GenerateUnicodeLocale.cpp 68 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622
  1. /*
  2. * Copyright (c) 2021-2022, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "GeneratorUtil.h"
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/Format.h>
  10. #include <AK/HashMap.h>
  11. #include <AK/JsonObject.h>
  12. #include <AK/JsonParser.h>
  13. #include <AK/JsonValue.h>
  14. #include <AK/LexicalPath.h>
  15. #include <AK/QuickSort.h>
  16. #include <AK/SourceGenerator.h>
  17. #include <AK/String.h>
  18. #include <AK/StringBuilder.h>
  19. #include <LibCore/ArgsParser.h>
  20. #include <LibCore/DirIterator.h>
  21. #include <LibCore/File.h>
  22. #include <LibCore/Stream.h>
  23. using StringIndexType = u32;
  24. constexpr auto s_string_index_type = "u32"sv;
  25. using DisplayPatternIndexType = u8;
  26. constexpr auto s_display_pattern_index_type = "u8"sv;
  27. using LanguageListIndexType = u8;
  28. constexpr auto s_language_list_index_type = "u8"sv;
  29. using TerritoryListIndexType = u8;
  30. constexpr auto s_territory_list_index_type = "u8"sv;
  31. using ScriptListIndexType = u8;
  32. constexpr auto s_script_list_index_type = "u8"sv;
  33. using CurrencyListIndexType = u16;
  34. constexpr auto s_currency_list_index_type = "u16"sv;
  35. using CalendarListIndexType = u8;
  36. constexpr auto s_calendar_list_index_type = "u8"sv;
  37. using DateFieldListIndexType = u8;
  38. constexpr auto s_date_field_list_index_type = "u8"sv;
  39. using KeywordListIndexType = u8;
  40. constexpr auto s_keyword_list_index_type = "u8"sv;
  41. using ListPatternIndexType = u16;
  42. constexpr auto s_list_pattern_index_type = "u16"sv;
  43. using ListPatternListIndexType = u8;
  44. constexpr auto s_list_pattern_list_index_type = "u8"sv;
  45. static String format_identifier(StringView owner, String identifier)
  46. {
  47. identifier = identifier.replace("-"sv, "_"sv, true);
  48. if (all_of(identifier, is_ascii_digit))
  49. return String::formatted("{}_{}", owner[0], identifier);
  50. if (is_ascii_lower_alpha(identifier[0]))
  51. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  52. return identifier;
  53. }
  54. struct DisplayPattern {
  55. unsigned hash() const
  56. {
  57. return pair_int_hash(locale_pattern, locale_separator);
  58. }
  59. bool operator==(DisplayPattern const& other) const
  60. {
  61. return (locale_pattern == other.locale_pattern)
  62. && (locale_separator == other.locale_separator);
  63. }
  64. StringIndexType locale_pattern { 0 };
  65. StringIndexType locale_separator { 0 };
  66. };
  67. template<>
  68. struct AK::Formatter<DisplayPattern> : Formatter<FormatString> {
  69. ErrorOr<void> format(FormatBuilder& builder, DisplayPattern const& patterns)
  70. {
  71. return Formatter<FormatString>::format(builder,
  72. "{{ {}, {} }}",
  73. patterns.locale_pattern,
  74. patterns.locale_separator);
  75. }
  76. };
  77. template<>
  78. struct AK::Traits<DisplayPattern> : public GenericTraits<DisplayPattern> {
  79. static unsigned hash(DisplayPattern const& p) { return p.hash(); }
  80. };
  81. struct ListPatterns {
  82. unsigned hash() const
  83. {
  84. auto hash = pair_int_hash(type.hash(), style.hash());
  85. hash = pair_int_hash(hash, start);
  86. hash = pair_int_hash(hash, middle);
  87. hash = pair_int_hash(hash, end);
  88. hash = pair_int_hash(hash, pair);
  89. return hash;
  90. }
  91. bool operator==(ListPatterns const& other) const
  92. {
  93. return (type == other.type)
  94. && (style == other.style)
  95. && (start == other.start)
  96. && (middle == other.middle)
  97. && (end == other.end)
  98. && (pair == other.pair);
  99. }
  100. StringView type;
  101. StringView style;
  102. StringIndexType start { 0 };
  103. StringIndexType middle { 0 };
  104. StringIndexType end { 0 };
  105. StringIndexType pair { 0 };
  106. };
  107. template<>
  108. struct AK::Formatter<ListPatterns> : Formatter<FormatString> {
  109. ErrorOr<void> format(FormatBuilder& builder, ListPatterns const& patterns)
  110. {
  111. return Formatter<FormatString>::format(builder,
  112. "{{ ListPatternType::{}, Style::{}, {}, {}, {}, {} }}",
  113. format_identifier({}, patterns.type),
  114. format_identifier({}, patterns.style),
  115. patterns.start,
  116. patterns.middle,
  117. patterns.end,
  118. patterns.pair);
  119. }
  120. };
  121. template<>
  122. struct AK::Traits<ListPatterns> : public GenericTraits<ListPatterns> {
  123. static unsigned hash(ListPatterns const& p) { return p.hash(); }
  124. };
  125. using LanguageList = Vector<StringIndexType>;
  126. using TerritoryList = Vector<StringIndexType>;
  127. using ScriptList = Vector<StringIndexType>;
  128. using CurrencyList = Vector<StringIndexType>;
  129. using CalendarList = Vector<StringIndexType>;
  130. using DateFieldList = Vector<StringIndexType>;
  131. using KeywordList = Vector<StringIndexType>;
  132. using ListPatternList = Vector<ListPatternIndexType>;
  133. struct Locale {
  134. String language;
  135. Optional<String> territory;
  136. Optional<String> variant;
  137. DisplayPatternIndexType display_patterns { 0 };
  138. LanguageListIndexType languages { 0 };
  139. TerritoryListIndexType territories { 0 };
  140. ScriptListIndexType scripts { 0 };
  141. CurrencyListIndexType long_currencies { 0 };
  142. CurrencyListIndexType short_currencies { 0 };
  143. CurrencyListIndexType narrow_currencies { 0 };
  144. CurrencyListIndexType numeric_currencies { 0 };
  145. CalendarListIndexType calendars { 0 };
  146. DateFieldListIndexType long_date_fields { 0 };
  147. DateFieldListIndexType short_date_fields { 0 };
  148. DateFieldListIndexType narrow_date_fields { 0 };
  149. KeywordListIndexType calendar_keywords { 0 };
  150. KeywordListIndexType collation_case_keywords { 0 };
  151. KeywordListIndexType collation_numeric_keywords { 0 };
  152. KeywordListIndexType number_system_keywords { 0 };
  153. ListPatternListIndexType list_patterns { 0 };
  154. };
  155. struct LanguageMapping {
  156. CanonicalLanguageID<StringIndexType> key {};
  157. CanonicalLanguageID<StringIndexType> alias {};
  158. };
  159. struct UnicodeLocaleData {
  160. UniqueStringStorage<StringIndexType> unique_strings;
  161. UniqueStorage<DisplayPattern, DisplayPatternIndexType> unique_display_patterns;
  162. UniqueStorage<LanguageList, LanguageListIndexType> unique_language_lists;
  163. UniqueStorage<TerritoryList, TerritoryListIndexType> unique_territory_lists;
  164. UniqueStorage<ScriptList, ScriptListIndexType> unique_script_lists;
  165. UniqueStorage<CurrencyList, CurrencyListIndexType> unique_currency_lists;
  166. UniqueStorage<CalendarList, CalendarListIndexType> unique_calendar_lists;
  167. UniqueStorage<DateFieldList, DateFieldListIndexType> unique_date_field_lists;
  168. UniqueStorage<KeywordList, KeywordListIndexType> unique_keyword_lists;
  169. UniqueStorage<ListPatterns, ListPatternIndexType> unique_list_patterns;
  170. UniqueStorage<ListPatternList, ListPatternListIndexType> unique_list_pattern_lists;
  171. HashMap<String, Locale> locales;
  172. Vector<Alias> locale_aliases;
  173. Vector<String> languages;
  174. Vector<String> territories;
  175. Vector<String> scripts;
  176. Vector<String> variants;
  177. Vector<String> currencies;
  178. Vector<String> date_fields;
  179. Vector<Alias> date_field_aliases {
  180. // ECMA-402 and the CLDR refer to some date fields with different names. Defining these aliases
  181. // means we can remain agnostic about the naming differences elsewhere.
