GenerateLocaleData.cpp 73 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777
  1. /*
  2. * Copyright (c) 2021-2022, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "../LibUnicode/GeneratorUtil.h" // FIXME: Move this somewhere common.
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/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. static String format_identifier(StringView owner, String identifier)
  24. {
  25. identifier = identifier.replace("-"sv, "_"sv, ReplaceMode::All);
  26. if (all_of(identifier, is_ascii_digit))
  27. return String::formatted("{}_{}", owner[0], identifier);
  28. if (is_ascii_lower_alpha(identifier[0]))
  29. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  30. return identifier;
  31. }
  32. struct DisplayPattern {
  33. unsigned hash() const
  34. {
  35. return pair_int_hash(locale_pattern, locale_separator);
  36. }
  37. bool operator==(DisplayPattern const& other) const
  38. {
  39. return (locale_pattern == other.locale_pattern)
  40. && (locale_separator == other.locale_separator);
  41. }
  42. size_t locale_pattern { 0 };
  43. size_t locale_separator { 0 };
  44. };
  45. template<>
  46. struct AK::Formatter<DisplayPattern> : Formatter<FormatString> {
  47. ErrorOr<void> format(FormatBuilder& builder, DisplayPattern const& patterns)
  48. {
  49. return Formatter<FormatString>::format(builder,
  50. "{{ {}, {} }}"sv,
  51. patterns.locale_pattern,
  52. patterns.locale_separator);
  53. }
  54. };
  55. template<>
  56. struct AK::Traits<DisplayPattern> : public GenericTraits<DisplayPattern> {
  57. static unsigned hash(DisplayPattern const& p) { return p.hash(); }
  58. };
  59. struct ListPatterns {
  60. unsigned hash() const
  61. {
  62. auto hash = pair_int_hash(type.hash(), style.hash());
  63. hash = pair_int_hash(hash, start);
  64. hash = pair_int_hash(hash, middle);
  65. hash = pair_int_hash(hash, end);
  66. hash = pair_int_hash(hash, pair);
  67. return hash;
  68. }
  69. bool operator==(ListPatterns const& other) const
  70. {
  71. return (type == other.type)
  72. && (style == other.style)
  73. && (start == other.start)
  74. && (middle == other.middle)
  75. && (end == other.end)
  76. && (pair == other.pair);
  77. }
  78. StringView type;
  79. StringView style;
  80. size_t start { 0 };
  81. size_t middle { 0 };
  82. size_t end { 0 };
  83. size_t pair { 0 };
  84. };
  85. template<>
  86. struct AK::Formatter<ListPatterns> : Formatter<FormatString> {
  87. ErrorOr<void> format(FormatBuilder& builder, ListPatterns const& patterns)
  88. {
  89. return Formatter<FormatString>::format(builder,
  90. "{{ ListPatternType::{}, Style::{}, {}, {}, {}, {} }}"sv,
  91. format_identifier({}, patterns.type),
  92. format_identifier({}, patterns.style),
  93. patterns.start,
  94. patterns.middle,
  95. patterns.end,
  96. patterns.pair);
  97. }
  98. };
  99. template<>
  100. struct AK::Traits<ListPatterns> : public GenericTraits<ListPatterns> {
  101. static unsigned hash(ListPatterns const& p) { return p.hash(); }
  102. };
  103. struct TextLayout {
  104. unsigned hash() const
  105. {
  106. return character_order.hash();
  107. }
  108. bool operator==(TextLayout const& other) const
  109. {
  110. return character_order == other.character_order;
  111. }
  112. StringView character_order;
  113. };
  114. template<>
  115. struct AK::Formatter<TextLayout> : Formatter<FormatString> {
  116. ErrorOr<void> format(FormatBuilder& builder, TextLayout const& patterns)
  117. {
  118. return Formatter<FormatString>::format(builder,
  119. "{{ CharacterOrder::{} }}"sv,
  120. format_identifier({}, patterns.character_order));
  121. }
  122. };
  123. template<>
  124. struct AK::Traits<TextLayout> : public GenericTraits<TextLayout> {
  125. static unsigned hash(TextLayout const& t) { return t.hash(); }
  126. };
  127. using LanguageList = Vector<size_t>;
  128. using TerritoryList = Vector<size_t>;
  129. using ScriptList = Vector<size_t>;
  130. using CurrencyList = Vector<size_t>;
  131. using CalendarList = Vector<size_t>;
  132. using DateFieldList = Vector<size_t>;
  133. using KeywordList = Vector<size_t>;
  134. using ListPatternList = Vector<size_t>;
  135. struct LocaleData {
  136. String language;
  137. Optional<String> territory;
  138. Optional<String> variant;
  139. size_t display_patterns { 0 };
  140. size_t languages { 0 };
  141. size_t territories { 0 };
  142. size_t scripts { 0 };
  143. size_t long_currencies { 0 };
  144. size_t short_currencies { 0 };
  145. size_t narrow_currencies { 0 };
  146. size_t numeric_currencies { 0 };
  147. size_t calendars { 0 };
  148. size_t long_date_fields { 0 };
  149. size_t short_date_fields { 0 };
  150. size_t narrow_date_fields { 0 };
  151. size_t calendar_keywords { 0 };
  152. size_t collation_case_keywords { 0 };
  153. size_t collation_numeric_keywords { 0 };
  154. size_t number_system_keywords { 0 };
  155. size_t list_patterns { 0 };
  156. size_t text_layout { 0 };
  157. };
  158. struct LanguageMapping {
  159. CanonicalLanguageID key {};
  160. CanonicalLanguageID alias {};
  161. };
  162. struct CLDR {
  163. UniqueStringStorage unique_strings;
  164. UniqueStorage<DisplayPattern> unique_display_patterns;
  165. UniqueStorage<LanguageList> unique_language_lists;
  166. UniqueStorage<TerritoryList> unique_territory_lists;
  167. UniqueStorage<ScriptList> unique_script_lists;
  168. UniqueStorage<CurrencyList> unique_currency_lists;
  169. UniqueStorage<CalendarList> unique_calendar_lists;
  170. UniqueStorage<DateFieldList> unique_date_field_lists;
  171. UniqueStorage<KeywordList> unique_keyword_lists;
  172. UniqueStorage<ListPatterns> unique_list_patterns;
  173. UniqueStorage<ListPatternList> unique_list_pattern_lists;
  174. UniqueStorage<TextLayout> unique_text_layouts;
  175. HashMap<String, LocaleData> locales;
  176. Vector<Alias> locale_aliases;
  177. Vector<String> languages;
  178. Vector<String> territories;
  179. Vector<String> scripts;
  180. Vector<String> variants;
  181. Vector<String> currencies;
  182. Vector<String> date_fields;
  183. Vector<Alias> date_field_aliases {
  184. // ECMA-402 and the CLDR refer to some date fields with different names. Defining these aliases
  185. // means we can remain agnostic about the naming differences elsewhere.
  186. { "dayperiod"sv, "dayPeriod"sv },
  187. { "week"sv, "weekOfYear"sv },
  188. { "zone"sv, "timeZoneName"sv },
  189. };
  190. HashMap<String, Vector<String>> keywords;
  191. HashMap<String, Vector<Alias>> keyword_aliases;
  192. HashMap<String, String> keyword_names;
  193. Vector<String> list_pattern_types;
  194. Vector<String> character_orders;
  195. HashMap<String, size_t> language_aliases;
  196. HashMap<String, size_t> territory_aliases;
  197. HashMap<String, size_t> script_aliases;
  198. HashMap<String, size_t> variant_aliases;
  199. HashMap<String, size_t> subdivision_aliases;
  200. Vector<LanguageMapping> complex_mappings;
  201. Vector<LanguageMapping> likely_subtags;
  202. size_t max_variant_size { 0 };
  203. };
  204. // Some parsing is expected to fail. For example, the CLDR contains language mappings
  205. // with locales such as "en-GB-oed" that are canonically invalid locale IDs.
