GenerateLocaleData.cpp 75 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786
  1. /*
  2. * Copyright (c) 2021-2022, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "../LibUnicode/GeneratorUtil.h" // FIXME: Move this somewhere common.
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/DeprecatedString.h>
  10. #include <AK/Error.h>
  11. #include <AK/Format.h>
  12. #include <AK/HashMap.h>
  13. #include <AK/JsonObject.h>
  14. #include <AK/JsonParser.h>
  15. #include <AK/JsonValue.h>
  16. #include <AK/LexicalPath.h>
  17. #include <AK/QuickSort.h>
  18. #include <AK/SourceGenerator.h>
  19. #include <AK/StringBuilder.h>
  20. #include <LibCore/ArgsParser.h>
  21. #include <LibCore/DirIterator.h>
  22. #include <LibCore/File.h>
  23. #include <LibCore/Stream.h>
  24. static DeprecatedString format_identifier(StringView owner, DeprecatedString identifier)
  25. {
  26. identifier = identifier.replace("-"sv, "_"sv, ReplaceMode::All);
  27. if (all_of(identifier, is_ascii_digit))
  28. return DeprecatedString::formatted("{}_{}", owner[0], identifier);
  29. if (is_ascii_lower_alpha(identifier[0]))
  30. return DeprecatedString::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  31. return identifier;
  32. }
  33. struct DisplayPattern {
  34. unsigned hash() const
  35. {
  36. return pair_int_hash(locale_pattern, locale_separator);
  37. }
  38. bool operator==(DisplayPattern const& other) const
  39. {
  40. return (locale_pattern == other.locale_pattern)
  41. && (locale_separator == other.locale_separator);
  42. }
  43. size_t locale_pattern { 0 };
  44. size_t locale_separator { 0 };
  45. };
  46. template<>
  47. struct AK::Formatter<DisplayPattern> : Formatter<FormatString> {
  48. ErrorOr<void> format(FormatBuilder& builder, DisplayPattern const& patterns)
  49. {
  50. return Formatter<FormatString>::format(builder,
  51. "{{ {}, {} }}"sv,
  52. patterns.locale_pattern,
  53. patterns.locale_separator);
  54. }
  55. };
  56. template<>
  57. struct AK::Traits<DisplayPattern> : public GenericTraits<DisplayPattern> {
  58. static unsigned hash(DisplayPattern const& p) { return p.hash(); }
  59. };
  60. struct ListPatterns {
  61. unsigned hash() const
  62. {
  63. auto hash = pair_int_hash(type.hash(), style.hash());
  64. hash = pair_int_hash(hash, start);
  65. hash = pair_int_hash(hash, middle);
  66. hash = pair_int_hash(hash, end);
  67. hash = pair_int_hash(hash, pair);
  68. return hash;
  69. }
  70. bool operator==(ListPatterns const& other) const
  71. {
  72. return (type == other.type)
  73. && (style == other.style)
  74. && (start == other.start)
  75. && (middle == other.middle)
  76. && (end == other.end)
  77. && (pair == other.pair);
  78. }
  79. StringView type;
  80. StringView style;
  81. size_t start { 0 };
  82. size_t middle { 0 };
  83. size_t end { 0 };
  84. size_t pair { 0 };
  85. };
  86. template<>
  87. struct AK::Formatter<ListPatterns> : Formatter<FormatString> {
  88. ErrorOr<void> format(FormatBuilder& builder, ListPatterns const& patterns)
  89. {
  90. return Formatter<FormatString>::format(builder,
  91. "{{ ListPatternType::{}, Style::{}, {}, {}, {}, {} }}"sv,
  92. format_identifier({}, patterns.type),
  93. format_identifier({}, patterns.style),
  94. patterns.start,
  95. patterns.middle,
  96. patterns.end,
  97. patterns.pair);
  98. }
  99. };
  100. template<>
  101. struct AK::Traits<ListPatterns> : public GenericTraits<ListPatterns> {
  102. static unsigned hash(ListPatterns const& p) { return p.hash(); }
  103. };
  104. struct TextLayout {
  105. unsigned hash() const
  106. {
  107. return character_order.hash();
  108. }
  109. bool operator==(TextLayout const& other) const
  110. {
  111. return character_order == other.character_order;
  112. }
  113. StringView character_order;
  114. };
  115. template<>
  116. struct AK::Formatter<TextLayout> : Formatter<FormatString> {
  117. ErrorOr<void> format(FormatBuilder& builder, TextLayout const& patterns)
  118. {
  119. return Formatter<FormatString>::format(builder,
  120. "{{ CharacterOrder::{} }}"sv,
  121. format_identifier({}, patterns.character_order));
  122. }
  123. };
  124. template<>
  125. struct AK::Traits<TextLayout> : public GenericTraits<TextLayout> {
  126. static unsigned hash(TextLayout const& t) { return t.hash(); }
  127. };
  128. using LanguageList = Vector<size_t>;
  129. using TerritoryList = Vector<size_t>;
  130. using ScriptList = Vector<size_t>;
  131. using CurrencyList = Vector<size_t>;