  182. { "dayperiod"sv, "dayPeriod"sv },
  183. { "week"sv, "weekOfYear"sv },
  184. { "zone"sv, "timeZoneName"sv },
  185. };
  186. HashMap<String, Vector<String>> keywords;
  187. HashMap<String, Vector<Alias>> keyword_aliases;
  188. HashMap<String, String> keyword_names;
  189. Vector<String> list_pattern_types;
  190. HashMap<String, StringIndexType> language_aliases;
  191. HashMap<String, StringIndexType> territory_aliases;
  192. HashMap<String, StringIndexType> script_aliases;
  193. HashMap<String, StringIndexType> variant_aliases;
  194. HashMap<String, StringIndexType> subdivision_aliases;
  195. Vector<LanguageMapping> complex_mappings;
  196. Vector<LanguageMapping> likely_subtags;
  197. size_t max_variant_size { 0 };
  198. };
  199. // Some parsing is expected to fail. For example, the CLDR contains language mappings
  200. // with locales such as "en-GB-oed" that are canonically invalid locale IDs.
  201. #define TRY_OR_DISCARD(expression) \
  202. ({ \
  203. auto _temporary_result = (expression); \
  204. if (_temporary_result.is_error()) \
  205. return; \
  206. _temporary_result.release_value(); \
  207. })
  208. static ErrorOr<LanguageMapping> parse_language_mapping(UnicodeLocaleData& locale_data, StringView key, StringView alias)
  209. {
  210. auto parsed_key = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, key));
  211. auto parsed_alias = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, alias));
  212. return LanguageMapping { move(parsed_key), move(parsed_alias) };
  213. }
  214. static ErrorOr<void> parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
  215. {
  216. LexicalPath core_aliases_path(move(core_supplemental_path));
  217. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  218. auto core_aliases = TRY(read_json_file(core_aliases_path.string()));
  219. auto const& supplemental_object = core_aliases.as_object().get("supplemental"sv);
  220. auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
  221. auto const& alias_object = metadata_object.as_object().get("alias"sv);
  222. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  223. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  224. auto alias = value.as_object().get("_replacement"sv).as_string();
  225. if (key.contains('-')) {
  226. auto mapping = TRY_OR_DISCARD(parse_language_mapping(locale_data, key, alias));
  227. locale_data.max_variant_size = max(mapping.key.variants.size(), locale_data.max_variant_size);
  228. locale_data.max_variant_size = max(mapping.alias.variants.size(), locale_data.max_variant_size);
  229. locale_data.complex_mappings.append(move(mapping));
  230. } else {
  231. alias_map.set(key, locale_data.unique_strings.ensure(alias));
  232. }
  233. });
  234. };
  235. append_aliases(alias_object.as_object().get("languageAlias"sv), locale_data.language_aliases);
  236. append_aliases(alias_object.as_object().get("territoryAlias"sv), locale_data.territory_aliases);
  237. append_aliases(alias_object.as_object().get("scriptAlias"sv), locale_data.script_aliases);
  238. append_aliases(alias_object.as_object().get("variantAlias"sv), locale_data.variant_aliases);
  239. append_aliases(alias_object.as_object().get("subdivisionAlias"sv), locale_data.subdivision_aliases);
  240. return {};
  241. }
  242. static ErrorOr<void> parse_likely_subtags(String core_supplemental_path, UnicodeLocaleData& locale_data)
  243. {
  244. LexicalPath likely_subtags_path(move(core_supplemental_path));
  245. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  246. auto likely_subtags = TRY(read_json_file(likely_subtags_path.string()));
  247. auto const& supplemental_object = likely_subtags.as_object().get("supplemental"sv);
  248. auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
  249. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  250. auto mapping = TRY_OR_DISCARD(parse_language_mapping(locale_data, key, value.as_string()));
  251. locale_data.max_variant_size = max(mapping.key.variants.size(), locale_data.max_variant_size);
  252. locale_data.max_variant_size = max(mapping.alias.variants.size(), locale_data.max_variant_size);
  253. locale_data.likely_subtags.append(move(mapping));
  254. });
  255. return {};
  256. }
  257. static ErrorOr<void> parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  258. {
  259. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  260. languages_path = languages_path.append("languages.json"sv);
  261. auto languages = TRY(read_json_file(languages_path.string()));
  262. auto const& main_object = languages.as_object().get("main"sv);
  263. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  264. auto const& identity_object = locale_object.as_object().get("identity"sv);
  265. auto const& language_string = identity_object.as_object().get("language"sv);
  266. auto const& territory_string = identity_object.as_object().get("territory"sv);
  267. auto const& script_string = identity_object.as_object().get("script"sv);
  268. auto const& variant_string = identity_object.as_object().get("variant"sv);
  269. locale.language = language_string.as_string();
  270. if (territory_string.is_string()) {
  271. locale.territory = territory_string.as_string();
  272. if (!locale_data.territories.contains_slow(*locale.territory))
  273. locale_data.territories.append(*locale.territory);
  274. }
  275. if (script_string.is_string()) {
  276. auto script = script_string.as_string();
  277. if (!locale_data.scripts.contains_slow(script))
  278. locale_data.scripts.append(script);
  279. }
  280. if (variant_string.is_string()) {
  281. locale.variant = variant_string.as_string();
  282. if (!locale_data.variants.contains_slow(*locale.variant))
  283. locale_data.variants.append(*locale.variant);
  284. }
  285. return {};
  286. }
  287. static ErrorOr<void> parse_locale_display_patterns(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  288. {
  289. LexicalPath locale_display_names_path(move(locale_path));
  290. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  291. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  292. auto const& main_object = locale_display_names.as_object().get("main"sv);
  293. auto const& locale_object = main_object.as_object().get(locale_display_names_path.parent().basename());
  294. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  295. auto const& locale_display_patterns_object = locale_display_names_object.as_object().get("localeDisplayPattern"sv);
  296. auto const& locale_pattern = locale_display_patterns_object.as_object().get("localePattern"sv);
  297. auto const& locale_separator = locale_display_patterns_object.as_object().get("localeSeparator"sv);
  298. DisplayPattern patterns {};
  299. patterns.locale_pattern = locale_data.unique_strings.ensure(locale_pattern.as_string());
  300. patterns.locale_separator = locale_data.unique_strings.ensure(locale_separator.as_string());
  301. locale.display_patterns = locale_data.unique_display_patterns.ensure(move(patterns));
  302. return {};
  303. }
  304. static ErrorOr<void> preprocess_languages(String locale_path, UnicodeLocaleData& locale_data)
  305. {
  306. LexicalPath languages_path(move(locale_path));
  307. languages_path = languages_path.append("languages.json"sv);
  308. auto locale_languages = TRY(read_json_file(languages_path.string()));
  309. auto const& main_object = locale_languages.as_object().get("main"sv);
  310. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  311. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  312. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  313. languages_object.as_object().for_each_member([&](auto const& key, auto const&) {
  314. if (!key.contains("-alt-"sv) && !locale_data.languages.contains_slow(key))
  315. locale_data.languages.append(key);
  316. });
  317. return {};
  318. }
  319. static ErrorOr<void> parse_unicode_extension_keywords(String bcp47_path, UnicodeLocaleData& locale_data)
  320. {
  321. constexpr auto desired_keywords = Array { "ca"sv, "kf"sv, "kn"sv, "nu"sv };
  322. auto keywords = TRY(read_json_file(bcp47_path));
  323. auto const& keyword_object = keywords.as_object().get("keyword"sv);
  324. auto const& unicode_object = keyword_object.as_object().get("u"sv);
  325. if (unicode_object.is_null())
  326. return {};
  327. unicode_object.as_object().for_each_member([&](auto const& key, auto const& value) {
  328. if (!desired_keywords.span().contains_slow(key))
  329. return;
  330. auto const& name = value.as_object().get("_alias");
  331. locale_data.keyword_names.set(key, name.as_string());
  332. auto& keywords = locale_data.keywords.ensure(key);
  333. value.as_object().for_each_member([&](auto const& keyword, auto const& properties) {
  334. if (!properties.is_object())
  335. return;
  336. if (auto const& preferred = properties.as_object().get("_preferred"sv); preferred.is_string()) {
  337. locale_data.keyword_aliases.ensure(key).append({ preferred.as_string(), keyword });
  338. return;
  339. }
  340. if (auto const& alias = properties.as_object().get("_alias"sv); alias.is_string())
  341. locale_data.keyword_aliases.ensure(key).append({ keyword, alias.as_string() });
  342. keywords.append(keyword);
  343. });
  344. });
  345. return {};
  346. }
  347. static Optional<String> find_keyword_alias(StringView key, StringView calendar, UnicodeLocaleData& locale_data)
  348. {
  349. auto it = locale_data.keyword_aliases.find(key);
  350. if (it == locale_data.keyword_aliases.end())
  351. return {};
  352. auto alias = it->value.find_if([&](auto const& alias) { return calendar == alias.alias; });
  353. if (alias == it->value.end())
  354. return {};
  355. return alias->name;
  356. }
  357. static ErrorOr<void> parse_locale_languages(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  358. {