  206. #define TRY_OR_DISCARD(expression) \
  207. ({ \
  208. auto _temporary_result = (expression); \
  209. if (_temporary_result.is_error()) \
  210. return; \
  211. _temporary_result.release_value(); \
  212. })
  213. static ErrorOr<LanguageMapping> parse_language_mapping(CLDR& cldr, StringView key, StringView alias)
  214. {
  215. auto parsed_key = TRY(CanonicalLanguageID::parse(cldr.unique_strings, key));
  216. auto parsed_alias = TRY(CanonicalLanguageID::parse(cldr.unique_strings, alias));
  217. return LanguageMapping { move(parsed_key), move(parsed_alias) };
  218. }
  219. static ErrorOr<void> parse_core_aliases(String core_supplemental_path, CLDR& cldr)
  220. {
  221. LexicalPath core_aliases_path(move(core_supplemental_path));
  222. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  223. auto core_aliases = TRY(read_json_file(core_aliases_path.string()));
  224. auto const& supplemental_object = core_aliases.as_object().get("supplemental"sv);
  225. auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
  226. auto const& alias_object = metadata_object.as_object().get("alias"sv);
  227. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  228. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  229. auto alias = value.as_object().get("_replacement"sv).as_string();
  230. if (key.contains('-')) {
  231. auto mapping = TRY_OR_DISCARD(parse_language_mapping(cldr, key, alias));
  232. cldr.max_variant_size = max(mapping.key.variants.size(), cldr.max_variant_size);
  233. cldr.max_variant_size = max(mapping.alias.variants.size(), cldr.max_variant_size);
  234. cldr.complex_mappings.append(move(mapping));
  235. } else {
  236. alias_map.set(key, cldr.unique_strings.ensure(alias));
  237. }
  238. });
  239. };
  240. append_aliases(alias_object.as_object().get("languageAlias"sv), cldr.language_aliases);
  241. append_aliases(alias_object.as_object().get("territoryAlias"sv), cldr.territory_aliases);
  242. append_aliases(alias_object.as_object().get("scriptAlias"sv), cldr.script_aliases);
  243. append_aliases(alias_object.as_object().get("variantAlias"sv), cldr.variant_aliases);
  244. append_aliases(alias_object.as_object().get("subdivisionAlias"sv), cldr.subdivision_aliases);
  245. return {};
  246. }
  247. static ErrorOr<void> parse_likely_subtags(String core_supplemental_path, CLDR& cldr)
  248. {
  249. LexicalPath likely_subtags_path(move(core_supplemental_path));
  250. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  251. auto likely_subtags = TRY(read_json_file(likely_subtags_path.string()));
  252. auto const& supplemental_object = likely_subtags.as_object().get("supplemental"sv);
  253. auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
  254. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  255. auto mapping = TRY_OR_DISCARD(parse_language_mapping(cldr, key, value.as_string()));
  256. cldr.max_variant_size = max(mapping.key.variants.size(), cldr.max_variant_size);
  257. cldr.max_variant_size = max(mapping.alias.variants.size(), cldr.max_variant_size);
  258. cldr.likely_subtags.append(move(mapping));
  259. });
  260. return {};
  261. }
  262. static ErrorOr<void> parse_identity(String locale_path, CLDR& cldr, LocaleData& locale)
  263. {
  264. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  265. languages_path = languages_path.append("languages.json"sv);
  266. auto languages = TRY(read_json_file(languages_path.string()));
  267. auto const& main_object = languages.as_object().get("main"sv);
  268. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  269. auto const& identity_object = locale_object.as_object().get("identity"sv);
  270. auto const& language_string = identity_object.as_object().get("language"sv);
  271. auto const& territory_string = identity_object.as_object().get("territory"sv);
  272. auto const& script_string = identity_object.as_object().get("script"sv);
  273. auto const& variant_string = identity_object.as_object().get("variant"sv);
  274. locale.language = language_string.as_string();
  275. if (territory_string.is_string()) {
  276. locale.territory = territory_string.as_string();
  277. if (!cldr.territories.contains_slow(*locale.territory))
  278. cldr.territories.append(*locale.territory);
  279. }
  280. if (script_string.is_string()) {
  281. auto script = script_string.as_string();
  282. if (!cldr.scripts.contains_slow(script))
  283. cldr.scripts.append(script);
  284. }
  285. if (variant_string.is_string()) {
  286. locale.variant = variant_string.as_string();
  287. if (!cldr.variants.contains_slow(*locale.variant))
  288. cldr.variants.append(*locale.variant);
  289. }
  290. return {};
  291. }
  292. static ErrorOr<void> parse_locale_display_patterns(String locale_path, CLDR& cldr, LocaleData& locale)
  293. {
  294. LexicalPath locale_display_names_path(move(locale_path));
  295. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  296. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  297. auto const& main_object = locale_display_names.as_object().get("main"sv);
  298. auto const& locale_object = main_object.as_object().get(locale_display_names_path.parent().basename());
  299. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  300. auto const& locale_display_patterns_object = locale_display_names_object.as_object().get("localeDisplayPattern"sv);
  301. auto const& locale_pattern = locale_display_patterns_object.as_object().get("localePattern"sv);
  302. auto const& locale_separator = locale_display_patterns_object.as_object().get("localeSeparator"sv);
  303. DisplayPattern patterns {};
  304. patterns.locale_pattern = cldr.unique_strings.ensure(locale_pattern.as_string());
  305. patterns.locale_separator = cldr.unique_strings.ensure(locale_separator.as_string());
  306. locale.display_patterns = cldr.unique_display_patterns.ensure(move(patterns));
  307. return {};
  308. }
  309. static ErrorOr<void> preprocess_languages(String locale_path, CLDR& cldr)
  310. {
  311. LexicalPath languages_path(move(locale_path));
  312. languages_path = languages_path.append("languages.json"sv);
  313. auto locale_languages = TRY(read_json_file(languages_path.string()));
  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) && !cldr.languages.contains_slow(key))
  320. cldr.languages.append(key);
  321. });
  322. return {};
  323. }
  324. static ErrorOr<void> parse_unicode_extension_keywords(String bcp47_path, CLDR& cldr)
  325. {
  326. constexpr auto desired_keywords = Array { "ca"sv, "co"sv, "hc"sv, "kf"sv, "kn"sv, "nu"sv };
  327. auto keywords = TRY(read_json_file(bcp47_path));
  328. auto const& keyword_object = keywords.as_object().get("keyword"sv);
  329. auto const& unicode_object = keyword_object.as_object().get("u"sv);
  330. if (unicode_object.is_null())
  331. return {};
  332. unicode_object.as_object().for_each_member([&](auto const& key, auto const& value) {
  333. if (!desired_keywords.span().contains_slow(key))
  334. return;
  335. auto const& name = value.as_object().get("_alias"sv);
  336. cldr.keyword_names.set(key, name.as_string());
  337. auto& keywords = cldr.keywords.ensure(key);
  338. // FIXME: ECMA-402 requires the list of supported collation types to include "default", but
  339. // that type does not appear in collation.json.
  340. if (key == "co" && !keywords.contains_slow("default"sv))
  341. keywords.append("default"sv);
  342. value.as_object().for_each_member([&](auto const& keyword, auto const& properties) {
  343. if (!properties.is_object())
  344. return;
  345. // Filter out values not permitted by ECMA-402.
