GenerateUnicodeLocale.cpp 65 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526
  1. /*
  2. * Copyright (c) 2021-2022, Tim Flynn <trflynn89@pm.me>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "GeneratorUtil.h"
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/Format.h>
  10. #include <AK/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. using StringIndexType = u32;
  23. constexpr auto s_string_index_type = "u32"sv;
  24. using DisplayPatternIndexType = u8;
  25. constexpr auto s_display_pattern_index_type = "u8"sv;
  26. using LanguageListIndexType = u8;
  27. constexpr auto s_language_list_index_type = "u8"sv;
  28. using TerritoryListIndexType = u8;
  29. constexpr auto s_territory_list_index_type = "u8"sv;
  30. using ScriptListIndexType = u8;
  31. constexpr auto s_script_list_index_type = "u8"sv;
  32. using CurrencyListIndexType = u16;
  33. constexpr auto s_currency_list_index_type = "u16"sv;
  34. using CalendarListIndexType = u8;
  35. constexpr auto s_calendar_list_index_type = "u8"sv;
  36. using DateFieldListIndexType = u8;
  37. constexpr auto s_date_field_list_index_type = "u8"sv;
  38. using KeywordListIndexType = u8;
  39. constexpr auto s_keyword_list_index_type = "u8"sv;
  40. using ListPatternIndexType = u16;
  41. constexpr auto s_list_pattern_index_type = "u16"sv;
  42. using ListPatternListIndexType = u8;
  43. constexpr auto s_list_pattern_list_index_type = "u8"sv;
  44. static String format_identifier(StringView owner, String identifier)
  45. {
  46. identifier = identifier.replace("-"sv, "_"sv, true);
  47. if (all_of(identifier, is_ascii_digit))
  48. return String::formatted("{}_{}", owner[0], identifier);
  49. if (is_ascii_lower_alpha(identifier[0]))
  50. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  51. return identifier;
  52. }
  53. struct DisplayPattern {
  54. unsigned hash() const
  55. {
  56. return pair_int_hash(locale_pattern, locale_separator);
  57. }
  58. bool operator==(DisplayPattern const& other) const
  59. {
  60. return (locale_pattern == other.locale_pattern)
  61. && (locale_separator == other.locale_separator);
  62. }
  63. StringIndexType locale_pattern { 0 };
  64. StringIndexType locale_separator { 0 };
  65. };
  66. template<>
  67. struct AK::Formatter<DisplayPattern> : Formatter<FormatString> {
  68. ErrorOr<void> format(FormatBuilder& builder, DisplayPattern const& patterns)
  69. {
  70. return Formatter<FormatString>::format(builder,
  71. "{{ {}, {} }}",
  72. patterns.locale_pattern,
  73. patterns.locale_separator);
  74. }
  75. };
  76. template<>
  77. struct AK::Traits<DisplayPattern> : public GenericTraits<DisplayPattern> {
  78. static unsigned hash(DisplayPattern const& p) { return p.hash(); }
  79. };
  80. struct ListPatterns {
  81. unsigned hash() const
  82. {
  83. auto hash = pair_int_hash(type.hash(), style.hash());
  84. hash = pair_int_hash(hash, start);
  85. hash = pair_int_hash(hash, middle);
  86. hash = pair_int_hash(hash, end);
  87. hash = pair_int_hash(hash, pair);
  88. return hash;
  89. }
  90. bool operator==(ListPatterns const& other) const
  91. {
  92. return (type == other.type)
  93. && (style == other.style)
  94. && (start == other.start)
  95. && (middle == other.middle)
  96. && (end == other.end)
  97. && (pair == other.pair);
  98. }
  99. StringView type;
  100. StringView style;
  101. StringIndexType start { 0 };
  102. StringIndexType middle { 0 };
  103. StringIndexType end { 0 };
  104. StringIndexType pair { 0 };
  105. };
  106. template<>
  107. struct AK::Formatter<ListPatterns> : Formatter<FormatString> {
  108. ErrorOr<void> format(FormatBuilder& builder, ListPatterns const& patterns)
  109. {
  110. return Formatter<FormatString>::format(builder,
  111. "{{ ListPatternType::{}, Style::{}, {}, {}, {}, {} }}",
  112. format_identifier({}, patterns.type),
  113. format_identifier({}, patterns.style),
  114. patterns.start,
  115. patterns.middle,
  116. patterns.end,
  117. patterns.pair);
  118. }
  119. };
  120. template<>
  121. struct AK::Traits<ListPatterns> : public GenericTraits<ListPatterns> {
  122. static unsigned hash(ListPatterns const& p) { return p.hash(); }
  123. };
  124. using LanguageList = Vector<StringIndexType>;
  125. using TerritoryList = Vector<StringIndexType>;
  126. using ScriptList = Vector<StringIndexType>;
  127. using CurrencyList = Vector<StringIndexType>;
  128. using CalendarList = Vector<StringIndexType>;
  129. using DateFieldList = Vector<StringIndexType>;
  130. using KeywordList = Vector<StringIndexType>;
  131. using ListPatternList = Vector<ListPatternIndexType>;
  132. struct Locale {
  133. String language;
  134. Optional<String> territory;
  135. Optional<String> variant;
  136. DisplayPatternIndexType display_patterns { 0 };
  137. LanguageListIndexType languages { 0 };
  138. TerritoryListIndexType territories { 0 };
  139. ScriptListIndexType scripts { 0 };
  140. CurrencyListIndexType long_currencies { 0 };
  141. CurrencyListIndexType short_currencies { 0 };
  142. CurrencyListIndexType narrow_currencies { 0 };
  143. CurrencyListIndexType numeric_currencies { 0 };
  144. CalendarListIndexType calendars { 0 };
  145. DateFieldListIndexType long_date_fields { 0 };
  146. DateFieldListIndexType short_date_fields { 0 };
  147. DateFieldListIndexType narrow_date_fields { 0 };
  148. KeywordListIndexType keywords { 0 };
  149. ListPatternListIndexType list_patterns { 0 };
  150. };
  151. struct LanguageMapping {
  152. CanonicalLanguageID<StringIndexType> key {};
  153. CanonicalLanguageID<StringIndexType> alias {};
  154. };
  155. struct UnicodeLocaleData {
  156. UniqueStringStorage<StringIndexType> unique_strings;
  157. UniqueStorage<DisplayPattern, DisplayPatternIndexType> unique_display_patterns;
  158. UniqueStorage<LanguageList, LanguageListIndexType> unique_language_lists;
  159. UniqueStorage<TerritoryList, TerritoryListIndexType> unique_territory_lists;
  160. UniqueStorage<ScriptList, ScriptListIndexType> unique_script_lists;
  161. UniqueStorage<CurrencyList, CurrencyListIndexType> unique_currency_lists;
  162. UniqueStorage<CalendarList, CalendarListIndexType> unique_calendar_lists;
  163. UniqueStorage<DateFieldList, DateFieldListIndexType> unique_date_field_lists;
  164. UniqueStorage<KeywordList, KeywordListIndexType> unique_keyword_lists;
  165. UniqueStorage<ListPatterns, ListPatternIndexType> unique_list_patterns;
  166. UniqueStorage<ListPatternList, ListPatternListIndexType> unique_list_pattern_lists;
  167. HashMap<String, Locale> locales;
  168. Vector<Alias> locale_aliases;
  169. Vector<String> languages;
  170. Vector<String> territories;
  171. Vector<String> scripts;
  172. Vector<String> variants;
  173. Vector<String> currencies;
  174. Vector<String> calendars;
  175. Vector<Alias> calendar_aliases {
  176. // FIXME: Aliases should come from BCP47. See: https://unicode-org.atlassian.net/browse/CLDR-15158
  177. { "ethiopic-amete-alem"sv, "ethioaa"sv },
  178. { "gregorian"sv, "gregory"sv },
  179. };
  180. Vector<String> date_fields;
  181. Vector<Alias> date_field_aliases {
  182. // ECMA-402 and the CLDR refer to some date fields with different names. Defining these aliases
  183. // means we can remain agnostic about the naming differences elsewhere.
