GenerateUnicodeLocale.cpp 63 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493
  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 keywords { 0 };
  150. ListPatternListIndexType list_patterns { 0 };
  151. };
  152. struct LanguageMapping {
  153. CanonicalLanguageID<StringIndexType> key {};
  154. CanonicalLanguageID<StringIndexType> alias {};
  155. };
  156. struct UnicodeLocaleData {
  157. UniqueStringStorage<StringIndexType> unique_strings;
  158. UniqueStorage<DisplayPattern, DisplayPatternIndexType> unique_display_patterns;
  159. UniqueStorage<LanguageList, LanguageListIndexType> unique_language_lists;
  160. UniqueStorage<TerritoryList, TerritoryListIndexType> unique_territory_lists;
  161. UniqueStorage<ScriptList, ScriptListIndexType> unique_script_lists;
  162. UniqueStorage<CurrencyList, CurrencyListIndexType> unique_currency_lists;
  163. UniqueStorage<CalendarList, CalendarListIndexType> unique_calendar_lists;
  164. UniqueStorage<DateFieldList, DateFieldListIndexType> unique_date_field_lists;
  165. UniqueStorage<KeywordList, KeywordListIndexType> unique_keyword_lists;
  166. UniqueStorage<ListPatterns, ListPatternIndexType> unique_list_patterns;
  167. UniqueStorage<ListPatternList, ListPatternListIndexType> unique_list_pattern_lists;
  168. HashMap<String, Locale> locales;
  169. Vector<Alias> locale_aliases;
  170. Vector<String> languages;
  171. Vector<String> territories;
  172. Vector<String> scripts;
  173. Vector<String> variants;
  174. Vector<String> currencies;
  175. Vector<String> calendars;
  176. Vector<Alias> calendar_aliases {
  177. // FIXME: Aliases should come from BCP47. See: https://unicode-org.atlassian.net/browse/CLDR-15158
  178. { "ethiopic-amete-alem"sv, "ethioaa"sv },
  179. { "gregorian"sv, "gregory"sv },
  180. };
  181. Vector<String> date_fields;
  182. Vector<Alias> date_field_aliases {
  183. // ECMA-402 and the CLDR refer to some date fields with different names. Defining these aliases
  184. // means we can remain agnostic about the naming differences elsewhere.
  185. { "dayperiod"sv, "dayPeriod"sv },
  186. { "week"sv, "weekOfYear"sv },
  187. { "zone"sv, "timeZoneName"sv },
  188. };
  189. 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
  190. Vector<String> list_pattern_types;
  191. HashMap<String, StringIndexType> language_aliases;
  192. HashMap<String, StringIndexType> territory_aliases;
  193. HashMap<String, StringIndexType> script_aliases;
  194. HashMap<String, StringIndexType> variant_aliases;
  195. HashMap<String, StringIndexType> subdivision_aliases;
  196. Vector<LanguageMapping> complex_mappings;
  197. Vector<LanguageMapping> likely_subtags;
  198. size_t max_variant_size { 0 };
  199. };
  200. // Some parsing is expected to fail. For example, the CLDR contains language mappings
  201. // with locales such as "en-GB-oed" that are canonically invalid locale IDs.
  202. #define TRY_OR_DISCARD(expression) \
  203. ({ \
  204. auto _temporary_result = (expression); \
  205. if (_temporary_result.is_error()) \
  206. return; \
  207. _temporary_result.release_value(); \
  208. })
  209. static ErrorOr<LanguageMapping> parse_language_mapping(UnicodeLocaleData& locale_data, StringView key, StringView alias)
  210. {
  211. auto parsed_key = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, key));
  212. auto parsed_alias = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, alias));
  213. return LanguageMapping { move(parsed_key), move(parsed_alias) };
  214. }
  215. static ErrorOr<void> parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
  216. {
  217. LexicalPath core_aliases_path(move(core_supplemental_path));
  218. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  219. auto core_aliases = TRY(read_json_file(core_aliases_path.string()));
  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 = TRY(read_json_file(likely_subtags_path.string()));
  248. auto const& supplemental_object = likely_subtags.as_object().get("supplemental"sv);
  249. auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
  250. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  251. auto mapping = TRY_OR_DISCARD(parse_language_mapping(locale_data, key, value.as_string()));
  252. locale_data.max_variant_size = max(mapping.key.variants.size(), locale_data.max_variant_size);
  253. locale_data.max_variant_size = max(mapping.alias.variants.size(), locale_data.max_variant_size);
  254. locale_data.likely_subtags.append(move(mapping));
  255. });
  256. return {};
  257. }
  258. static ErrorOr<void> parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  259. {
  260. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  261. languages_path = languages_path.append("languages.json"sv);
  262. auto languages = TRY(read_json_file(languages_path.string()));
  263. auto const& main_object = languages.as_object().get("main"sv);
  264. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  265. auto const& identity_object = locale_object.as_object().get("identity"sv);
  266. auto const& language_string = identity_object.as_object().get("language"sv);
  267. auto const& territory_string = identity_object.as_object().get("territory"sv);
  268. auto const& script_string = identity_object.as_object().get("script"sv);
  269. auto const& variant_string = identity_object.as_object().get("variant"sv);
  270. locale.language = language_string.as_string();
  271. if (territory_string.is_string()) {
  272. locale.territory = territory_string.as_string();
  273. if (!locale_data.territories.contains_slow(*locale.territory))
  274. locale_data.territories.append(*locale.territory);
  275. }
  276. if (script_string.is_string()) {
  277. auto script = script_string.as_string();
  278. if (!locale_data.scripts.contains_slow(script))
  279. locale_data.scripts.append(script);
  280. }
  281. if (variant_string.is_string()) {
  282. locale.variant = variant_string.as_string();
  283. if (!locale_data.variants.contains_slow(*locale.variant))
  284. locale_data.variants.append(*locale.variant);
  285. }
  286. return {};
  287. }
  288. static ErrorOr<void> parse_locale_display_patterns(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  289. {
  290. LexicalPath locale_display_names_path(move(locale_path));