  346. // https://tc39.es/ecma402/#sec-intl-collator-internal-slots
  347. if (key == "co"sv && keyword.is_one_of("search"sv, "standard"sv))
  348. return;
  349. // https://tc39.es/ecma402/#sec-intl.numberformat-internal-slots
  350. if (key == "nu"sv && keyword.is_one_of("finance"sv, "native"sv, "traditio"sv))
  351. return;
  352. if (auto const& preferred = properties.as_object().get("_preferred"sv); preferred.is_string()) {
  353. cldr.keyword_aliases.ensure(key).append({ preferred.as_string(), keyword });
  354. return;
  355. }
  356. if (auto const& alias = properties.as_object().get("_alias"sv); alias.is_string())
  357. cldr.keyword_aliases.ensure(key).append({ keyword, alias.as_string() });
  358. keywords.append(keyword);
  359. });
  360. });
  361. return {};
  362. }
  363. static Optional<String> find_keyword_alias(StringView key, StringView calendar, CLDR& cldr)
  364. {
  365. auto it = cldr.keyword_aliases.find(key);
  366. if (it == cldr.keyword_aliases.end())
  367. return {};
  368. auto alias = it->value.find_if([&](auto const& alias) { return calendar == alias.alias; });
  369. if (alias == it->value.end())
  370. return {};
  371. return alias->name;
  372. }
  373. static ErrorOr<void> parse_locale_languages(String locale_path, CLDR& cldr, LocaleData& locale)
  374. {
  375. LexicalPath languages_path(move(locale_path));
  376. languages_path = languages_path.append("languages.json"sv);
  377. auto locale_languages = TRY(read_json_file(languages_path.string()));
  378. auto const& main_object = locale_languages.as_object().get("main"sv);
  379. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  380. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  381. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  382. LanguageList languages;
  383. languages.resize(cldr.languages.size());
  384. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  385. if (key.contains("-alt-"sv))
  386. return;
  387. auto index = cldr.languages.find_first_index(key).value();
  388. languages[index] = cldr.unique_strings.ensure(value.as_string());
  389. });
  390. locale.languages = cldr.unique_language_lists.ensure(move(languages));
  391. return {};
  392. }
  393. static ErrorOr<void> parse_locale_territories(String locale_path, CLDR& cldr, LocaleData& locale)
  394. {
  395. LexicalPath territories_path(move(locale_path));
  396. territories_path = territories_path.append("territories.json"sv);
  397. auto locale_territories = TRY(read_json_file(territories_path.string()));
  398. auto const& main_object = locale_territories.as_object().get("main"sv);
  399. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  400. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  401. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  402. TerritoryList territories;
  403. territories.resize(cldr.territories.size());
  404. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  405. if (auto index = cldr.territories.find_first_index(key); index.has_value())
  406. territories[*index] = cldr.unique_strings.ensure(value.as_string());
  407. });
  408. locale.territories = cldr.unique_territory_lists.ensure(move(territories));
  409. return {};
  410. }
  411. static ErrorOr<void> parse_locale_scripts(String locale_path, CLDR& cldr, LocaleData& locale)
  412. {
  413. LexicalPath scripts_path(move(locale_path));
  414. scripts_path = scripts_path.append("scripts.json"sv);
  415. auto locale_scripts = TRY(read_json_file(scripts_path.string()));
  416. auto const& main_object = locale_scripts.as_object().get("main"sv);
  417. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  418. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  419. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  420. ScriptList scripts;
  421. scripts.resize(cldr.scripts.size());
  422. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  423. if (auto index = cldr.scripts.find_first_index(key); index.has_value())
  424. scripts[*index] = cldr.unique_strings.ensure(value.as_string());
  425. });
  426. locale.scripts = cldr.unique_script_lists.ensure(move(scripts));
  427. return {};
  428. }
  429. static ErrorOr<void> parse_locale_list_patterns(String misc_path, CLDR& cldr, LocaleData& locale)
  430. {
  431. LexicalPath list_patterns_path(move(misc_path));
  432. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  433. auto locale_list_patterns = TRY(read_json_file(list_patterns_path.string()));
  434. auto const& main_object = locale_list_patterns.as_object().get("main"sv);
  435. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  436. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  437. auto list_pattern_type = [](StringView key) {
  438. if (key.contains("type-standard"sv))
  439. return "conjunction"sv;
  440. if (key.contains("type-or"sv))
  441. return "disjunction"sv;
  442. if (key.contains("type-unit"sv))
  443. return "unit"sv;
  444. VERIFY_NOT_REACHED();
  445. };
  446. auto list_pattern_style = [](StringView key) {
  447. if (key.contains("short"sv))
  448. return "short"sv;
  449. if (key.contains("narrow"sv))
  450. return "narrow"sv;
  451. return "long"sv;
  452. };
  453. ListPatternList list_patterns;
  454. list_patterns.ensure_capacity(list_patterns_object.as_object().size());
  455. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  456. auto type = list_pattern_type(key);
  457. auto style = list_pattern_style(key);
  458. auto start = cldr.unique_strings.ensure(value.as_object().get("start"sv).as_string());
  459. auto middle = cldr.unique_strings.ensure(value.as_object().get("middle"sv).as_string());
  460. auto end = cldr.unique_strings.ensure(value.as_object().get("end"sv).as_string());
  461. auto pair = cldr.unique_strings.ensure(value.as_object().get("2"sv).as_string());
  462. if (!cldr.list_pattern_types.contains_slow(type))
  463. cldr.list_pattern_types.append(type);
  464. ListPatterns list_pattern { type, style, start, middle, end, pair };
  465. list_patterns.append(cldr.unique_list_patterns.ensure(move(list_pattern)));
  466. });
  467. locale.list_patterns = cldr.unique_list_pattern_lists.ensure(move(list_patterns));
  468. return {};
  469. }
  470. static ErrorOr<void> parse_locale_layout(String misc_path, CLDR& cldr, LocaleData& locale)
  471. {
  472. LexicalPath layout_path(move(misc_path));
  473. layout_path = layout_path.append("layout.json"sv);
  474. auto locale_layout = TRY(read_json_file(layout_path.string()));
  475. auto const& main_object = locale_layout.as_object().get("main"sv);
  476. auto const& locale_object = main_object.as_object().get(layout_path.parent().basename());
  477. auto const& layout_object = locale_object.as_object().get("layout"sv);
  478. auto const& orientation_object = layout_object.as_object().get("orientation"sv);
  479. auto text_layout_character_order = [](StringView key) {
  480. if (key == "left-to-right"sv)
  481. return "ltr"sv;
  482. if (key == "right-to-left"sv)
  483. return "rtl"sv;
  484. VERIFY_NOT_REACHED();
  485. };
  486. auto const& character_order_string = orientation_object.as_object().get("characterOrder"sv);
  487. auto const& character_order = character_order_string.as_string();
  488. TextLayout layout {};
  489. layout.character_order = text_layout_character_order(character_order);
  490. if (!cldr.character_orders.contains_slow(layout.character_order))
  491. cldr.character_orders.append(layout.character_order);
  492. locale.text_layout = cldr.unique_text_layouts.ensure(move(layout));
  493. return {};
  494. }
  495. static ErrorOr<void> parse_locale_currencies(String numbers_path, CLDR& cldr, LocaleData& locale)
  496. {
  497. LexicalPath currencies_path(move(numbers_path));
  498. currencies_path = currencies_path.append("currencies.json"sv);
  499. auto locale_currencies = TRY(read_json_file(currencies_path.string()));
  500. auto const& main_object = locale_currencies.as_object().get("main"sv);
  501. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  502. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  503. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  504. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const&) {
  505. if (!cldr.currencies.contains_slow(key))
  506. cldr.currencies.append(key);
  507. });
  508. CurrencyList long_currencies {};
  509. long_currencies.resize(cldr.currencies.size());
  510. CurrencyList short_currencies {};
  511. short_currencies.resize(cldr.currencies.size());
  512. CurrencyList narrow_currencies {};
  513. narrow_currencies.resize(cldr.currencies.size());
  514. CurrencyList numeric_currencies {};
  515. numeric_currencies.resize(cldr.currencies.size());
  516. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  517. auto const& long_name = value.as_object().get("displayName"sv);
  518. auto const& short_name = value.as_object().get("symbol"sv);
  519. auto const& narrow_name = value.as_object().get("symbol-alt-narrow"sv);
  520. auto const& numeric_name = value.as_object().get("displayName-count-other"sv);
  521. auto index = cldr.currencies.find_first_index(key).value();
  522. long_currencies[index] = cldr.unique_strings.ensure(long_name.as_string());
  523. short_currencies[index] = cldr.unique_strings.ensure(short_name.as_string());
  524. narrow_currencies[index] = narrow_name.is_null() ? 0 : cldr.unique_strings.ensure(narrow_name.as_string());
  525. numeric_currencies[index] = cldr.unique_strings.ensure(numeric_name.is_null() ? long_name.as_string() : numeric_name.as_string());
  526. });
  527. locale.long_currencies = cldr.unique_currency_lists.ensure(move(long_currencies));
  528. locale.short_currencies = cldr.unique_currency_lists.ensure(move(short_currencies));
  529. locale.narrow_currencies = cldr.unique_currency_lists.ensure(move(narrow_currencies));
  530. locale.numeric_currencies = cldr.unique_currency_lists.ensure(move(numeric_currencies));
  531. return {};
  532. }
  533. static ErrorOr<void> parse_locale_calendars(String locale_path, CLDR& cldr, LocaleData& locale)
  534. {
  535. LexicalPath locale_display_names_path(move(locale_path));
  536. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  537. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  538. auto const& main_object = locale_display_names.as_object().get("main"sv);
  539. auto const& locale_object = main_object.as_object().get(locale_display_names_path.parent().basename());
  540. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  541. auto const& types_object = locale_display_names_object.as_object().get("types"sv);
  542. auto const& calendar_object = types_object.as_object().get("calendar"sv);
  543. auto const& supported_calendars = cldr.keywords.find("ca"sv)->value;
  544. CalendarList calendars;
  545. calendars.resize(supported_calendars.size());
  546. calendar_object.as_object().for_each_member([&](auto const& key, auto const& calendar) {
  547. auto index = supported_calendars.find_first_index(key);
  548. if (!index.has_value()) {
  549. auto alias = find_keyword_alias("ca"sv, key, cldr);
  550. index = supported_calendars.find_first_index(*alias);
  551. }
  552. calendars[*index] = cldr.unique_strings.ensure(calendar.as_string());
  553. });
  554. locale.calendars = cldr.unique_calendar_lists.ensure(move(calendars));
  555. return {};
  556. }
  557. static ErrorOr<void> parse_locale_date_fields(String dates_path, CLDR& cldr, LocaleData& locale)
  558. {
  559. LexicalPath date_fields_path(move(dates_path));
  560. date_fields_path = date_fields_path.append("dateFields.json"sv);
  561. auto locale_date_fields = TRY(read_json_file(date_fields_path.string()));
  562. auto const& main_object = locale_date_fields.as_object().get("main"sv);
  563. auto const& locale_object = main_object.as_object().get(date_fields_path.parent().basename());
  564. auto const& dates_object = locale_object.as_object().get("dates"sv);
  565. auto const& fields_object = dates_object.as_object().get("fields"sv);
  566. auto is_sanctioned_field = [](StringView field) {
  567. // This is a copy of the units sanctioned for use within ECMA-402, with names adjusted for the names used by the CLDR.