  359. LexicalPath languages_path(move(locale_path));
  360. languages_path = languages_path.append("languages.json"sv);
  361. auto locale_languages = TRY(read_json_file(languages_path.string()));
  362. auto const& main_object = locale_languages.as_object().get("main"sv);
  363. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  364. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  365. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  366. LanguageList languages;
  367. languages.resize(locale_data.languages.size());
  368. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  369. if (key.contains("-alt-"sv))
  370. return;
  371. auto index = locale_data.languages.find_first_index(key).value();
  372. languages[index] = locale_data.unique_strings.ensure(value.as_string());
  373. });
  374. locale.languages = locale_data.unique_language_lists.ensure(move(languages));
  375. return {};
  376. }
  377. static ErrorOr<void> parse_locale_territories(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  378. {
  379. LexicalPath territories_path(move(locale_path));
  380. territories_path = territories_path.append("territories.json"sv);
  381. auto locale_territories = TRY(read_json_file(territories_path.string()));
  382. auto const& main_object = locale_territories.as_object().get("main"sv);
  383. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  384. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  385. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  386. TerritoryList territories;
  387. territories.resize(locale_data.territories.size());
  388. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  389. if (auto index = locale_data.territories.find_first_index(key); index.has_value())
  390. territories[*index] = locale_data.unique_strings.ensure(value.as_string());
  391. });
  392. locale.territories = locale_data.unique_territory_lists.ensure(move(territories));
  393. return {};
  394. }
  395. static ErrorOr<void> parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  396. {
  397. LexicalPath scripts_path(move(locale_path));
  398. scripts_path = scripts_path.append("scripts.json"sv);
  399. auto locale_scripts = TRY(read_json_file(scripts_path.string()));
  400. auto const& main_object = locale_scripts.as_object().get("main"sv);
  401. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  402. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  403. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  404. ScriptList scripts;
  405. scripts.resize(locale_data.scripts.size());
  406. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  407. if (auto index = locale_data.scripts.find_first_index(key); index.has_value())
  408. scripts[*index] = locale_data.unique_strings.ensure(value.as_string());
  409. });
  410. locale.scripts = locale_data.unique_script_lists.ensure(move(scripts));
  411. return {};
  412. }
  413. static ErrorOr<void> parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
  414. {
  415. LexicalPath list_patterns_path(move(misc_path));
  416. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  417. auto locale_list_patterns = TRY(read_json_file(list_patterns_path.string()));
  418. auto const& main_object = locale_list_patterns.as_object().get("main"sv);
  419. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  420. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  421. auto list_pattern_type = [](StringView key) {
  422. if (key.contains("type-standard"sv))
  423. return "conjunction"sv;
  424. if (key.contains("type-or"sv))
  425. return "disjunction"sv;
  426. if (key.contains("type-unit"sv))
  427. return "unit"sv;
  428. VERIFY_NOT_REACHED();
  429. };
  430. auto list_pattern_style = [](StringView key) {
  431. if (key.contains("short"sv))
  432. return "short"sv;
  433. if (key.contains("narrow"sv))
  434. return "narrow"sv;
  435. return "long"sv;
  436. };
  437. ListPatternList list_patterns;
  438. list_patterns.ensure_capacity(list_patterns_object.as_object().size());
  439. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  440. auto type = list_pattern_type(key);
  441. auto style = list_pattern_style(key);
  442. auto start = locale_data.unique_strings.ensure(value.as_object().get("start"sv).as_string());
  443. auto middle = locale_data.unique_strings.ensure(value.as_object().get("middle"sv).as_string());
  444. auto end = locale_data.unique_strings.ensure(value.as_object().get("end"sv).as_string());
  445. auto pair = locale_data.unique_strings.ensure(value.as_object().get("2"sv).as_string());
  446. if (!locale_data.list_pattern_types.contains_slow(type))
  447. locale_data.list_pattern_types.append(type);
  448. ListPatterns list_pattern { type, style, start, middle, end, pair };
  449. list_patterns.append(locale_data.unique_list_patterns.ensure(move(list_pattern)));
  450. });
  451. locale.list_patterns = locale_data.unique_list_pattern_lists.ensure(move(list_patterns));
  452. return {};
  453. }
  454. static ErrorOr<void> parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  455. {
  456. LexicalPath currencies_path(move(numbers_path));
  457. currencies_path = currencies_path.append("currencies.json"sv);
  458. auto locale_currencies = TRY(read_json_file(currencies_path.string()));
  459. auto const& main_object = locale_currencies.as_object().get("main"sv);
  460. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  461. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  462. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  463. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const&) {
  464. if (!locale_data.currencies.contains_slow(key))
  465. locale_data.currencies.append(key);
  466. });
  467. CurrencyList long_currencies {};
  468. long_currencies.resize(locale_data.currencies.size());
  469. CurrencyList short_currencies {};
  470. short_currencies.resize(locale_data.currencies.size());
  471. CurrencyList narrow_currencies {};
  472. narrow_currencies.resize(locale_data.currencies.size());
  473. CurrencyList numeric_currencies {};
  474. numeric_currencies.resize(locale_data.currencies.size());
  475. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  476. auto const& long_name = value.as_object().get("displayName"sv);
  477. auto const& short_name = value.as_object().get("symbol"sv);
  478. auto const& narrow_name = value.as_object().get("symbol-alt-narrow"sv);
  479. auto const& numeric_name = value.as_object().get("displayName-count-other"sv);
  480. auto index = locale_data.currencies.find_first_index(key).value();
  481. long_currencies[index] = locale_data.unique_strings.ensure(long_name.as_string());
  482. short_currencies[index] = locale_data.unique_strings.ensure(short_name.as_string());
  483. narrow_currencies[index] = narrow_name.is_null() ? 0 : locale_data.unique_strings.ensure(narrow_name.as_string());
  484. numeric_currencies[index] = locale_data.unique_strings.ensure(numeric_name.is_null() ? long_name.as_string() : numeric_name.as_string());
  485. });
  486. locale.long_currencies = locale_data.unique_currency_lists.ensure(move(long_currencies));
  487. locale.short_currencies = locale_data.unique_currency_lists.ensure(move(short_currencies));
  488. locale.narrow_currencies = locale_data.unique_currency_lists.ensure(move(narrow_currencies));
  489. locale.numeric_currencies = locale_data.unique_currency_lists.ensure(move(numeric_currencies));
  490. return {};
  491. }
  492. static ErrorOr<void> parse_locale_calendars(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  493. {
  494. LexicalPath locale_display_names_path(move(locale_path));
  495. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  496. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  497. auto const& main_object = locale_display_names.as_object().get("main"sv);
  498. auto const& locale_object = main_object.as_object().get(locale_display_names_path.parent().basename());
  499. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  500. auto const& types_object = locale_display_names_object.as_object().get("types"sv);
  501. auto const& calendar_object = types_object.as_object().get("calendar"sv);
  502. auto const& supported_calendars = locale_data.keywords.find("ca"sv)->value;
  503. CalendarList calendars;
  504. calendars.resize(supported_calendars.size());
  505. calendar_object.as_object().for_each_member([&](auto const& key, auto const& calendar) {
  506. auto index = supported_calendars.find_first_index(key);
  507. if (!index.has_value()) {
  508. auto alias = find_keyword_alias("ca"sv, key, locale_data);
  509. index = supported_calendars.find_first_index(*alias);
  510. }
  511. calendars[*index] = locale_data.unique_strings.ensure(calendar.as_string());
  512. });
  513. locale.calendars = locale_data.unique_calendar_lists.ensure(move(calendars));
  514. return {};
  515. }
  516. static ErrorOr<void> parse_locale_date_fields(String dates_path, UnicodeLocaleData& locale_data, Locale& locale)
  517. {
  518. LexicalPath date_fields_path(move(dates_path));
  519. date_fields_path = date_fields_path.append("dateFields.json"sv);
  520. auto locale_date_fields = TRY(read_json_file(date_fields_path.string()));
  521. auto const& main_object = locale_date_fields.as_object().get("main"sv);
  522. auto const& locale_object = main_object.as_object().get(date_fields_path.parent().basename());
  523. auto const& dates_object = locale_object.as_object().get("dates"sv);
  524. auto const& fields_object = dates_object.as_object().get("fields"sv);
  525. auto is_sanctioned_field = [](StringView field) {
  526. // This is a copy of the units sanctioned for use within ECMA-402, with names adjusted for the names used by the CLDR.