  291. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  292. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  293. auto const& main_object = locale_display_names.as_object().get("main"sv);
  294. auto const& locale_object = main_object.as_object().get(locale_display_names_path.parent().basename());
  295. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  296. auto const& locale_display_patterns_object = locale_display_names_object.as_object().get("localeDisplayPattern"sv);
  297. auto const& locale_pattern = locale_display_patterns_object.as_object().get("localePattern"sv);
  298. auto const& locale_separator = locale_display_patterns_object.as_object().get("localeSeparator"sv);
  299. DisplayPattern patterns {};
  300. patterns.locale_pattern = locale_data.unique_strings.ensure(locale_pattern.as_string());
  301. patterns.locale_separator = locale_data.unique_strings.ensure(locale_separator.as_string());
  302. locale.display_patterns = locale_data.unique_display_patterns.ensure(move(patterns));
  303. return {};
  304. }
  305. static ErrorOr<void> preprocess_languages(String locale_path, UnicodeLocaleData& locale_data)
  306. {
  307. LexicalPath languages_path(move(locale_path));
  308. languages_path = languages_path.append("languages.json"sv);
  309. auto locale_languages = TRY(read_json_file(languages_path.string()));
  310. auto const& main_object = locale_languages.as_object().get("main"sv);
  311. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  312. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  313. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  314. languages_object.as_object().for_each_member([&](auto const& key, auto const&) {
  315. if (!key.contains("-alt-"sv) && !locale_data.languages.contains_slow(key))
  316. locale_data.languages.append(key);
  317. });
  318. return {};
  319. }
  320. static ErrorOr<void> parse_locale_languages(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  321. {
  322. LexicalPath languages_path(move(locale_path));
  323. languages_path = languages_path.append("languages.json"sv);
  324. auto locale_languages = TRY(read_json_file(languages_path.string()));
  325. auto const& main_object = locale_languages.as_object().get("main"sv);
  326. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  327. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  328. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  329. LanguageList languages;
  330. languages.resize(locale_data.languages.size());
  331. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  332. if (key.contains("-alt-"sv))
  333. return;
  334. auto index = locale_data.languages.find_first_index(key).value();
  335. languages[index] = locale_data.unique_strings.ensure(value.as_string());
  336. });
  337. locale.languages = locale_data.unique_language_lists.ensure(move(languages));
  338. return {};
  339. }
  340. static ErrorOr<void> parse_locale_territories(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  341. {
  342. LexicalPath territories_path(move(locale_path));
  343. territories_path = territories_path.append("territories.json"sv);
  344. auto locale_territories = TRY(read_json_file(territories_path.string()));
  345. auto const& main_object = locale_territories.as_object().get("main"sv);
  346. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  347. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  348. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  349. TerritoryList territories;
  350. territories.resize(locale_data.territories.size());
  351. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  352. if (auto index = locale_data.territories.find_first_index(key); index.has_value())
  353. territories[*index] = locale_data.unique_strings.ensure(value.as_string());
  354. });
  355. locale.territories = locale_data.unique_territory_lists.ensure(move(territories));
  356. return {};
  357. }
  358. static ErrorOr<void> parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  359. {
  360. LexicalPath scripts_path(move(locale_path));
  361. scripts_path = scripts_path.append("scripts.json"sv);
  362. auto locale_scripts = TRY(read_json_file(scripts_path.string()));
  363. auto const& main_object = locale_scripts.as_object().get("main"sv);
  364. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  365. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  366. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  367. ScriptList scripts;
  368. scripts.resize(locale_data.scripts.size());
  369. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  370. if (auto index = locale_data.scripts.find_first_index(key); index.has_value())
  371. scripts[*index] = locale_data.unique_strings.ensure(value.as_string());
  372. });
  373. locale.scripts = locale_data.unique_script_lists.ensure(move(scripts));
  374. return {};
  375. }
  376. static ErrorOr<void> parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
  377. {
  378. LexicalPath list_patterns_path(move(misc_path));
  379. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  380. auto locale_list_patterns = TRY(read_json_file(list_patterns_path.string()));
  381. auto const& main_object = locale_list_patterns.as_object().get("main"sv);
  382. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  383. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  384. auto list_pattern_type = [](StringView key) {
  385. if (key.contains("type-standard"sv))
  386. return "conjunction"sv;
  387. if (key.contains("type-or"sv))
  388. return "disjunction"sv;
  389. if (key.contains("type-unit"sv))
  390. return "unit"sv;
  391. VERIFY_NOT_REACHED();
  392. };
  393. auto list_pattern_style = [](StringView key) {
  394. if (key.contains("short"sv))
  395. return "short"sv;
  396. if (key.contains("narrow"sv))
  397. return "narrow"sv;
  398. return "long"sv;
  399. };
  400. ListPatternList list_patterns;
  401. list_patterns.ensure_capacity(list_patterns_object.as_object().size());
  402. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  403. auto type = list_pattern_type(key);
  404. auto style = list_pattern_style(key);
  405. auto start = locale_data.unique_strings.ensure(value.as_object().get("start"sv).as_string());
  406. auto middle = locale_data.unique_strings.ensure(value.as_object().get("middle"sv).as_string());
  407. auto end = locale_data.unique_strings.ensure(value.as_object().get("end"sv).as_string());
