GenerateLocaleData.cpp 78 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838
  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. Vector<DeprecatedString> territories;
  179. Vector<DeprecatedString> scripts;
  180. Vector<DeprecatedString> variants;
  181. Vector<DeprecatedString> currencies;
  182. Vector<DeprecatedString> date_fields;
  183. Vector<Alias> date_field_aliases {
  184. // ECMA-402 and the CLDR refer to some date fields with different names. Defining these aliases
  185. // means we can remain agnostic about the naming differences elsewhere.
  186. { "dayperiod"sv, "dayPeriod"sv },
  187. { "week"sv, "weekOfYear"sv },
  188. { "zone"sv, "timeZoneName"sv },
  189. };
  190. HashMap<DeprecatedString, Vector<DeprecatedString>> keywords;
  191. HashMap<DeprecatedString, Vector<Alias>> keyword_aliases;
  192. HashMap<DeprecatedString, DeprecatedString> keyword_names;
  193. Vector<DeprecatedString> list_pattern_types;
  194. Vector<DeprecatedString> character_orders;
  195. HashMap<DeprecatedString, size_t> language_aliases;
  196. HashMap<DeprecatedString, size_t> territory_aliases;
  197. HashMap<DeprecatedString, size_t> script_aliases;
  198. HashMap<DeprecatedString, size_t> variant_aliases;
  199. HashMap<DeprecatedString, size_t> subdivision_aliases;
  200. Vector<LanguageMapping> complex_mappings;
  201. Vector<LanguageMapping> likely_subtags;
  202. size_t max_variant_size { 0 };
  203. };
  204. // Some parsing is expected to fail. For example, the CLDR contains language mappings
  205. // with locales such as "en-GB-oed" that are canonically invalid locale IDs.
  206. #define TRY_OR_DISCARD(expression) \
  207. ({ \
  208. auto&& _temporary_result = (expression); \
  209. if (_temporary_result.is_error()) \
  210. return; \
  211. static_assert(!::AK::Detail::IsLvalueReference<decltype(_temporary_result.release_value())>, \
  212. "Do not return a reference from a fallible expression"); \
  213. _temporary_result.release_value(); \
  214. })
  215. static ErrorOr<LanguageMapping> parse_language_mapping(CLDR& cldr, StringView key, StringView alias)
  216. {
  217. auto parsed_key = TRY(CanonicalLanguageID::parse(cldr.unique_strings, key));
  218. auto parsed_alias = TRY(CanonicalLanguageID::parse(cldr.unique_strings, alias));
  219. return LanguageMapping { move(parsed_key), move(parsed_alias) };
  220. }
  221. static ErrorOr<void> parse_core_aliases(DeprecatedString core_supplemental_path, CLDR& cldr)
  222. {
  223. LexicalPath core_aliases_path(move(core_supplemental_path));
  224. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  225. auto core_aliases = TRY(read_json_file(core_aliases_path.string()));
  226. auto const& supplemental_object = core_aliases.as_object().get_object("supplemental"sv).value();
  227. auto const& metadata_object = supplemental_object.get_object("metadata"sv).value();
  228. auto const& alias_object = metadata_object.get_object("alias"sv).value();
  229. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  230. alias_object.for_each_member([&](auto const& key, JsonValue const& value) {
  231. auto alias = value.as_object().get_deprecated_string("_replacement"sv).value();
  232. if (key.contains('-')) {
  233. auto mapping = TRY_OR_DISCARD(parse_language_mapping(cldr, key, alias));
  234. cldr.max_variant_size = max(mapping.key.variants.size(), cldr.max_variant_size);
  235. cldr.max_variant_size = max(mapping.alias.variants.size(), cldr.max_variant_size);
  236. cldr.complex_mappings.append(move(mapping));
  237. } else {
  238. alias_map.set(key, cldr.unique_strings.ensure(alias));
  239. }
  240. });
  241. };
  242. append_aliases(alias_object.get_object("languageAlias"sv).value(), cldr.language_aliases);
  243. append_aliases(alias_object.get_object("territoryAlias"sv).value(), cldr.territory_aliases);
  244. append_aliases(alias_object.get_object("scriptAlias"sv).value(), cldr.script_aliases);
  245. append_aliases(alias_object.get_object("variantAlias"sv).value(), cldr.variant_aliases);
  246. append_aliases(alias_object.get_object("subdivisionAlias"sv).value(), cldr.subdivision_aliases);
  247. return {};
  248. }
  249. static ErrorOr<void> parse_likely_subtags(DeprecatedString core_supplemental_path, CLDR& cldr)
  250. {
  251. LexicalPath likely_subtags_path(move(core_supplemental_path));
  252. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  253. auto likely_subtags = TRY(read_json_file(likely_subtags_path.string()));
  254. auto const& supplemental_object = likely_subtags.as_object().get_object("supplemental"sv).value();
  255. auto const& likely_subtags_object = supplemental_object.get_object("likelySubtags"sv).value();
  256. likely_subtags_object.for_each_member([&](auto const& key, JsonValue const& value) {
  257. auto mapping = TRY_OR_DISCARD(parse_language_mapping(cldr, key, value.as_string()));
  258. cldr.max_variant_size = max(mapping.key.variants.size(), cldr.max_variant_size);
  259. cldr.max_variant_size = max(mapping.alias.variants.size(), cldr.max_variant_size);
  260. cldr.likely_subtags.append(move(mapping));
  261. });
  262. return {};
  263. }
  264. static ErrorOr<void> parse_identity(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  265. {
  266. LexicalPath locale_display_names_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  267. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  268. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  269. auto const& main_object = locale_display_names.as_object().get_object("main"sv).value();
  270. auto const& locale_object = main_object.get_object(locale_display_names_path.parent().basename()).value();
  271. auto const& identity_object = locale_object.get_object("identity"sv).value();
  272. auto const& language_string = identity_object.get_deprecated_string("language"sv).value();
  273. auto const& territory_string = identity_object.get_deprecated_string("territory"sv);
  274. auto const& script_string = identity_object.get_deprecated_string("script"sv);
  275. auto const& variant_string = identity_object.get_deprecated_string("variant"sv);
  276. locale.language = language_string;
  277. if (territory_string.has_value()) {
  278. locale.territory = territory_string.value();
  279. if (!cldr.territories.contains_slow(*locale.territory))
  280. cldr.territories.append(*locale.territory);
  281. }
  282. if (script_string.has_value()) {
  283. auto script = script_string.value();
  284. if (!cldr.scripts.contains_slow(script))
  285. cldr.scripts.append(script);
  286. }
  287. if (variant_string.has_value()) {
  288. locale.variant = variant_string.value();
  289. if (!cldr.variants.contains_slow(*locale.variant))
  290. cldr.variants.append(*locale.variant);
  291. }
  292. return {};
  293. }
  294. static ErrorOr<void> parse_locale_display_patterns(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  295. {
  296. LexicalPath locale_display_names_path(move(locale_path));
  297. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  298. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  299. auto const& main_object = locale_display_names.as_object().get_object("main"sv).value();
  300. auto const& locale_object = main_object.get_object(locale_display_names_path.parent().basename()).value();
  301. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  302. auto const& locale_display_patterns_object = locale_display_names_object.get_object("localeDisplayPattern"sv).value();
  303. auto const& locale_pattern = locale_display_patterns_object.get_deprecated_string("localePattern"sv).value();
  304. auto const& locale_separator = locale_display_patterns_object.get_deprecated_string("localeSeparator"sv).value();
  305. DisplayPattern patterns {};
  306. patterns.locale_pattern = cldr.unique_strings.ensure(locale_pattern);
  307. patterns.locale_separator = cldr.unique_strings.ensure(locale_separator);
  308. locale.display_patterns = cldr.unique_display_patterns.ensure(move(patterns));
  309. return {};
  310. }
  311. static ErrorOr<void> preprocess_languages(DeprecatedString locale_path, CLDR& cldr)
  312. {
  313. LexicalPath languages_path(move(locale_path));
  314. languages_path = languages_path.append("languages.json"sv);
  315. if (!FileSystem::exists(languages_path.string()))
  316. return {};
  317. auto locale_languages = TRY(read_json_file(languages_path.string()));
  318. auto const& main_object = locale_languages.as_object().get_object("main"sv).value();
  319. auto const& locale_object = main_object.get_object(languages_path.parent().basename()).value();
  320. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  321. auto const& languages_object = locale_display_names_object.get_object("languages"sv).value();
  322. languages_object.for_each_member([&](auto const& key, auto const&) {
  323. if (!key.contains("-alt-"sv) && !cldr.languages.contains_slow(key))
  324. cldr.languages.append(key);
  325. });
  326. return {};
  327. }
  328. static ErrorOr<void> preprocess_currencies(DeprecatedString numbers_path, CLDR& cldr)
  329. {
  330. LexicalPath currencies_path(move(numbers_path));
  331. currencies_path = currencies_path.append("currencies.json"sv);
  332. auto locale_currencies = TRY(read_json_file(currencies_path.string()));
  333. auto const& main_object = locale_currencies.as_object().get_object("main"sv).value();
  334. auto const& locale_object = main_object.get_object(currencies_path.parent().basename()).value();
  335. auto const& locale_numbers_object = locale_object.get_object("numbers"sv).value();
  336. auto const& currencies_object = locale_numbers_object.get_object("currencies"sv).value();
  337. currencies_object.for_each_member([&](auto const& key, JsonValue const&) {
  338. if (!cldr.currencies.contains_slow(key))
  339. cldr.currencies.append(key);
  340. });
  341. return {};
  342. }
  343. static ErrorOr<void> parse_unicode_extension_keywords(DeprecatedString bcp47_path, CLDR& cldr)
  344. {
  345. constexpr auto desired_keywords = Array { "ca"sv, "co"sv, "hc"sv, "kf"sv, "kn"sv, "nu"sv };
  346. auto keywords = TRY(read_json_file(bcp47_path));
  347. auto const& keyword_object = keywords.as_object().get_object("keyword"sv).value();
  348. auto unicode_object = keyword_object.get_object("u"sv);
  349. if (!unicode_object.has_value())
  350. return {};
  351. unicode_object->for_each_member([&](auto const& key, auto const& value) {
  352. if (!desired_keywords.span().contains_slow(key))
  353. return;
  354. auto const& name = value.as_object().get_deprecated_string("_alias"sv).value();
  355. cldr.keyword_names.set(key, name);
  356. auto& keywords = cldr.keywords.ensure(key);
  357. // FIXME: ECMA-402 requires the list of supported collation types to include "default", but
  358. // that type does not appear in collation.json.