  527. // https://tc39.es/ecma402/#table-validcodeforDateField
  528. 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);
  529. };
  530. fields_object.as_object().for_each_member([&](auto const& key, JsonValue const&) {
  531. if (!is_sanctioned_field(key))
  532. return;
  533. if (!locale_data.date_fields.contains_slow(key))
  534. locale_data.date_fields.append(key);
  535. });
  536. quick_sort(locale_data.date_fields);
  537. DateFieldList long_date_fields {};
  538. long_date_fields.resize(locale_data.date_fields.size());
  539. DateFieldList short_date_fields {};
  540. short_date_fields.resize(locale_data.date_fields.size());
  541. DateFieldList narrow_date_fields {};
  542. narrow_date_fields.resize(locale_data.date_fields.size());
  543. fields_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  544. if (!is_sanctioned_field(key))
  545. return;
  546. auto const& long_name = value.as_object().get("displayName"sv);
  547. auto const& short_name = fields_object.as_object().get(String::formatted("{}-short", key)).as_object().get("displayName"sv);
  548. auto const& narrow_name = fields_object.as_object().get(String::formatted("{}-narrow", key)).as_object().get("displayName"sv);
  549. auto index = locale_data.date_fields.find_first_index(key).value();
  550. long_date_fields[index] = locale_data.unique_strings.ensure(long_name.as_string());
  551. short_date_fields[index] = locale_data.unique_strings.ensure(short_name.as_string());
  552. narrow_date_fields[index] = locale_data.unique_strings.ensure(narrow_name.as_string());
  553. });
  554. locale.long_date_fields = locale_data.unique_date_field_lists.ensure(move(long_date_fields));
  555. locale.short_date_fields = locale_data.unique_date_field_lists.ensure(move(short_date_fields));
  556. locale.narrow_date_fields = locale_data.unique_date_field_lists.ensure(move(narrow_date_fields));
  557. return {};
  558. }
  559. static ErrorOr<void> parse_number_system_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  560. {
  561. LexicalPath numbers_path(move(locale_numbers_path));
  562. numbers_path = numbers_path.append("numbers.json"sv);
  563. auto numbers = TRY(read_json_file(numbers_path.string()));
  564. auto const& main_object = numbers.as_object().get("main"sv);
  565. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  566. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  567. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  568. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  569. KeywordList keywords {};
  570. auto append_numbering_system = [&](String system_name) {
  571. if (auto system_alias = find_keyword_alias("nu"sv, system_name, locale_data); system_alias.has_value())
  572. system_name = system_alias.release_value();
  573. auto index = locale_data.unique_strings.ensure(move(system_name));
  574. if (!keywords.contains_slow(index))
  575. keywords.append(move(index));
  576. };
  577. append_numbering_system(default_numbering_system_object.as_string());
  578. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  579. append_numbering_system(value.as_string());
  580. });
  581. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  582. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  583. return;
  584. append_numbering_system(value.as_string());
  585. });
  586. locale.number_system_keywords = locale_data.unique_keyword_lists.ensure(move(keywords));
  587. return {};
  588. }
  589. static ErrorOr<void> parse_calendar_keywords(String locale_dates_path, UnicodeLocaleData& locale_data, Locale& locale)
  590. {
  591. auto calendars_iterator = TRY(path_to_dir_iterator(locale_dates_path, {}));
  592. KeywordList keywords {};
  593. while (calendars_iterator.has_next()) {
  594. auto locale_calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  595. LexicalPath calendars_path(move(locale_calendars_path));
  596. if (!calendars_path.basename().starts_with("ca-"sv))
  597. continue;
  598. auto calendars = TRY(read_json_file(calendars_path.string()));
  599. auto const& main_object = calendars.as_object().get("main"sv);
  600. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  601. auto const& dates_object = locale_object.as_object().get("dates"sv);
  602. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  603. calendars_object.as_object().for_each_member([&](auto calendar_name, JsonValue const&) {
  604. // The generic calendar is not a supported Unicode calendar key, so skip it:
  605. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  606. if (calendar_name == "generic"sv)
  607. return;
  608. if (auto calendar_alias = find_keyword_alias("ca"sv, calendar_name, locale_data); calendar_alias.has_value())
  609. calendar_name = calendar_alias.release_value();
  610. keywords.append(locale_data.unique_strings.ensure(calendar_name));
  611. });
  612. }
  613. locale.calendar_keywords = locale_data.unique_keyword_lists.ensure(move(keywords));
  614. return {};
  615. }
  616. static void fill_in_collation_keywords(UnicodeLocaleData& locale_data, Locale& locale)
  617. {
  618. // FIXME: If collation data becomes available in the CLDR, parse per-locale ordering from there.
  619. auto create_list_with_default_first = [&](auto key, auto default_value) {
  620. auto& values = locale_data.keywords.find(key)->value;
  621. quick_sort(values, [&](auto const& lhs, auto const& rhs) {
  622. if (lhs == default_value)
  623. return true;
  624. if (rhs == default_value)
  625. return false;
  626. return lhs < rhs;
  627. });
  628. KeywordList keywords;
  629. keywords.ensure_capacity(values.size());
  630. for (auto const& value : values)
  631. keywords.append(locale_data.unique_strings.ensure(value));
  632. return locale_data.unique_keyword_lists.ensure(move(keywords));
  633. };
  634. static auto kf_index = create_list_with_default_first("kf"sv, "upper"sv);
  635. static auto kn_index = create_list_with_default_first("kn"sv, "true"sv);
  636. locale.collation_case_keywords = kf_index;
  637. locale.collation_numeric_keywords = kn_index;
  638. }
  639. static ErrorOr<void> parse_default_content_locales(String core_path, UnicodeLocaleData& locale_data)
  640. {
  641. LexicalPath default_content_path(move(core_path));
  642. default_content_path = default_content_path.append("defaultContent.json"sv);
  643. auto default_content = TRY(read_json_file(default_content_path.string()));
  644. auto const& default_content_array = default_content.as_object().get("defaultContent"sv);
  645. default_content_array.as_array().for_each([&](JsonValue const& value) {
  646. auto locale = value.as_string();
  647. StringView default_locale = locale;
  648. while (true) {
  649. if (locale_data.locales.contains(default_locale))
  650. break;
  651. auto pos = default_locale.find_last('-');
  652. if (!pos.has_value())
  653. return;
  654. default_locale = default_locale.substring_view(0, *pos);
  655. }
  656. if (default_locale != locale)
  657. locale_data.locale_aliases.append({ default_locale, move(locale) });
  658. });
  659. return {};
  660. }
  661. static ErrorOr<void> define_aliases_without_scripts(UnicodeLocaleData& locale_data)
  662. {
  663. // From ECMA-402: https://tc39.es/ecma402/#sec-internal-slots
  664. //
  665. // For locales that include a script subtag in addition to language and region, the
  666. // corresponding locale without a script subtag must also be supported.