  408. auto pair = locale_data.unique_strings.ensure(value.as_object().get("2"sv).as_string());
  409. if (!locale_data.list_pattern_types.contains_slow(type))
  410. locale_data.list_pattern_types.append(type);
  411. ListPatterns list_pattern { type, style, start, middle, end, pair };
  412. list_patterns.append(locale_data.unique_list_patterns.ensure(move(list_pattern)));
  413. });
  414. locale.list_patterns = locale_data.unique_list_pattern_lists.ensure(move(list_patterns));
  415. return {};
  416. }
  417. static ErrorOr<void> parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  418. {
  419. LexicalPath currencies_path(move(numbers_path));
  420. currencies_path = currencies_path.append("currencies.json"sv);
  421. auto locale_currencies = TRY(read_json_file(currencies_path.string()));
  422. auto const& main_object = locale_currencies.as_object().get("main"sv);
  423. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  424. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  425. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  426. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const&) {
  427. if (!locale_data.currencies.contains_slow(key))
  428. locale_data.currencies.append(key);
  429. });
  430. CurrencyList long_currencies {};
  431. long_currencies.resize(locale_data.currencies.size());
  432. CurrencyList short_currencies {};
  433. short_currencies.resize(locale_data.currencies.size());
  434. CurrencyList narrow_currencies {};
  435. narrow_currencies.resize(locale_data.currencies.size());
  436. CurrencyList numeric_currencies {};
  437. numeric_currencies.resize(locale_data.currencies.size());
  438. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  439. auto const& long_name = value.as_object().get("displayName"sv);
  440. auto const& short_name = value.as_object().get("symbol"sv);
  441. auto const& narrow_name = value.as_object().get("symbol-alt-narrow"sv);
  442. auto const& numeric_name = value.as_object().get("displayName-count-other"sv);
  443. auto index = locale_data.currencies.find_first_index(key).value();
  444. long_currencies[index] = locale_data.unique_strings.ensure(long_name.as_string());
  445. short_currencies[index] = locale_data.unique_strings.ensure(short_name.as_string());
  446. narrow_currencies[index] = narrow_name.is_null() ? 0 : locale_data.unique_strings.ensure(narrow_name.as_string());
  447. numeric_currencies[index] = locale_data.unique_strings.ensure(numeric_name.is_null() ? long_name.as_string() : numeric_name.as_string());
  448. });
  449. locale.long_currencies = locale_data.unique_currency_lists.ensure(move(long_currencies));
  450. locale.short_currencies = locale_data.unique_currency_lists.ensure(move(short_currencies));
  451. locale.narrow_currencies = locale_data.unique_currency_lists.ensure(move(narrow_currencies));
  452. locale.numeric_currencies = locale_data.unique_currency_lists.ensure(move(numeric_currencies));
  453. return {};
  454. }
  455. static ErrorOr<void> parse_locale_calendars(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  456. {
  457. LexicalPath locale_display_names_path(move(locale_path));
  458. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  459. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  460. auto const& main_object = locale_display_names.as_object().get("main"sv);
  461. auto const& locale_object = main_object.as_object().get(locale_display_names_path.parent().basename());
  462. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  463. auto const& types_object = locale_display_names_object.as_object().get("types"sv);
  464. auto const& calendar_object = types_object.as_object().get("calendar"sv);
  465. calendar_object.as_object().for_each_member([&](auto const& key, auto const&) {
  466. if (!locale_data.calendars.contains_slow(key))
  467. locale_data.calendars.append(key);
  468. });
  469. CalendarList calendars;
  470. calendars.resize(locale_data.calendars.size());
  471. calendar_object.as_object().for_each_member([&](auto const& key, auto const& calendar) {
  472. auto index = locale_data.calendars.find_first_index(key).value();
  473. calendars[index] = locale_data.unique_strings.ensure(calendar.as_string());
  474. });
  475. locale.calendars = locale_data.unique_calendar_lists.ensure(move(calendars));
  476. return {};
  477. }
  478. static ErrorOr<void> parse_locale_date_fields(String dates_path, UnicodeLocaleData& locale_data, Locale& locale)
  479. {
  480. LexicalPath date_fields_path(move(dates_path));
  481. date_fields_path = date_fields_path.append("dateFields.json"sv);
  482. auto locale_date_fields = TRY(read_json_file(date_fields_path.string()));
  483. auto const& main_object = locale_date_fields.as_object().get("main"sv);
  484. auto const& locale_object = main_object.as_object().get(date_fields_path.parent().basename());
  485. auto const& dates_object = locale_object.as_object().get("dates"sv);
  486. auto const& fields_object = dates_object.as_object().get("fields"sv);
  487. auto is_sanctioned_field = [](StringView field) {
  488. // This is a copy of the units sanctioned for use within ECMA-402, with names adjusted for the names used by the CLDR.
  489. // https://tc39.es/ecma402/#table-validcodeforDateField
  490. 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);
  491. };
  492. fields_object.as_object().for_each_member([&](auto const& key, JsonValue const&) {
  493. if (!is_sanctioned_field(key))
  494. return;
  495. if (!locale_data.date_fields.contains_slow(key))
  496. locale_data.date_fields.append(key);
  497. });
  498. quick_sort(locale_data.date_fields);
  499. DateFieldList long_date_fields {};
  500. long_date_fields.resize(locale_data.date_fields.size());
  501. DateFieldList short_date_fields {};
  502. short_date_fields.resize(locale_data.date_fields.size());
  503. DateFieldList narrow_date_fields {};
  504. narrow_date_fields.resize(locale_data.date_fields.size());
  505. fields_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  506. if (!is_sanctioned_field(key))
  507. return;
  508. auto const& long_name = value.as_object().get("displayName"sv);
  509. auto const& short_name = fields_object.as_object().get(String::formatted("{}-short", key)).as_object().get("displayName"sv);