  184. { "dayperiod"sv, "dayPeriod"sv },
  185. { "week"sv, "weekOfYear"sv },
  186. { "zone"sv, "timeZoneName"sv },
  187. };
  188. Vector<String> keywords { "ca"sv, "kf"sv, "kn"sv, "nu"sv }; // FIXME: These should be parsed from BCP47. https://unicode-org.atlassian.net/browse/CLDR-15158
  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_file = TRY(Core::File::open(core_aliases_path.string(), Core::OpenMode::ReadOnly));
  219. auto core_aliases = TRY(JsonValue::from_string(core_aliases_file->read_all()));
  220. auto const& supplemental_object = core_aliases.as_object().get("supplemental"sv);
  221. auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
  222. auto const& alias_object = metadata_object.as_object().get("alias"sv);
  223. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  224. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  225. auto alias = value.as_object().get("_replacement"sv).as_string();
  226. if (key.contains('-')) {
  227. auto mapping = TRY_OR_DISCARD(parse_language_mapping(locale_data, key, alias));
  228. locale_data.max_variant_size = max(mapping.key.variants.size(), locale_data.max_variant_size);
  229. locale_data.max_variant_size = max(mapping.alias.variants.size(), locale_data.max_variant_size);
  230. locale_data.complex_mappings.append(move(mapping));
  231. } else {
  232. alias_map.set(key, locale_data.unique_strings.ensure(alias));
  233. }
  234. });
  235. };
  236. append_aliases(alias_object.as_object().get("languageAlias"sv), locale_data.language_aliases);
  237. append_aliases(alias_object.as_object().get("territoryAlias"sv), locale_data.territory_aliases);
  238. append_aliases(alias_object.as_object().get("scriptAlias"sv), locale_data.script_aliases);
  239. append_aliases(alias_object.as_object().get("variantAlias"sv), locale_data.variant_aliases);
  240. append_aliases(alias_object.as_object().get("subdivisionAlias"sv), locale_data.subdivision_aliases);
  241. return {};
  242. }
  243. static ErrorOr<void> parse_likely_subtags(String core_supplemental_path, UnicodeLocaleData& locale_data)
  244. {
  245. LexicalPath likely_subtags_path(move(core_supplemental_path));
  246. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  247. auto likely_subtags_file = TRY(Core::File::open(likely_subtags_path.string(), Core::OpenMode::ReadOnly));
  248. auto likely_subtags = TRY(JsonValue::from_string(likely_subtags_file->read_all()));
  249. auto const& supplemental_object = likely_subtags.as_object().get("supplemental"sv);
  250. auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
  251. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  252. auto mapping = TRY_OR_DISCARD(parse_language_mapping(locale_data, key, value.as_string()));
  253. locale_data.max_variant_size = max(mapping.key.variants.size(), locale_data.max_variant_size);
  254. locale_data.max_variant_size = max(mapping.alias.variants.size(), locale_data.max_variant_size);
  255. locale_data.likely_subtags.append(move(mapping));
  256. });
  257. return {};
  258. }
  259. static ErrorOr<void> parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  260. {
  261. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  262. languages_path = languages_path.append("languages.json"sv);
  263. auto languages_file = TRY(Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly));
  264. auto languages = TRY(JsonValue::from_string(languages_file->read_all()));
  265. auto const& main_object = languages.as_object().get("main"sv);
  266. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  267. auto const& identity_object = locale_object.as_object().get("identity"sv);
  268. auto const& language_string = identity_object.as_object().get("language"sv);
  269. auto const& territory_string = identity_object.as_object().get("territory"sv);
  270. auto const& script_string = identity_object.as_object().get("script"sv);
  271. auto const& variant_string = identity_object.as_object().get("variant"sv);
  272. locale.language = language_string.as_string();
  273. if (territory_string.is_string()) {
  274. locale.territory = territory_string.as_string();
  275. if (!locale_data.territories.contains_slow(*locale.territory))
  276. locale_data.territories.append(*locale.territory);
  277. }
  278. if (script_string.is_string()) {
  279. auto script = script_string.as_string();
  280. if (!locale_data.scripts.contains_slow(script))
  281. locale_data.scripts.append(script);
  282. }
  283. if (variant_string.is_string()) {
  284. locale.variant = variant_string.as_string();
  285. if (!locale_data.variants.contains_slow(*locale.variant))
  286. locale_data.variants.append(*locale.variant);
  287. }
  288. return {};
  289. }
  290. static ErrorOr<void> parse_locale_display_patterns(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  291. {
  292. LexicalPath locale_display_names_path(move(locale_path));
  293. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  294. auto locale_display_names_file = TRY(Core::File::open(locale_display_names_path.string(), Core::OpenMode::ReadOnly));
  295. auto locale_display_names = TRY(JsonValue::from_string(locale_display_names_file->read_all()));
  296. auto const& main_object = locale_display_names.as_object().get("main"sv);
  297. auto const& locale_object = main_object.as_object().get(locale_display_names_path.parent().basename());
  298. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  299. auto const& locale_display_patterns_object = locale_display_names_object.as_object().get("localeDisplayPattern"sv);
  300. auto const& locale_pattern = locale_display_patterns_object.as_object().get("localePattern"sv);
  301. auto const& locale_separator = locale_display_patterns_object.as_object().get("localeSeparator"sv);
  302. DisplayPattern patterns {};
  303. patterns.locale_pattern = locale_data.unique_strings.ensure(locale_pattern.as_string());
  304. patterns.locale_separator = locale_data.unique_strings.ensure(locale_separator.as_string());
  305. locale.display_patterns = locale_data.unique_display_patterns.ensure(move(patterns));
  306. return {};
  307. }
  308. static ErrorOr<void> preprocess_languages(String locale_path, UnicodeLocaleData& locale_data)
  309. {
  310. LexicalPath languages_path(move(locale_path));
  311. languages_path = languages_path.append("languages.json"sv);
  312. auto languages_file = TRY(Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly));
  313. auto locale_languages = TRY(JsonValue::from_string(languages_file->read_all()));
  314. auto const& main_object = locale_languages.as_object().get("main"sv);
  315. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  316. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  317. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  318. languages_object.as_object().for_each_member([&](auto const& key, auto const&) {
  319. if (!key.contains("-alt-"sv) && !locale_data.languages.contains_slow(key))
  320. locale_data.languages.append(key);
  321. });
  322. return {};
  323. }
  324. static ErrorOr<void> parse_locale_languages(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  325. {
  326. LexicalPath languages_path(move(locale_path));
  327. languages_path = languages_path.append("languages.json"sv);
  328. auto languages_file = TRY(Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly));
  329. auto locale_languages = TRY(JsonValue::from_string(languages_file->read_all()));
  330. auto const& main_object = locale_languages.as_object().get("main"sv);
  331. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  332. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  333. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  334. LanguageList languages;
  335. languages.resize(locale_data.languages.size());
  336. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  337. if (key.contains("-alt-"sv))
  338. return;
  339. auto index = locale_data.languages.find_first_index(key).value();
  340. languages[index] = locale_data.unique_strings.ensure(value.as_string());
  341. });
  342. locale.languages = locale_data.unique_language_lists.ensure(move(languages));
  343. return {};
  344. }
  345. static ErrorOr<void> parse_locale_territories(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  346. {
  347. LexicalPath territories_path(move(locale_path));
  348. territories_path = territories_path.append("territories.json"sv);
