
  960. #include <LibLocale/LocaleData.h>
  961. #include <LibUnicode/CurrencyCode.h>
  962. namespace Locale {
  963. )~~~");
  964. cldr.unique_strings.generate(generator);
  965. generator.append(R"~~~(
  966. struct DisplayPatternImpl {
  967. DisplayPattern to_display_pattern() const
  968. {
  969. DisplayPattern display_patterns {};
  970. display_patterns.locale_pattern = decode_string(locale_pattern);
  971. display_patterns.locale_separator = decode_string(locale_separator);
  972. return display_patterns;
  973. }
  974. @string_index_type@ locale_pattern { 0 };
  975. @string_index_type@ locale_separator { 0 };
  976. };
  977. struct Patterns {
  978. ListPatternType type;
  979. Style style;
  980. @string_index_type@ start { 0 };
  981. @string_index_type@ middle { 0 };
  982. @string_index_type@ end { 0 };
  983. @string_index_type@ pair { 0 };
  984. };
  985. struct TextLayout {
  986. CharacterOrder character_order;
  987. };
  988. )~~~");
  989. generate_available_values(generator, "get_available_calendars"sv, cldr.keywords.find("ca"sv)->value, cldr.keyword_aliases.find("ca"sv)->value,
  990. [](auto calendar) {
  991. // FIXME: Remove this filter when we support all calendars.
  992. return calendar.is_one_of("gregory"sv, "iso8601"sv);
  993. });
  994. generate_available_values(generator, "get_available_collation_case_orderings"sv, cldr.keywords.find("kf"sv)->value, cldr.keyword_aliases.find("kf"sv)->value);
  995. generate_available_values(generator, "get_available_collation_numeric_orderings"sv, cldr.keywords.find("kn"sv)->value, cldr.keyword_aliases.find("kn"sv)->value);
  996. generate_available_values(generator, "get_available_collation_types"sv, cldr.keywords.find("co"sv)->value, cldr.keyword_aliases.find("co"sv)->value,
  997. [](auto collation) {
  998. // FIXME: Remove this filter when we support all collation types.
  999. return collation == "default"sv;
  1000. });
  1001. generate_available_values(generator, "get_available_hour_cycles"sv, cldr.keywords.find("hc"sv)->value);
  1002. generate_available_values(generator, "get_available_number_systems"sv, cldr.keywords.find("nu"sv)->value);
  1003. generate_available_values(generator, "get_available_currencies"sv, cldr.currencies);
  1004. generator.append(R"~~~(
  1005. ReadonlySpan<StringView> get_available_keyword_values(StringView key)
  1006. {
  1007. auto key_value = key_from_string(key);
  1008. if (!key_value.has_value())
  1009. return {};
  1010. switch (*key_value) {
  1011. case Key::Ca:
  1012. return get_available_calendars();
  1013. case Key::Co:
  1014. return get_available_collation_types();
  1015. case Key::Hc:
  1016. return get_available_hour_cycles();
  1017. case Key::Kf:
  1018. return get_available_collation_case_orderings();
  1019. case Key::Kn:
  1020. return get_available_collation_numeric_orderings();
  1021. case Key::Nu:
  1022. return get_available_number_systems();
  1023. }
  1024. VERIFY_NOT_REACHED();
  1025. }
  1026. )~~~");
  1027. cldr.unique_display_patterns.generate(generator, "DisplayPatternImpl"sv, "s_display_patterns"sv, 30);
  1028. cldr.unique_language_lists.generate(generator, string_index_type, "s_language_lists"sv);
  1029. cldr.unique_territory_lists.generate(generator, string_index_type, "s_territory_lists"sv);
  1030. cldr.unique_script_lists.generate(generator, string_index_type, "s_script_lists"sv);
  1031. cldr.unique_currency_lists.generate(generator, string_index_type, "s_currency_lists"sv);
  1032. cldr.unique_calendar_lists.generate(generator, string_index_type, "s_calendar_lists"sv);
  1033. cldr.unique_date_field_lists.generate(generator, string_index_type, "s_date_field_lists"sv);
  1034. cldr.unique_keyword_lists.generate(generator, string_index_type, "s_keyword_lists"sv);
  1035. cldr.unique_list_patterns.generate(generator, "Patterns"sv, "s_list_patterns"sv, 10);
  1036. cldr.unique_list_pattern_lists.generate(generator, cldr.unique_list_patterns.type_that_fits(), "s_list_pattern_lists"sv);
  1037. cldr.unique_text_layouts.generate(generator, "TextLayout"sv, "s_text_layouts"sv, 30);