  568. // https://tc39.es/ecma402/#table-validcodeforDateField
  569. 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);
  570. };
  571. fields_object.as_object().for_each_member([&](auto const& key, JsonValue const&) {
  572. if (!is_sanctioned_field(key))
  573. return;
  574. if (!cldr.date_fields.contains_slow(key))
  575. cldr.date_fields.append(key);
  576. });
  577. quick_sort(cldr.date_fields);
  578. DateFieldList long_date_fields {};
  579. long_date_fields.resize(cldr.date_fields.size());
  580. DateFieldList short_date_fields {};
  581. short_date_fields.resize(cldr.date_fields.size());
  582. DateFieldList narrow_date_fields {};
  583. narrow_date_fields.resize(cldr.date_fields.size());
  584. fields_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  585. if (!is_sanctioned_field(key))
  586. return;
  587. auto const& long_name = value.as_object().get("displayName"sv);
  588. auto const& short_name = fields_object.as_object().get(String::formatted("{}-short", key)).as_object().get("displayName"sv);
  589. auto const& narrow_name = fields_object.as_object().get(String::formatted("{}-narrow", key)).as_object().get("displayName"sv);
  590. auto index = cldr.date_fields.find_first_index(key).value();
  591. long_date_fields[index] = cldr.unique_strings.ensure(long_name.as_string());
  592. short_date_fields[index] = cldr.unique_strings.ensure(short_name.as_string());
  593. narrow_date_fields[index] = cldr.unique_strings.ensure(narrow_name.as_string());
  594. });
  595. locale.long_date_fields = cldr.unique_date_field_lists.ensure(move(long_date_fields));
  596. locale.short_date_fields = cldr.unique_date_field_lists.ensure(move(short_date_fields));
  597. locale.narrow_date_fields = cldr.unique_date_field_lists.ensure(move(narrow_date_fields));
  598. return {};
  599. }
  600. static ErrorOr<void> parse_number_system_keywords(String locale_numbers_path, CLDR& cldr, LocaleData& locale)
  601. {
  602. LexicalPath numbers_path(move(locale_numbers_path));
  603. numbers_path = numbers_path.append("numbers.json"sv);
  604. auto numbers = TRY(read_json_file(numbers_path.string()));
  605. auto const& main_object = numbers.as_object().get("main"sv);
  606. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  607. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  608. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  609. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  610. KeywordList keywords {};
  611. auto append_numbering_system = [&](String system_name) {
  612. if (auto system_alias = find_keyword_alias("nu"sv, system_name, cldr); system_alias.has_value())
  613. system_name = system_alias.release_value();
  614. auto index = cldr.unique_strings.ensure(move(system_name));
  615. if (!keywords.contains_slow(index))
  616. keywords.append(move(index));
  617. };
  618. append_numbering_system(default_numbering_system_object.as_string());
  619. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  620. append_numbering_system(value.as_string());
  621. });
  622. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  623. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  624. return;
  625. append_numbering_system(value.as_string());
  626. });
  627. locale.number_system_keywords = cldr.unique_keyword_lists.ensure(move(keywords));
  628. return {};
  629. }
  630. static ErrorOr<void> parse_calendar_keywords(String locale_dates_path, CLDR& cldr, LocaleData& locale)
  631. {
  632. auto calendars_iterator = TRY(path_to_dir_iterator(locale_dates_path, {}));
  633. KeywordList keywords {};
  634. while (calendars_iterator.has_next()) {
  635. auto locale_calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  636. LexicalPath calendars_path(move(locale_calendars_path));
  637. if (!calendars_path.basename().starts_with("ca-"sv))
  638. continue;
  639. auto calendars = TRY(read_json_file(calendars_path.string()));
  640. auto const& main_object = calendars.as_object().get("main"sv);
  641. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  642. auto const& dates_object = locale_object.as_object().get("dates"sv);
  643. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  644. calendars_object.as_object().for_each_member([&](auto calendar_name, JsonValue const&) {
  645. // The generic calendar is not a supported Unicode calendar key, so skip it:
  646. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  647. if (calendar_name == "generic"sv)
  648. return;
  649. if (auto calendar_alias = find_keyword_alias("ca"sv, calendar_name, cldr); calendar_alias.has_value())
  650. calendar_name = calendar_alias.release_value();
  651. keywords.append(cldr.unique_strings.ensure(calendar_name));
  652. });
  653. }
  654. locale.calendar_keywords = cldr.unique_keyword_lists.ensure(move(keywords));
  655. return {};
  656. }
  657. static void fill_in_collation_keywords(CLDR& cldr, LocaleData& locale)
  658. {
  659. // FIXME: If collation data becomes available in the CLDR, parse per-locale ordering from there.
  660. auto create_list_with_default_first = [&](auto key, auto default_value) {
  661. auto& values = cldr.keywords.find(key)->value;
  662. quick_sort(values, [&](auto const& lhs, auto const& rhs) {
  663. if (lhs == default_value)
  664. return true;
  665. if (rhs == default_value)
  666. return false;
  667. return lhs < rhs;
  668. });
  669. KeywordList keywords;
  670. keywords.ensure_capacity(values.size());
  671. for (auto const& value : values)
  672. keywords.append(cldr.unique_strings.ensure(value));
  673. return cldr.unique_keyword_lists.ensure(move(keywords));
  674. };
  675. static auto kf_index = create_list_with_default_first("kf"sv, "upper"sv);
  676. static auto kn_index = create_list_with_default_first("kn"sv, "true"sv);
  677. locale.collation_case_keywords = kf_index;
  678. locale.collation_numeric_keywords = kn_index;
  679. }
  680. static ErrorOr<void> parse_default_content_locales(String core_path, CLDR& cldr)
  681. {
  682. LexicalPath default_content_path(move(core_path));
  683. default_content_path = default_content_path.append("defaultContent.json"sv);
  684. auto default_content = TRY(read_json_file(default_content_path.string()));
  685. auto const& default_content_array = default_content.as_object().get("defaultContent"sv);
  686. default_content_array.as_array().for_each([&](JsonValue const& value) {
  687. auto locale = value.as_string();
  688. StringView default_locale = locale;
  689. while (true) {
  690. if (cldr.locales.contains(default_locale))
  691. break;
  692. auto pos = default_locale.find_last('-');
  693. if (!pos.has_value())
  694. return;
  695. default_locale = default_locale.substring_view(0, *pos);
  696. }
  697. if (default_locale != locale)
  698. cldr.locale_aliases.append({ default_locale, move(locale) });
  699. });
  700. return {};
  701. }
  702. static ErrorOr<void> define_aliases_without_scripts(CLDR& cldr)
  703. {
  704. // From ECMA-402: https://tc39.es/ecma402/#sec-internal-slots
  705. //
  706. // For locales that include a script subtag in addition to language and region, the
  707. // corresponding locale without a script subtag must also be supported.
  708. //
  709. // So we define aliases for locales that contain all three subtags, but we must also take
  710. // care to handle when the locale itself or the locale without a script subtag are an alias
  711. // by way of default-content locales.