  667. //
  668. // So we define aliases for locales that contain all three subtags, but we must also take
  669. // care to handle when the locale itself or the locale without a script subtag are an alias
  670. // by way of default-content locales.
  671. auto find_alias = [&](auto const& locale) {
  672. return locale_data.locale_aliases.find_if([&](auto const& alias) { return locale == alias.alias; });
  673. };
  674. auto append_alias_without_script = [&](auto const& locale) -> ErrorOr<void> {
  675. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, locale));
  676. if ((parsed_locale.language == 0) || (parsed_locale.script == 0) || (parsed_locale.region == 0))
  677. return {};
  678. auto locale_without_script = String::formatted("{}-{}",
  679. locale_data.unique_strings.get(parsed_locale.language),
  680. locale_data.unique_strings.get(parsed_locale.region));
  681. if (locale_data.locales.contains(locale_without_script))
  682. return {};
  683. if (find_alias(locale_without_script) != locale_data.locale_aliases.end())
  684. return {};
  685. if (auto it = find_alias(locale); it != locale_data.locale_aliases.end())
  686. locale_data.locale_aliases.append({ it->name, locale_without_script });
  687. else
  688. locale_data.locale_aliases.append({ locale, locale_without_script });
  689. return {};
  690. };
  691. for (auto const& locale : locale_data.locales)
  692. TRY(append_alias_without_script(locale.key));
  693. for (auto const& locale : locale_data.locale_aliases)
  694. TRY(append_alias_without_script(locale.alias));
  695. return {};
  696. }
  697. static ErrorOr<void> parse_all_locales(String bcp47_path, String core_path, String locale_names_path, String misc_path, String numbers_path, String dates_path, UnicodeLocaleData& locale_data)
  698. {
  699. auto bcp47_iterator = TRY(path_to_dir_iterator(move(bcp47_path), "bcp47"sv));
  700. auto identity_iterator = TRY(path_to_dir_iterator(locale_names_path));
  701. auto preprocess_iterator = TRY(path_to_dir_iterator(locale_names_path));
  702. auto locale_names_iterator = TRY(path_to_dir_iterator(move(locale_names_path)));
  703. auto misc_iterator = TRY(path_to_dir_iterator(move(misc_path)));
  704. auto numbers_iterator = TRY(path_to_dir_iterator(move(numbers_path)));
  705. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  706. LexicalPath core_supplemental_path(core_path);
  707. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  708. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  709. TRY(parse_core_aliases(core_supplemental_path.string(), locale_data));
  710. TRY(parse_likely_subtags(core_supplemental_path.string(), locale_data));
  711. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  712. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  713. StringBuilder builder;
  714. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  715. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  716. builder.appendff("-{}", script);
  717. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  718. builder.appendff("-{}", region);
  719. return builder.build();
  720. };
  721. while (identity_iterator.has_next()) {
  722. auto locale_path = TRY(next_path_from_dir_iterator(identity_iterator));
  723. auto language = TRY(remove_variants_from_path(locale_path));
  724. auto& locale = locale_data.locales.ensure(language);
  725. TRY(parse_identity(locale_path, locale_data, locale));
  726. }
  727. while (preprocess_iterator.has_next()) {
  728. auto locale_path = TRY(next_path_from_dir_iterator(preprocess_iterator));
  729. TRY(preprocess_languages(locale_path, locale_data));
  730. }
  731. quick_sort(locale_data.languages);
  732. quick_sort(locale_data.territories);
  733. quick_sort(locale_data.scripts);
  734. while (bcp47_iterator.has_next()) {
  735. auto bcp47_path = TRY(next_path_from_dir_iterator(bcp47_iterator));
  736. TRY(parse_unicode_extension_keywords(move(bcp47_path), locale_data));
  737. }
  738. while (locale_names_iterator.has_next()) {
  739. auto locale_path = TRY(next_path_from_dir_iterator(locale_names_iterator));
  740. auto language = TRY(remove_variants_from_path(locale_path));
  741. auto& locale = locale_data.locales.ensure(language);
  742. TRY(parse_locale_display_patterns(locale_path, locale_data, locale));
  743. TRY(parse_locale_languages(locale_path, locale_data, locale));
  744. TRY(parse_locale_territories(locale_path, locale_data, locale));
  745. TRY(parse_locale_scripts(locale_path, locale_data, locale));
  746. TRY(parse_locale_calendars(locale_path, locale_data, locale));
  747. }
  748. while (misc_iterator.has_next()) {
  749. auto misc_path = TRY(next_path_from_dir_iterator(misc_iterator));
  750. auto language = TRY(remove_variants_from_path(misc_path));
  751. auto& locale = locale_data.locales.ensure(language);
  752. TRY(parse_locale_list_patterns(misc_path, locale_data, locale));
  753. }
  754. while (numbers_iterator.has_next()) {
  755. auto numbers_path = TRY(next_path_from_dir_iterator(numbers_iterator));
  756. auto language = TRY(remove_variants_from_path(numbers_path));
  757. auto& locale = locale_data.locales.ensure(language);
  758. TRY(parse_locale_currencies(numbers_path, locale_data, locale));
  759. TRY(parse_number_system_keywords(numbers_path, locale_data, locale));
  760. fill_in_collation_keywords(locale_data, locale);
  761. }
  762. while (dates_iterator.has_next()) {
  763. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  764. auto language = TRY(remove_variants_from_path(dates_path));
  765. auto& locale = locale_data.locales.ensure(language);
  766. TRY(parse_locale_date_fields(dates_path, locale_data, locale));
  767. TRY(parse_calendar_keywords(dates_path, locale_data, locale));
  768. }
  769. TRY(parse_default_content_locales(move(core_path), locale_data));
  770. TRY(define_aliases_without_scripts(locale_data));
  771. return {};
  772. }
  773. static ErrorOr<void> generate_unicode_locale_header(Core::Stream::BufferedFile& file, UnicodeLocaleData& locale_data)
  774. {
  775. StringBuilder builder;
  776. SourceGenerator generator { builder };
  777. generator.append(R"~~~(
  778. #pragma once
  779. #include <AK/Types.h>
  780. namespace Unicode {
  781. )~~~");
  782. auto locales = locale_data.locales.keys();
  783. auto keywords = locale_data.keywords.keys();
  784. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, locale_data.locale_aliases);
  785. generate_enum(generator, format_identifier, "Language"sv, {}, locale_data.languages);
  786. generate_enum(generator, format_identifier, "Territory"sv, {}, locale_data.territories);
  787. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, locale_data.scripts);
  788. generate_enum(generator, format_identifier, "Currency"sv, {}, locale_data.currencies);
  789. generate_enum(generator, format_identifier, "DateField"sv, {}, locale_data.date_fields, locale_data.date_field_aliases);
  790. generate_enum(generator, format_identifier, "Variant"sv, {}, locale_data.variants);
  791. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, locale_data.list_pattern_types);
  792. generate_enum(generator, format_identifier, "Key"sv, {}, keywords);
  793. for (auto& keyword : locale_data.keywords) {
  794. auto const& keyword_name = locale_data.keyword_names.find(keyword.key)->value;
  795. auto enum_name = String::formatted("Keyword{}", format_identifier({}, keyword_name));
  796. if (auto aliases = locale_data.keyword_aliases.find(keyword.key); aliases != locale_data.keyword_aliases.end())
  797. generate_enum(generator, format_identifier, enum_name, {}, keyword.value, aliases->value);
  798. else
  799. generate_enum(generator, format_identifier, enum_name, {}, keyword.value);
  800. }
  801. generator.append(R"~~~(
  802. }
  803. )~~~");
  804. TRY(file.write(generator.as_string_view().bytes()));
  805. return {};
  806. }
  807. static ErrorOr<void> generate_unicode_locale_implementation(Core::Stream::BufferedFile& file, UnicodeLocaleData& locale_data)
  808. {
  809. StringBuilder builder;
  810. SourceGenerator generator { builder };
  811. generator.set("string_index_type"sv, s_string_index_type);
  812. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  813. generator.set("territories_size", String::number(locale_data.territories.size()));