  359. if (key == "co" && !keywords.contains_slow("default"sv))
  360. keywords.append("default"sv);
  361. value.as_object().for_each_member([&](auto const& keyword, auto const& properties) {
  362. if (!properties.is_object())
  363. return;
  364. // Filter out values not permitted by ECMA-402.
  365. // https://tc39.es/ecma402/#sec-intl-collator-internal-slots
  366. if (key == "co"sv && keyword.is_one_of("search"sv, "standard"sv))
  367. return;
  368. // https://tc39.es/ecma402/#sec-intl.numberformat-internal-slots
  369. if (key == "nu"sv && keyword.is_one_of("finance"sv, "native"sv, "traditio"sv))
  370. return;
  371. if (auto const& preferred = properties.as_object().get_deprecated_string("_preferred"sv); preferred.has_value()) {
  372. cldr.keyword_aliases.ensure(key).append({ preferred.value(), keyword });
  373. return;
  374. }
  375. if (auto const& alias = properties.as_object().get_deprecated_string("_alias"sv); alias.has_value())
  376. cldr.keyword_aliases.ensure(key).append({ keyword, alias.value() });
  377. keywords.append(keyword);
  378. });
  379. });
  380. return {};
  381. }
  382. static Optional<DeprecatedString> find_keyword_alias(StringView key, StringView calendar, CLDR& cldr)
  383. {
  384. auto it = cldr.keyword_aliases.find(key);
  385. if (it == cldr.keyword_aliases.end())
  386. return {};
  387. auto alias = it->value.find_if([&](auto const& alias) { return calendar == alias.alias; });
  388. if (alias == it->value.end())
  389. return {};
  390. return alias->name;
  391. }
  392. static ErrorOr<void> parse_locale_languages(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  393. {
  394. LexicalPath languages_path(move(locale_path));
  395. languages_path = languages_path.append("languages.json"sv);
  396. LanguageList languages;
  397. languages.resize(cldr.languages.size());
  398. if (!FileSystem::exists(languages_path.string())) {
  399. for (size_t i = 0; i < languages.size(); ++i)
  400. languages[i] = cldr.unique_strings.ensure(cldr.languages[i]);
  401. locale.languages = cldr.unique_language_lists.ensure(move(languages));
  402. return {};
  403. }
  404. auto locale_languages = TRY(read_json_file(languages_path.string()));
  405. auto const& main_object = locale_languages.as_object().get_object("main"sv).value();
  406. auto const& locale_object = main_object.get_object(languages_path.parent().basename()).value();
  407. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  408. auto const& languages_object = locale_display_names_object.get_object("languages"sv).value();
  409. languages_object.for_each_member([&](auto const& key, JsonValue const& value) {
  410. if (key.contains("-alt-"sv))
  411. return;
  412. auto index = cldr.languages.find_first_index(key).value();
  413. languages[index] = cldr.unique_strings.ensure(value.as_string());
  414. });
  415. locale.languages = cldr.unique_language_lists.ensure(move(languages));
  416. return {};
  417. }
  418. static ErrorOr<void> parse_locale_territories(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  419. {
  420. LexicalPath territories_path(move(locale_path));
  421. territories_path = territories_path.append("territories.json"sv);
  422. TerritoryList territories;
  423. territories.resize(cldr.territories.size());
  424. if (!FileSystem::exists(territories_path.string())) {
  425. for (size_t i = 0; i < territories.size(); ++i)
  426. territories[i] = cldr.unique_strings.ensure(cldr.territories[i]);
  427. locale.territories = cldr.unique_territory_lists.ensure(move(territories));
  428. return {};
  429. }
  430. auto locale_territories = TRY(read_json_file(territories_path.string()));
  431. auto const& main_object = locale_territories.as_object().get_object("main"sv).value();
  432. auto const& locale_object = main_object.get_object(territories_path.parent().basename()).value();
  433. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  434. auto const& territories_object = locale_display_names_object.get_object("territories"sv).value();
  435. territories_object.for_each_member([&](auto const& key, JsonValue const& value) {
  436. if (auto index = cldr.territories.find_first_index(key); index.has_value())
  437. territories[*index] = cldr.unique_strings.ensure(value.as_string());
  438. });
  439. locale.territories = cldr.unique_territory_lists.ensure(move(territories));
  440. return {};
  441. }
  442. static ErrorOr<void> parse_locale_scripts(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  443. {
  444. LexicalPath scripts_path(move(locale_path));
  445. scripts_path = scripts_path.append("scripts.json"sv);
  446. ScriptList scripts;
  447. scripts.resize(cldr.scripts.size());
  448. if (!FileSystem::exists(scripts_path.string())) {
  449. for (size_t i = 0; i < scripts.size(); ++i)
  450. scripts[i] = cldr.unique_strings.ensure(cldr.scripts[i]);
  451. locale.scripts = cldr.unique_script_lists.ensure(move(scripts));
  452. return {};
  453. }
  454. auto locale_scripts = TRY(read_json_file(scripts_path.string()));
  455. auto const& main_object = locale_scripts.as_object().get_object("main"sv).value();
  456. auto const& locale_object = main_object.get_object(scripts_path.parent().basename()).value();
  457. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  458. auto const& scripts_object = locale_display_names_object.get_object("scripts"sv).value();
  459. scripts_object.for_each_member([&](auto const& key, JsonValue const& value) {
  460. if (auto index = cldr.scripts.find_first_index(key); index.has_value())
  461. scripts[*index] = cldr.unique_strings.ensure(value.as_string());
  462. });
  463. locale.scripts = cldr.unique_script_lists.ensure(move(scripts));
  464. return {};
  465. }
  466. static ErrorOr<void> parse_locale_list_patterns(DeprecatedString misc_path, CLDR& cldr, LocaleData& locale)
  467. {
  468. LexicalPath list_patterns_path(move(misc_path));
  469. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  470. auto locale_list_patterns = TRY(read_json_file(list_patterns_path.string()));
  471. auto const& main_object = locale_list_patterns.as_object().get_object("main"sv).value();
  472. auto const& locale_object = main_object.get_object(list_patterns_path.parent().basename()).value();
  473. auto const& list_patterns_object = locale_object.get_object("listPatterns"sv).value();
  474. auto list_pattern_type = [](StringView key) {
  475. if (key.contains("type-standard"sv))
  476. return "conjunction"sv;
  477. if (key.contains("type-or"sv))
  478. return "disjunction"sv;
  479. if (key.contains("type-unit"sv))
  480. return "unit"sv;
  481. VERIFY_NOT_REACHED();
  482. };
  483. auto list_pattern_style = [](StringView key) {
  484. if (key.contains("short"sv))
  485. return "short"sv;
  486. if (key.contains("narrow"sv))
  487. return "narrow"sv;
  488. return "long"sv;
  489. };
  490. ListPatternList list_patterns;
  491. list_patterns.ensure_capacity(list_patterns_object.size());
  492. list_patterns_object.for_each_member([&](auto const& key, JsonValue const& value) {