  712. auto find_alias = [&](auto const& locale) {
  713. return cldr.locale_aliases.find_if([&](auto const& alias) { return locale == alias.alias; });
  714. };
  715. auto append_alias_without_script = [&](auto const& locale) -> ErrorOr<void> {
  716. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, locale));
  717. if ((parsed_locale.language == 0) || (parsed_locale.script == 0) || (parsed_locale.region == 0))
  718. return {};
  719. auto locale_without_script = String::formatted("{}-{}",
  720. cldr.unique_strings.get(parsed_locale.language),
  721. cldr.unique_strings.get(parsed_locale.region));
  722. if (cldr.locales.contains(locale_without_script))
  723. return {};
  724. if (find_alias(locale_without_script) != cldr.locale_aliases.end())
  725. return {};
  726. if (auto it = find_alias(locale); it != cldr.locale_aliases.end())
  727. cldr.locale_aliases.append({ it->name, locale_without_script });
  728. else
  729. cldr.locale_aliases.append({ locale, locale_without_script });
  730. return {};
  731. };
  732. for (auto const& locale : cldr.locales)
  733. TRY(append_alias_without_script(locale.key));
  734. for (auto const& locale : cldr.locale_aliases)
  735. TRY(append_alias_without_script(locale.alias));
  736. return {};
  737. }
  738. static ErrorOr<void> parse_all_locales(String bcp47_path, String core_path, String locale_names_path, String misc_path, String numbers_path, String dates_path, CLDR& cldr)
  739. {
  740. auto bcp47_iterator = TRY(path_to_dir_iterator(move(bcp47_path), "bcp47"sv));
  741. auto identity_iterator = TRY(path_to_dir_iterator(locale_names_path));
  742. auto preprocess_iterator = TRY(path_to_dir_iterator(locale_names_path));
  743. auto locale_names_iterator = TRY(path_to_dir_iterator(move(locale_names_path)));
  744. auto misc_iterator = TRY(path_to_dir_iterator(move(misc_path)));
  745. auto numbers_iterator = TRY(path_to_dir_iterator(move(numbers_path)));
  746. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  747. LexicalPath core_supplemental_path(core_path);
  748. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  749. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  750. TRY(parse_core_aliases(core_supplemental_path.string(), cldr));
  751. TRY(parse_likely_subtags(core_supplemental_path.string(), cldr));
  752. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  753. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, LexicalPath::basename(path)));
  754. StringBuilder builder;
  755. builder.append(cldr.unique_strings.get(parsed_locale.language));
  756. if (auto script = cldr.unique_strings.get(parsed_locale.script); !script.is_empty())
  757. builder.appendff("-{}", script);
  758. if (auto region = cldr.unique_strings.get(parsed_locale.region); !region.is_empty())
  759. builder.appendff("-{}", region);
  760. return builder.build();
  761. };
  762. while (identity_iterator.has_next()) {
  763. auto locale_path = TRY(next_path_from_dir_iterator(identity_iterator));
  764. auto language = TRY(remove_variants_from_path(locale_path));
  765. auto& locale = cldr.locales.ensure(language);
  766. TRY(parse_identity(locale_path, cldr, locale));
  767. }
  768. while (preprocess_iterator.has_next()) {
  769. auto locale_path = TRY(next_path_from_dir_iterator(preprocess_iterator));
  770. TRY(preprocess_languages(locale_path, cldr));
  771. }
  772. quick_sort(cldr.languages);
  773. quick_sort(cldr.territories);
  774. quick_sort(cldr.scripts);
  775. while (bcp47_iterator.has_next()) {
  776. auto bcp47_path = TRY(next_path_from_dir_iterator(bcp47_iterator));
  777. TRY(parse_unicode_extension_keywords(move(bcp47_path), cldr));
  778. }
  779. while (locale_names_iterator.has_next()) {
  780. auto locale_path = TRY(next_path_from_dir_iterator(locale_names_iterator));
  781. auto language = TRY(remove_variants_from_path(locale_path));
  782. auto& locale = cldr.locales.ensure(language);
  783. TRY(parse_locale_display_patterns(locale_path, cldr, locale));
  784. TRY(parse_locale_languages(locale_path, cldr, locale));
  785. TRY(parse_locale_territories(locale_path, cldr, locale));
  786. TRY(parse_locale_scripts(locale_path, cldr, locale));
  787. TRY(parse_locale_calendars(locale_path, cldr, locale));
  788. }
  789. while (misc_iterator.has_next()) {
  790. auto misc_path = TRY(next_path_from_dir_iterator(misc_iterator));
  791. auto language = TRY(remove_variants_from_path(misc_path));
  792. auto& locale = cldr.locales.ensure(language);
  793. TRY(parse_locale_list_patterns(misc_path, cldr, locale));
  794. TRY(parse_locale_layout(misc_path, cldr, locale));
  795. }
  796. while (numbers_iterator.has_next()) {
  797. auto numbers_path = TRY(next_path_from_dir_iterator(numbers_iterator));
  798. auto language = TRY(remove_variants_from_path(numbers_path));
  799. auto& locale = cldr.locales.ensure(language);
  800. TRY(parse_locale_currencies(numbers_path, cldr, locale));
  801. TRY(parse_number_system_keywords(numbers_path, cldr, locale));
  802. fill_in_collation_keywords(cldr, locale);
  803. }
  804. while (dates_iterator.has_next()) {
  805. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  806. auto language = TRY(remove_variants_from_path(dates_path));
  807. auto& locale = cldr.locales.ensure(language);
  808. TRY(parse_locale_date_fields(dates_path, cldr, locale));
  809. TRY(parse_calendar_keywords(dates_path, cldr, locale));
  810. }
  811. TRY(parse_default_content_locales(move(core_path), cldr));
  812. TRY(define_aliases_without_scripts(cldr));
  813. return {};
  814. }
  815. static ErrorOr<void> generate_unicode_locale_header(Core::Stream::BufferedFile& file, CLDR& cldr)
  816. {
  817. StringBuilder builder;
  818. SourceGenerator generator { builder };
  819. generator.append(R"~~~(
  820. #pragma once
  821. #include <AK/Types.h>
  822. namespace Locale {
  823. )~~~");
  824. auto locales = cldr.locales.keys();
  825. auto keywords = cldr.keywords.keys();
  826. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, cldr.locale_aliases);
  827. generate_enum(generator, format_identifier, "Language"sv, {}, cldr.languages);
  828. generate_enum(generator, format_identifier, "Territory"sv, {}, cldr.territories);
  829. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, cldr.scripts);
  830. generate_enum(generator, format_identifier, "Currency"sv, {}, cldr.currencies);
  831. generate_enum(generator, format_identifier, "DateField"sv, {}, cldr.date_fields, cldr.date_field_aliases);
  832. generate_enum(generator, format_identifier, "Variant"sv, {}, cldr.variants);
  833. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, cldr.list_pattern_types);
  834. generate_enum(generator, format_identifier, "CharacterOrder"sv, {}, cldr.character_orders);
  835. generate_enum(generator, format_identifier, "Key"sv, {}, keywords);
  836. for (auto& keyword : cldr.keywords) {
  837. auto const& keyword_name = cldr.keyword_names.find(keyword.key)->value;
  838. auto enum_name = String::formatted("Keyword{}", format_identifier({}, keyword_name));
  839. if (auto aliases = cldr.keyword_aliases.find(keyword.key); aliases != cldr.keyword_aliases.end())
  840. generate_enum(generator, format_identifier, enum_name, {}, keyword.value, aliases->value);
  841. else
  842. generate_enum(generator, format_identifier, enum_name, {}, keyword.value);
  843. }
  844. generator.append(R"~~~(
  845. }
  846. )~~~");
  847. TRY(file.write(generator.as_string_view().bytes()));
  848. return {};
  849. }
  850. static ErrorOr<void> generate_unicode_locale_implementation(Core::Stream::BufferedFile& file, CLDR& cldr)
  851. {
  852. auto string_index_type = cldr.unique_strings.type_that_fits();
  853. StringBuilder builder;
  854. SourceGenerator generator { builder };
  855. generator.set("string_index_type"sv, string_index_type);
  856. generator.set("locales_size"sv, String::number(cldr.locales.size()));
  857. generator.set("territories_size", String::number(cldr.territories.size()));
  858. generator.set("variants_size", String::number(cldr.max_variant_size));
  859. generator.append(R"~~~(
  860. #include <AK/Array.h>
  861. #include <AK/BinarySearch.h>
  862. #include <AK/Optional.h>
  863. #include <AK/Span.h>
  864. #include <AK/StringView.h>
  865. #include <AK/Vector.h>
  866. #include <LibLocale/DateTimeFormat.h>
  867. #include <LibLocale/Locale.h>
  868. #include <LibLocale/LocaleData.h>
  869. #include <LibUnicode/CurrencyCode.h>
  870. namespace Locale {
  871. )~~~");
  872. cldr.unique_strings.generate(generator);
  873. generator.append(R"~~~(
  874. struct DisplayPatternImpl {
  875. DisplayPattern to_display_pattern() const
  876. {
  877. DisplayPattern display_patterns {};
  878. display_patterns.locale_pattern = decode_string(locale_pattern);
  879. display_patterns.locale_separator = decode_string(locale_separator);
  880. return display_patterns;
  881. }
  882. @string_index_type@ locale_pattern { 0 };
  883. @string_index_type@ locale_separator { 0 };
  884. };
  885. struct Patterns {
  886. ListPatternType type;
  887. Style style;
  888. @string_index_type@ start { 0 };
  889. @string_index_type@ middle { 0 };
  890. @string_index_type@ end { 0 };
  891. @string_index_type@ pair { 0 };
  892. };
  893. struct TextLayout {
  894. CharacterOrder character_order;
  895. };
  896. )~~~");
  897. generate_available_values(generator, "get_available_calendars"sv, cldr.keywords.find("ca"sv)->value, cldr.keyword_aliases.find("ca"sv)->value,
  898. [](auto calendar) {
  899. // FIXME: Remove this filter when we support all calendars.
  900. return calendar.is_one_of("gregory"sv, "iso8601"sv);
  901. });
  902. generate_available_values(generator, "get_available_collation_case_orderings"sv, cldr.keywords.find("kf"sv)->value, cldr.keyword_aliases.find("kf"sv)->value);
  903. generate_available_values(generator, "get_available_collation_numeric_orderings"sv, cldr.keywords.find("kn"sv)->value, cldr.keyword_aliases.find("kn"sv)->value);
  904. generate_available_values(generator, "get_available_collation_types"sv, cldr.keywords.find("co"sv)->value, cldr.keyword_aliases.find("co"sv)->value,
  905. [](auto collation) {
  906. // FIXME: Remove this filter when we support all collation types.