  1038. auto append_index = [&](auto index) {
  1039. generator.append(DeprecatedString::formatted(", {}", index));
  1040. };
  1041. auto append_list_and_size = [&](auto const& list) {
  1042. if (list.is_empty()) {
  1043. generator.append(", {}, 0");
  1044. return;
  1045. }
  1046. bool first = true;
  1047. generator.append(", {");
  1048. for (auto const& item : list) {
  1049. generator.append(first ? " "sv : ", "sv);
  1050. generator.append(DeprecatedString::number(item));
  1051. first = false;
  1052. }
  1053. generator.append(DeprecatedString::formatted(" }}, {}", list.size()));
  1054. };
  1055. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  1056. generator.set("type", type);
  1057. generator.set("name", name);
  1058. generator.set("size", DeprecatedString::number(keys.size()));
  1059. generator.append(R"~~~(
  1060. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1061. bool first = true;
  1062. for (auto const& key : keys) {
  1063. auto const& value = map.find(key)->value;
  1064. auto mapping = mapping_getter(value);
  1065. generator.append(first ? " "sv : ", "sv);
  1066. generator.append(DeprecatedString::number(mapping));
  1067. first = false;
  1068. }
  1069. generator.append(" } };");
  1070. };
  1071. auto locales = cldr.locales.keys();
  1072. quick_sort(locales);
  1073. append_mapping(locales, cldr.locales, cldr.unique_display_patterns.type_that_fits(), "s_locale_display_patterns"sv, [&](auto const& locale) { return locale.display_patterns; });
  1074. append_mapping(locales, cldr.locales, cldr.unique_language_lists.type_that_fits(), "s_languages"sv, [&](auto const& locale) { return locale.languages; });
  1075. append_mapping(locales, cldr.locales, cldr.unique_territory_lists.type_that_fits(), "s_territories"sv, [&](auto const& locale) { return locale.territories; });
  1076. append_mapping(locales, cldr.locales, cldr.unique_script_lists.type_that_fits(), "s_scripts"sv, [&](auto const& locale) { return locale.scripts; });
  1077. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_long_currencies"sv, [&](auto const& locale) { return locale.long_currencies; });
  1078. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_short_currencies"sv, [&](auto const& locale) { return locale.short_currencies; });
  1079. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_narrow_currencies"sv, [&](auto const& locale) { return locale.narrow_currencies; });
  1080. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_numeric_currencies"sv, [&](auto const& locale) { return locale.numeric_currencies; });
  1081. append_mapping(locales, cldr.locales, cldr.unique_calendar_lists.type_that_fits(), "s_calendars"sv, [&](auto const& locale) { return locale.calendars; });
  1082. append_mapping(locales, cldr.locales, cldr.unique_date_field_lists.type_that_fits(), "s_long_date_fields"sv, [&](auto const& locale) { return locale.long_date_fields; });
  1083. append_mapping(locales, cldr.locales, cldr.unique_date_field_lists.type_that_fits(), "s_short_date_fields"sv, [&](auto const& locale) { return locale.short_date_fields; });
  1084. append_mapping(locales, cldr.locales, cldr.unique_date_field_lists.type_that_fits(), "s_narrow_date_fields"sv, [&](auto const& locale) { return locale.narrow_date_fields; });
  1085. append_mapping(locales, cldr.locales, cldr.unique_keyword_lists.type_that_fits(), "s_calendar_keywords"sv, [&](auto const& locale) { return locale.calendar_keywords; });
  1086. append_mapping(locales, cldr.locales, cldr.unique_keyword_lists.type_that_fits(), "s_collation_case_keywords"sv, [&](auto const& locale) { return locale.collation_case_keywords; });
  1087. append_mapping(locales, cldr.locales, cldr.unique_keyword_lists.type_that_fits(), "s_collation_numeric_keywords"sv, [&](auto const& locale) { return locale.collation_numeric_keywords; });
  1088. append_mapping(locales, cldr.locales, cldr.unique_keyword_lists.type_that_fits(), "s_number_system_keywords"sv, [&](auto const& locale) { return locale.number_system_keywords; });