  349. auto territories_file = TRY(Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly));
  350. auto locale_territories = TRY(JsonValue::from_string(territories_file->read_all()));
  351. auto const& main_object = locale_territories.as_object().get("main"sv);
  352. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  353. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  354. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  355. TerritoryList territories;
  356. territories.resize(locale_data.territories.size());
  357. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  358. if (auto index = locale_data.territories.find_first_index(key); index.has_value())
  359. territories[*index] = locale_data.unique_strings.ensure(value.as_string());
  360. });
  361. locale.territories = locale_data.unique_territory_lists.ensure(move(territories));
  362. return {};
  363. }
  364. static ErrorOr<void> parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  365. {
  366. LexicalPath scripts_path(move(locale_path));
  367. scripts_path = scripts_path.append("scripts.json"sv);
  368. auto scripts_file = TRY(Core::File::open(scripts_path.string(), Core::OpenMode::ReadOnly));
  369. auto locale_scripts = TRY(JsonValue::from_string(scripts_file->read_all()));
  370. auto const& main_object = locale_scripts.as_object().get("main"sv);
  371. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  372. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  373. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  374. ScriptList scripts;
  375. scripts.resize(locale_data.scripts.size());
  376. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  377. if (auto index = locale_data.scripts.find_first_index(key); index.has_value())
  378. scripts[*index] = locale_data.unique_strings.ensure(value.as_string());
  379. });
  380. locale.scripts = locale_data.unique_script_lists.ensure(move(scripts));
  381. return {};
  382. }
  383. static ErrorOr<void> parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
  384. {
  385. LexicalPath list_patterns_path(move(misc_path));
  386. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  387. auto list_patterns_file = TRY(Core::File::open(list_patterns_path.string(), Core::OpenMode::ReadOnly));
  388. auto locale_list_patterns = TRY(JsonValue::from_string(list_patterns_file->read_all()));
  389. auto const& main_object = locale_list_patterns.as_object().get("main"sv);
  390. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  391. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  392. auto list_pattern_type = [](StringView key) {
  393. if (key.contains("type-standard"sv))
  394. return "conjunction"sv;
  395. if (key.contains("type-or"sv))
  396. return "disjunction"sv;
  397. if (key.contains("type-unit"sv))
  398. return "unit"sv;
  399. VERIFY_NOT_REACHED();
  400. };
  401. auto list_pattern_style = [](StringView key) {
  402. if (key.contains("short"sv))
  403. return "short"sv;
  404. if (key.contains("narrow"sv))
  405. return "narrow"sv;
  406. return "long"sv;
  407. };
  408. ListPatternList list_patterns;
  409. list_patterns.ensure_capacity(list_patterns_object.as_object().size());
  410. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  411. auto type = list_pattern_type(key);
  412. auto style = list_pattern_style(key);
  413. auto start = locale_data.unique_strings.ensure(value.as_object().get("start"sv).as_string());
  414. auto middle = locale_data.unique_strings.ensure(value.as_object().get("middle"sv).as_string());
  415. auto end = locale_data.unique_strings.ensure(value.as_object().get("end"sv).as_string());
  416. auto pair = locale_data.unique_strings.ensure(value.as_object().get("2"sv).as_string());
  417. if (!locale_data.list_pattern_types.contains_slow(type))
  418. locale_data.list_pattern_types.append(type);
  419. ListPatterns list_pattern { type, style, start, middle, end, pair };
  420. list_patterns.append(locale_data.unique_list_patterns.ensure(move(list_pattern)));
  421. });
  422. locale.list_patterns = locale_data.unique_list_pattern_lists.ensure(move(list_patterns));
  423. return {};
  424. }
  425. static ErrorOr<void> parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  426. {
  427. LexicalPath currencies_path(move(numbers_path));
  428. currencies_path = currencies_path.append("currencies.json"sv);
  429. auto currencies_file = TRY(Core::File::open(currencies_path.string(), Core::OpenMode::ReadOnly));
  430. auto locale_currencies = TRY(JsonValue::from_string(currencies_file->read_all()));
  431. auto const& main_object = locale_currencies.as_object().get("main"sv);
  432. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  433. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  434. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  435. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const&) {
  436. if (!locale_data.currencies.contains_slow(key))
  437. locale_data.currencies.append(key);
  438. });
  439. CurrencyList long_currencies {};
  440. long_currencies.resize(locale_data.currencies.size());
  441. CurrencyList short_currencies {};
  442. short_currencies.resize(locale_data.currencies.size());
  443. CurrencyList narrow_currencies {};
  444. narrow_currencies.resize(locale_data.currencies.size());
  445. CurrencyList numeric_currencies {};
  446. numeric_currencies.resize(locale_data.currencies.size());
  447. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  448. auto const& long_name = value.as_object().get("displayName"sv);
  449. auto const& short_name = value.as_object().get("symbol"sv);
  450. auto const& narrow_name = value.as_object().get("symbol-alt-narrow"sv);
  451. auto const& numeric_name = value.as_object().get("displayName-count-other"sv);
  452. auto index = locale_data.currencies.find_first_index(key).value();
  453. long_currencies[index] = locale_data.unique_strings.ensure(long_name.as_string());
  454. short_currencies[index] = locale_data.unique_strings.ensure(short_name.as_string());
  455. narrow_currencies[index] = narrow_name.is_null() ? 0 : locale_data.unique_strings.ensure(narrow_name.as_string());
  456. numeric_currencies[index] = locale_data.unique_strings.ensure(numeric_name.is_null() ? long_name.as_string() : numeric_name.as_string());
  457. });
  458. locale.long_currencies = locale_data.unique_currency_lists.ensure(move(long_currencies));
  459. locale.short_currencies = locale_data.unique_currency_lists.ensure(move(short_currencies));
  460. locale.narrow_currencies = locale_data.unique_currency_lists.ensure(move(narrow_currencies));
  461. locale.numeric_currencies = locale_data.unique_currency_lists.ensure(move(numeric_currencies));
  462. return {};
  463. }
  464. static ErrorOr<void> parse_locale_calendars(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  465. {
  466. LexicalPath locale_display_names_path(move(locale_path));
  467. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  468. auto locale_display_names_file = TRY(Core::File::open(locale_display_names_path.string(), Core::OpenMode::ReadOnly));
  469. auto locale_display_names = TRY(JsonValue::from_string(locale_display_names_file->read_all()));
  470. auto const& main_object = locale_display_names.as_object().get("main"sv);
  471. auto const& locale_object = main_object.as_object().get(locale_display_names_path.parent().basename());
  472. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  473. auto const& types_object = locale_display_names_object.as_object().get("types"sv);
  474. auto const& calendar_object = types_object.as_object().get("calendar"sv);
  475. calendar_object.as_object().for_each_member([&](auto const& key, auto const&) {
  476. if (!locale_data.calendars.contains_slow(key))
  477. locale_data.calendars.append(key);
  478. });
  479. CalendarList calendars;
  480. calendars.resize(locale_data.calendars.size());
  481. calendar_object.as_object().for_each_member([&](auto const& key, auto const& calendar) {
  482. auto index = locale_data.calendars.find_first_index(key).value();
  483. calendars[index] = locale_data.unique_strings.ensure(calendar.as_string());
  484. });
  485. locale.calendars = locale_data.unique_calendar_lists.ensure(move(calendars));
  486. return {};
  487. }
  488. static ErrorOr<void> parse_locale_date_fields(String dates_path, UnicodeLocaleData& locale_data, Locale& locale)