  814. generator.set("variants_size", String::number(locale_data.max_variant_size));
  815. generator.append(R"~~~(
  816. #include <AK/Array.h>
  817. #include <AK/BinarySearch.h>
  818. #include <AK/Optional.h>
  819. #include <AK/Span.h>
  820. #include <AK/StringView.h>
  821. #include <AK/Vector.h>
  822. #include <LibUnicode/CurrencyCode.h>
  823. #include <LibUnicode/DateTimeFormat.h>
  824. #include <LibUnicode/Locale.h>
  825. #include <LibUnicode/UnicodeLocale.h>
  826. namespace Unicode {
  827. )~~~");
  828. locale_data.unique_strings.generate(generator);
  829. generator.append(R"~~~(
  830. struct DisplayPatternImpl {
  831. DisplayPattern to_display_pattern() const
  832. {
  833. DisplayPattern display_patterns {};
  834. display_patterns.locale_pattern = s_string_list[locale_pattern];
  835. display_patterns.locale_separator = s_string_list[locale_separator];
  836. return display_patterns;
  837. }
  838. @string_index_type@ locale_pattern { 0 };
  839. @string_index_type@ locale_separator { 0 };
  840. };
  841. struct Patterns {
  842. ListPatternType type;
  843. Style style;
  844. @string_index_type@ start { 0 };
  845. @string_index_type@ middle { 0 };
  846. @string_index_type@ end { 0 };
  847. @string_index_type@ pair { 0 };
  848. };
  849. )~~~");
  850. generate_available_values(generator, "get_available_currencies"sv, locale_data.currencies);
  851. locale_data.unique_display_patterns.generate(generator, "DisplayPatternImpl"sv, "s_display_patterns"sv, 30);
  852. locale_data.unique_language_lists.generate(generator, s_string_index_type, "s_language_lists"sv);
  853. locale_data.unique_territory_lists.generate(generator, s_string_index_type, "s_territory_lists"sv);
  854. locale_data.unique_script_lists.generate(generator, s_string_index_type, "s_script_lists"sv);
  855. locale_data.unique_currency_lists.generate(generator, s_string_index_type, "s_currency_lists"sv);
  856. locale_data.unique_calendar_lists.generate(generator, s_string_index_type, "s_calendar_lists"sv);
  857. locale_data.unique_date_field_lists.generate(generator, s_string_index_type, "s_date_field_lists"sv);
  858. locale_data.unique_keyword_lists.generate(generator, s_string_index_type, "s_keyword_lists"sv);
  859. locale_data.unique_list_patterns.generate(generator, "Patterns"sv, "s_list_patterns"sv, 10);
  860. locale_data.unique_list_pattern_lists.generate(generator, s_list_pattern_index_type, "s_list_pattern_lists"sv);
  861. auto append_index = [&](auto index) {
  862. generator.append(String::formatted(", {}", index));
  863. };
  864. auto append_list_and_size = [&](auto const& list) {
  865. if (list.is_empty()) {
  866. generator.append(", {}, 0");
  867. return;
  868. }
  869. bool first = true;
  870. generator.append(", {");
  871. for (auto const& item : list) {
  872. generator.append(first ? " " : ", ");
  873. generator.append(String::number(item));
  874. first = false;
  875. }
  876. generator.append(String::formatted(" }}, {}", list.size()));
  877. };
  878. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  879. generator.set("type", type);
  880. generator.set("name", name);
  881. generator.set("size", String::number(keys.size()));
  882. generator.append(R"~~~(
  883. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  884. bool first = true;
  885. for (auto const& key : keys) {
  886. auto const& value = map.find(key)->value;
  887. auto mapping = mapping_getter(value);
  888. generator.append(first ? " " : ", ");
  889. generator.append(String::number(mapping));
  890. first = false;
  891. }
  892. generator.append(" } };");
  893. };
  894. auto locales = locale_data.locales.keys();
  895. quick_sort(locales);
  896. append_mapping(locales, locale_data.locales, s_display_pattern_index_type, "s_locale_display_patterns"sv, [&](auto const& locale) { return locale.display_patterns; });
  897. append_mapping(locales, locale_data.locales, s_language_list_index_type, "s_languages"sv, [&](auto const& locale) { return locale.languages; });
  898. append_mapping(locales, locale_data.locales, s_territory_list_index_type, "s_territories"sv, [&](auto const& locale) { return locale.territories; });
  899. append_mapping(locales, locale_data.locales, s_script_list_index_type, "s_scripts"sv, [&](auto const& locale) { return locale.scripts; });
  900. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_long_currencies"sv, [&](auto const& locale) { return locale.long_currencies; });
  901. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_short_currencies"sv, [&](auto const& locale) { return locale.short_currencies; });
  902. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_narrow_currencies"sv, [&](auto const& locale) { return locale.narrow_currencies; });
  903. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_numeric_currencies"sv, [&](auto const& locale) { return locale.numeric_currencies; });
  904. append_mapping(locales, locale_data.locales, s_calendar_list_index_type, "s_calendars"sv, [&](auto const& locale) { return locale.calendars; });
  905. append_mapping(locales, locale_data.locales, s_date_field_list_index_type, "s_long_date_fields"sv, [&](auto const& locale) { return locale.long_date_fields; });
  906. append_mapping(locales, locale_data.locales, s_date_field_list_index_type, "s_short_date_fields"sv, [&](auto const& locale) { return locale.short_date_fields; });
  907. append_mapping(locales, locale_data.locales, s_date_field_list_index_type, "s_narrow_date_fields"sv, [&](auto const& locale) { return locale.narrow_date_fields; });
  908. append_mapping(locales, locale_data.locales, s_keyword_list_index_type, "s_calendar_keywords"sv, [&](auto const& locale) { return locale.calendar_keywords; });
  909. append_mapping(locales, locale_data.locales, s_keyword_list_index_type, "s_collation_case_keywords"sv, [&](auto const& locale) { return locale.collation_case_keywords; });
  910. append_mapping(locales, locale_data.locales, s_keyword_list_index_type, "s_collation_numeric_keywords"sv, [&](auto const& locale) { return locale.collation_numeric_keywords; });
  911. append_mapping(locales, locale_data.locales, s_keyword_list_index_type, "s_number_system_keywords"sv, [&](auto const& locale) { return locale.number_system_keywords; });
  912. append_mapping(locales, locale_data.locales, s_list_pattern_list_index_type, "s_locale_list_patterns"sv, [&](auto const& locale) { return locale.list_patterns; });
  913. generator.append(R"~~~(
  914. struct CanonicalLanguageID {
  915. LanguageID to_unicode_language_id() const
  916. {
  917. LanguageID language_id {};
  918. language_id.variants.ensure_capacity(variants_size);
  919. language_id.language = s_string_list[language];
  920. if (script != 0)
  921. language_id.script = s_string_list[script];
  922. if (region != 0)
  923. language_id.region = s_string_list[region];
  924. for (size_t i = 0; i < variants_size; ++i)
  925. language_id.variants.append(s_string_list[variants[i]]);
  926. return language_id;
  927. }
  928. bool matches_variants(Vector<String> const& other_variants) const {
  929. if (variants_size == 0)
  930. return true;
  931. if (other_variants.size() != variants_size)
  932. return false;
  933. for (size_t i = 0; i < variants_size; ++i) {
  934. if (s_string_list[variants[i]] != other_variants[i])
  935. return false;
  936. }
  937. return true;
  938. };
  939. @string_index_type@ language { 0 };
  940. @string_index_type@ script { 0 };
  941. @string_index_type@ region { 0 };
  942. Array<@string_index_type@, @variants_size@> variants {};
  943. size_t variants_size { 0 };
  944. };
  945. struct LanguageMapping {
  946. CanonicalLanguageID key;
  947. CanonicalLanguageID alias;
  948. };
  949. )~~~");
  950. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  951. generator.set("size", String::number(mappings.size()));
  952. generator.set("name"sv, name);
  953. generator.append(R"~~~(
  954. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  955. )~~~");
  956. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  957. auto const& lhs_language = locale_data.unique_strings.get(lhs.key.language);
  958. auto const& rhs_language = locale_data.unique_strings.get(rhs.key.language);
  959. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  960. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  961. return false;