  1089. append_mapping(locales, cldr.locales, cldr.unique_list_pattern_lists.type_that_fits(), "s_locale_list_patterns"sv, [&](auto const& locale) { return locale.list_patterns; });
  1090. append_mapping(locales, cldr.locales, cldr.unique_text_layouts.type_that_fits(), "s_locale_text_layouts"sv, [&](auto const& locale) { return locale.text_layout; });
  1091. generator.append(R"~~~(
  1092. struct CanonicalLanguageID
  1093. {
  1094. LanguageID to_unicode_language_id() const
  1095. {
  1096. LanguageID language_id {};
  1097. language_id.variants.ensure_capacity(variants_size);
  1098. language_id.language = MUST(String::from_utf8(decode_string(language)));
  1099. if (script != 0)
  1100. language_id.script = MUST(String::from_utf8(decode_string(script)));
  1101. if (region != 0)
  1102. language_id.region = MUST(String::from_utf8(decode_string(region)));
  1103. for (size_t i = 0; i < variants_size; ++i)
  1104. language_id.variants.append(MUST(String::from_utf8(decode_string(variants[i]))));
  1105. return language_id;
  1106. }
  1107. bool matches_variants(Vector<String> const& other_variants) const {
  1108. if (variants_size == 0)
  1109. return true;
  1110. if (other_variants.size() != variants_size)
  1111. return false;
  1112. for (size_t i = 0; i < variants_size; ++i) {
  1113. if (decode_string(variants[i]) != other_variants[i])
  1114. return false;
  1115. }
  1116. return true;
  1117. };
  1118. @string_index_type@ language { 0 };
  1119. @string_index_type@ script { 0 };
  1120. @string_index_type@ region { 0 };
  1121. Array<@string_index_type@, @variants_size@> variants {};
  1122. size_t variants_size { 0 };
  1123. };
  1124. struct LanguageMapping {
  1125. CanonicalLanguageID key;
  1126. CanonicalLanguageID alias;
  1127. };
  1128. )~~~");
  1129. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  1130. generator.set("size", DeprecatedString::number(mappings.size()));
  1131. generator.set("name"sv, name);
  1132. generator.append(R"~~~(
  1133. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  1134. )~~~");
  1135. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  1136. auto const& lhs_language = cldr.unique_strings.get(lhs.key.language);
  1137. auto const& rhs_language = cldr.unique_strings.get(rhs.key.language);
  1138. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  1139. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  1140. return false;
  1141. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  1142. return true;
  1143. return lhs_language < rhs_language;
  1144. });
  1145. for (auto const& mapping : mappings) {
  1146. generator.set("language"sv, DeprecatedString::number(mapping.key.language));
  1147. generator.append(" { { @language@");
  1148. append_index(mapping.key.script);
  1149. append_index(mapping.key.region);
  1150. append_list_and_size(mapping.key.variants);
  1151. generator.set("language"sv, DeprecatedString::number(mapping.alias.language));
  1152. generator.append(" }, { @language@");
  1153. append_index(mapping.alias.script);
  1154. append_index(mapping.alias.region);
  1155. append_list_and_size(mapping.alias.variants);
  1156. generator.append(" } },\n");
  1157. }
  1158. generator.append("} };\n");
  1159. };
  1160. append_complex_mapping("complex_alias"sv, cldr.complex_mappings);
  1161. append_complex_mapping("likely_subtags"sv, cldr.likely_subtags);
  1162. generator.append(R"~~~(
  1163. static LanguageMapping const* resolve_likely_subtag(LanguageID const& language_id)
  1164. {
  1165. // https://unicode.org/reports/tr35/#Likely_Subtags
  1166. enum class State {
  1167. LanguageScriptRegion,
  1168. LanguageRegion,
  1169. LanguageScript,
  1170. Language,
  1171. UndScript,
  1172. Done,
  1173. };
  1174. auto state = State::LanguageScriptRegion;
  1175. while (state != State::Done) {
  1176. LanguageID search_key;
  1177. switch (state) {
  1178. case State::LanguageScriptRegion:
  1179. state = State::LanguageRegion;
  1180. if (!language_id.script.has_value() || !language_id.region.has_value())
  1181. continue;
  1182. search_key.language = *language_id.language;