  489. {
  490. LexicalPath date_fields_path(move(dates_path));
  491. date_fields_path = date_fields_path.append("dateFields.json"sv);
  492. auto date_fields_file = TRY(Core::File::open(date_fields_path.string(), Core::OpenMode::ReadOnly));
  493. auto locale_date_fields = TRY(JsonValue::from_string(date_fields_file->read_all()));
  494. auto const& main_object = locale_date_fields.as_object().get("main"sv);
  495. auto const& locale_object = main_object.as_object().get(date_fields_path.parent().basename());
  496. auto const& dates_object = locale_object.as_object().get("dates"sv);
  497. auto const& fields_object = dates_object.as_object().get("fields"sv);
  498. auto is_sanctioned_field = [](StringView field) {
  499. // This is a copy of the units sanctioned for use within ECMA-402, with names adjusted for the names used by the CLDR.
  500. // https://tc39.es/ecma402/#table-validcodeforDateField
  501. 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);
  502. };
  503. fields_object.as_object().for_each_member([&](auto const& key, JsonValue const&) {
  504. if (!is_sanctioned_field(key))
  505. return;
  506. if (!locale_data.date_fields.contains_slow(key))
  507. locale_data.date_fields.append(key);
  508. });
  509. quick_sort(locale_data.date_fields);
  510. DateFieldList long_date_fields {};
  511. long_date_fields.resize(locale_data.date_fields.size());
  512. DateFieldList short_date_fields {};
  513. short_date_fields.resize(locale_data.date_fields.size());
  514. DateFieldList narrow_date_fields {};
  515. narrow_date_fields.resize(locale_data.date_fields.size());
  516. fields_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  517. if (!is_sanctioned_field(key))
  518. return;
  519. auto const& long_name = value.as_object().get("displayName"sv);
  520. auto const& short_name = fields_object.as_object().get(String::formatted("{}-short", key)).as_object().get("displayName"sv);
  521. auto const& narrow_name = fields_object.as_object().get(String::formatted("{}-narrow", key)).as_object().get("displayName"sv);
  522. auto index = locale_data.date_fields.find_first_index(key).value();
  523. long_date_fields[index] = locale_data.unique_strings.ensure(long_name.as_string());
  524. short_date_fields[index] = locale_data.unique_strings.ensure(short_name.as_string());
  525. narrow_date_fields[index] = locale_data.unique_strings.ensure(narrow_name.as_string());
  526. });
  527. locale.long_date_fields = locale_data.unique_date_field_lists.ensure(move(long_date_fields));
  528. locale.short_date_fields = locale_data.unique_date_field_lists.ensure(move(short_date_fields));
  529. locale.narrow_date_fields = locale_data.unique_date_field_lists.ensure(move(narrow_date_fields));
  530. return {};
  531. }
  532. static ErrorOr<void> parse_numeric_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, KeywordList& keywords)
  533. {
  534. static constexpr StringView key = "nu"sv;
  535. LexicalPath numbers_path(move(locale_numbers_path));
  536. numbers_path = numbers_path.append("numbers.json"sv);
  537. auto numbers_file = TRY(Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly));
  538. auto numbers = TRY(JsonValue::from_string(numbers_file->read_all()));
  539. auto const& main_object = numbers.as_object().get("main"sv);
  540. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  541. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  542. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  543. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  544. Vector<String> keyword_values {};
  545. keyword_values.append(default_numbering_system_object.as_string());
  546. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  547. auto keyword_value = value.as_string();
  548. if (!keyword_values.contains_slow(keyword_value))
  549. keyword_values.append(move(keyword_value));
  550. });
  551. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  552. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  553. return;
  554. auto keyword_value = value.as_string();
  555. if (!keyword_values.contains_slow(keyword_value))
  556. keyword_values.append(move(keyword_value));
  557. });
  558. StringBuilder builder;
  559. builder.join(',', keyword_values);
  560. auto index = locale_data.keywords.find_first_index(key).value();
  561. keywords[index] = locale_data.unique_strings.ensure(builder.build());
  562. return {};
  563. }
  564. static ErrorOr<void> parse_calendar_keywords(String locale_dates_path, UnicodeLocaleData& locale_data, KeywordList& keywords)
  565. {
  566. static constexpr StringView key = "ca"sv;
  567. auto calendars_iterator = TRY(path_to_dir_iterator(locale_dates_path, {}));
  568. Vector<String> keyword_values {};
  569. while (calendars_iterator.has_next()) {
  570. auto locale_calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  571. LexicalPath calendars_path(move(locale_calendars_path));
  572. if (!calendars_path.basename().starts_with("ca-"sv))
  573. continue;
  574. auto calendars_file = TRY(Core::File::open(calendars_path.string(), Core::OpenMode::ReadOnly));
  575. auto calendars = TRY(JsonValue::from_string(calendars_file->read_all()));
  576. auto const& main_object = calendars.as_object().get("main"sv);
  577. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  578. auto const& dates_object = locale_object.as_object().get("dates"sv);
  579. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  580. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const&) {
  581. // The generic calendar is not a supported Unicode calendar key, so skip it:
  582. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  583. if (calendar_name == "generic"sv)
  584. return;
  585. // FIXME: Similar to the calendar aliases defined in GenerateUnicodeDateTimeFormat, this
  586. // should be parsed from BCP47. https://unicode-org.atlassian.net/browse/CLDR-15158
  587. if (calendar_name == "gregorian"sv)
  588. keyword_values.append("gregory"sv);
  589. else
  590. keyword_values.append(calendar_name);
  591. });
  592. }
  593. StringBuilder builder;
  594. builder.join(',', keyword_values);
  595. auto index = locale_data.keywords.find_first_index(key).value();
  596. keywords[index] = locale_data.unique_strings.ensure(builder.build());
  597. return {};
  598. }
  599. static void fill_in_bcp47_keywords(UnicodeLocaleData& locale_data, KeywordList& keywords)
  600. {
  601. // FIXME: These should be parsed from BCP47. They are only available in this XML file:
  602. // https://github.com/unicode-org/cldr/blob/main/common/bcp47/collation.xml
  603. auto kf_index = locale_data.keywords.find_first_index("kf"sv).value();
  604. keywords[kf_index] = locale_data.unique_strings.ensure("upper,lower,false"sv);
  605. auto kn_index = locale_data.keywords.find_first_index("kn"sv).value();
  606. keywords[kn_index] = locale_data.unique_strings.ensure("true,false"sv);
  607. }
  608. static ErrorOr<void> parse_default_content_locales(String core_path, UnicodeLocaleData& locale_data)
  609. {
  610. LexicalPath default_content_path(move(core_path));
  611. default_content_path = default_content_path.append("defaultContent.json"sv);
  612. auto default_content_file = TRY(Core::File::open(default_content_path.string(), Core::OpenMode::ReadOnly));
  613. auto default_content = TRY(JsonValue::from_string(default_content_file->read_all()));
  614. auto const& default_content_array = default_content.as_object().get("defaultContent"sv);
  615. default_content_array.as_array().for_each([&](JsonValue const& value) {
  616. auto locale = value.as_string();
  617. StringView default_locale = locale;
  618. while (true) {
  619. if (locale_data.locales.contains(default_locale))
  620. break;
  621. auto pos = default_locale.find_last('-');
  622. if (!pos.has_value())
  623. return;
  624. default_locale = default_locale.substring_view(0, *pos);
  625. }
  626. if (default_locale != locale)
  627. locale_data.locale_aliases.append({ default_locale, move(locale) });
  628. });
  629. return {};
  630. }
  631. static ErrorOr<void> define_aliases_without_scripts(UnicodeLocaleData& locale_data)
  632. {
  633. // From ECMA-402: https://tc39.es/ecma402/#sec-internal-slots
  634. //
  635. // For locales that include a script subtag in addition to language and region, the
  636. // corresponding locale without a script subtag must also be supported.