  493. auto type = list_pattern_type(key);
  494. auto style = list_pattern_style(key);
  495. auto start = cldr.unique_strings.ensure(value.as_object().get_deprecated_string("start"sv).value());
  496. auto middle = cldr.unique_strings.ensure(value.as_object().get_deprecated_string("middle"sv).value());
  497. auto end = cldr.unique_strings.ensure(value.as_object().get_deprecated_string("end"sv).value());
  498. auto pair = cldr.unique_strings.ensure(value.as_object().get_deprecated_string("2"sv).value());
  499. if (!cldr.list_pattern_types.contains_slow(type))
  500. cldr.list_pattern_types.append(type);
  501. ListPatterns list_pattern { type, style, start, middle, end, pair };
  502. list_patterns.append(cldr.unique_list_patterns.ensure(move(list_pattern)));
  503. });
  504. locale.list_patterns = cldr.unique_list_pattern_lists.ensure(move(list_patterns));
  505. return {};
  506. }
  507. static ErrorOr<void> parse_locale_layout(DeprecatedString misc_path, CLDR& cldr, LocaleData& locale)
  508. {
  509. LexicalPath layout_path(move(misc_path));
  510. layout_path = layout_path.append("layout.json"sv);
  511. auto locale_layout = TRY(read_json_file(layout_path.string()));
  512. auto const& main_object = locale_layout.as_object().get_object("main"sv).value();
  513. auto const& locale_object = main_object.get_object(layout_path.parent().basename()).value();
  514. auto const& layout_object = locale_object.get_object("layout"sv).value();
  515. auto const& orientation_object = layout_object.get_object("orientation"sv).value();
  516. auto text_layout_character_order = [](StringView key) {
  517. if (key == "left-to-right"sv)
  518. return "ltr"sv;
  519. if (key == "right-to-left"sv)
  520. return "rtl"sv;
  521. VERIFY_NOT_REACHED();
  522. };
  523. auto character_order = orientation_object.get_deprecated_string("characterOrder"sv).value();
  524. TextLayout layout {};
  525. layout.character_order = text_layout_character_order(character_order);
  526. if (!cldr.character_orders.contains_slow(layout.character_order))
  527. cldr.character_orders.append(layout.character_order);
  528. locale.text_layout = cldr.unique_text_layouts.ensure(move(layout));
  529. return {};
  530. }
  531. static ErrorOr<void> parse_locale_currencies(DeprecatedString numbers_path, CLDR& cldr, LocaleData& locale)
  532. {
  533. LexicalPath currencies_path(move(numbers_path));
  534. currencies_path = currencies_path.append("currencies.json"sv);
  535. auto locale_currencies = TRY(read_json_file(currencies_path.string()));
  536. auto const& main_object = locale_currencies.as_object().get_object("main"sv).value();
  537. auto const& locale_object = main_object.get_object(currencies_path.parent().basename()).value();
  538. auto const& locale_numbers_object = locale_object.get_object("numbers"sv).value();
  539. auto const& currencies_object = locale_numbers_object.get_object("currencies"sv).value();
  540. CurrencyList long_currencies {};
  541. long_currencies.resize(cldr.currencies.size());
  542. CurrencyList short_currencies {};
  543. short_currencies.resize(cldr.currencies.size());
  544. CurrencyList narrow_currencies {};
  545. narrow_currencies.resize(cldr.currencies.size());
  546. CurrencyList numeric_currencies {};
  547. numeric_currencies.resize(cldr.currencies.size());
  548. currencies_object.for_each_member([&](auto const& key, JsonValue const& value) {
  549. auto long_name = value.as_object().get_deprecated_string("displayName"sv).value_or(key);
  550. auto short_name = value.as_object().get_deprecated_string("symbol"sv).value_or(key);
  551. auto narrow_name = value.as_object().get_deprecated_string("symbol-alt-narrow"sv);
  552. auto numeric_name = value.as_object().get_deprecated_string("displayName-count-other"sv);
  553. auto index = cldr.currencies.find_first_index(key).value();
  554. long_currencies[index] = cldr.unique_strings.ensure(move(long_name));
  555. short_currencies[index] = cldr.unique_strings.ensure(move(short_name));
  556. narrow_currencies[index] = narrow_name.has_value() ? cldr.unique_strings.ensure(narrow_name.release_value()) : 0;
  557. numeric_currencies[index] = numeric_name.has_value() ? cldr.unique_strings.ensure(numeric_name.release_value()) : long_currencies[index];
  558. });
  559. locale.long_currencies = cldr.unique_currency_lists.ensure(move(long_currencies));
  560. locale.short_currencies = cldr.unique_currency_lists.ensure(move(short_currencies));
  561. locale.narrow_currencies = cldr.unique_currency_lists.ensure(move(narrow_currencies));
  562. locale.numeric_currencies = cldr.unique_currency_lists.ensure(move(numeric_currencies));
  563. return {};
  564. }
  565. static ErrorOr<void> parse_locale_calendars(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  566. {
  567. LexicalPath locale_display_names_path(move(locale_path));
  568. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  569. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  570. auto const& main_object = locale_display_names.as_object().get_object("main"sv).value();
  571. auto const& locale_object = main_object.get_object(locale_display_names_path.parent().basename()).value();
  572. auto const& locale_display_names_object = locale_object.get_object("localeDisplayNames"sv).value();
  573. if (!locale_display_names_object.has_object("types"sv))
  574. return {};
  575. auto const& types_object = locale_display_names_object.get_object("types"sv).value();
  576. auto const& calendar_object = types_object.get_object("calendar"sv).value();
  577. auto const& supported_calendars = cldr.keywords.find("ca"sv)->value;
  578. CalendarList calendars;
  579. calendars.resize(supported_calendars.size());
  580. calendar_object.for_each_member([&](auto const& key, auto const& calendar) {
  581. auto index = supported_calendars.find_first_index(key);
  582. if (!index.has_value()) {
  583. auto alias = find_keyword_alias("ca"sv, key, cldr);
  584. index = supported_calendars.find_first_index(*alias);
  585. }
  586. calendars[*index] = cldr.unique_strings.ensure(calendar.as_string());
  587. });
  588. locale.calendars = cldr.unique_calendar_lists.ensure(move(calendars));
  589. return {};
  590. }
  591. static ErrorOr<void> parse_locale_date_fields(DeprecatedString dates_path, CLDR& cldr, LocaleData& locale)
  592. {
  593. LexicalPath date_fields_path(move(dates_path));
  594. date_fields_path = date_fields_path.append("dateFields.json"sv);
  595. auto locale_date_fields = TRY(read_json_file(date_fields_path.string()));
  596. auto const& main_object = locale_date_fields.as_object().get_object("main"sv).value();
  597. auto const& locale_object = main_object.get_object(date_fields_path.parent().basename()).value();
  598. auto const& dates_object = locale_object.get_object("dates"sv).value();
  599. auto const& fields_object = dates_object.get_object("fields"sv).value();
  600. auto is_sanctioned_field = [](StringView field) {
  601. // This is a copy of the units sanctioned for use within ECMA-402, with names adjusted for the names used by the CLDR.