  1183. search_key.script = *language_id.script;
  1184. search_key.region = *language_id.region;
  1185. break;
  1186. case State::LanguageRegion:
  1187. state = State::LanguageScript;
  1188. if (!language_id.region.has_value())
  1189. continue;
  1190. search_key.language = *language_id.language;
  1191. search_key.region = *language_id.region;
  1192. break;
  1193. case State::LanguageScript:
  1194. state = State::Language;
  1195. if (!language_id.script.has_value())
  1196. continue;
  1197. search_key.language = *language_id.language;
  1198. search_key.script = *language_id.script;
  1199. break;
  1200. case State::Language:
  1201. state = State::UndScript;
  1202. search_key.language = *language_id.language;
  1203. break;
  1204. case State::UndScript:
  1205. state = State::Done;
  1206. if (!language_id.script.has_value())
  1207. continue;
  1208. search_key.language = "und"_string;
  1209. search_key.script = *language_id.script;
  1210. break;
  1211. default:
  1212. VERIFY_NOT_REACHED();
  1213. }
  1214. for (auto const& map : s_likely_subtags) {
  1215. auto const& key_language = decode_string(map.key.language);
  1216. auto const& key_script = decode_string(map.key.script);
  1217. auto const& key_region = decode_string(map.key.region);
  1218. if (key_language != search_key.language)
  1219. continue;
  1220. if (!key_script.is_empty() || search_key.script.has_value()) {
  1221. if (key_script != search_key.script)
  1222. continue;
  1223. }
  1224. if (!key_region.is_empty() || search_key.region.has_value()) {
  1225. if (key_region != search_key.region)
  1226. continue;
  1227. }
  1228. return &map;
  1229. }
  1230. }
  1231. return nullptr;
  1232. }
  1233. )~~~");
  1234. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name, StringView unique_list) {
  1235. generator.set("enum_snake", enum_snake);
  1236. generator.set("from_string_name", from_string_name);
  1237. generator.set("collection_name", collection_name);
  1238. generator.set("unique_list", unique_list);
  1239. generator.append(R"~~~(
  1240. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  1241. {
  1242. auto locale_value = locale_from_string(locale);
  1243. if (!locale_value.has_value())
  1244. return {};
  1245. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  1246. if (!@enum_snake@_value.has_value())
  1247. return {};
  1248. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1249. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  1250. auto mapping_index = @collection_name@.at(locale_index);
  1251. auto const& mappings = @unique_list@.at(mapping_index);
  1252. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  1253. auto @enum_snake@_mapping = decode_string(@enum_snake@_string_index);
  1254. if (@enum_snake@_mapping.is_empty())
  1255. return {};
  1256. return @enum_snake@_mapping;
  1257. }
  1258. )~~~");
  1259. };
  1260. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) -> ErrorOr<void> {
  1261. HashValueMap<DeprecatedString> hashes;
  1262. TRY(hashes.try_ensure_capacity(values.size()));
  1263. for (auto const& value : values)
  1264. hashes.set(value.hash(), format_identifier(enum_title, value));
  1265. for (auto const& alias : aliases)
  1266. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1267. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1268. return {};
  1269. };
  1270. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) -> ErrorOr<void> {
  1271. HashValueMap<size_t> hashes;
  1272. TRY(hashes.try_ensure_capacity(aliases.size()));
  1273. for (auto const& alias : aliases)
  1274. hashes.set(alias.key.hash(), alias.value);
  1275. ValueFromStringOptions options {};
  1276. options.return_type = "StringView"sv;
  1277. options.return_format = "decode_string({})"sv;
  1278. generate_value_from_string(generator, "resolve_{}_alias"sv, string_index_type, enum_snake, move(hashes), options);
  1279. return {};
  1280. };
  1281. TRY(append_from_string("Locale"sv, "locale"sv, cldr.locales.keys(), cldr.locale_aliases));