  637. //
  638. // So we define aliases for locales that contain all three subtags, but we must also take
  639. // care to handle when the locale itself or the locale without a script subtag are an alias
  640. // by way of default-content locales.
  641. auto find_alias = [&](auto const& locale) {
  642. return locale_data.locale_aliases.find_if([&](auto const& alias) { return locale == alias.alias; });
  643. };
  644. auto append_alias_without_script = [&](auto const& locale) -> ErrorOr<void> {
  645. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, locale));
  646. if ((parsed_locale.language == 0) || (parsed_locale.script == 0) || (parsed_locale.region == 0))
  647. return {};
  648. auto locale_without_script = String::formatted("{}-{}",
  649. locale_data.unique_strings.get(parsed_locale.language),
  650. locale_data.unique_strings.get(parsed_locale.region));
  651. if (locale_data.locales.contains(locale_without_script))
  652. return {};
  653. if (find_alias(locale_without_script) != locale_data.locale_aliases.end())
  654. return {};
  655. if (auto it = find_alias(locale); it != locale_data.locale_aliases.end())
  656. locale_data.locale_aliases.append({ it->name, locale_without_script });
  657. else
  658. locale_data.locale_aliases.append({ locale, locale_without_script });
  659. return {};
  660. };
  661. for (auto const& locale : locale_data.locales)
  662. TRY(append_alias_without_script(locale.key));
  663. for (auto const& locale : locale_data.locale_aliases)
  664. TRY(append_alias_without_script(locale.alias));
  665. return {};
  666. }
  667. static ErrorOr<void> parse_all_locales(String core_path, String locale_names_path, String misc_path, String numbers_path, String dates_path, UnicodeLocaleData& locale_data)
  668. {
  669. auto identity_iterator = TRY(path_to_dir_iterator(locale_names_path));
  670. auto preprocess_iterator = TRY(path_to_dir_iterator(locale_names_path));
  671. auto locale_names_iterator = TRY(path_to_dir_iterator(move(locale_names_path)));
  672. auto misc_iterator = TRY(path_to_dir_iterator(move(misc_path)));
  673. auto numbers_iterator = TRY(path_to_dir_iterator(move(numbers_path)));
  674. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  675. LexicalPath core_supplemental_path(core_path);
  676. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  677. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  678. TRY(parse_core_aliases(core_supplemental_path.string(), locale_data));
  679. TRY(parse_likely_subtags(core_supplemental_path.string(), locale_data));
  680. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  681. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  682. StringBuilder builder;
  683. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  684. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  685. builder.appendff("-{}", script);
  686. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  687. builder.appendff("-{}", region);
  688. return builder.build();
  689. };
  690. while (identity_iterator.has_next()) {
  691. auto locale_path = TRY(next_path_from_dir_iterator(identity_iterator));
  692. auto language = TRY(remove_variants_from_path(locale_path));
  693. auto& locale = locale_data.locales.ensure(language);
  694. TRY(parse_identity(locale_path, locale_data, locale));
  695. }
  696. while (preprocess_iterator.has_next()) {
  697. auto locale_path = TRY(next_path_from_dir_iterator(preprocess_iterator));
  698. TRY(preprocess_languages(locale_path, locale_data));
  699. }
  700. quick_sort(locale_data.languages);
  701. quick_sort(locale_data.territories);
  702. quick_sort(locale_data.scripts);
  703. HashMap<String, KeywordList> keywords;
  704. auto ensure_keyword_list = [&](auto const& language) -> KeywordList& {
  705. return keywords.ensure(language, [&]() {
  706. KeywordList keywords;
  707. keywords.resize(locale_data.keywords.size());
  708. return keywords;
  709. });
  710. };
  711. while (locale_names_iterator.has_next()) {
  712. auto locale_path = TRY(next_path_from_dir_iterator(locale_names_iterator));
  713. auto language = TRY(remove_variants_from_path(locale_path));
  714. auto& locale = locale_data.locales.ensure(language);
  715. TRY(parse_locale_display_patterns(locale_path, locale_data, locale));
  716. TRY(parse_locale_languages(locale_path, locale_data, locale));
  717. TRY(parse_locale_territories(locale_path, locale_data, locale));
  718. TRY(parse_locale_scripts(locale_path, locale_data, locale));
  719. TRY(parse_locale_calendars(locale_path, locale_data, locale));
  720. }
  721. while (misc_iterator.has_next()) {
  722. auto misc_path = TRY(next_path_from_dir_iterator(misc_iterator));
  723. auto language = TRY(remove_variants_from_path(misc_path));
  724. auto& locale = locale_data.locales.ensure(language);
  725. TRY(parse_locale_list_patterns(misc_path, locale_data, locale));
  726. }
  727. while (numbers_iterator.has_next()) {
  728. auto numbers_path = TRY(next_path_from_dir_iterator(numbers_iterator));
  729. auto language = TRY(remove_variants_from_path(numbers_path));
  730. auto& locale = locale_data.locales.ensure(language);
  731. TRY(parse_locale_currencies(numbers_path, locale_data, locale));
  732. auto& keywords = ensure_keyword_list(language);
  733. TRY(parse_numeric_keywords(numbers_path, locale_data, keywords));
  734. fill_in_bcp47_keywords(locale_data, keywords);
  735. }
  736. while (dates_iterator.has_next()) {
  737. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  738. auto language = TRY(remove_variants_from_path(dates_path));
  739. auto& locale = locale_data.locales.ensure(language);
  740. TRY(parse_locale_date_fields(dates_path, locale_data, locale));
  741. auto& keywords = ensure_keyword_list(language);
  742. TRY(parse_calendar_keywords(dates_path, locale_data, keywords));
  743. }
  744. TRY(parse_default_content_locales(move(core_path), locale_data));
  745. TRY(define_aliases_without_scripts(locale_data));
  746. for (auto& list : keywords) {
  747. auto& locale = locale_data.locales.find(list.key)->value;
  748. locale.keywords = locale_data.unique_keyword_lists.ensure(move(list.value));
  749. }
  750. return {};
  751. }
  752. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  753. {
  754. StringBuilder builder;
  755. SourceGenerator generator { builder };
  756. generator.append(R"~~~(
  757. #pragma once
  758. #include <AK/Types.h>
  759. namespace Unicode {
  760. )~~~");
  761. auto locales = locale_data.locales.keys();
  762. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, locale_data.locale_aliases);
  763. generate_enum(generator, format_identifier, "Language"sv, {}, locale_data.languages);
  764. generate_enum(generator, format_identifier, "Territory"sv, {}, locale_data.territories);
  765. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, locale_data.scripts);
  766. generate_enum(generator, format_identifier, "Currency"sv, {}, locale_data.currencies);
  767. generate_enum(generator, format_identifier, "CalendarName"sv, {}, locale_data.calendars, locale_data.calendar_aliases);
  768. generate_enum(generator, format_identifier, "DateField"sv, {}, locale_data.date_fields, locale_data.date_field_aliases);