  907. return collation == "default"sv;
  908. });
  909. generate_available_values(generator, "get_available_hour_cycles"sv, cldr.keywords.find("hc"sv)->value);
  910. generate_available_values(generator, "get_available_number_systems"sv, cldr.keywords.find("nu"sv)->value);
  911. generate_available_values(generator, "get_available_currencies"sv, cldr.currencies);
  912. generator.append(R"~~~(
  913. Span<StringView const> get_available_keyword_values(StringView key)
  914. {
  915. auto key_value = key_from_string(key);
  916. if (!key_value.has_value())
  917. return {};
  918. switch (*key_value) {
  919. case Key::Ca:
  920. return get_available_calendars();
  921. case Key::Co:
  922. return get_available_collation_types();
  923. case Key::Hc:
  924. return get_available_hour_cycles();
  925. case Key::Kf:
  926. return get_available_collation_case_orderings();
  927. case Key::Kn:
  928. return get_available_collation_numeric_orderings();
  929. case Key::Nu:
  930. return get_available_number_systems();
  931. }
  932. VERIFY_NOT_REACHED();
  933. }
  934. )~~~");
  935. cldr.unique_display_patterns.generate(generator, "DisplayPatternImpl"sv, "s_display_patterns"sv, 30);
  936. cldr.unique_language_lists.generate(generator, string_index_type, "s_language_lists"sv);
  937. cldr.unique_territory_lists.generate(generator, string_index_type, "s_territory_lists"sv);
  938. cldr.unique_script_lists.generate(generator, string_index_type, "s_script_lists"sv);
  939. cldr.unique_currency_lists.generate(generator, string_index_type, "s_currency_lists"sv);
  940. cldr.unique_calendar_lists.generate(generator, string_index_type, "s_calendar_lists"sv);
  941. cldr.unique_date_field_lists.generate(generator, string_index_type, "s_date_field_lists"sv);
  942. cldr.unique_keyword_lists.generate(generator, string_index_type, "s_keyword_lists"sv);
  943. cldr.unique_list_patterns.generate(generator, "Patterns"sv, "s_list_patterns"sv, 10);
  944. cldr.unique_list_pattern_lists.generate(generator, cldr.unique_list_patterns.type_that_fits(), "s_list_pattern_lists"sv);
  945. cldr.unique_text_layouts.generate(generator, "TextLayout"sv, "s_text_layouts"sv, 30);
  946. auto append_index = [&](auto index) {
  947. generator.append(String::formatted(", {}", index));
  948. };
  949. auto append_list_and_size = [&](auto const& list) {
  950. if (list.is_empty()) {
  951. generator.append(", {}, 0");
  952. return;
  953. }
  954. bool first = true;
  955. generator.append(", {");
  956. for (auto const& item : list) {
  957. generator.append(first ? " "sv : ", "sv);
  958. generator.append(String::number(item));
  959. first = false;
  960. }
  961. generator.append(String::formatted(" }}, {}", list.size()));
  962. };
  963. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  964. generator.set("type", type);
  965. generator.set("name", name);
  966. generator.set("size", String::number(keys.size()));
  967. generator.append(R"~~~(
  968. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  969. bool first = true;
  970. for (auto const& key : keys) {
  971. auto const& value = map.find(key)->value;
  972. auto mapping = mapping_getter(value);
  973. generator.append(first ? " "sv : ", "sv);
  974. generator.append(String::number(mapping));
  975. first = false;
  976. }
  977. generator.append(" } };");
  978. };
  979. auto locales = cldr.locales.keys();
  980. quick_sort(locales);
  981. append_mapping(locales, cldr.locales, cldr.unique_display_patterns.type_that_fits(), "s_locale_display_patterns"sv, [&](auto const& locale) { return locale.display_patterns; });
  982. append_mapping(locales, cldr.locales, cldr.unique_language_lists.type_that_fits(), "s_languages"sv, [&](auto const& locale) { return locale.languages; });
  983. append_mapping(locales, cldr.locales, cldr.unique_territory_lists.type_that_fits(), "s_territories"sv, [&](auto const& locale) { return locale.territories; });
  984. append_mapping(locales, cldr.locales, cldr.unique_script_lists.type_that_fits(), "s_scripts"sv, [&](auto const& locale) { return locale.scripts; });
  985. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_long_currencies"sv, [&](auto const& locale) { return locale.long_currencies; });
  986. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_short_currencies"sv, [&](auto const& locale) { return locale.short_currencies; });
  987. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_narrow_currencies"sv, [&](auto const& locale) { return locale.narrow_currencies; });
  988. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_numeric_currencies"sv, [&](auto const& locale) { return locale.numeric_currencies; });
  989. append_mapping(locales, cldr.locales, cldr.unique_calendar_lists.type_that_fits(), "s_calendars"sv, [&](auto const& locale) { return locale.calendars; });
  990. append_mapping(locales, cldr.locales, cldr.unique_date_field_lists.type_that_fits(), "s_long_date_fields"sv, [&](auto const& locale) { return locale.long_date_fields; });
  991. append_mapping(locales, cldr.locales, cldr.unique_date_field_lists.type_that_fits(), "s_short_date_fields"sv, [&](auto const& locale) { return locale.short_date_fields; });
  992. append_mapping(locales, cldr.locales, cldr.unique_date_field_lists.type_that_fits(), "s_narrow_date_fields"sv, [&](auto const& locale) { return locale.narrow_date_fields; });
  993. append_mapping(locales, cldr.locales, cldr.unique_keyword_lists.type_that_fits(), "s_calendar_keywords"sv, [&](auto const& locale) { return locale.calendar_keywords; });
  994. append_mapping(locales, cldr.locales, cldr.unique_keyword_lists.type_that_fits(), "s_collation_case_keywords"sv, [&](auto const& locale) { return locale.collation_case_keywords; });
  995. append_mapping(locales, cldr.locales, cldr.unique_keyword_lists.type_that_fits(), "s_collation_numeric_keywords"sv, [&](auto const& locale) { return locale.collation_numeric_keywords; });
  996. append_mapping(locales, cldr.locales, cldr.unique_keyword_lists.type_that_fits(), "s_number_system_keywords"sv, [&](auto const& locale) { return locale.number_system_keywords; });
  997. append_mapping(locales, cldr.locales, cldr.unique_list_pattern_lists.type_that_fits(), "s_locale_list_patterns"sv, [&](auto const& locale) { return locale.list_patterns; });
  998. append_mapping(locales, cldr.locales, cldr.unique_text_layouts.type_that_fits(), "s_locale_text_layouts"sv, [&](auto const& locale) { return locale.text_layout; });
  999. generator.append(R"~~~(
  1000. struct CanonicalLanguageID {
  1001. LanguageID to_unicode_language_id() const
  1002. {
  1003. LanguageID language_id {};
  1004. language_id.variants.ensure_capacity(variants_size);
  1005. language_id.language = decode_string(language);
  1006. if (script != 0)
  1007. language_id.script = decode_string(script);
  1008. if (region != 0)
  1009. language_id.region = decode_string(region);
  1010. for (size_t i = 0; i < variants_size; ++i)
  1011. language_id.variants.append(decode_string(variants[i]));
  1012. return language_id;
  1013. }
  1014. bool matches_variants(Vector<String> const& other_variants) const {
  1015. if (variants_size == 0)
  1016. return true;
  1017. if (other_variants.size() != variants_size)
  1018. return false;
  1019. for (size_t i = 0; i < variants_size; ++i) {
  1020. if (decode_string(variants[i]) != other_variants[i])
  1021. return false;
  1022. }
  1023. return true;
  1024. };
  1025. @string_index_type@ language { 0 };
  1026. @string_index_type@ script { 0 };
  1027. @string_index_type@ region { 0 };
  1028. Array<@string_index_type@, @variants_size@> variants {};
  1029. size_t variants_size { 0 };
  1030. };
  1031. struct LanguageMapping {
  1032. CanonicalLanguageID key;
  1033. CanonicalLanguageID alias;
  1034. };
  1035. )~~~");
  1036. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  1037. generator.set("size", String::number(mappings.size()));
  1038. generator.set("name"sv, name);
  1039. generator.append(R"~~~(
  1040. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  1041. )~~~");
  1042. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  1043. auto const& lhs_language = cldr.unique_strings.get(lhs.key.language);
  1044. auto const& rhs_language = cldr.unique_strings.get(rhs.key.language);
  1045. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  1046. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  1047. return false;
  1048. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  1049. return true;
  1050. return lhs_language < rhs_language;
  1051. });
  1052. for (auto const& mapping : mappings) {
  1053. generator.set("language"sv, String::number(mapping.key.language));
  1054. generator.append(" { { @language@");
  1055. append_index(mapping.key.script);
  1056. append_index(mapping.key.region);
  1057. append_list_and_size(mapping.key.variants);
  1058. generator.set("language"sv, String::number(mapping.alias.language));
  1059. generator.append(" }, { @language@");
  1060. append_index(mapping.alias.script);