  1282. TRY(append_from_string("Language"sv, "language"sv, cldr.languages));
  1283. append_mapping_search("language"sv, "language"sv, "s_languages"sv, "s_language_lists"sv);
  1284. TRY(append_alias_search("language"sv, cldr.language_aliases));
  1285. TRY(append_from_string("Territory"sv, "territory"sv, cldr.territories));
  1286. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv, "s_territory_lists"sv);
  1287. TRY(append_alias_search("territory"sv, cldr.territory_aliases));
  1288. TRY(append_from_string("ScriptTag"sv, "script_tag"sv, cldr.scripts));
  1289. append_mapping_search("script"sv, "script_tag"sv, "s_scripts"sv, "s_script_lists"sv);
  1290. TRY(append_alias_search("script_tag"sv, cldr.script_aliases));
  1291. TRY(append_from_string("Currency"sv, "currency"sv, cldr.currencies));
  1292. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv, "s_currency_lists"sv);
  1293. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv, "s_currency_lists"sv);
  1294. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv, "s_currency_lists"sv);
  1295. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv, "s_currency_lists"sv);
  1296. TRY(append_from_string("DateField"sv, "date_field"sv, cldr.date_fields, cldr.date_field_aliases));
  1297. append_mapping_search("long_date_field"sv, "date_field"sv, "s_long_date_fields"sv, "s_date_field_lists"sv);
  1298. append_mapping_search("short_date_field"sv, "date_field"sv, "s_short_date_fields"sv, "s_date_field_lists"sv);
  1299. append_mapping_search("narrow_date_field"sv, "date_field"sv, "s_narrow_date_fields"sv, "s_date_field_lists"sv);
  1300. TRY(append_from_string("Key"sv, "key"sv, cldr.keywords.keys()));
  1301. for (auto const& keyword : cldr.keywords) {
  1302. auto const& keyword_name = cldr.keyword_names.find(keyword.key)->value;
  1303. auto enum_name = DeprecatedString::formatted("Keyword{}", format_identifier({}, keyword_name));
  1304. auto enum_snake = DeprecatedString::formatted("keyword_{}", keyword.key);
  1305. if (auto aliases = cldr.keyword_aliases.find(keyword.key); aliases != cldr.keyword_aliases.end())
  1306. TRY(append_from_string(enum_name, enum_snake, keyword.value, aliases->value));
  1307. else
  1308. TRY(append_from_string(enum_name, enum_snake, keyword.value));
  1309. }
  1310. append_mapping_search("calendar"sv, "keyword_ca"sv, "s_calendars"sv, "s_calendar_lists"sv);
  1311. TRY(append_alias_search("variant"sv, cldr.variant_aliases));
  1312. TRY(append_alias_search("subdivision"sv, cldr.subdivision_aliases));
  1313. TRY(append_from_string("ListPatternType"sv, "list_pattern_type"sv, cldr.list_pattern_types));
  1314. TRY(append_from_string("CharacterOrder"sv, "character_order"sv, cldr.character_orders));
  1315. generate_value_to_string(generator, "{}_to_string"sv, "CharacterOrder"sv, "character_order"sv, format_identifier, cldr.character_orders);
  1316. generator.append(R"~~~(
  1317. static ReadonlySpan<@string_index_type@> find_keyword_indices(StringView locale, StringView key)
  1318. {
  1319. auto locale_value = locale_from_string(locale);
  1320. if (!locale_value.has_value())
  1321. return {};
  1322. auto key_value = key_from_string(key);
  1323. if (!key_value.has_value())
  1324. return {};
  1325. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1326. size_t keywords_index = 0;
  1327. switch (*key_value) {
  1328. case Key::Ca:
  1329. keywords_index = s_calendar_keywords.at(locale_index);
  1330. break;
  1331. case Key::Kf:
  1332. keywords_index = s_collation_case_keywords.at(locale_index);
  1333. break;
  1334. case Key::Kn:
  1335. keywords_index = s_collation_numeric_keywords.at(locale_index);
  1336. break;
  1337. case Key::Nu:
  1338. keywords_index = s_number_system_keywords.at(locale_index);
  1339. break;
  1340. default:
  1341. VERIFY_NOT_REACHED();
  1342. }
  1343. return s_keyword_lists.at(keywords_index);
  1344. }
  1345. Optional<StringView> get_preferred_keyword_value_for_locale(StringView locale, StringView key)
  1346. {
  1347. // Hour cycle keywords are region-based rather than locale-based, so they need to be handled specially.