  769. generate_enum(generator, format_identifier, "Key"sv, {}, locale_data.keywords);
  770. generate_enum(generator, format_identifier, "Variant"sv, {}, locale_data.variants);
  771. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, locale_data.list_pattern_types);
  772. generator.append(R"~~~(
  773. }
  774. )~~~");
  775. VERIFY(file.write(generator.as_string_view()));
  776. }
  777. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  778. {
  779. StringBuilder builder;
  780. SourceGenerator generator { builder };
  781. generator.set("string_index_type"sv, s_string_index_type);
  782. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  783. generator.set("territories_size", String::number(locale_data.territories.size()));
  784. generator.set("variants_size", String::number(locale_data.max_variant_size));
  785. generator.append(R"~~~(
  786. #include <AK/Array.h>
  787. #include <AK/BinarySearch.h>
  788. #include <AK/Optional.h>
  789. #include <AK/Span.h>
  790. #include <AK/StringView.h>
  791. #include <AK/Vector.h>
  792. #include <LibUnicode/CurrencyCode.h>
  793. #include <LibUnicode/Locale.h>
  794. #include <LibUnicode/UnicodeLocale.h>
  795. namespace Unicode {
  796. )~~~");
  797. locale_data.unique_strings.generate(generator);
  798. generator.append(R"~~~(
  799. struct DisplayPatternImpl {
  800. DisplayPattern to_display_pattern() const
  801. {
  802. DisplayPattern display_patterns {};
  803. display_patterns.locale_pattern = s_string_list[locale_pattern];
  804. display_patterns.locale_separator = s_string_list[locale_separator];
  805. return display_patterns;
  806. }
  807. @string_index_type@ locale_pattern { 0 };
  808. @string_index_type@ locale_separator { 0 };
  809. };
  810. struct Patterns {
  811. ListPatternType type;
  812. Style style;
  813. @string_index_type@ start { 0 };
  814. @string_index_type@ middle { 0 };
  815. @string_index_type@ end { 0 };
  816. @string_index_type@ pair { 0 };
  817. };
  818. )~~~");
  819. generate_available_values(generator, "get_available_currencies"sv, locale_data.currencies);
  820. locale_data.unique_display_patterns.generate(generator, "DisplayPatternImpl"sv, "s_display_patterns"sv, 30);
  821. locale_data.unique_language_lists.generate(generator, s_string_index_type, "s_language_lists"sv);
  822. locale_data.unique_territory_lists.generate(generator, s_string_index_type, "s_territory_lists"sv);
  823. locale_data.unique_script_lists.generate(generator, s_string_index_type, "s_script_lists"sv);
  824. locale_data.unique_currency_lists.generate(generator, s_string_index_type, "s_currency_lists"sv);
  825. locale_data.unique_calendar_lists.generate(generator, s_string_index_type, "s_calendar_lists"sv);
  826. locale_data.unique_date_field_lists.generate(generator, s_string_index_type, "s_date_field_lists"sv);
  827. locale_data.unique_keyword_lists.generate(generator, s_string_index_type, "s_keyword_lists"sv);
  828. locale_data.unique_list_patterns.generate(generator, "Patterns"sv, "s_list_patterns"sv, 10);
  829. locale_data.unique_list_pattern_lists.generate(generator, s_list_pattern_index_type, "s_list_pattern_lists"sv);
  830. auto append_index = [&](auto index) {
  831. generator.append(String::formatted(", {}", index));
  832. };
  833. auto append_list_and_size = [&](auto const& list) {
  834. if (list.is_empty()) {
  835. generator.append(", {}, 0");
  836. return;
  837. }
  838. bool first = true;
  839. generator.append(", {");
  840. for (auto const& item : list) {
  841. generator.append(first ? " " : ", ");
  842. generator.append(String::number(item));
  843. first = false;
  844. }
  845. generator.append(String::formatted(" }}, {}", list.size()));
  846. };
  847. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  848. generator.set("type", type);
  849. generator.set("name", name);
  850. generator.set("size", String::number(keys.size()));
  851. generator.append(R"~~~(
  852. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  853. bool first = true;
  854. for (auto const& key : keys) {
  855. auto const& value = map.find(key)->value;
  856. auto mapping = mapping_getter(value);
  857. generator.append(first ? " " : ", ");
  858. generator.append(String::number(mapping));
  859. first = false;
  860. }
  861. generator.append(" } };");
  862. };
  863. auto locales = locale_data.locales.keys();
  864. quick_sort(locales);
  865. append_mapping(locales, locale_data.locales, s_display_pattern_index_type, "s_locale_display_patterns"sv, [&](auto const& locale) { return locale.display_patterns; });
  866. append_mapping(locales, locale_data.locales, s_language_list_index_type, "s_languages"sv, [&](auto const& locale) { return locale.languages; });
  867. append_mapping(locales, locale_data.locales, s_territory_list_index_type, "s_territories"sv, [&](auto const& locale) { return locale.territories; });
  868. append_mapping(locales, locale_data.locales, s_script_list_index_type, "s_scripts"sv, [&](auto const& locale) { return locale.scripts; });
  869. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_long_currencies"sv, [&](auto const& locale) { return locale.long_currencies; });
  870. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_short_currencies"sv, [&](auto const& locale) { return locale.short_currencies; });
  871. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_narrow_currencies"sv, [&](auto const& locale) { return locale.narrow_currencies; });
  872. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_numeric_currencies"sv, [&](auto const& locale) { return locale.numeric_currencies; });
  873. append_mapping(locales, locale_data.locales, s_calendar_list_index_type, "s_calendars"sv, [&](auto const& locale) { return locale.calendars; });
  874. 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; });
  875. 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; });
  876. 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; });
  877. append_mapping(locales, locale_data.locales, s_keyword_list_index_type, "s_keywords"sv, [&](auto const& locale) { return locale.keywords; });
  878. append_mapping(locales, locale_data.locales, s_list_pattern_list_index_type, "s_locale_list_patterns"sv, [&](auto const& locale) { return locale.list_patterns; });
  879. generator.append(R"~~~(
  880. struct CanonicalLanguageID {
  881. LanguageID to_unicode_language_id() const
  882. {
  883. LanguageID language_id {};
  884. language_id.variants.ensure_capacity(variants_size);
  885. language_id.language = s_string_list[language];
  886. if (script != 0)
  887. language_id.script = s_string_list[script];
  888. if (region != 0)
  889. language_id.region = s_string_list[region];
  890. for (size_t i = 0; i < variants_size; ++i)
  891. language_id.variants.append(s_string_list[variants[i]]);
  892. return language_id;
  893. }
  894. bool matches_variants(Vector<String> const& other_variants) const {
  895. if (variants_size == 0)
  896. return true;
  897. if (other_variants.size() != variants_size)
  898. return false;
  899. for (size_t i = 0; i < variants_size; ++i) {
  900. if (s_string_list[variants[i]] != other_variants[i])
  901. return false;
  902. }
  903. return true;
  904. };