  1061. append_index(mapping.alias.region);
  1062. append_list_and_size(mapping.alias.variants);
  1063. generator.append(" } },\n");
  1064. }
  1065. generator.append("} };\n");
  1066. };
  1067. append_complex_mapping("complex_alias"sv, cldr.complex_mappings);
  1068. append_complex_mapping("likely_subtags"sv, cldr.likely_subtags);
  1069. generator.append(R"~~~(
  1070. static LanguageMapping const* resolve_likely_subtag(LanguageID const& language_id)
  1071. {
  1072. // https://unicode.org/reports/tr35/#Likely_Subtags
  1073. enum class State {
  1074. LanguageScriptRegion,
  1075. LanguageRegion,
  1076. LanguageScript,
  1077. Language,
  1078. UndScript,
  1079. Done,
  1080. };
  1081. auto state = State::LanguageScriptRegion;
  1082. while (state != State::Done) {
  1083. LanguageID search_key;
  1084. switch (state) {
  1085. case State::LanguageScriptRegion:
  1086. state = State::LanguageRegion;
  1087. if (!language_id.script.has_value() || !language_id.region.has_value())
  1088. continue;
  1089. search_key.language = *language_id.language;
  1090. search_key.script = *language_id.script;
  1091. search_key.region = *language_id.region;
  1092. break;
  1093. case State::LanguageRegion:
  1094. state = State::LanguageScript;
  1095. if (!language_id.region.has_value())
  1096. continue;
  1097. search_key.language = *language_id.language;
  1098. search_key.region = *language_id.region;
  1099. break;
  1100. case State::LanguageScript:
  1101. state = State::Language;
  1102. if (!language_id.script.has_value())
  1103. continue;
  1104. search_key.language = *language_id.language;
  1105. search_key.script = *language_id.script;
  1106. break;
  1107. case State::Language:
  1108. state = State::UndScript;
  1109. search_key.language = *language_id.language;
  1110. break;
  1111. case State::UndScript:
  1112. state = State::Done;
  1113. if (!language_id.script.has_value())
  1114. continue;
  1115. search_key.language = "und"sv;
  1116. search_key.script = *language_id.script;
  1117. break;
  1118. default:
  1119. VERIFY_NOT_REACHED();
  1120. }
  1121. for (auto const& map : s_likely_subtags) {
  1122. auto const& key_language = decode_string(map.key.language);
  1123. auto const& key_script = decode_string(map.key.script);
  1124. auto const& key_region = decode_string(map.key.region);
  1125. if (key_language != search_key.language)
  1126. continue;
  1127. if (!key_script.is_empty() || search_key.script.has_value()) {
  1128. if (key_script != search_key.script)
  1129. continue;
  1130. }
  1131. if (!key_region.is_empty() || search_key.region.has_value()) {
  1132. if (key_region != search_key.region)
  1133. continue;
  1134. }
  1135. return &map;
  1136. }
  1137. }
  1138. return nullptr;
  1139. }
  1140. )~~~");
  1141. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name, StringView unique_list) {
  1142. generator.set("enum_snake", enum_snake);
  1143. generator.set("from_string_name", from_string_name);
  1144. generator.set("collection_name", collection_name);
  1145. generator.set("unique_list", unique_list);
  1146. generator.append(R"~~~(
  1147. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  1148. {
  1149. auto locale_value = locale_from_string(locale);
  1150. if (!locale_value.has_value())
  1151. return {};
  1152. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  1153. if (!@enum_snake@_value.has_value())
  1154. return {};
  1155. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1156. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  1157. auto mapping_index = @collection_name@.at(locale_index);
  1158. auto const& mappings = @unique_list@.at(mapping_index);
  1159. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  1160. auto @enum_snake@_mapping = decode_string(@enum_snake@_string_index);
  1161. if (@enum_snake@_mapping.is_empty())
  1162. return {};
  1163. return @enum_snake@_mapping;
  1164. }
  1165. )~~~");
  1166. };
  1167. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  1168. HashValueMap<String> hashes;
  1169. hashes.ensure_capacity(values.size());
  1170. for (auto const& value : values)
  1171. hashes.set(value.hash(), format_identifier(enum_title, value));
  1172. for (auto const& alias : aliases)
  1173. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1174. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1175. };
  1176. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  1177. HashValueMap<size_t> hashes;
  1178. hashes.ensure_capacity(aliases.size());
  1179. for (auto const& alias : aliases)
  1180. hashes.set(alias.key.hash(), alias.value);
  1181. ValueFromStringOptions options {};
  1182. options.return_type = "StringView"sv;
  1183. options.return_format = "decode_string({})"sv;
  1184. generate_value_from_string(generator, "resolve_{}_alias"sv, string_index_type, enum_snake, move(hashes), options);
  1185. };
  1186. append_from_string("Locale"sv, "locale"sv, cldr.locales.keys(), cldr.locale_aliases);
  1187. append_from_string("Language"sv, "language"sv, cldr.languages);
  1188. append_mapping_search("language"sv, "language"sv, "s_languages"sv, "s_language_lists"sv);
  1189. append_alias_search("language"sv, cldr.language_aliases);
  1190. append_from_string("Territory"sv, "territory"sv, cldr.territories);
  1191. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv, "s_territory_lists"sv);
  1192. append_alias_search("territory"sv, cldr.territory_aliases);
  1193. append_from_string("ScriptTag"sv, "script_tag"sv, cldr.scripts);
  1194. append_mapping_search("script"sv, "script_tag"sv, "s_scripts"sv, "s_script_lists"sv);
  1195. append_alias_search("script_tag"sv, cldr.script_aliases);
  1196. append_from_string("Currency"sv, "currency"sv, cldr.currencies);
  1197. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv, "s_currency_lists"sv);
  1198. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv, "s_currency_lists"sv);
  1199. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv, "s_currency_lists"sv);
  1200. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv, "s_currency_lists"sv);
  1201. append_from_string("DateField"sv, "date_field"sv, cldr.date_fields, cldr.date_field_aliases);
  1202. append_mapping_search("long_date_field"sv, "date_field"sv, "s_long_date_fields"sv, "s_date_field_lists"sv);
  1203. append_mapping_search("short_date_field"sv, "date_field"sv, "s_short_date_fields"sv, "s_date_field_lists"sv);
  1204. append_mapping_search("narrow_date_field"sv, "date_field"sv, "s_narrow_date_fields"sv, "s_date_field_lists"sv);
  1205. append_from_string("Key"sv, "key"sv, cldr.keywords.keys());
  1206. for (auto const& keyword : cldr.keywords) {
  1207. auto const& keyword_name = cldr.keyword_names.find(keyword.key)->value;
  1208. auto enum_name = String::formatted("Keyword{}", format_identifier({}, keyword_name));
  1209. auto enum_snake = String::formatted("keyword_{}", keyword.key);
  1210. if (auto aliases = cldr.keyword_aliases.find(keyword.key); aliases != cldr.keyword_aliases.end())
  1211. append_from_string(enum_name, enum_snake, keyword.value, aliases->value);
  1212. else
  1213. append_from_string(enum_name, enum_snake, keyword.value);
  1214. }
  1215. append_mapping_search("calendar"sv, "keyword_ca"sv, "s_calendars"sv, "s_calendar_lists"sv);
  1216. append_alias_search("variant"sv, cldr.variant_aliases);
  1217. append_alias_search("subdivision"sv, cldr.subdivision_aliases);
  1218. append_from_string("ListPatternType"sv, "list_pattern_type"sv, cldr.list_pattern_types);
  1219. append_from_string("CharacterOrder"sv, "character_order"sv, cldr.character_orders);
  1220. generate_value_to_string(generator, "{}_to_string"sv, "CharacterOrder"sv, "character_order"sv, format_identifier, cldr.character_orders);
  1221. generator.append(R"~~~(
  1222. static Span<@string_index_type@ const> find_keyword_indices(StringView locale, StringView key)
  1223. {
  1224. auto locale_value = locale_from_string(locale);
  1225. if (!locale_value.has_value())
  1226. return {};
  1227. auto key_value = key_from_string(key);
  1228. if (!key_value.has_value())
  1229. return {};
  1230. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1231. size_t keywords_index = 0;
  1232. switch (*key_value) {
  1233. case Key::Ca:
  1234. keywords_index = s_calendar_keywords.at(locale_index);
  1235. break;
  1236. case Key::Kf:
  1237. keywords_index = s_collation_case_keywords.at(locale_index);
  1238. break;
  1239. case Key::Kn:
  1240. keywords_index = s_collation_numeric_keywords.at(locale_index);
  1241. break;
  1242. case Key::Nu:
  1243. keywords_index = s_number_system_keywords.at(locale_index);
  1244. break;
  1245. default:
  1246. VERIFY_NOT_REACHED();
  1247. }
  1248. return s_keyword_lists.at(keywords_index);
  1249. }
  1250. Optional<StringView> get_preferred_keyword_value_for_locale(StringView locale, StringView key)
  1251. {
  1252. // Hour cycle keywords are region-based rather than locale-based, so they need to be handled specially.