  1348. // FIXME: Calendar keywords are also region-based, and will need to be handled here when we support non-Gregorian calendars:
  1349. // https://github.com/unicode-org/cldr-json/blob/main/cldr-json/cldr-core/supplemental/calendarPreferenceData.json
  1350. if (key == "hc"sv) {
  1351. auto hour_cycles = MUST(get_locale_hour_cycles(locale));
  1352. if (hour_cycles.is_empty())
  1353. return OptionalNone {};
  1354. return Optional<StringView> { hour_cycle_to_string(hour_cycles[0]) };
  1355. }
  1356. // FIXME: Generate locale-preferred collation data when available in the CLDR.
  1357. if (key == "co"sv) {
  1358. auto collations = get_available_collation_types();
  1359. if (collations.is_empty())
  1360. return OptionalNone {};
  1361. return Optional<StringView> { collations[0] };
  1362. }
  1363. auto keyword_indices = find_keyword_indices(locale, key);
  1364. if (keyword_indices.is_empty())
  1365. return OptionalNone {};
  1366. return Optional<StringView> { decode_string(keyword_indices[0]) };
  1367. }
  1368. Vector<StringView> get_keywords_for_locale(StringView locale, StringView key)
  1369. {
  1370. // Hour cycle keywords are region-based rather than locale-based, so they need to be handled specially.
  1371. // FIXME: Calendar keywords are also region-based, and will need to be handled here when we support non-Gregorian calendars:
  1372. // https://github.com/unicode-org/cldr-json/blob/main/cldr-json/cldr-core/supplemental/calendarPreferenceData.json
  1373. if (key == "hc"sv) {
  1374. auto hour_cycles = MUST(get_locale_hour_cycles(locale));
  1375. Vector<StringView> values;
  1376. values.ensure_capacity(hour_cycles.size());
  1377. for (auto hour_cycle : hour_cycles)
  1378. values.unchecked_append(hour_cycle_to_string(hour_cycle));
  1379. return values;
  1380. }
  1381. // FIXME: Generate locale-preferred collation data when available in the CLDR.
  1382. if (key == "co"sv)
  1383. return Vector<StringView> { get_available_collation_types() };
  1384. auto keyword_indices = find_keyword_indices(locale, key);
  1385. Vector<StringView> keywords;
  1386. keywords.ensure_capacity(keyword_indices.size());
  1387. for (auto keyword : keyword_indices)
  1388. keywords.unchecked_append(decode_string(keyword));
  1389. return keywords;
  1390. }
  1391. Optional<DisplayPattern> get_locale_display_patterns(StringView locale)
  1392. {
  1393. auto locale_value = locale_from_string(locale);
  1394. if (!locale_value.has_value())
  1395. return {};
  1396. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1397. auto display_patterns_index = s_locale_display_patterns.at(locale_index);
  1398. auto const& display_patterns = s_display_patterns.at(display_patterns_index);
  1399. return display_patterns.to_display_pattern();
  1400. }
  1401. Optional<ListPatterns> get_locale_list_patterns(StringView locale, StringView list_pattern_type, Style list_pattern_style)
  1402. {
  1403. auto locale_value = locale_from_string(locale);
  1404. if (!locale_value.has_value())
  1405. return {};
  1406. auto type_value = list_pattern_type_from_string(list_pattern_type);
  1407. if (!type_value.has_value())
  1408. return {};
  1409. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1410. auto list_patterns_list_index = s_locale_list_patterns.at(locale_index);
  1411. auto const& locale_list_patterns = s_list_pattern_lists.at(list_patterns_list_index);
  1412. for (auto list_patterns_index : locale_list_patterns) {
  1413. auto const& list_patterns = s_list_patterns.at(list_patterns_index);
  1414. if ((list_patterns.type == type_value) && (list_patterns.style == list_pattern_style)) {
  1415. auto const& start = decode_string(list_patterns.start);
  1416. auto const& middle = decode_string(list_patterns.middle);
  1417. auto const& end = decode_string(list_patterns.end);
  1418. auto const& pair = decode_string(list_patterns.pair);
  1419. return ListPatterns { start, middle, end, pair };
  1420. }
  1421. }
  1422. return {};
  1423. }
  1424. static Optional<TextLayout> text_layout_for_locale(StringView locale)
  1425. {
  1426. auto locale_value = locale_from_string(locale);
  1427. if (!locale_value.has_value())
  1428. return {};
  1429. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1430. auto text_layouts_index = s_locale_text_layouts.at(locale_index);
  1431. return s_text_layouts.at(text_layouts_index);
  1432. }