  905. @string_index_type@ language { 0 };
  906. @string_index_type@ script { 0 };
  907. @string_index_type@ region { 0 };
  908. Array<@string_index_type@, @variants_size@> variants {};
  909. size_t variants_size { 0 };
  910. };
  911. struct LanguageMapping {
  912. CanonicalLanguageID key;
  913. CanonicalLanguageID alias;
  914. };
  915. )~~~");
  916. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  917. generator.set("size", String::number(mappings.size()));
  918. generator.set("name"sv, name);
  919. generator.append(R"~~~(
  920. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  921. )~~~");
  922. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  923. auto const& lhs_language = locale_data.unique_strings.get(lhs.key.language);
  924. auto const& rhs_language = locale_data.unique_strings.get(rhs.key.language);
  925. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  926. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  927. return false;
  928. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  929. return true;
  930. return lhs_language < rhs_language;
  931. });
  932. for (auto const& mapping : mappings) {
  933. generator.set("language"sv, String::number(mapping.key.language));
  934. generator.append(" { { @language@");
  935. append_index(mapping.key.script);
  936. append_index(mapping.key.region);
  937. append_list_and_size(mapping.key.variants);
  938. generator.set("language"sv, String::number(mapping.alias.language));
  939. generator.append(" }, { @language@");
  940. append_index(mapping.alias.script);
  941. append_index(mapping.alias.region);
  942. append_list_and_size(mapping.alias.variants);
  943. generator.append(" } },\n");
  944. }
  945. generator.append("} };\n");
  946. };
  947. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  948. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  949. generator.append(R"~~~(
  950. static LanguageMapping const* resolve_likely_subtag(LanguageID const& language_id)
  951. {
  952. // https://unicode.org/reports/tr35/#Likely_Subtags
  953. enum class State {
  954. LanguageScriptRegion,
  955. LanguageRegion,
  956. LanguageScript,
  957. Language,
  958. UndScript,
  959. Done,
  960. };
  961. auto state = State::LanguageScriptRegion;
  962. while (state != State::Done) {
  963. LanguageID search_key;
  964. switch (state) {
  965. case State::LanguageScriptRegion:
  966. state = State::LanguageRegion;
  967. if (!language_id.script.has_value() || !language_id.region.has_value())
  968. continue;
  969. search_key.language = *language_id.language;
  970. search_key.script = *language_id.script;
  971. search_key.region = *language_id.region;
  972. break;
  973. case State::LanguageRegion:
  974. state = State::LanguageScript;
  975. if (!language_id.region.has_value())
  976. continue;
  977. search_key.language = *language_id.language;
  978. search_key.region = *language_id.region;
  979. break;
  980. case State::LanguageScript:
  981. state = State::Language;
  982. if (!language_id.script.has_value())
  983. continue;
  984. search_key.language = *language_id.language;
  985. search_key.script = *language_id.script;
  986. break;
  987. case State::Language:
  988. state = State::UndScript;
  989. search_key.language = *language_id.language;
  990. break;
  991. case State::UndScript:
  992. state = State::Done;
  993. if (!language_id.script.has_value())
  994. continue;
  995. search_key.language = "und"sv;
  996. search_key.script = *language_id.script;
  997. break;
  998. default:
  999. VERIFY_NOT_REACHED();
  1000. }
  1001. for (auto const& map : s_likely_subtags) {
  1002. auto const& key_language = s_string_list[map.key.language];
  1003. auto const& key_script = s_string_list[map.key.script];
  1004. auto const& key_region = s_string_list[map.key.region];
  1005. if (key_language != search_key.language)
  1006. continue;
  1007. if (!key_script.is_empty() || search_key.script.has_value()) {
  1008. if (key_script != search_key.script)
  1009. continue;
  1010. }
  1011. if (!key_region.is_empty() || search_key.region.has_value()) {
  1012. if (key_region != search_key.region)
  1013. continue;
  1014. }
  1015. return &map;
  1016. }
  1017. }
  1018. return nullptr;
  1019. }
  1020. )~~~");
  1021. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name, StringView unique_list) {
  1022. generator.set("enum_snake", enum_snake);
  1023. generator.set("from_string_name", from_string_name);
  1024. generator.set("collection_name", collection_name);
  1025. generator.set("unique_list", unique_list);
  1026. generator.append(R"~~~(
  1027. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  1028. {
  1029. auto locale_value = locale_from_string(locale);
  1030. if (!locale_value.has_value())
  1031. return {};
  1032. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  1033. if (!@enum_snake@_value.has_value())
  1034. return {};
  1035. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1036. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  1037. auto mapping_index = @collection_name@.at(locale_index);
  1038. auto const& mappings = @unique_list@.at(mapping_index);
  1039. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  1040. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  1041. if (@enum_snake@_mapping.is_empty())
  1042. return {};
  1043. return @enum_snake@_mapping;
  1044. }
  1045. )~~~");
  1046. };
  1047. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1048. HashValueMap<String> hashes;
  1049. hashes.ensure_capacity(values.size());
  1050. for (auto const& value : values)
  1051. hashes.set(value.hash(), format_identifier(enum_title, value));
  1052. for (auto const& alias : aliases)
  1053. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1054. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1055. };
  1056. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  1057. HashValueMap<StringIndexType> hashes;
  1058. hashes.ensure_capacity(aliases.size());
  1059. for (auto const& alias : aliases)
  1060. hashes.set(alias.key.hash(), alias.value);
  1061. ValueFromStringOptions options {};
  1062. options.return_type = "StringView"sv;
  1063. options.return_format = "s_string_list[{}]"sv;
  1064. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), options);
  1065. };
  1066. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys(), locale_data.locale_aliases);
  1067. append_from_string("Language"sv, "language"sv, locale_data.languages);
  1068. append_mapping_search("language"sv, "language"sv, "s_languages"sv, "s_language_lists"sv);
  1069. append_alias_search("language"sv, locale_data.language_aliases);
  1070. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  1071. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv, "s_territory_lists"sv);
  1072. append_alias_search("territory"sv, locale_data.territory_aliases);
  1073. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  1074. append_mapping_search("script"sv, "script_tag"sv, "s_scripts"sv, "s_script_lists"sv);
  1075. append_alias_search("script_tag"sv, locale_data.script_aliases);
  1076. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  1077. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv, "s_currency_lists"sv);