  510. auto const& narrow_name = fields_object.as_object().get(String::formatted("{}-narrow", key)).as_object().get("displayName"sv);
  511. auto index = locale_data.date_fields.find_first_index(key).value();
  512. long_date_fields[index] = locale_data.unique_strings.ensure(long_name.as_string());
  513. short_date_fields[index] = locale_data.unique_strings.ensure(short_name.as_string());
  514. narrow_date_fields[index] = locale_data.unique_strings.ensure(narrow_name.as_string());
  515. });
  516. locale.long_date_fields = locale_data.unique_date_field_lists.ensure(move(long_date_fields));
  517. locale.short_date_fields = locale_data.unique_date_field_lists.ensure(move(short_date_fields));
  518. locale.narrow_date_fields = locale_data.unique_date_field_lists.ensure(move(narrow_date_fields));
  519. return {};
  520. }
  521. static ErrorOr<void> parse_numeric_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, KeywordList& keywords)
  522. {
  523. static constexpr StringView key = "nu"sv;
  524. LexicalPath numbers_path(move(locale_numbers_path));
  525. numbers_path = numbers_path.append("numbers.json"sv);
  526. auto numbers = TRY(read_json_file(numbers_path.string()));
  527. auto const& main_object = numbers.as_object().get("main"sv);
  528. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  529. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  530. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  531. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  532. Vector<String> keyword_values {};
  533. keyword_values.append(default_numbering_system_object.as_string());
  534. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  535. auto keyword_value = value.as_string();
  536. if (!keyword_values.contains_slow(keyword_value))
  537. keyword_values.append(move(keyword_value));
  538. });
  539. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  540. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  541. return;
  542. auto keyword_value = value.as_string();
  543. if (!keyword_values.contains_slow(keyword_value))
  544. keyword_values.append(move(keyword_value));
  545. });
  546. StringBuilder builder;
  547. builder.join(',', keyword_values);
  548. auto index = locale_data.keywords.find_first_index(key).value();
  549. keywords[index] = locale_data.unique_strings.ensure(builder.build());
  550. return {};
  551. }
  552. static ErrorOr<void> parse_calendar_keywords(String locale_dates_path, UnicodeLocaleData& locale_data, KeywordList& keywords)
  553. {
  554. static constexpr StringView key = "ca"sv;
  555. auto calendars_iterator = TRY(path_to_dir_iterator(locale_dates_path, {}));
  556. Vector<String> keyword_values {};
  557. while (calendars_iterator.has_next()) {
  558. auto locale_calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  559. LexicalPath calendars_path(move(locale_calendars_path));
  560. if (!calendars_path.basename().starts_with("ca-"sv))
  561. continue;
  562. auto calendars = TRY(read_json_file(calendars_path.string()));
  563. auto const& main_object = calendars.as_object().get("main"sv);
  564. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  565. auto const& dates_object = locale_object.as_object().get("dates"sv);
  566. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  567. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const&) {
  568. // The generic calendar is not a supported Unicode calendar key, so skip it:
  569. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  570. if (calendar_name == "generic"sv)
  571. return;
  572. // FIXME: Similar to the calendar aliases defined in GenerateUnicodeDateTimeFormat, this
  573. // should be parsed from BCP47. https://unicode-org.atlassian.net/browse/CLDR-15158
  574. if (calendar_name == "gregorian"sv)
  575. keyword_values.append("gregory"sv);
  576. else
  577. keyword_values.append(calendar_name);
  578. });
  579. }
  580. StringBuilder builder;
  581. builder.join(',', keyword_values);
  582. auto index = locale_data.keywords.find_first_index(key).value();
  583. keywords[index] = locale_data.unique_strings.ensure(builder.build());
  584. return {};
  585. }
  586. static void fill_in_bcp47_keywords(UnicodeLocaleData& locale_data, KeywordList& keywords)
  587. {
  588. // FIXME: These should be parsed from BCP47. They are only available in this XML file:
  589. // https://github.com/unicode-org/cldr/blob/main/common/bcp47/collation.xml
  590. auto kf_index = locale_data.keywords.find_first_index("kf"sv).value();
  591. keywords[kf_index] = locale_data.unique_strings.ensure("upper,lower,false"sv);
  592. auto kn_index = locale_data.keywords.find_first_index("kn"sv).value();
  593. keywords[kn_index] = locale_data.unique_strings.ensure("true,false"sv);
  594. }
  595. static ErrorOr<void> parse_default_content_locales(String core_path, UnicodeLocaleData& locale_data)
  596. {
  597. LexicalPath default_content_path(move(core_path));
  598. default_content_path = default_content_path.append("defaultContent.json"sv);
  599. auto default_content = TRY(read_json_file(default_content_path.string()));
  600. auto const& default_content_array = default_content.as_object().get("defaultContent"sv);
  601. default_content_array.as_array().for_each([&](JsonValue const& value) {
  602. auto locale = value.as_string();
  603. StringView default_locale = locale;
  604. while (true) {
  605. if (locale_data.locales.contains(default_locale))
  606. break;
  607. auto pos = default_locale.find_last('-');
  608. if (!pos.has_value())
  609. return;
  610. default_locale = default_locale.substring_view(0, *pos);
  611. }
  612. if (default_locale != locale)
  613. locale_data.locale_aliases.append({ default_locale, move(locale) });
  614. });
  615. return {};
  616. }
  617. static ErrorOr<void> define_aliases_without_scripts(UnicodeLocaleData& locale_data)
  618. {
  619. // From ECMA-402: https://tc39.es/ecma402/#sec-internal-slots
  620. //
  621. // For locales that include a script subtag in addition to language and region, the
  622. // corresponding locale without a script subtag must also be supported.
  623. //
  624. // So we define aliases for locales that contain all three subtags, but we must also take
  625. // care to handle when the locale itself or the locale without a script subtag are an alias
  626. // by way of default-content locales.