  1433. Optional<CharacterOrder> character_order_for_locale(StringView locale)
  1434. {
  1435. if (auto text_layout = text_layout_for_locale(locale); text_layout.has_value())
  1436. return text_layout->character_order;
  1437. return {};
  1438. }
  1439. void resolve_complex_language_aliases(LanguageID& language_id)
  1440. {
  1441. for (auto const& map : s_complex_alias) {
  1442. auto key_language = decode_string(map.key.language);
  1443. auto key_script = decode_string(map.key.script);
  1444. auto key_region = decode_string(map.key.region);
  1445. if ((key_language != language_id.language) && (key_language != "und"sv))
  1446. continue;
  1447. if (!key_script.is_empty() && (key_script != language_id.script))
  1448. continue;
  1449. if (!key_region.is_empty() && (key_region != language_id.region))
  1450. continue;
  1451. if (!map.key.matches_variants(language_id.variants))
  1452. continue;
  1453. auto alias = map.alias.to_unicode_language_id();
  1454. if (alias.language == "und"sv)
  1455. alias.language = move(language_id.language);
  1456. if (key_script.is_empty() && !alias.script.has_value())
  1457. alias.script = move(language_id.script);
  1458. if (key_region.is_empty() && !alias.region.has_value())
  1459. alias.region = move(language_id.region);
  1460. if (map.key.variants_size == 0 && alias.variants.is_empty())
  1461. alias.variants = move(language_id.variants);
  1462. language_id = move(alias);
  1463. break;
  1464. }
  1465. }
  1466. Optional<LanguageID> add_likely_subtags(LanguageID const& language_id)
  1467. {
  1468. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  1469. auto const* likely_subtag = resolve_likely_subtag(language_id);
  1470. if (likely_subtag == nullptr)
  1471. return OptionalNone {};
  1472. auto maximized = language_id;
  1473. auto key_script = decode_string(likely_subtag->key.script);
  1474. auto key_region = decode_string(likely_subtag->key.region);
  1475. auto alias_language = decode_string(likely_subtag->alias.language);
  1476. auto alias_script = decode_string(likely_subtag->alias.script);
  1477. auto alias_region = decode_string(likely_subtag->alias.region);
  1478. if (maximized.language == "und"sv)
  1479. maximized.language = MUST(String::from_utf8(alias_language));
  1480. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  1481. maximized.script = MUST(String::from_utf8(alias_script));
  1482. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  1483. maximized.region = MUST(String::from_utf8(alias_region));
  1484. return maximized;
  1485. }
  1486. Optional<String> resolve_most_likely_territory(LanguageID const& language_id)
  1487. {
  1488. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  1489. return MUST(String::from_utf8(decode_string(likely_subtag->alias.region)));
  1490. return OptionalNone {};
  1491. }
  1492. }
  1493. )~~~");
  1494. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  1495. return {};
  1496. }
  1497. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1498. {
  1499. StringView generated_header_path;
  1500. StringView generated_implementation_path;
  1501. StringView bcp47_path;
  1502. StringView core_path;
  1503. StringView locale_names_path;
  1504. StringView misc_path;
  1505. StringView numbers_path;
  1506. StringView dates_path;
  1507. Core::ArgsParser args_parser;
  1508. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1509. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1510. args_parser.add_option(bcp47_path, "Path to cldr-bcp47 directory", "bcp47-path", 'b', "bcp47-path");
  1511. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1512. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  1513. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  1514. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  1515. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1516. args_parser.parse(arguments);
  1517. auto generated_header_file = TRY(open_file(generated_header_path, Core::File::OpenMode::Write));
  1518. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::File::OpenMode::Write));
  1519. CLDR cldr;
  1520. TRY(parse_all_locales(bcp47_path, core_path, locale_names_path, misc_path, numbers_path, dates_path, cldr));
  1521. TRY(generate_unicode_locale_header(*generated_header_file, cldr));
  1522. TRY(generate_unicode_locale_implementation(*generated_implementation_file, cldr));
  1523. return 0;
  1524. }