  602. // https://tc39.es/ecma402/#table-validcodeforDateField
  603. 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);
  604. };
  605. fields_object.for_each_member([&](auto const& key, JsonValue const&) {
  606. if (!is_sanctioned_field(key))
  607. return;
  608. if (!cldr.date_fields.contains_slow(key))
  609. cldr.date_fields.append(key);
  610. });
  611. quick_sort(cldr.date_fields);
  612. DateFieldList long_date_fields {};
  613. long_date_fields.resize(cldr.date_fields.size());
  614. DateFieldList short_date_fields {};
  615. short_date_fields.resize(cldr.date_fields.size());
  616. DateFieldList narrow_date_fields {};
  617. narrow_date_fields.resize(cldr.date_fields.size());
  618. fields_object.for_each_member([&](auto const& key, JsonValue const& value) {
  619. if (!is_sanctioned_field(key))
  620. return;
  621. auto const& long_name = value.as_object().get_deprecated_string("displayName"sv).value();
  622. auto const& short_name = fields_object.get_object(DeprecatedString::formatted("{}-short", key))->get_deprecated_string("displayName"sv).value();
  623. auto const& narrow_name = fields_object.get_object(DeprecatedString::formatted("{}-narrow", key))->get_deprecated_string("displayName"sv).value();
  624. auto index = cldr.date_fields.find_first_index(key).value();
  625. long_date_fields[index] = cldr.unique_strings.ensure(long_name);
  626. short_date_fields[index] = cldr.unique_strings.ensure(short_name);
  627. narrow_date_fields[index] = cldr.unique_strings.ensure(narrow_name);
  628. });
  629. locale.long_date_fields = cldr.unique_date_field_lists.ensure(move(long_date_fields));
  630. locale.short_date_fields = cldr.unique_date_field_lists.ensure(move(short_date_fields));
  631. locale.narrow_date_fields = cldr.unique_date_field_lists.ensure(move(narrow_date_fields));
  632. return {};
  633. }
  634. static ErrorOr<void> parse_number_system_keywords(DeprecatedString locale_numbers_path, CLDR& cldr, LocaleData& locale)
  635. {
  636. LexicalPath numbers_path(move(locale_numbers_path));
  637. numbers_path = numbers_path.append("numbers.json"sv);
  638. auto numbers = TRY(read_json_file(numbers_path.string()));
  639. auto const& main_object = numbers.as_object().get_object("main"sv).value();
  640. auto const& locale_object = main_object.get_object(numbers_path.parent().basename()).value();
  641. auto const& locale_numbers_object = locale_object.get_object("numbers"sv).value();
  642. auto const& default_numbering_system_object = locale_numbers_object.get_deprecated_string("defaultNumberingSystem"sv).value();
  643. auto const& other_numbering_systems_object = locale_numbers_object.get_object("otherNumberingSystems"sv).value();
  644. KeywordList keywords {};
  645. auto append_numbering_system = [&](DeprecatedString system_name) {
  646. if (auto system_alias = find_keyword_alias("nu"sv, system_name, cldr); system_alias.has_value())
  647. system_name = system_alias.release_value();
  648. auto index = cldr.unique_strings.ensure(move(system_name));
  649. if (!keywords.contains_slow(index))
  650. keywords.append(move(index));
  651. };
  652. append_numbering_system(default_numbering_system_object);
  653. other_numbering_systems_object.for_each_member([&](auto const&, JsonValue const& value) {
  654. append_numbering_system(value.as_string());
  655. });
  656. locale_numbers_object.for_each_member([&](auto const& key, JsonValue const& value) {
  657. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  658. return;
  659. append_numbering_system(value.as_string());
  660. });
  661. locale.number_system_keywords = cldr.unique_keyword_lists.ensure(move(keywords));
  662. return {};
  663. }
  664. static ErrorOr<void> parse_calendar_keywords(DeprecatedString locale_dates_path, CLDR& cldr, LocaleData& locale)
  665. {
  666. KeywordList keywords {};
  667. TRY(Core::Directory::for_each_entry(locale_dates_path, Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  668. auto locale_calendars_path = LexicalPath::join(directory.path().string(), entry.name).string();
  669. LexicalPath calendars_path(move(locale_calendars_path));
  670. if (!calendars_path.basename().starts_with("ca-"sv))
  671. return IterationDecision::Continue;
  672. auto calendars = TRY(read_json_file(calendars_path.string()));
  673. auto const& main_object = calendars.as_object().get_object("main"sv).value();
  674. auto const& locale_object = main_object.get_object(calendars_path.parent().basename()).value();
  675. auto const& dates_object = locale_object.get_object("dates"sv).value();
  676. auto const& calendars_object = dates_object.get_object("calendars"sv).value();
  677. calendars_object.for_each_member([&](auto calendar_name, JsonValue const&) {
  678. // The generic calendar is not a supported Unicode calendar key, so skip it:
  679. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  680. if (calendar_name == "generic"sv)
  681. return;
  682. if (auto calendar_alias = find_keyword_alias("ca"sv, calendar_name, cldr); calendar_alias.has_value())
  683. calendar_name = calendar_alias.release_value();
  684. keywords.append(cldr.unique_strings.ensure(calendar_name));
  685. });
  686. return IterationDecision::Continue;
  687. }));
  688. locale.calendar_keywords = cldr.unique_keyword_lists.ensure(move(keywords));
  689. return {};
  690. }
  691. static void fill_in_collation_keywords(CLDR& cldr, LocaleData& locale)
  692. {
  693. // FIXME: If collation data becomes available in the CLDR, parse per-locale ordering from there.
  694. auto create_list_with_default_first = [&](auto key, auto default_value) {
  695. auto& values = cldr.keywords.find(key)->value;
  696. quick_sort(values, [&](auto const& lhs, auto const& rhs) {
  697. if (lhs == default_value)
  698. return true;
  699. if (rhs == default_value)
  700. return false;
  701. return lhs < rhs;
  702. });
  703. KeywordList keywords;
  704. keywords.ensure_capacity(values.size());
  705. for (auto const& value : values)
  706. keywords.append(cldr.unique_strings.ensure(value));
  707. return cldr.unique_keyword_lists.ensure(move(keywords));
  708. };
  709. static auto kf_index = create_list_with_default_first("kf"sv, "upper"sv);
  710. static auto kn_index = create_list_with_default_first("kn"sv, "true"sv);
  711. locale.collation_case_keywords = kf_index;
  712. locale.collation_numeric_keywords = kn_index;
  713. }
  714. static ErrorOr<void> parse_default_content_locales(DeprecatedString core_path, CLDR& cldr)
  715. {
  716. LexicalPath default_content_path(move(core_path));
  717. default_content_path = default_content_path.append("defaultContent.json"sv);
  718. auto default_content = TRY(read_json_file(default_content_path.string()));
  719. auto const& default_content_array = default_content.as_object().get_array("defaultContent"sv).value();
  720. default_content_array.for_each([&](JsonValue const& value) {
  721. auto locale = value.as_string();
  722. StringView default_locale = locale;
  723. while (true) {
  724. if (cldr.locales.contains(default_locale))
  725. break;
  726. auto pos = default_locale.find_last('-');
  727. if (!pos.has_value())
  728. return;
  729. default_locale = default_locale.substring_view(0, *pos);
  730. }
  731. if (default_locale != locale)
  732. cldr.locale_aliases.append({ default_locale, move(locale) });
  733. });
  734. return {};
  735. }
  736. static ErrorOr<void> define_aliases_without_scripts(CLDR& cldr)
  737. {
  738. // From ECMA-402: https://tc39.es/ecma402/#sec-internal-slots
  739. //
  740. // For locales that include a script subtag in addition to language and region, the
  741. // corresponding locale without a script subtag must also be supported.
  742. //
  743. // So we define aliases for locales that contain all three subtags, but we must also take
  744. // care to handle when the locale itself or the locale without a script subtag are an alias
  745. // by way of default-content locales.