  1078. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv, "s_currency_lists"sv);
  1079. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv, "s_currency_lists"sv);
  1080. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv, "s_currency_lists"sv);
  1081. append_from_string("CalendarName"sv, "calendar_name"sv, locale_data.calendars, locale_data.calendar_aliases);
  1082. append_mapping_search("calendar"sv, "calendar_name"sv, "s_calendars"sv, "s_calendar_lists"sv);
  1083. append_from_string("DateField"sv, "date_field"sv, locale_data.date_fields, locale_data.date_field_aliases);
  1084. append_mapping_search("long_date_field"sv, "date_field"sv, "s_long_date_fields"sv, "s_date_field_lists"sv);
  1085. append_mapping_search("short_date_field"sv, "date_field"sv, "s_short_date_fields"sv, "s_date_field_lists"sv);
  1086. append_mapping_search("narrow_date_field"sv, "date_field"sv, "s_narrow_date_fields"sv, "s_date_field_lists"sv);
  1087. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  1088. append_mapping_search("key"sv, "key"sv, "s_keywords"sv, "s_keyword_lists"sv);
  1089. append_alias_search("variant"sv, locale_data.variant_aliases);
  1090. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  1091. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  1092. generator.append(R"~~~(
  1093. Optional<DisplayPattern> get_locale_display_patterns(StringView locale)
  1094. {
  1095. auto locale_value = locale_from_string(locale);
  1096. if (!locale_value.has_value())
  1097. return {};
  1098. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1099. auto display_patterns_index = s_locale_display_patterns.at(locale_index);
  1100. auto const& display_patterns = s_display_patterns.at(display_patterns_index);
  1101. return display_patterns.to_display_pattern();
  1102. }
  1103. Optional<ListPatterns> get_locale_list_patterns(StringView locale, StringView list_pattern_type, Style list_pattern_style)
  1104. {
  1105. auto locale_value = locale_from_string(locale);
  1106. if (!locale_value.has_value())
  1107. return {};
  1108. auto type_value = list_pattern_type_from_string(list_pattern_type);
  1109. if (!type_value.has_value())
  1110. return {};
  1111. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1112. auto list_patterns_list_index = s_locale_list_patterns.at(locale_index);
  1113. auto const& locale_list_patterns = s_list_pattern_lists.at(list_patterns_list_index);
  1114. for (auto list_patterns_index : locale_list_patterns) {
  1115. auto const& list_patterns = s_list_patterns.at(list_patterns_index);
  1116. if ((list_patterns.type == type_value) && (list_patterns.style == list_pattern_style)) {
  1117. auto const& start = s_string_list[list_patterns.start];
  1118. auto const& middle = s_string_list[list_patterns.middle];
  1119. auto const& end = s_string_list[list_patterns.end];
  1120. auto const& pair = s_string_list[list_patterns.pair];
  1121. return ListPatterns { start, middle, end, pair };
  1122. }
  1123. }
  1124. return {};
  1125. }
  1126. void resolve_complex_language_aliases(LanguageID& language_id)
  1127. {
  1128. for (auto const& map : s_complex_alias) {
  1129. auto const& key_language = s_string_list[map.key.language];
  1130. auto const& key_script = s_string_list[map.key.script];
  1131. auto const& key_region = s_string_list[map.key.region];
  1132. if ((key_language != language_id.language) && (key_language != "und"sv))
  1133. continue;
  1134. if (!key_script.is_empty() && (key_script != language_id.script))
  1135. continue;
  1136. if (!key_region.is_empty() && (key_region != language_id.region))
  1137. continue;
  1138. if (!map.key.matches_variants(language_id.variants))
  1139. continue;
  1140. auto alias = map.alias.to_unicode_language_id();
  1141. if (alias.language == "und"sv)
  1142. alias.language = move(language_id.language);
  1143. if (key_script.is_empty() && !alias.script.has_value())
  1144. alias.script = move(language_id.script);
  1145. if (key_region.is_empty() && !alias.region.has_value())
  1146. alias.region = move(language_id.region);
  1147. if (map.key.variants_size == 0 && alias.variants.is_empty())
  1148. alias.variants = move(language_id.variants);
  1149. language_id = move(alias);
  1150. break;
  1151. }
  1152. }
  1153. Optional<LanguageID> add_likely_subtags(LanguageID const& language_id)
  1154. {
  1155. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  1156. auto const* likely_subtag = resolve_likely_subtag(language_id);
  1157. if (likely_subtag == nullptr)
  1158. return {};
  1159. auto maximized = language_id;
  1160. auto const& key_script = s_string_list[likely_subtag->key.script];
  1161. auto const& key_region = s_string_list[likely_subtag->key.region];
  1162. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  1163. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  1164. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  1165. if (maximized.language == "und"sv)
  1166. maximized.language = alias_language;
  1167. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  1168. maximized.script = alias_script;
  1169. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  1170. maximized.region = alias_region;
  1171. return maximized;
  1172. }
  1173. Optional<String> resolve_most_likely_territory(LanguageID const& language_id)
  1174. {
  1175. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  1176. return s_string_list[likely_subtag->alias.region];
  1177. return {};
  1178. }
  1179. }
  1180. )~~~");
  1181. VERIFY(file.write(generator.as_string_view()));
  1182. }
  1183. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1184. {
  1185. StringView generated_header_path;
  1186. StringView generated_implementation_path;
  1187. StringView core_path;
  1188. StringView locale_names_path;
  1189. StringView misc_path;
  1190. StringView numbers_path;
  1191. StringView dates_path;
  1192. Core::ArgsParser args_parser;
  1193. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1194. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1195. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1196. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  1197. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  1198. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  1199. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1200. args_parser.parse(arguments);
  1201. auto open_file = [&](StringView path) -> ErrorOr<NonnullRefPtr<Core::File>> {
  1202. if (path.is_empty()) {
  1203. args_parser.print_usage(stderr, arguments.argv[0]);
  1204. return Error::from_string_literal("Must provide all command line options"sv);
  1205. }
  1206. return Core::File::open(path, Core::OpenMode::ReadWrite);
  1207. };
  1208. auto generated_header_file = TRY(open_file(generated_header_path));
  1209. auto generated_implementation_file = TRY(open_file(generated_implementation_path));
  1210. UnicodeLocaleData locale_data;
  1211. TRY(parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, dates_path, locale_data));
  1212. generate_unicode_locale_header(generated_header_file, locale_data);
  1213. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  1214. return 0;
  1215. }