  1253. // FIXME: Calendar keywords are also region-based, and will need to be handled here when we support non-Gregorian calendars:
  1254. // https://github.com/unicode-org/cldr-json/blob/main/cldr-json/cldr-core/supplemental/calendarPreferenceData.json
  1255. if (key == "hc"sv) {
  1256. auto hour_cycles = get_locale_hour_cycles(locale);
  1257. if (hour_cycles.is_empty())
  1258. return {};
  1259. return hour_cycle_to_string(hour_cycles[0]);
  1260. }
  1261. // FIXME: Generate locale-preferred collation data when available in the CLDR.
  1262. if (key == "co"sv) {
  1263. auto collations = get_available_collation_types();
  1264. if (collations.is_empty())
  1265. return {};
  1266. return collations[0];
  1267. }
  1268. auto keyword_indices = find_keyword_indices(locale, key);
  1269. if (keyword_indices.is_empty())
  1270. return {};
  1271. return decode_string(keyword_indices[0]);
  1272. }
  1273. Vector<StringView> get_keywords_for_locale(StringView locale, StringView key)
  1274. {
  1275. // Hour cycle keywords are region-based rather than locale-based, so they need to be handled specially.
  1276. // FIXME: Calendar keywords are also region-based, and will need to be handled here when we support non-Gregorian calendars:
  1277. // https://github.com/unicode-org/cldr-json/blob/main/cldr-json/cldr-core/supplemental/calendarPreferenceData.json
  1278. if (key == "hc"sv) {
  1279. auto hour_cycles = get_locale_hour_cycles(locale);
  1280. Vector<StringView> values;
  1281. values.ensure_capacity(hour_cycles.size());
  1282. for (auto hour_cycle : hour_cycles)
  1283. values.unchecked_append(hour_cycle_to_string(hour_cycle));
  1284. return values;
  1285. }
  1286. // FIXME: Generate locale-preferred collation data when available in the CLDR.
  1287. if (key == "co"sv)
  1288. return Vector<StringView> { get_available_collation_types() };
  1289. auto keyword_indices = find_keyword_indices(locale, key);
  1290. Vector<StringView> keywords;
  1291. keywords.ensure_capacity(keyword_indices.size());
  1292. for (auto keyword : keyword_indices)
  1293. keywords.unchecked_append(decode_string(keyword));
  1294. return keywords;
  1295. }
  1296. Optional<DisplayPattern> get_locale_display_patterns(StringView locale)
  1297. {
  1298. auto locale_value = locale_from_string(locale);
  1299. if (!locale_value.has_value())
  1300. return {};
  1301. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1302. auto display_patterns_index = s_locale_display_patterns.at(locale_index);
  1303. auto const& display_patterns = s_display_patterns.at(display_patterns_index);
  1304. return display_patterns.to_display_pattern();
  1305. }
  1306. Optional<ListPatterns> get_locale_list_patterns(StringView locale, StringView list_pattern_type, Style list_pattern_style)
  1307. {
  1308. auto locale_value = locale_from_string(locale);
  1309. if (!locale_value.has_value())
  1310. return {};
  1311. auto type_value = list_pattern_type_from_string(list_pattern_type);
  1312. if (!type_value.has_value())
  1313. return {};
  1314. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1315. auto list_patterns_list_index = s_locale_list_patterns.at(locale_index);
  1316. auto const& locale_list_patterns = s_list_pattern_lists.at(list_patterns_list_index);
  1317. for (auto list_patterns_index : locale_list_patterns) {
  1318. auto const& list_patterns = s_list_patterns.at(list_patterns_index);
  1319. if ((list_patterns.type == type_value) && (list_patterns.style == list_pattern_style)) {
  1320. auto const& start = decode_string(list_patterns.start);
  1321. auto const& middle = decode_string(list_patterns.middle);
  1322. auto const& end = decode_string(list_patterns.end);
  1323. auto const& pair = decode_string(list_patterns.pair);
  1324. return ListPatterns { start, middle, end, pair };
  1325. }
  1326. }
  1327. return {};
  1328. }
  1329. static Optional<TextLayout> text_layout_for_locale(StringView locale)
  1330. {
  1331. auto locale_value = locale_from_string(locale);
  1332. if (!locale_value.has_value())
  1333. return {};
  1334. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1335. auto text_layouts_index = s_locale_text_layouts.at(locale_index);
  1336. return s_text_layouts.at(text_layouts_index);
  1337. }
  1338. Optional<CharacterOrder> character_order_for_locale(StringView locale)
  1339. {
  1340. if (auto text_layout = text_layout_for_locale(locale); text_layout.has_value())
  1341. return text_layout->character_order;
  1342. return {};
  1343. }
  1344. void resolve_complex_language_aliases(LanguageID& language_id)
  1345. {
  1346. for (auto const& map : s_complex_alias) {
  1347. auto const& key_language = decode_string(map.key.language);
  1348. auto const& key_script = decode_string(map.key.script);
  1349. auto const& key_region = decode_string(map.key.region);
  1350. if ((key_language != language_id.language) && (key_language != "und"sv))
  1351. continue;
  1352. if (!key_script.is_empty() && (key_script != language_id.script))
  1353. continue;
  1354. if (!key_region.is_empty() && (key_region != language_id.region))
  1355. continue;
  1356. if (!map.key.matches_variants(language_id.variants))
  1357. continue;
  1358. auto alias = map.alias.to_unicode_language_id();
  1359. if (alias.language == "und"sv)
  1360. alias.language = move(language_id.language);
  1361. if (key_script.is_empty() && !alias.script.has_value())
  1362. alias.script = move(language_id.script);
  1363. if (key_region.is_empty() && !alias.region.has_value())
  1364. alias.region = move(language_id.region);
  1365. if (map.key.variants_size == 0 && alias.variants.is_empty())
  1366. alias.variants = move(language_id.variants);
  1367. language_id = move(alias);
  1368. break;
  1369. }
  1370. }
  1371. Optional<LanguageID> add_likely_subtags(LanguageID const& language_id)
  1372. {
  1373. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  1374. auto const* likely_subtag = resolve_likely_subtag(language_id);
  1375. if (likely_subtag == nullptr)
  1376. return {};
  1377. auto maximized = language_id;
  1378. auto const& key_script = decode_string(likely_subtag->key.script);
  1379. auto const& key_region = decode_string(likely_subtag->key.region);
  1380. auto const& alias_language = decode_string(likely_subtag->alias.language);
  1381. auto const& alias_script = decode_string(likely_subtag->alias.script);
  1382. auto const& alias_region = decode_string(likely_subtag->alias.region);
  1383. if (maximized.language == "und"sv)
  1384. maximized.language = alias_language;
  1385. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  1386. maximized.script = alias_script;
  1387. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  1388. maximized.region = alias_region;
  1389. return maximized;
  1390. }
  1391. Optional<String> resolve_most_likely_territory(LanguageID const& language_id)
  1392. {
  1393. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  1394. return decode_string(likely_subtag->alias.region);
  1395. return {};
  1396. }
  1397. }
  1398. )~~~");
  1399. TRY(file.write(generator.as_string_view().bytes()));
  1400. return {};
  1401. }
  1402. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1403. {
  1404. StringView generated_header_path;
  1405. StringView generated_implementation_path;
  1406. StringView bcp47_path;
  1407. StringView core_path;
  1408. StringView locale_names_path;
  1409. StringView misc_path;
  1410. StringView numbers_path;
  1411. StringView dates_path;
  1412. Core::ArgsParser args_parser;
  1413. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1414. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1415. args_parser.add_option(bcp47_path, "Path to cldr-bcp47 directory", "bcp47-path", 'b', "bcp47-path");
  1416. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1417. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  1418. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  1419. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  1420. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1421. args_parser.parse(arguments);
  1422. auto generated_header_file = TRY(open_file(generated_header_path, Core::Stream::OpenMode::Write));
  1423. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::Stream::OpenMode::Write));
  1424. CLDR cldr;
  1425. TRY(parse_all_locales(bcp47_path, core_path, locale_names_path, misc_path, numbers_path, dates_path, cldr));
  1426. TRY(generate_unicode_locale_header(*generated_header_file, cldr));
  1427. TRY(generate_unicode_locale_implementation(*generated_implementation_file, cldr));
  1428. return 0;
  1429. }