  746. auto find_alias = [&](auto const& locale) {
  747. return cldr.locale_aliases.find_if([&](auto const& alias) { return locale == alias.alias; });
  748. };
  749. auto append_alias_without_script = [&](auto const& locale) -> ErrorOr<void> {
  750. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, locale));
  751. if ((parsed_locale.language == 0) || (parsed_locale.script == 0) || (parsed_locale.region == 0))
  752. return {};
  753. auto locale_without_script = DeprecatedString::formatted("{}-{}",
  754. cldr.unique_strings.get(parsed_locale.language),
  755. cldr.unique_strings.get(parsed_locale.region));
  756. if (cldr.locales.contains(locale_without_script))
  757. return {};
  758. if (find_alias(locale_without_script) != cldr.locale_aliases.end())
  759. return {};
  760. if (auto it = find_alias(locale); it != cldr.locale_aliases.end())
  761. cldr.locale_aliases.append({ it->name, locale_without_script });
  762. else
  763. cldr.locale_aliases.append({ locale, locale_without_script });
  764. return {};
  765. };
  766. for (auto const& locale : cldr.locales)
  767. TRY(append_alias_without_script(locale.key));
  768. for (auto const& locale : cldr.locale_aliases)
  769. TRY(append_alias_without_script(locale.alias));
  770. return {};
  771. }
  772. 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)
  773. {
  774. LexicalPath core_supplemental_path(core_path);
  775. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  776. VERIFY(FileSystem::is_directory(core_supplemental_path.string()));
  777. TRY(parse_core_aliases(core_supplemental_path.string(), cldr));
  778. TRY(parse_likely_subtags(core_supplemental_path.string(), cldr));
  779. auto remove_variants_from_path = [&](DeprecatedString path) -> ErrorOr<DeprecatedString> {
  780. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, LexicalPath::basename(path)));
  781. StringBuilder builder;
  782. builder.append(cldr.unique_strings.get(parsed_locale.language));
  783. if (auto script = cldr.unique_strings.get(parsed_locale.script); !script.is_empty())
  784. builder.appendff("-{}", script);
  785. if (auto region = cldr.unique_strings.get(parsed_locale.region); !region.is_empty())
  786. builder.appendff("-{}", region);
  787. return builder.to_deprecated_string();
  788. };
  789. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", locale_names_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  790. auto locale_path = LexicalPath::join(directory.path().string(), entry.name).string();
  791. auto language = TRY(remove_variants_from_path(locale_path));
  792. auto& locale = cldr.locales.ensure(language);
  793. TRY(parse_identity(locale_path, cldr, locale));
  794. return IterationDecision::Continue;
  795. }));
  796. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", locale_names_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  797. auto locale_path = LexicalPath::join(directory.path().string(), entry.name).string();
  798. TRY(preprocess_languages(locale_path, cldr));
  799. return IterationDecision::Continue;
  800. }));
  801. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", numbers_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  802. auto numbers_path = LexicalPath::join(directory.path().string(), entry.name).string();
  803. TRY(preprocess_currencies(numbers_path, cldr));
  804. return IterationDecision::Continue;
  805. }));
  806. quick_sort(cldr.languages);
  807. quick_sort(cldr.territories);
  808. quick_sort(cldr.scripts);
  809. quick_sort(cldr.currencies);
  810. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/bcp47", bcp47_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  811. auto bcp47_path = LexicalPath::join(directory.path().string(), entry.name).string();
  812. TRY(parse_unicode_extension_keywords(move(bcp47_path), cldr));
  813. return IterationDecision::Continue;
  814. }));
  815. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", locale_names_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  816. auto locale_path = LexicalPath::join(directory.path().string(), entry.name).string();
  817. auto language = TRY(remove_variants_from_path(locale_path));
  818. auto& locale = cldr.locales.ensure(language);
  819. TRY(parse_locale_display_patterns(locale_path, cldr, locale));
  820. TRY(parse_locale_languages(locale_path, cldr, locale));
  821. TRY(parse_locale_territories(locale_path, cldr, locale));
  822. TRY(parse_locale_scripts(locale_path, cldr, locale));
  823. TRY(parse_locale_calendars(locale_path, cldr, locale));
  824. return IterationDecision::Continue;
  825. }));
  826. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", misc_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  827. auto misc_path = LexicalPath::join(directory.path().string(), entry.name).string();
  828. auto language = TRY(remove_variants_from_path(misc_path));
  829. auto& locale = cldr.locales.ensure(language);
  830. TRY(parse_locale_list_patterns(misc_path, cldr, locale));
  831. TRY(parse_locale_layout(misc_path, cldr, locale));
  832. return IterationDecision::Continue;
  833. }));
  834. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", numbers_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  835. auto numbers_path = LexicalPath::join(directory.path().string(), entry.name).string();
  836. auto language = TRY(remove_variants_from_path(numbers_path));
  837. auto& locale = cldr.locales.ensure(language);
  838. TRY(parse_locale_currencies(numbers_path, cldr, locale));
  839. TRY(parse_number_system_keywords(numbers_path, cldr, locale));
  840. fill_in_collation_keywords(cldr, locale);
  841. return IterationDecision::Continue;
  842. }));
  843. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", dates_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  844. auto dates_path = LexicalPath::join(directory.path().string(), entry.name).string();
  845. auto language = TRY(remove_variants_from_path(dates_path));
  846. auto& locale = cldr.locales.ensure(language);
  847. TRY(parse_locale_date_fields(dates_path, cldr, locale));
  848. TRY(parse_calendar_keywords(dates_path, cldr, locale));
  849. return IterationDecision::Continue;
  850. }));
  851. TRY(parse_default_content_locales(move(core_path), cldr));
  852. TRY(define_aliases_without_scripts(cldr));
  853. return {};
  854. }
  855. static ErrorOr<void> generate_unicode_locale_header(Core::BufferedFile& file, CLDR& cldr)
  856. {
  857. StringBuilder builder;
  858. SourceGenerator generator { builder };
  859. generator.append(R"~~~(
  860. #pragma once
  861. #include <AK/Types.h>
  862. namespace Locale {
  863. )~~~");
  864. auto locales = cldr.locales.keys();
  865. auto keywords = cldr.keywords.keys();
  866. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, cldr.locale_aliases);
  867. generate_enum(generator, format_identifier, "Language"sv, {}, cldr.languages);
  868. generate_enum(generator, format_identifier, "Territory"sv, {}, cldr.territories);
  869. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, cldr.scripts);
  870. generate_enum(generator, format_identifier, "Currency"sv, {}, cldr.currencies);
  871. generate_enum(generator, format_identifier, "DateField"sv, {}, cldr.date_fields, cldr.date_field_aliases);
  872. generate_enum(generator, format_identifier, "Variant"sv, {}, cldr.variants);
  873. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, cldr.list_pattern_types);
  874. generate_enum(generator, format_identifier, "CharacterOrder"sv, {}, cldr.character_orders);
  875. generate_enum(generator, format_identifier, "Key"sv, {}, keywords);
  876. for (auto& keyword : cldr.keywords) {
  877. auto const& keyword_name = cldr.keyword_names.find(keyword.key)->value;
  878. auto enum_name = DeprecatedString::formatted("Keyword{}", format_identifier({}, keyword_name));
  879. if (auto aliases = cldr.keyword_aliases.find(keyword.key); aliases != cldr.keyword_aliases.end())
  880. generate_enum(generator, format_identifier, enum_name, {}, keyword.value, aliases->value);
  881. else
  882. generate_enum(generator, format_identifier, enum_name, {}, keyword.value);
  883. }
  884. generator.append(R"~~~(
  885. }
  886. )~~~");
  887. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  888. return {};
  889. }
  890. static ErrorOr<void> generate_unicode_locale_implementation(Core::BufferedFile& file, CLDR& cldr)
  891. {
  892. auto string_index_type = cldr.unique_strings.type_that_fits();
  893. StringBuilder builder;
  894. SourceGenerator generator { builder };
  895. generator.set("string_index_type"sv, string_index_type);
  896. generator.set("locales_size"sv, DeprecatedString::number(cldr.locales.size()));
  897. generator.set("territories_size", DeprecatedString::number(cldr.territories.size()));
  898. generator.set("variants_size", DeprecatedString::number(cldr.max_variant_size));
  899. generator.append(R"~~~(
  900. #include <AK/Array.h>
  901. #include <AK/BinarySearch.h>
  902. #include <AK/Optional.h>
  903. #include <AK/Span.h>
  904. #include <AK/String.h>
  905. #include <AK/StringView.h>
  906. #include <AK/Vector.h>
  907. #include <LibLocale/DateTimeFormat.h>
  908. #include <LibLocale/Locale.h>
  909. #include <LibLocale/LocaleData.h>
  910. #include <LibUnicode/CurrencyCode.h>
  911. namespace Locale {
  912. )~~~");
  913. cldr.unique_strings.generate(generator);
  914. generator.append(R"~~~(
  915. struct DisplayPatternImpl {
  916. DisplayPattern to_display_pattern() const
  917. {
  918. DisplayPattern display_patterns {};
  919. display_patterns.locale_pattern = decode_string(locale_pattern);
  920. display_patterns.locale_separator = decode_string(locale_separator);
  921. return display_patterns;
  922. }
  923. @string_index_type@ locale_pattern { 0 };
  924. @string_index_type@ locale_separator { 0 };
  925. };
  926. struct Patterns {
  927. ListPatternType type;
  928. Style style;
  929. @string_index_type@ start { 0 };
  930. @string_index_type@ middle { 0 };
  931. @string_index_type@ end { 0 };
  932. @string_index_type@ pair { 0 };
  933. };
  934. struct TextLayout {
  935. CharacterOrder character_order;
  936. };
  937. )~~~");
  938. generate_available_values(generator, "get_available_calendars"sv, cldr.keywords.find("ca"sv)->value, cldr.keyword_aliases.find("ca"sv)->value,
  939. [](auto calendar) {
  940. // FIXME: Remove this filter when we support all calendars.
  941. return calendar.is_one_of("gregory"sv, "iso8601"sv);
  942. });
  943. generate_available_values(generator, "get_available_collation_case_orderings"sv, cldr.keywords.find("kf"sv)->value, cldr.keyword_aliases.find("kf"sv)->value);
  944. generate_available_values(generator, "get_available_collation_numeric_orderings"sv, cldr.keywords.find("kn"sv)->value, cldr.keyword_aliases.find("kn"sv)->value);
  945. generate_available_values(generator, "get_available_collation_types"sv, cldr.keywords.find("co"sv)->value, cldr.keyword_aliases.find("co"sv)->value,
  946. [](auto collation) {
  947. // FIXME: Remove this filter when we support all collation types.