  627. auto find_alias = [&](auto const& locale) {
  628. return locale_data.locale_aliases.find_if([&](auto const& alias) { return locale == alias.alias; });
  629. };
  630. auto append_alias_without_script = [&](auto const& locale) -> ErrorOr<void> {
  631. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, locale));
  632. if ((parsed_locale.language == 0) || (parsed_locale.script == 0) || (parsed_locale.region == 0))
  633. return {};
  634. auto locale_without_script = String::formatted("{}-{}",
  635. locale_data.unique_strings.get(parsed_locale.language),
  636. locale_data.unique_strings.get(parsed_locale.region));
  637. if (locale_data.locales.contains(locale_without_script))
  638. return {};
  639. if (find_alias(locale_without_script) != locale_data.locale_aliases.end())
  640. return {};
  641. if (auto it = find_alias(locale); it != locale_data.locale_aliases.end())
  642. locale_data.locale_aliases.append({ it->name, locale_without_script });
  643. else
  644. locale_data.locale_aliases.append({ locale, locale_without_script });
  645. return {};
  646. };
  647. for (auto const& locale : locale_data.locales)
  648. TRY(append_alias_without_script(locale.key));
  649. for (auto const& locale : locale_data.locale_aliases)
  650. TRY(append_alias_without_script(locale.alias));
  651. return {};
  652. }
  653. 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)
  654. {
  655. auto identity_iterator = TRY(path_to_dir_iterator(locale_names_path));
  656. auto preprocess_iterator = TRY(path_to_dir_iterator(locale_names_path));
  657. auto locale_names_iterator = TRY(path_to_dir_iterator(move(locale_names_path)));
  658. auto misc_iterator = TRY(path_to_dir_iterator(move(misc_path)));
  659. auto numbers_iterator = TRY(path_to_dir_iterator(move(numbers_path)));
  660. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  661. LexicalPath core_supplemental_path(core_path);
  662. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  663. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  664. TRY(parse_core_aliases(core_supplemental_path.string(), locale_data));
  665. TRY(parse_likely_subtags(core_supplemental_path.string(), locale_data));
  666. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  667. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  668. StringBuilder builder;
  669. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  670. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  671. builder.appendff("-{}", script);
  672. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  673. builder.appendff("-{}", region);
  674. return builder.build();
  675. };
  676. while (identity_iterator.has_next()) {
  677. auto locale_path = TRY(next_path_from_dir_iterator(identity_iterator));
  678. auto language = TRY(remove_variants_from_path(locale_path));
  679. auto& locale = locale_data.locales.ensure(language);
  680. TRY(parse_identity(locale_path, locale_data, locale));
  681. }
  682. while (preprocess_iterator.has_next()) {
  683. auto locale_path = TRY(next_path_from_dir_iterator(preprocess_iterator));
  684. TRY(preprocess_languages(locale_path, locale_data));
  685. }
  686. quick_sort(locale_data.languages);
  687. quick_sort(locale_data.territories);
  688. quick_sort(locale_data.scripts);
  689. HashMap<String, KeywordList> keywords;
  690. auto ensure_keyword_list = [&](auto const& language) -> KeywordList& {
  691. return keywords.ensure(language, [&]() {
  692. KeywordList keywords;
  693. keywords.resize(locale_data.keywords.size());
  694. return keywords;
  695. });
  696. };
  697. while (locale_names_iterator.has_next()) {
  698. auto locale_path = TRY(next_path_from_dir_iterator(locale_names_iterator));
  699. auto language = TRY(remove_variants_from_path(locale_path));
  700. auto& locale = locale_data.locales.ensure(language);
  701. TRY(parse_locale_display_patterns(locale_path, locale_data, locale));
  702. TRY(parse_locale_languages(locale_path, locale_data, locale));
  703. TRY(parse_locale_territories(locale_path, locale_data, locale));
  704. TRY(parse_locale_scripts(locale_path, locale_data, locale));
  705. TRY(parse_locale_calendars(locale_path, locale_data, locale));
  706. }
  707. while (misc_iterator.has_next()) {
  708. auto misc_path = TRY(next_path_from_dir_iterator(misc_iterator));
  709. auto language = TRY(remove_variants_from_path(misc_path));
  710. auto& locale = locale_data.locales.ensure(language);
  711. TRY(parse_locale_list_patterns(misc_path, locale_data, locale));
  712. }
  713. while (numbers_iterator.has_next()) {
  714. auto numbers_path = TRY(next_path_from_dir_iterator(numbers_iterator));
  715. auto language = TRY(remove_variants_from_path(numbers_path));
  716. auto& locale = locale_data.locales.ensure(language);
  717. TRY(parse_locale_currencies(numbers_path, locale_data, locale));
  718. auto& keywords = ensure_keyword_list(language);
  719. TRY(parse_numeric_keywords(numbers_path, locale_data, keywords));
  720. fill_in_bcp47_keywords(locale_data, keywords);
  721. }
  722. while (dates_iterator.has_next()) {
  723. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  724. auto language = TRY(remove_variants_from_path(dates_path));
  725. auto& locale = locale_data.locales.ensure(language);
  726. TRY(parse_locale_date_fields(dates_path, locale_data, locale));
  727. auto& keywords = ensure_keyword_list(language);
  728. TRY(parse_calendar_keywords(dates_path, locale_data, keywords));
  729. }
  730. TRY(parse_default_content_locales(move(core_path), locale_data));
  731. TRY(define_aliases_without_scripts(locale_data));
  732. for (auto& list : keywords) {
  733. auto& locale = locale_data.locales.find(list.key)->value;
  734. locale.keywords = locale_data.unique_keyword_lists.ensure(move(list.value));
  735. }
  736. return {};
  737. }
  738. static ErrorOr<void> generate_unicode_locale_header(Core::Stream::BufferedFile& file, UnicodeLocaleData& locale_data)
  739. {
  740. StringBuilder builder;
  741. SourceGenerator generator { builder };
  742. generator.append(R"~~~(
  743. #pragma once
  744. #include <AK/Types.h>
  745. namespace Unicode {
  746. )~~~");
  747. auto locales = locale_data.locales.keys();
  748. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, locale_data.locale_aliases);
  749. generate_enum(generator, format_identifier, "Language"sv, {}, locale_data.languages);
  750. generate_enum(generator, format_identifier, "Territory"sv, {}, locale_data.territories);
  751. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, locale_data.scripts);
  752. generate_enum(generator, format_identifier, "Currency"sv, {}, locale_data.currencies);
  753. generate_enum(generator, format_identifier, "CalendarName"sv, {}, locale_data.calendars, locale_data.calendar_aliases);
  754. generate_enum(generator, format_identifier, "DateField"sv, {}, locale_data.date_fields, locale_data.date_field_aliases);
  755. generate_enum(generator, format_identifier, "Key"sv, {}, locale_data.keywords);
  756. generate_enum(generator, format_identifier, "Variant"sv, {}, locale_data.variants);
  757. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, locale_data.list_pattern_types);