  962. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  963. return true;
  964. return lhs_language < rhs_language;
  965. });
  966. for (auto const& mapping : mappings) {
  967. generator.set("language"sv, String::number(mapping.key.language));
  968. generator.append(" { { @language@");
  969. append_index(mapping.key.script);
  970. append_index(mapping.key.region);
  971. append_list_and_size(mapping.key.variants);
  972. generator.set("language"sv, String::number(mapping.alias.language));
  973. generator.append(" }, { @language@");
  974. append_index(mapping.alias.script);
  975. append_index(mapping.alias.region);
  976. append_list_and_size(mapping.alias.variants);
  977. generator.append(" } },\n");
  978. }
  979. generator.append("} };\n");
  980. };
  981. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  982. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  983. generator.append(R"~~~(
  984. static LanguageMapping const* resolve_likely_subtag(LanguageID const& language_id)
  985. {
  986. // https://unicode.org/reports/tr35/#Likely_Subtags
  987. enum class State {
  988. LanguageScriptRegion,
  989. LanguageRegion,
  990. LanguageScript,
  991. Language,
  992. UndScript,
  993. Done,
  994. };
  995. auto state = State::LanguageScriptRegion;
  996. while (state != State::Done) {
  997. LanguageID search_key;
  998. switch (state) {
  999. case State::LanguageScriptRegion:
  1000. state = State::LanguageRegion;
  1001. if (!language_id.script.has_value() || !language_id.region.has_value())
  1002. continue;
  1003. search_key.language = *language_id.language;
  1004. search_key.script = *language_id.script;
  1005. search_key.region = *language_id.region;
  1006. break;
  1007. case State::LanguageRegion:
  1008. state = State::LanguageScript;
  1009. if (!language_id.region.has_value())
  1010. continue;
  1011. search_key.language = *language_id.language;
  1012. search_key.region = *language_id.region;
  1013. break;
  1014. case State::LanguageScript:
  1015. state = State::Language;
  1016. if (!language_id.script.has_value())
  1017. continue;
  1018. search_key.language = *language_id.language;
  1019. search_key.script = *language_id.script;
  1020. break;
  1021. case State::Language:
  1022. state = State::UndScript;
  1023. search_key.language = *language_id.language;
  1024. break;
  1025. case State::UndScript:
  1026. state = State::Done;
  1027. if (!language_id.script.has_value())
  1028. continue;
  1029. search_key.language = "und"sv;
  1030. search_key.script = *language_id.script;
  1031. break;
  1032. default:
  1033. VERIFY_NOT_REACHED();
  1034. }
  1035. for (auto const& map : s_likely_subtags) {
  1036. auto const& key_language = s_string_list[map.key.language];
  1037. auto const& key_script = s_string_list[map.key.script];
  1038. auto const& key_region = s_string_list[map.key.region];
  1039. if (key_language != search_key.language)
  1040. continue;
  1041. if (!key_script.is_empty() || search_key.script.has_value()) {
  1042. if (key_script != search_key.script)
  1043. continue;
  1044. }
  1045. if (!key_region.is_empty() || search_key.region.has_value()) {
  1046. if (key_region != search_key.region)
  1047. continue;
  1048. }
  1049. return &map;
  1050. }
  1051. }
  1052. return nullptr;
  1053. }
  1054. )~~~");
  1055. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name, StringView unique_list) {
  1056. generator.set("enum_snake", enum_snake);
  1057. generator.set("from_string_name", from_string_name);
  1058. generator.set("collection_name", collection_name);
  1059. generator.set("unique_list", unique_list);
  1060. generator.append(R"~~~(
  1061. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  1062. {
  1063. auto locale_value = locale_from_string(locale);
  1064. if (!locale_value.has_value())
  1065. return {};
  1066. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  1067. if (!@enum_snake@_value.has_value())
  1068. return {};
  1069. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1070. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  1071. auto mapping_index = @collection_name@.at(locale_index);
  1072. auto const& mappings = @unique_list@.at(mapping_index);
  1073. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  1074. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  1075. if (@enum_snake@_mapping.is_empty())
  1076. return {};
  1077. return @enum_snake@_mapping;
  1078. }
  1079. )~~~");
  1080. };
  1081. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1082. HashValueMap<String> hashes;
  1083. hashes.ensure_capacity(values.size());
  1084. for (auto const& value : values)
  1085. hashes.set(value.hash(), format_identifier(enum_title, value));
  1086. for (auto const& alias : aliases)
  1087. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1088. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1089. };
  1090. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  1091. HashValueMap<StringIndexType> hashes;
  1092. hashes.ensure_capacity(aliases.size());
  1093. for (auto const& alias : aliases)
  1094. hashes.set(alias.key.hash(), alias.value);
  1095. ValueFromStringOptions options {};
  1096. options.return_type = "StringView"sv;
  1097. options.return_format = "s_string_list[{}]"sv;
  1098. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), options);
  1099. };
  1100. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys(), locale_data.locale_aliases);
  1101. append_from_string("Language"sv, "language"sv, locale_data.languages);
  1102. append_mapping_search("language"sv, "language"sv, "s_languages"sv, "s_language_lists"sv);
  1103. append_alias_search("language"sv, locale_data.language_aliases);
  1104. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  1105. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv, "s_territory_lists"sv);
  1106. append_alias_search("territory"sv, locale_data.territory_aliases);
  1107. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  1108. append_mapping_search("script"sv, "script_tag"sv, "s_scripts"sv, "s_script_lists"sv);
  1109. append_alias_search("script_tag"sv, locale_data.script_aliases);
  1110. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  1111. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv, "s_currency_lists"sv);
  1112. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv, "s_currency_lists"sv);
  1113. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv, "s_currency_lists"sv);
  1114. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv, "s_currency_lists"sv);
  1115. append_from_string("DateField"sv, "date_field"sv, locale_data.date_fields, locale_data.date_field_aliases);
  1116. append_mapping_search("long_date_field"sv, "date_field"sv, "s_long_date_fields"sv, "s_date_field_lists"sv);
  1117. append_mapping_search("short_date_field"sv, "date_field"sv, "s_short_date_fields"sv, "s_date_field_lists"sv);
  1118. append_mapping_search("narrow_date_field"sv, "date_field"sv, "s_narrow_date_fields"sv, "s_date_field_lists"sv);
  1119. append_from_string("Key"sv, "key"sv, locale_data.keywords.keys());
  1120. for (auto const& keyword : locale_data.keywords) {
  1121. auto const& keyword_name = locale_data.keyword_names.find(keyword.key)->value;
  1122. auto enum_name = String::formatted("Keyword{}", format_identifier({}, keyword_name));
  1123. auto enum_snake = String::formatted("keyword_{}", keyword.key);
  1124. if (auto aliases = locale_data.keyword_aliases.find(keyword.key); aliases != locale_data.keyword_aliases.end())
  1125. append_from_string(enum_name, enum_snake, keyword.value, aliases->value);
  1126. else
  1127. append_from_string(enum_name, enum_snake, keyword.value);
  1128. }
  1129. append_mapping_search("calendar"sv, "keyword_ca"sv, "s_calendars"sv, "s_calendar_lists"sv);
  1130. append_alias_search("variant"sv, locale_data.variant_aliases);
  1131. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  1132. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  1133. generator.append(R"~~~(
  1134. Vector<StringView> get_keywords_for_locale(StringView locale, StringView key)
  1135. {
  1136. // Hour cycle keywords are region-based rather than locale-based, so they need to be handled specially.