  948. return collation == "default"sv;
  949. });
  950. generate_available_values(generator, "get_available_hour_cycles"sv, cldr.keywords.find("hc"sv)->value);
  951. generate_available_values(generator, "get_available_number_systems"sv, cldr.keywords.find("nu"sv)->value);
  952. generate_available_values(generator, "get_available_currencies"sv, cldr.currencies);
  953. generator.append(R"~~~(
  954. ReadonlySpan<StringView> get_available_keyword_values(StringView key)
  955. {
  956. auto key_value = key_from_string(key);
  957. if (!key_value.has_value())
  958. return {};
  959. switch (*key_value) {
  960. case Key::Ca:
  961. return get_available_calendars();
  962. case Key::Co:
  963. return get_available_collation_types();
  964. case Key::Hc:
  965. return get_available_hour_cycles();
  966. case Key::Kf:
  967. return get_available_collation_case_orderings();
  968. case Key::Kn:
  969. return get_available_collation_numeric_orderings();
  970. case Key::Nu:
  971. return get_available_number_systems();
  972. }
  973. VERIFY_NOT_REACHED();
  974. }
  975. )~~~");
  976. cldr.unique_display_patterns.generate(generator, "DisplayPatternImpl"sv, "s_display_patterns"sv, 30);
  977. cldr.unique_language_lists.generate(generator, string_index_type, "s_language_lists"sv);
  978. cldr.unique_territory_lists.generate(generator, string_index_type, "s_territory_lists"sv);
  979. cldr.unique_script_lists.generate(generator, string_index_type, "s_script_lists"sv);
  980. cldr.unique_currency_lists.generate(generator, string_index_type, "s_currency_lists"sv);
  981. cldr.unique_calendar_lists.generate(generator, string_index_type, "s_calendar_lists"sv);
  982. cldr.unique_date_field_lists.generate(generator, string_index_type, "s_date_field_lists"sv);
  983. cldr.unique_keyword_lists.generate(generator, string_index_type, "s_keyword_lists"sv);
  984. cldr.unique_list_patterns.generate(generator, "Patterns"sv, "s_list_patterns"sv, 10);
  985. cldr.unique_list_pattern_lists.generate(generator, cldr.unique_list_patterns.type_that_fits(), "s_list_pattern_lists"sv);
  986. cldr.unique_text_layouts.generate(generator, "TextLayout"sv, "s_text_layouts"sv, 30);
  987. auto append_index = [&](auto index) {
  988. generator.append(DeprecatedString::formatted(", {}", index));
  989. };
  990. auto append_list_and_size = [&](auto const& list) {
  991. if (list.is_empty()) {
  992. generator.append(", {}, 0");
  993. return;
  994. }
  995. bool first = true;
  996. generator.append(", {");
  997. for (auto const& item : list) {
  998. generator.append(first ? " "sv : ", "sv);
  999. generator.append(DeprecatedString::number(item));
  1000. first = false;
  1001. }
  1002. generator.append(DeprecatedString::formatted(" }}, {}", list.size()));
  1003. };
  1004. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  1005. generator.set("type", type);
  1006. generator.set("name", name);
  1007. generator.set("size", DeprecatedString::number(keys.size()));
  1008. generator.append(R"~~~(
  1009. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  1010. bool first = true;
  1011. for (auto const& key : keys) {
  1012. auto const& value = map.find(key)->value;
  1013. auto mapping = mapping_getter(value);
  1014. generator.append(first ? " "sv : ", "sv);
  1015. generator.append(DeprecatedString::number(mapping));
  1016. first = false;
  1017. }
  1018. generator.append(" } };");
  1019. };
  1020. auto locales = cldr.locales.keys();
  1021. quick_sort(locales);
  1022. append_mapping(locales, cldr.locales, cldr.unique_display_patterns.type_that_fits(), "s_locale_display_patterns"sv, [&](auto const& locale) { return locale.display_patterns; });
  1023. append_mapping(locales, cldr.locales, cldr.unique_language_lists.type_that_fits(), "s_languages"sv, [&](auto const& locale) { return locale.languages; });
  1024. append_mapping(locales, cldr.locales, cldr.unique_territory_lists.type_that_fits(), "s_territories"sv, [&](auto const& locale) { return locale.territories; });
  1025. append_mapping(locales, cldr.locales, cldr.unique_script_lists.type_that_fits(), "s_scripts"sv, [&](auto const& locale) { return locale.scripts; });
  1026. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_long_currencies"sv, [&](auto const& locale) { return locale.long_currencies; });
  1027. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_short_currencies"sv, [&](auto const& locale) { return locale.short_currencies; });
  1028. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_narrow_currencies"sv, [&](auto const& locale) { return locale.narrow_currencies; });
  1029. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_numeric_currencies"sv, [&](auto const& locale) { return locale.numeric_currencies; });
  1030. append_mapping(locales, cldr.locales, cldr.unique_calendar_lists.type_that_fits(), "s_calendars"sv, [&](auto const& locale) { return locale.calendars; });
  1031. 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; });
  1032. 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; });
  1033. 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; });
  1034. append_mapping(locales, cldr.locales, cldr.unique_keyword_lists.type_that_fits(), "s_calendar_keywords"sv, [&](auto const& locale) { return locale.calendar_keywords; });
  1035. 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; });
  1036. 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; });
  1037. 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; });
  1038. 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; });
  1039. append_mapping(locales, cldr.locales, cldr.unique_text_layouts.type_that_fits(), "s_locale_text_layouts"sv, [&](auto const& locale) { return locale.text_layout; });
  1040. generator.append(R"~~~(
  1041. struct CanonicalLanguageID
  1042. {
  1043. ErrorOr<LanguageID> to_unicode_language_id() const
  1044. {
  1045. LanguageID language_id {};
  1046. TRY(language_id.variants.try_ensure_capacity(variants_size));
  1047. language_id.language = TRY(String::from_utf8(decode_string(language)));
  1048. if (script != 0)
  1049. language_id.script = TRY(String::from_utf8(decode_string(script)));
  1050. if (region != 0)
  1051. language_id.region = TRY(String::from_utf8(decode_string(region)));
  1052. for (size_t i = 0; i < variants_size; ++i)
  1053. language_id.variants.append(TRY(String::from_utf8(decode_string(variants[i]))));
  1054. return language_id;
  1055. }
  1056. bool matches_variants(Vector<String> const& other_variants) const {
  1057. if (variants_size == 0)
  1058. return true;
  1059. if (other_variants.size() != variants_size)
  1060. return false;
  1061. for (size_t i = 0; i < variants_size; ++i) {
  1062. if (decode_string(variants[i]) != other_variants[i])
  1063. return false;
  1064. }
  1065. return true;
  1066. };
  1067. @string_index_type@ language { 0 };
  1068. @string_index_type@ script { 0 };
  1069. @string_index_type@ region { 0 };
  1070. Array<@string_index_type@, @variants_size@> variants {};
  1071. size_t variants_size { 0 };
  1072. };
  1073. struct LanguageMapping {
  1074. CanonicalLanguageID key;
  1075. CanonicalLanguageID alias;
  1076. };
  1077. )~~~");
  1078. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  1079. generator.set("size", DeprecatedString::number(mappings.size()));
  1080. generator.set("name"sv, name);
  1081. generator.append(R"~~~(
  1082. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  1083. )~~~");
  1084. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  1085. auto const& lhs_language = cldr.unique_strings.get(lhs.key.language);
  1086. auto const& rhs_language = cldr.unique_strings.get(rhs.key.language);
  1087. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  1088. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  1089. return false;
  1090. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  1091. return true;
  1092. return lhs_language < rhs_language;
  1093. });
  1094. for (auto const& mapping : mappings) {
  1095. generator.set("language"sv, DeprecatedString::number(mapping.key.language));
  1096. generator.append(" { { @language@");
  1097. append_index(mapping.key.script);
  1098. append_index(mapping.key.region);
  1099. append_list_and_size(mapping.key.variants);
  1100. generator.set("language"sv, DeprecatedString::number(mapping.alias.language));
  1101. generator.append(" }, { @language@");
  1102. append_index(mapping.alias.script);
  1103. append_index(mapping.alias.region);
  1104. append_list_and_size(mapping.alias.variants);
  1105. generator.append(" } },\n");
  1106. }
  1107. generator.append("} };\n");
  1108. };
  1109. append_complex_mapping("complex_alias"sv, cldr.complex_mappings);
  1110. append_complex_mapping("likely_subtags"sv, cldr.likely_subtags);
  1111. generator.append(R"~~~(