  758. generator.append(R"~~~(
  759. }
  760. )~~~");
  761. TRY(file.write(generator.as_string_view().bytes()));
  762. return {};
  763. }
  764. static ErrorOr<void> generate_unicode_locale_implementation(Core::Stream::BufferedFile& file, UnicodeLocaleData& locale_data)
  765. {
  766. StringBuilder builder;
  767. SourceGenerator generator { builder };
  768. generator.set("string_index_type"sv, s_string_index_type);
  769. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  770. generator.set("territories_size", String::number(locale_data.territories.size()));
  771. generator.set("variants_size", String::number(locale_data.max_variant_size));
  772. generator.append(R"~~~(
  773. #include <AK/Array.h>
  774. #include <AK/BinarySearch.h>
  775. #include <AK/Optional.h>
  776. #include <AK/Span.h>
  777. #include <AK/StringView.h>
  778. #include <AK/Vector.h>
  779. #include <LibUnicode/CurrencyCode.h>
  780. #include <LibUnicode/Locale.h>
  781. #include <LibUnicode/UnicodeLocale.h>
  782. namespace Unicode {
  783. )~~~");
  784. locale_data.unique_strings.generate(generator);
  785. generator.append(R"~~~(
  786. struct DisplayPatternImpl {
  787. DisplayPattern to_display_pattern() const
  788. {
  789. DisplayPattern display_patterns {};
  790. display_patterns.locale_pattern = s_string_list[locale_pattern];
  791. display_patterns.locale_separator = s_string_list[locale_separator];
  792. return display_patterns;
  793. }
  794. @string_index_type@ locale_pattern { 0 };
  795. @string_index_type@ locale_separator { 0 };
  796. };
  797. struct Patterns {
  798. ListPatternType type;
  799. Style style;
  800. @string_index_type@ start { 0 };
  801. @string_index_type@ middle { 0 };
  802. @string_index_type@ end { 0 };
  803. @string_index_type@ pair { 0 };
  804. };
  805. )~~~");
  806. generate_available_values(generator, "get_available_currencies"sv, locale_data.currencies);
  807. locale_data.unique_display_patterns.generate(generator, "DisplayPatternImpl"sv, "s_display_patterns"sv, 30);
  808. locale_data.unique_language_lists.generate(generator, s_string_index_type, "s_language_lists"sv);
  809. locale_data.unique_territory_lists.generate(generator, s_string_index_type, "s_territory_lists"sv);
  810. locale_data.unique_script_lists.generate(generator, s_string_index_type, "s_script_lists"sv);
  811. locale_data.unique_currency_lists.generate(generator, s_string_index_type, "s_currency_lists"sv);
  812. locale_data.unique_calendar_lists.generate(generator, s_string_index_type, "s_calendar_lists"sv);
  813. locale_data.unique_date_field_lists.generate(generator, s_string_index_type, "s_date_field_lists"sv);
  814. locale_data.unique_keyword_lists.generate(generator, s_string_index_type, "s_keyword_lists"sv);
  815. locale_data.unique_list_patterns.generate(generator, "Patterns"sv, "s_list_patterns"sv, 10);
  816. locale_data.unique_list_pattern_lists.generate(generator, s_list_pattern_index_type, "s_list_pattern_lists"sv);
  817. auto append_index = [&](auto index) {
  818. generator.append(String::formatted(", {}", index));
  819. };
  820. auto append_list_and_size = [&](auto const& list) {
  821. if (list.is_empty()) {
  822. generator.append(", {}, 0");
  823. return;
  824. }
  825. bool first = true;
  826. generator.append(", {");
  827. for (auto const& item : list) {
  828. generator.append(first ? " " : ", ");
  829. generator.append(String::number(item));
  830. first = false;
  831. }
  832. generator.append(String::formatted(" }}, {}", list.size()));
  833. };
  834. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  835. generator.set("type", type);
  836. generator.set("name", name);
  837. generator.set("size", String::number(keys.size()));
  838. generator.append(R"~~~(
  839. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  840. bool first = true;
  841. for (auto const& key : keys) {
  842. auto const& value = map.find(key)->value;
  843. auto mapping = mapping_getter(value);
  844. generator.append(first ? " " : ", ");
  845. generator.append(String::number(mapping));
  846. first = false;
  847. }
  848. generator.append(" } };");
  849. };
  850. auto locales = locale_data.locales.keys();
  851. quick_sort(locales);
  852. append_mapping(locales, locale_data.locales, s_display_pattern_index_type, "s_locale_display_patterns"sv, [&](auto const& locale) { return locale.display_patterns; });
  853. append_mapping(locales, locale_data.locales, s_language_list_index_type, "s_languages"sv, [&](auto const& locale) { return locale.languages; });
  854. append_mapping(locales, locale_data.locales, s_territory_list_index_type, "s_territories"sv, [&](auto const& locale) { return locale.territories; });
  855. append_mapping(locales, locale_data.locales, s_script_list_index_type, "s_scripts"sv, [&](auto const& locale) { return locale.scripts; });
  856. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_long_currencies"sv, [&](auto const& locale) { return locale.long_currencies; });
  857. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_short_currencies"sv, [&](auto const& locale) { return locale.short_currencies; });
  858. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_narrow_currencies"sv, [&](auto const& locale) { return locale.narrow_currencies; });
  859. append_mapping(locales, locale_data.locales, s_currency_list_index_type, "s_numeric_currencies"sv, [&](auto const& locale) { return locale.numeric_currencies; });
  860. append_mapping(locales, locale_data.locales, s_calendar_list_index_type, "s_calendars"sv, [&](auto const& locale) { return locale.calendars; });
  861. 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; });
  862. 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; });
  863. 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; });
  864. append_mapping(locales, locale_data.locales, s_keyword_list_index_type, "s_keywords"sv, [&](auto const& locale) { return locale.keywords; });
  865. append_mapping(locales, locale_data.locales, s_list_pattern_list_index_type, "s_locale_list_patterns"sv, [&](auto const& locale) { return locale.list_patterns; });
  866. generator.append(R"~~~(
  867. struct CanonicalLanguageID {
  868. LanguageID to_unicode_language_id() const
  869. {
  870. LanguageID language_id {};
  871. language_id.variants.ensure_capacity(variants_size);
  872. language_id.language = s_string_list[language];
  873. if (script != 0)
  874. language_id.script = s_string_list[script];
  875. if (region != 0)
  876. language_id.region = s_string_list[region];
  877. for (size_t i = 0; i < variants_size; ++i)
  878. language_id.variants.append(s_string_list[variants[i]]);
  879. return language_id;
  880. }
  881. bool matches_variants(Vector<String> const& other_variants) const {
  882. if (variants_size == 0)
  883. return true;
  884. if (other_variants.size() != variants_size)
  885. return false;
  886. for (size_t i = 0; i < variants_size; ++i) {
  887. if (s_string_list[variants[i]] != other_variants[i])
  888. return false;
  889. }
  890. return true;
  891. };
  892. @string_index_type@ language { 0 };