  1137. // FIXME: Calendar keywords are also region-based, and will need to be handled here when we support non-Gregorian calendars:
  1138. // https://github.com/unicode-org/cldr-json/blob/main/cldr-json/cldr-core/supplemental/calendarPreferenceData.json
  1139. if (key == "hc"sv) {
  1140. auto hour_cycles = get_locale_hour_cycles(locale);
  1141. Vector<StringView> values;
  1142. values.ensure_capacity(hour_cycles.size());
  1143. for (auto hour_cycle : hour_cycles)
  1144. values.unchecked_append(hour_cycle_to_string(hour_cycle));
  1145. return values;
  1146. }
  1147. auto locale_value = locale_from_string(locale);
  1148. if (!locale_value.has_value())
  1149. return {};
  1150. auto key_value = key_from_string(key);
  1151. if (!key_value.has_value())
  1152. return {};
  1153. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1154. size_t keywords_index = 0;
  1155. switch (*key_value) {
  1156. case Key::Ca:
  1157. keywords_index = s_calendar_keywords.at(locale_index);
  1158. break;
  1159. case Key::Kf:
  1160. keywords_index = s_collation_case_keywords.at(locale_index);
  1161. break;
  1162. case Key::Kn:
  1163. keywords_index = s_collation_numeric_keywords.at(locale_index);
  1164. break;
  1165. case Key::Nu:
  1166. keywords_index = s_number_system_keywords.at(locale_index);
  1167. break;
  1168. default:
  1169. VERIFY_NOT_REACHED();
  1170. }
  1171. auto keyword_indices = s_keyword_lists.at(keywords_index);
  1172. Vector<StringView> keywords;
  1173. keywords.ensure_capacity(keyword_indices.size());
  1174. for (auto keyword : keyword_indices)
  1175. keywords.unchecked_append(s_string_list[keyword]);
  1176. return keywords;
  1177. }
  1178. Optional<DisplayPattern> get_locale_display_patterns(StringView locale)
  1179. {
  1180. auto locale_value = locale_from_string(locale);
  1181. if (!locale_value.has_value())
  1182. return {};
  1183. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1184. auto display_patterns_index = s_locale_display_patterns.at(locale_index);
  1185. auto const& display_patterns = s_display_patterns.at(display_patterns_index);
  1186. return display_patterns.to_display_pattern();
  1187. }
  1188. Optional<ListPatterns> get_locale_list_patterns(StringView locale, StringView list_pattern_type, Style list_pattern_style)
  1189. {
  1190. auto locale_value = locale_from_string(locale);
  1191. if (!locale_value.has_value())
  1192. return {};
  1193. auto type_value = list_pattern_type_from_string(list_pattern_type);
  1194. if (!type_value.has_value())
  1195. return {};
  1196. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1197. auto list_patterns_list_index = s_locale_list_patterns.at(locale_index);
  1198. auto const& locale_list_patterns = s_list_pattern_lists.at(list_patterns_list_index);
  1199. for (auto list_patterns_index : locale_list_patterns) {
  1200. auto const& list_patterns = s_list_patterns.at(list_patterns_index);
  1201. if ((list_patterns.type == type_value) && (list_patterns.style == list_pattern_style)) {
  1202. auto const& start = s_string_list[list_patterns.start];
  1203. auto const& middle = s_string_list[list_patterns.middle];
  1204. auto const& end = s_string_list[list_patterns.end];
  1205. auto const& pair = s_string_list[list_patterns.pair];
  1206. return ListPatterns { start, middle, end, pair };
  1207. }
  1208. }
  1209. return {};
  1210. }
  1211. void resolve_complex_language_aliases(LanguageID& language_id)
  1212. {
  1213. for (auto const& map : s_complex_alias) {
  1214. auto const& key_language = s_string_list[map.key.language];
  1215. auto const& key_script = s_string_list[map.key.script];
  1216. auto const& key_region = s_string_list[map.key.region];
  1217. if ((key_language != language_id.language) && (key_language != "und"sv))
  1218. continue;
  1219. if (!key_script.is_empty() && (key_script != language_id.script))
  1220. continue;
  1221. if (!key_region.is_empty() && (key_region != language_id.region))
  1222. continue;
  1223. if (!map.key.matches_variants(language_id.variants))
  1224. continue;
  1225. auto alias = map.alias.to_unicode_language_id();
  1226. if (alias.language == "und"sv)
  1227. alias.language = move(language_id.language);
  1228. if (key_script.is_empty() && !alias.script.has_value())
  1229. alias.script = move(language_id.script);
  1230. if (key_region.is_empty() && !alias.region.has_value())
  1231. alias.region = move(language_id.region);
  1232. if (map.key.variants_size == 0 && alias.variants.is_empty())
  1233. alias.variants = move(language_id.variants);
  1234. language_id = move(alias);
  1235. break;
  1236. }
  1237. }
  1238. Optional<LanguageID> add_likely_subtags(LanguageID const& language_id)
  1239. {
  1240. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  1241. auto const* likely_subtag = resolve_likely_subtag(language_id);
  1242. if (likely_subtag == nullptr)
  1243. return {};
  1244. auto maximized = language_id;
  1245. auto const& key_script = s_string_list[likely_subtag->key.script];
  1246. auto const& key_region = s_string_list[likely_subtag->key.region];
  1247. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  1248. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  1249. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  1250. if (maximized.language == "und"sv)
  1251. maximized.language = alias_language;
  1252. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  1253. maximized.script = alias_script;
  1254. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  1255. maximized.region = alias_region;
  1256. return maximized;
  1257. }
  1258. Optional<String> resolve_most_likely_territory(LanguageID const& language_id)
  1259. {
  1260. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  1261. return s_string_list[likely_subtag->alias.region];
  1262. return {};
  1263. }
  1264. }
  1265. )~~~");
  1266. TRY(file.write(generator.as_string_view().bytes()));
  1267. return {};
  1268. }
  1269. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1270. {
  1271. StringView generated_header_path;
  1272. StringView generated_implementation_path;
  1273. StringView bcp47_path;
  1274. StringView core_path;
  1275. StringView locale_names_path;
  1276. StringView misc_path;
  1277. StringView numbers_path;
  1278. StringView dates_path;
  1279. Core::ArgsParser args_parser;
  1280. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1281. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1282. args_parser.add_option(bcp47_path, "Path to cldr-bcp47 directory", "bcp47-path", 'b', "bcp47-path");
  1283. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1284. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  1285. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  1286. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  1287. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1288. args_parser.parse(arguments);
  1289. auto generated_header_file = TRY(open_file(generated_header_path, Core::Stream::OpenMode::Write));
  1290. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::Stream::OpenMode::Write));
  1291. UnicodeLocaleData locale_data;
  1292. TRY(parse_all_locales(bcp47_path, core_path, locale_names_path, misc_path, numbers_path, dates_path, locale_data));
  1293. TRY(generate_unicode_locale_header(*generated_header_file, locale_data));
  1294. TRY(generate_unicode_locale_implementation(*generated_implementation_file, locale_data));
  1295. return 0;
  1296. }