  1112. static LanguageMapping const* resolve_likely_subtag(LanguageID const& language_id)
  1113. {
  1114. // https://unicode.org/reports/tr35/#Likely_Subtags
  1115. enum class State {
  1116. LanguageScriptRegion,
  1117. LanguageRegion,
  1118. LanguageScript,
  1119. Language,
  1120. UndScript,
  1121. Done,
  1122. };
  1123. auto state = State::LanguageScriptRegion;
  1124. while (state != State::Done) {
  1125. LanguageID search_key;
  1126. switch (state) {
  1127. case State::LanguageScriptRegion:
  1128. state = State::LanguageRegion;
  1129. if (!language_id.script.has_value() || !language_id.region.has_value())
  1130. continue;
  1131. search_key.language = *language_id.language;
  1132. search_key.script = *language_id.script;
  1133. search_key.region = *language_id.region;
  1134. break;
  1135. case State::LanguageRegion:
  1136. state = State::LanguageScript;
  1137. if (!language_id.region.has_value())
  1138. continue;
  1139. search_key.language = *language_id.language;
  1140. search_key.region = *language_id.region;
  1141. break;
  1142. case State::LanguageScript:
  1143. state = State::Language;
  1144. if (!language_id.script.has_value())
  1145. continue;
  1146. search_key.language = *language_id.language;
  1147. search_key.script = *language_id.script;
  1148. break;
  1149. case State::Language:
  1150. state = State::UndScript;
  1151. search_key.language = *language_id.language;
  1152. break;
  1153. case State::UndScript:
  1154. state = State::Done;
  1155. if (!language_id.script.has_value())
  1156. continue;
  1157. search_key.language = "und"_short_string;
  1158. search_key.script = *language_id.script;
  1159. break;
  1160. default:
  1161. VERIFY_NOT_REACHED();
  1162. }
  1163. for (auto const& map : s_likely_subtags) {
  1164. auto const& key_language = decode_string(map.key.language);
  1165. auto const& key_script = decode_string(map.key.script);
  1166. auto const& key_region = decode_string(map.key.region);
  1167. if (key_language != search_key.language)
  1168. continue;
  1169. if (!key_script.is_empty() || search_key.script.has_value()) {
  1170. if (key_script != search_key.script)
  1171. continue;
  1172. }
  1173. if (!key_region.is_empty() || search_key.region.has_value()) {
  1174. if (key_region != search_key.region)
  1175. continue;
  1176. }
  1177. return &map;
  1178. }
  1179. }
  1180. return nullptr;
  1181. }
  1182. )~~~");
  1183. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name, StringView unique_list) {
  1184. generator.set("enum_snake", enum_snake);
  1185. generator.set("from_string_name", from_string_name);
  1186. generator.set("collection_name", collection_name);
  1187. generator.set("unique_list", unique_list);
  1188. generator.append(R"~~~(
  1189. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  1190. {
  1191. auto locale_value = locale_from_string(locale);
  1192. if (!locale_value.has_value())
  1193. return {};
  1194. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  1195. if (!@enum_snake@_value.has_value())
  1196. return {};
  1197. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1198. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  1199. auto mapping_index = @collection_name@.at(locale_index);
  1200. auto const& mappings = @unique_list@.at(mapping_index);
  1201. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  1202. auto @enum_snake@_mapping = decode_string(@enum_snake@_string_index);
  1203. if (@enum_snake@_mapping.is_empty())
  1204. return {};
  1205. return @enum_snake@_mapping;
  1206. }
  1207. )~~~");
  1208. };
  1209. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) -> ErrorOr<void> {
  1210. HashValueMap<DeprecatedString> hashes;
  1211. TRY(hashes.try_ensure_capacity(values.size()));
  1212. for (auto const& value : values)
  1213. hashes.set(value.hash(), format_identifier(enum_title, value));
  1214. for (auto const& alias : aliases)
  1215. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1216. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1217. return {};
  1218. };
  1219. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) -> ErrorOr<void> {
  1220. HashValueMap<size_t> hashes;
  1221. TRY(hashes.try_ensure_capacity(aliases.size()));
  1222. for (auto const& alias : aliases)
  1223. hashes.set(alias.key.hash(), alias.value);
  1224. ValueFromStringOptions options {};
  1225. options.return_type = "StringView"sv;
  1226. options.return_format = "decode_string({})"sv;
  1227. generate_value_from_string(generator, "resolve_{}_alias"sv, string_index_type, enum_snake, move(hashes), options);
  1228. return {};
  1229. };
  1230. TRY(append_from_string("Locale"sv, "locale"sv, cldr.locales.keys(), cldr.locale_aliases));
  1231. TRY(append_from_string("Language"sv, "language"sv, cldr.languages));
  1232. append_mapping_search("language"sv, "language"sv, "s_languages"sv, "s_language_lists"sv);
  1233. TRY(append_alias_search("language"sv, cldr.language_aliases));
  1234. TRY(append_from_string("Territory"sv, "territory"sv, cldr.territories));
  1235. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv, "s_territory_lists"sv);
  1236. TRY(append_alias_search("territory"sv, cldr.territory_aliases));
  1237. TRY(append_from_string("ScriptTag"sv, "script_tag"sv, cldr.scripts));
  1238. append_mapping_search("script"sv, "script_tag"sv, "s_scripts"sv, "s_script_lists"sv);
  1239. TRY(append_alias_search("script_tag"sv, cldr.script_aliases));
  1240. TRY(append_from_string("Currency"sv, "currency"sv, cldr.currencies));
  1241. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv, "s_currency_lists"sv);
  1242. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv, "s_currency_lists"sv);
  1243. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv, "s_currency_lists"sv);
  1244. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv, "s_currency_lists"sv);
  1245. TRY(append_from_string("DateField"sv, "date_field"sv, cldr.date_fields, cldr.date_field_aliases));
  1246. append_mapping_search("long_date_field"sv, "date_field"sv, "s_long_date_fields"sv, "s_date_field_lists"sv);
  1247. append_mapping_search("short_date_field"sv, "date_field"sv, "s_short_date_fields"sv, "s_date_field_lists"sv);
  1248. append_mapping_search("narrow_date_field"sv, "date_field"sv, "s_narrow_date_fields"sv, "s_date_field_lists"sv);
  1249. TRY(append_from_string("Key"sv, "key"sv, cldr.keywords.keys()));
  1250. for (auto const& keyword : cldr.keywords) {
  1251. auto const& keyword_name = cldr.keyword_names.find(keyword.key)->value;
  1252. auto enum_name = DeprecatedString::formatted("Keyword{}", format_identifier({}, keyword_name));
  1253. auto enum_snake = DeprecatedString::formatted("keyword_{}", keyword.key);
  1254. if (auto aliases = cldr.keyword_aliases.find(keyword.key); aliases != cldr.keyword_aliases.end())
  1255. TRY(append_from_string(enum_name, enum_snake, keyword.value, aliases->value));
  1256. else
  1257. TRY(append_from_string(enum_name, enum_snake, keyword.value));
  1258. }
  1259. append_mapping_search("calendar"sv, "keyword_ca"sv, "s_calendars"sv, "s_calendar_lists"sv);
  1260. TRY(append_alias_search("variant"sv, cldr.variant_aliases));
  1261. TRY(append_alias_search("subdivision"sv, cldr.subdivision_aliases));
  1262. TRY(append_from_string("ListPatternType"sv, "list_pattern_type"sv, cldr.list_pattern_types));
  1263. TRY(append_from_string("CharacterOrder"sv, "character_order"sv, cldr.character_orders));
  1264. generate_value_to_string(generator, "{}_to_string"sv, "CharacterOrder"sv, "character_order"sv, format_identifier, cldr.character_orders);
  1265. generator.append(R"~~~(
  1266. static ReadonlySpan<@string_index_type@> find_keyword_indices(StringView locale, StringView key)
  1267. {
  1268. auto locale_value = locale_from_string(locale);
  1269. if (!locale_value.has_value())
  1270. return {};
  1271. auto key_value = key_from_string(key);
  1272. if (!key_value.has_value())
  1273. return {};
  1274. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1275. size_t keywords_index = 0;
  1276. switch (*key_value) {
  1277. case Key::Ca:
  1278. keywords_index = s_calendar_keywords.at(locale_index);
  1279. break;
  1280. case Key::Kf:
  1281. keywords_index = s_collation_case_keywords.at(locale_index);
  1282. break;
  1283. case Key::Kn:
  1284. keywords_index = s_collation_numeric_keywords.at(locale_index);
  1285. break;
  1286. case Key::Nu:
  1287. keywords_index = s_number_system_keywords.at(locale_index);
  1288. break;
  1289. default:
  1290. VERIFY_NOT_REACHED();
  1291. }
  1292. return s_keyword_lists.at(keywords_index);
  1293. }
  1294. ErrorOr<Optional<StringView>> get_preferred_keyword_value_for_locale(StringView locale, StringView key)
  1295. {
  1296. // Hour cycle keywords are region-based rather than locale-based, so they need to be handled specially.