  893. @string_index_type@ script { 0 };
  894. @string_index_type@ region { 0 };
  895. Array<@string_index_type@, @variants_size@> variants {};
  896. size_t variants_size { 0 };
  897. };
  898. struct LanguageMapping {
  899. CanonicalLanguageID key;
  900. CanonicalLanguageID alias;
  901. };
  902. )~~~");
  903. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  904. generator.set("size", String::number(mappings.size()));
  905. generator.set("name"sv, name);
  906. generator.append(R"~~~(
  907. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  908. )~~~");
  909. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  910. auto const& lhs_language = locale_data.unique_strings.get(lhs.key.language);
  911. auto const& rhs_language = locale_data.unique_strings.get(rhs.key.language);
  912. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  913. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  914. return false;
  915. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  916. return true;
  917. return lhs_language < rhs_language;
  918. });
  919. for (auto const& mapping : mappings) {
  920. generator.set("language"sv, String::number(mapping.key.language));
  921. generator.append(" { { @language@");
  922. append_index(mapping.key.script);
  923. append_index(mapping.key.region);
  924. append_list_and_size(mapping.key.variants);
  925. generator.set("language"sv, String::number(mapping.alias.language));
  926. generator.append(" }, { @language@");
  927. append_index(mapping.alias.script);
  928. append_index(mapping.alias.region);
  929. append_list_and_size(mapping.alias.variants);
  930. generator.append(" } },\n");
  931. }
  932. generator.append("} };\n");
  933. };
  934. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  935. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  936. generator.append(R"~~~(
  937. static LanguageMapping const* resolve_likely_subtag(LanguageID const& language_id)
  938. {
  939. // https://unicode.org/reports/tr35/#Likely_Subtags
  940. enum class State {
  941. LanguageScriptRegion,
  942. LanguageRegion,
  943. LanguageScript,
  944. Language,
  945. UndScript,
  946. Done,
  947. };
  948. auto state = State::LanguageScriptRegion;
  949. while (state != State::Done) {
  950. LanguageID search_key;
  951. switch (state) {
  952. case State::LanguageScriptRegion:
  953. state = State::LanguageRegion;
  954. if (!language_id.script.has_value() || !language_id.region.has_value())
  955. continue;
  956. search_key.language = *language_id.language;
  957. search_key.script = *language_id.script;
  958. search_key.region = *language_id.region;
  959. break;
  960. case State::LanguageRegion:
  961. state = State::LanguageScript;
  962. if (!language_id.region.has_value())
  963. continue;
  964. search_key.language = *language_id.language;
  965. search_key.region = *language_id.region;
  966. break;
  967. case State::LanguageScript:
  968. state = State::Language;
  969. if (!language_id.script.has_value())
  970. continue;
  971. search_key.language = *language_id.language;
  972. search_key.script = *language_id.script;
  973. break;
  974. case State::Language:
  975. state = State::UndScript;
  976. search_key.language = *language_id.language;
  977. break;
  978. case State::UndScript:
  979. state = State::Done;
  980. if (!language_id.script.has_value())
  981. continue;
  982. search_key.language = "und"sv;
  983. search_key.script = *language_id.script;
  984. break;
  985. default:
  986. VERIFY_NOT_REACHED();
  987. }
  988. for (auto const& map : s_likely_subtags) {
  989. auto const& key_language = s_string_list[map.key.language];
  990. auto const& key_script = s_string_list[map.key.script];
  991. auto const& key_region = s_string_list[map.key.region];
  992. if (key_language != search_key.language)
  993. continue;
  994. if (!key_script.is_empty() || search_key.script.has_value()) {
  995. if (key_script != search_key.script)
  996. continue;
  997. }
  998. if (!key_region.is_empty() || search_key.region.has_value()) {
  999. if (key_region != search_key.region)
  1000. continue;
  1001. }
  1002. return &map;
  1003. }
  1004. }
  1005. return nullptr;
  1006. }
  1007. )~~~");
  1008. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name, StringView unique_list) {
  1009. generator.set("enum_snake", enum_snake);
  1010. generator.set("from_string_name", from_string_name);
  1011. generator.set("collection_name", collection_name);
  1012. generator.set("unique_list", unique_list);
  1013. generator.append(R"~~~(
  1014. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  1015. {
  1016. auto locale_value = locale_from_string(locale);
  1017. if (!locale_value.has_value())
  1018. return {};
  1019. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  1020. if (!@enum_snake@_value.has_value())
  1021. return {};
  1022. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1023. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  1024. auto mapping_index = @collection_name@.at(locale_index);
  1025. auto const& mappings = @unique_list@.at(mapping_index);
  1026. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  1027. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  1028. if (@enum_snake@_mapping.is_empty())
  1029. return {};
  1030. return @enum_snake@_mapping;
  1031. }
  1032. )~~~");
  1033. };
  1034. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1035. HashValueMap<String> hashes;
  1036. hashes.ensure_capacity(values.size());
  1037. for (auto const& value : values)
  1038. hashes.set(value.hash(), format_identifier(enum_title, value));
  1039. for (auto const& alias : aliases)
  1040. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1041. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1042. };
  1043. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  1044. HashValueMap<StringIndexType> hashes;
  1045. hashes.ensure_capacity(aliases.size());
  1046. for (auto const& alias : aliases)
  1047. hashes.set(alias.key.hash(), alias.value);
  1048. ValueFromStringOptions options {};
  1049. options.return_type = "StringView"sv;
  1050. options.return_format = "s_string_list[{}]"sv;
  1051. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), options);
  1052. };
  1053. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys(), locale_data.locale_aliases);
  1054. append_from_string("Language"sv, "language"sv, locale_data.languages);
  1055. append_mapping_search("language"sv, "language"sv, "s_languages"sv, "s_language_lists"sv);
  1056. append_alias_search("language"sv, locale_data.language_aliases);
  1057. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  1058. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv, "s_territory_lists"sv);
  1059. append_alias_search("territory"sv, locale_data.territory_aliases);
  1060. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  1061. append_mapping_search("script"sv, "script_tag"sv, "s_scripts"sv, "s_script_lists"sv);
  1062. append_alias_search("script_tag"sv, locale_data.script_aliases);
  1063. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  1064. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv, "s_currency_lists"sv);
  1065. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv, "s_currency_lists"sv);
  1066. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv, "s_currency_lists"sv);
  1067. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv, "s_currency_lists"sv);
  1068. append_from_string("CalendarName"sv, "calendar_name"sv, locale_data.calendars, locale_data.calendar_aliases);
  1069. append_mapping_search("calendar"sv, "calendar_name"sv, "s_calendars"sv, "s_calendar_lists"sv);
  1070. append_from_string("DateField"sv, "date_field"sv, locale_data.date_fields, locale_data.date_field_aliases);
  1071. append_mapping_search("long_date_field"sv, "date_field"sv, "s_long_date_fields"sv, "s_date_field_lists"sv);
  1072. append_mapping_search("short_date_field"sv, "date_field"sv, "s_short_date_fields"sv, "s_date_field_lists"sv);
  1073. append_mapping_search("narrow_date_field"sv, "date_field"sv, "s_narrow_date_fields"sv, "s_date_field_lists"sv);
  1074. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  1075. append_mapping_search("key"sv, "key"sv, "s_keywords"sv, "s_keyword_lists"sv);
  1076. append_alias_search("variant"sv, locale_data.variant_aliases);
  1077. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  1078. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  1079. generator.append(R"~~~(
  1080. Optional<DisplayPattern> get_locale_display_patterns(StringView locale)
  1081. {
  1082. auto locale_value = locale_from_string(locale);
  1083. if (!locale_value.has_value())
  1084. return {};
  1085. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1086. auto display_patterns_index = s_locale_display_patterns.at(locale_index);
  1087. auto const& display_patterns = s_display_patterns.at(display_patterns_index);
  1088. return display_patterns.to_display_pattern();
  1089. }
  1090. Optional<ListPatterns> get_locale_list_patterns(StringView locale, StringView list_pattern_type, Style list_pattern_style)
  1091. {
  1092. auto locale_value = locale_from_string(locale);
  1093. if (!locale_value.has_value())
  1094. return {};
  1095. auto type_value = list_pattern_type_from_string(list_pattern_type);
  1096. if (!type_value.has_value())
  1097. return {};
  1098. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1099. auto list_patterns_list_index = s_locale_list_patterns.at(locale_index);
  1100. auto const& locale_list_patterns = s_list_pattern_lists.at(list_patterns_list_index);
  1101. for (auto list_patterns_index : locale_list_patterns) {
  1102. auto const& list_patterns = s_list_patterns.at(list_patterns_index);
  1103. if ((list_patterns.type == type_value) && (list_patterns.style == list_pattern_style)) {
  1104. auto const& start = s_string_list[list_patterns.start];
  1105. auto const& middle = s_string_list[list_patterns.middle];
  1106. auto const& end = s_string_list[list_patterns.end];
  1107. auto const& pair = s_string_list[list_patterns.pair];
  1108. return ListPatterns { start, middle, end, pair };
  1109. }
  1110. }
  1111. return {};
  1112. }
  1113. void resolve_complex_language_aliases(LanguageID& language_id)
  1114. {
  1115. for (auto const& map : s_complex_alias) {
  1116. auto const& key_language = s_string_list[map.key.language];
  1117. auto const& key_script = s_string_list[map.key.script];
  1118. auto const& key_region = s_string_list[map.key.region];
  1119. if ((key_language != language_id.language) && (key_language != "und"sv))
  1120. continue;
  1121. if (!key_script.is_empty() && (key_script != language_id.script))
  1122. continue;
  1123. if (!key_region.is_empty() && (key_region != language_id.region))
  1124. continue;
  1125. if (!map.key.matches_variants(language_id.variants))
  1126. continue;
  1127. auto alias = map.alias.to_unicode_language_id();
  1128. if (alias.language == "und"sv)
  1129. alias.language = move(language_id.language);
  1130. if (key_script.is_empty() && !alias.script.has_value())
  1131. alias.script = move(language_id.script);
  1132. if (key_region.is_empty() && !alias.region.has_value())
  1133. alias.region = move(language_id.region);
  1134. if (map.key.variants_size == 0 && alias.variants.is_empty())
  1135. alias.variants = move(language_id.variants);
  1136. language_id = move(alias);
  1137. break;
  1138. }
  1139. }
  1140. Optional<LanguageID> add_likely_subtags(LanguageID const& language_id)
  1141. {
  1142. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  1143. auto const* likely_subtag = resolve_likely_subtag(language_id);
  1144. if (likely_subtag == nullptr)
  1145. return {};
  1146. auto maximized = language_id;
  1147. auto const& key_script = s_string_list[likely_subtag->key.script];
  1148. auto const& key_region = s_string_list[likely_subtag->key.region];
  1149. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  1150. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  1151. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  1152. if (maximized.language == "und"sv)
  1153. maximized.language = alias_language;
  1154. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  1155. maximized.script = alias_script;
  1156. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  1157. maximized.region = alias_region;
  1158. return maximized;
  1159. }
  1160. Optional<String> resolve_most_likely_territory(LanguageID const& language_id)
  1161. {
  1162. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  1163. return s_string_list[likely_subtag->alias.region];
  1164. return {};
  1165. }
  1166. }
  1167. )~~~");
  1168. TRY(file.write(generator.as_string_view().bytes()));
  1169. return {};
  1170. }
  1171. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1172. {
  1173. StringView generated_header_path;
  1174. StringView generated_implementation_path;
  1175. StringView bcp47_path;
  1176. StringView core_path;
  1177. StringView locale_names_path;
  1178. StringView misc_path;
  1179. StringView numbers_path;
  1180. StringView dates_path;
  1181. Core::ArgsParser args_parser;
  1182. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1183. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1184. args_parser.add_option(bcp47_path, "Path to cldr-bcp47 directory", "bcp47-path", 'b', "bcp47-path");
  1185. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1186. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  1187. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  1188. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  1189. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1190. args_parser.parse(arguments);
  1191. auto generated_header_file = TRY(open_file(generated_header_path, Core::Stream::OpenMode::Write));
  1192. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::Stream::OpenMode::Write));
  1193. UnicodeLocaleData locale_data;
  1194. TRY(parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, dates_path, locale_data));
  1195. TRY(generate_unicode_locale_header(*generated_header_file, locale_data));
  1196. TRY(generate_unicode_locale_implementation(*generated_implementation_file, locale_data));
  1197. return 0;
  1198. }