  1297. // FIXME: Calendar keywords are also region-based, and will need to be handled here when we support non-Gregorian calendars:
  1298. // https://github.com/unicode-org/cldr-json/blob/main/cldr-json/cldr-core/supplemental/calendarPreferenceData.json
  1299. if (key == "hc"sv) {
  1300. auto hour_cycles = TRY(get_locale_hour_cycles(locale));
  1301. if (hour_cycles.is_empty())
  1302. return OptionalNone {};
  1303. return Optional<StringView> { hour_cycle_to_string(hour_cycles[0]) };
  1304. }
  1305. // FIXME: Generate locale-preferred collation data when available in the CLDR.
  1306. if (key == "co"sv) {
  1307. auto collations = get_available_collation_types();
  1308. if (collations.is_empty())
  1309. return OptionalNone {};
  1310. return Optional<StringView> { collations[0] };
  1311. }
  1312. auto keyword_indices = find_keyword_indices(locale, key);
  1313. if (keyword_indices.is_empty())
  1314. return OptionalNone {};
  1315. return Optional<StringView> { decode_string(keyword_indices[0]) };
  1316. }
  1317. ErrorOr<Vector<StringView>> get_keywords_for_locale(StringView locale, StringView key)
  1318. {
  1319. // Hour cycle keywords are region-based rather than locale-based, so they need to be handled specially.
  1320. // FIXME: Calendar keywords are also region-based, and will need to be handled here when we support non-Gregorian calendars:
  1321. // https://github.com/unicode-org/cldr-json/blob/main/cldr-json/cldr-core/supplemental/calendarPreferenceData.json
  1322. if (key == "hc"sv) {
  1323. auto hour_cycles = TRY(get_locale_hour_cycles(locale));
  1324. Vector<StringView> values;
  1325. TRY(values.try_ensure_capacity(hour_cycles.size()));
  1326. for (auto hour_cycle : hour_cycles)
  1327. values.unchecked_append(hour_cycle_to_string(hour_cycle));
  1328. return values;
  1329. }
  1330. // FIXME: Generate locale-preferred collation data when available in the CLDR.
  1331. if (key == "co"sv)
  1332. return Vector<StringView> { get_available_collation_types() };
  1333. auto keyword_indices = find_keyword_indices(locale, key);
  1334. Vector<StringView> keywords;
  1335. TRY(keywords.try_ensure_capacity(keyword_indices.size()));
  1336. for (auto keyword : keyword_indices)
  1337. keywords.unchecked_append(decode_string(keyword));
  1338. return keywords;
  1339. }
  1340. Optional<DisplayPattern> get_locale_display_patterns(StringView locale)
  1341. {
  1342. auto locale_value = locale_from_string(locale);
  1343. if (!locale_value.has_value())
  1344. return {};
  1345. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1346. auto display_patterns_index = s_locale_display_patterns.at(locale_index);
  1347. auto const& display_patterns = s_display_patterns.at(display_patterns_index);
  1348. return display_patterns.to_display_pattern();
  1349. }
  1350. Optional<ListPatterns> get_locale_list_patterns(StringView locale, StringView list_pattern_type, Style list_pattern_style)
  1351. {
  1352. auto locale_value = locale_from_string(locale);
  1353. if (!locale_value.has_value())
  1354. return {};
  1355. auto type_value = list_pattern_type_from_string(list_pattern_type);
  1356. if (!type_value.has_value())
  1357. return {};
  1358. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1359. auto list_patterns_list_index = s_locale_list_patterns.at(locale_index);
  1360. auto const& locale_list_patterns = s_list_pattern_lists.at(list_patterns_list_index);
  1361. for (auto list_patterns_index : locale_list_patterns) {
  1362. auto const& list_patterns = s_list_patterns.at(list_patterns_index);
  1363. if ((list_patterns.type == type_value) && (list_patterns.style == list_pattern_style)) {
  1364. auto const& start = decode_string(list_patterns.start);
  1365. auto const& middle = decode_string(list_patterns.middle);
  1366. auto const& end = decode_string(list_patterns.end);
  1367. auto const& pair = decode_string(list_patterns.pair);
  1368. return ListPatterns { start, middle, end, pair };
  1369. }
  1370. }
  1371. return {};
  1372. }
  1373. static Optional<TextLayout> text_layout_for_locale(StringView locale)
  1374. {
  1375. auto locale_value = locale_from_string(locale);
  1376. if (!locale_value.has_value())
  1377. return {};
  1378. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1379. auto text_layouts_index = s_locale_text_layouts.at(locale_index);
  1380. return s_text_layouts.at(text_layouts_index);
  1381. }
  1382. Optional<CharacterOrder> character_order_for_locale(StringView locale)
  1383. {
  1384. if (auto text_layout = text_layout_for_locale(locale); text_layout.has_value())
  1385. return text_layout->character_order;
  1386. return {};
  1387. }
  1388. ErrorOr<void> resolve_complex_language_aliases(LanguageID& language_id)
  1389. {
  1390. for (auto const& map : s_complex_alias) {
  1391. auto key_language = decode_string(map.key.language);
  1392. auto key_script = decode_string(map.key.script);
  1393. auto key_region = decode_string(map.key.region);
  1394. if ((key_language != language_id.language) && (key_language != "und"sv))
  1395. continue;
  1396. if (!key_script.is_empty() && (key_script != language_id.script))
  1397. continue;
  1398. if (!key_region.is_empty() && (key_region != language_id.region))
  1399. continue;
  1400. if (!map.key.matches_variants(language_id.variants))
  1401. continue;
  1402. auto alias = TRY(map.alias.to_unicode_language_id());
  1403. if (alias.language == "und"sv)
  1404. alias.language = move(language_id.language);
  1405. if (key_script.is_empty() && !alias.script.has_value())
  1406. alias.script = move(language_id.script);
  1407. if (key_region.is_empty() && !alias.region.has_value())
  1408. alias.region = move(language_id.region);
  1409. if (map.key.variants_size == 0 && alias.variants.is_empty())
  1410. alias.variants = move(language_id.variants);
  1411. language_id = move(alias);
  1412. break;
  1413. }
  1414. return {};
  1415. }
  1416. ErrorOr<Optional<LanguageID>> add_likely_subtags(LanguageID const& language_id)
  1417. {
  1418. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  1419. auto const* likely_subtag = resolve_likely_subtag(language_id);
  1420. if (likely_subtag == nullptr)
  1421. return OptionalNone {};
  1422. auto maximized = language_id;
  1423. auto key_script = decode_string(likely_subtag->key.script);
  1424. auto key_region = decode_string(likely_subtag->key.region);
  1425. auto alias_language = decode_string(likely_subtag->alias.language);
  1426. auto alias_script = decode_string(likely_subtag->alias.script);
  1427. auto alias_region = decode_string(likely_subtag->alias.region);
  1428. if (maximized.language == "und"sv)
  1429. maximized.language = TRY(String::from_utf8(alias_language));
  1430. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  1431. maximized.script = TRY(String::from_utf8(alias_script));
  1432. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  1433. maximized.region = TRY(String::from_utf8(alias_region));
  1434. return maximized;
  1435. }
  1436. ErrorOr<Optional<String>> resolve_most_likely_territory(LanguageID const& language_id)
  1437. {
  1438. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  1439. return String::from_utf8(decode_string(likely_subtag->alias.region));
  1440. return OptionalNone {};
  1441. }
  1442. }
  1443. )~~~");
  1444. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  1445. return {};
  1446. }
  1447. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1448. {
  1449. StringView generated_header_path;
  1450. StringView generated_implementation_path;
  1451. StringView bcp47_path;
  1452. StringView core_path;
  1453. StringView locale_names_path;
  1454. StringView misc_path;
  1455. StringView numbers_path;
  1456. StringView dates_path;
  1457. Core::ArgsParser args_parser;
  1458. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1459. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1460. args_parser.add_option(bcp47_path, "Path to cldr-bcp47 directory", "bcp47-path", 'b', "bcp47-path");
  1461. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1462. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  1463. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  1464. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  1465. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1466. args_parser.parse(arguments);
  1467. auto generated_header_file = TRY(open_file(generated_header_path, Core::File::OpenMode::Write));
  1468. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::File::OpenMode::Write));
  1469. CLDR cldr;
  1470. TRY(parse_all_locales(bcp47_path, core_path, locale_names_path, misc_path, numbers_path, dates_path, cldr));
  1471. TRY(generate_unicode_locale_header(*generated_header_file, cldr));
  1472. TRY(generate_unicode_locale_implementation(*generated_implementation_file, cldr));
  1473. return 0;
  1474. }