  132. using CalendarList = Vector<size_t>;
  133. using DateFieldList = Vector<size_t>;
  134. using KeywordList = Vector<size_t>;
  135. using ListPatternList = Vector<size_t>;
  136. struct LocaleData {
  137. DeprecatedString language;
  138. Optional<DeprecatedString> territory;
  139. Optional<DeprecatedString> variant;
  140. size_t display_patterns { 0 };
  141. size_t languages { 0 };
  142. size_t territories { 0 };
  143. size_t scripts { 0 };
  144. size_t long_currencies { 0 };
  145. size_t short_currencies { 0 };
  146. size_t narrow_currencies { 0 };
  147. size_t numeric_currencies { 0 };
  148. size_t calendars { 0 };
  149. size_t long_date_fields { 0 };
  150. size_t short_date_fields { 0 };
  151. size_t narrow_date_fields { 0 };
  152. size_t calendar_keywords { 0 };
  153. size_t collation_case_keywords { 0 };
  154. size_t collation_numeric_keywords { 0 };
  155. size_t number_system_keywords { 0 };
  156. size_t list_patterns { 0 };
  157. size_t text_layout { 0 };
  158. };
  159. struct LanguageMapping {
  160. CanonicalLanguageID key {};
  161. CanonicalLanguageID alias {};
  162. };
  163. struct CLDR {
  164. UniqueStringStorage unique_strings;
  165. UniqueStorage<DisplayPattern> unique_display_patterns;
  166. UniqueStorage<LanguageList> unique_language_lists;
  167. UniqueStorage<TerritoryList> unique_territory_lists;
  168. UniqueStorage<ScriptList> unique_script_lists;
  169. UniqueStorage<CurrencyList> unique_currency_lists;
  170. UniqueStorage<CalendarList> unique_calendar_lists;
  171. UniqueStorage<DateFieldList> unique_date_field_lists;
  172. UniqueStorage<KeywordList> unique_keyword_lists;
  173. UniqueStorage<ListPatterns> unique_list_patterns;
  174. UniqueStorage<ListPatternList> unique_list_pattern_lists;
  175. UniqueStorage<TextLayout> unique_text_layouts;
  176. HashMap<DeprecatedString, LocaleData> locales;
  177. Vector<Alias> locale_aliases;
  178. Vector<DeprecatedString> languages;
  179. Vector<DeprecatedString> territories;
  180. Vector<DeprecatedString> scripts;
  181. Vector<DeprecatedString> variants;
  182. Vector<DeprecatedString> currencies;
  183. Vector<DeprecatedString> date_fields;
  184. Vector<Alias> date_field_aliases {
  185. // ECMA-402 and the CLDR refer to some date fields with different names. Defining these aliases
  186. // means we can remain agnostic about the naming differences elsewhere.
  187. { "dayperiod"sv, "dayPeriod"sv },
  188. { "week"sv, "weekOfYear"sv },
  189. { "zone"sv, "timeZoneName"sv },
  190. };
  191. HashMap<DeprecatedString, Vector<DeprecatedString>> keywords;
  192. HashMap<DeprecatedString, Vector<Alias>> keyword_aliases;
  193. HashMap<DeprecatedString, DeprecatedString> keyword_names;
  194. Vector<DeprecatedString> list_pattern_types;
  195. Vector<DeprecatedString> character_orders;
  196. HashMap<DeprecatedString, size_t> language_aliases;
  197. HashMap<DeprecatedString, size_t> territory_aliases;
  198. HashMap<DeprecatedString, size_t> script_aliases;
  199. HashMap<DeprecatedString, size_t> variant_aliases;
  200. HashMap<DeprecatedString, size_t> subdivision_aliases;
  201. Vector<LanguageMapping> complex_mappings;
  202. Vector<LanguageMapping> likely_subtags;
  203. size_t max_variant_size { 0 };
  204. };
  205. // Some parsing is expected to fail. For example, the CLDR contains language mappings
  206. // with locales such as "en-GB-oed" that are canonically invalid locale IDs.
  207. #define TRY_OR_DISCARD(expression) \
  208. ({ \
  209. auto _temporary_result = (expression); \
  210. if (_temporary_result.is_error()) \
  211. return; \
  212. static_assert(!::AK::Detail::IsLvalueReference<decltype(_temporary_result.release_value())>, \
  213. "Do not return a reference from a fallible expression"); \
  214. _temporary_result.release_value(); \
  215. })
  216. static ErrorOr<LanguageMapping> parse_language_mapping(CLDR& cldr, StringView key, StringView alias)
  217. {
  218. auto parsed_key = TRY(CanonicalLanguageID::parse(cldr.unique_strings, key));
  219. auto parsed_alias = TRY(CanonicalLanguageID::parse(cldr.unique_strings, alias));
  220. return LanguageMapping { move(parsed_key), move(parsed_alias) };
  221. }
  222. static ErrorOr<void> parse_core_aliases(DeprecatedString core_supplemental_path, CLDR& cldr)
  223. {
  224. LexicalPath core_aliases_path(move(core_supplemental_path));
  225. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  226. auto core_aliases = TRY(read_json_file(core_aliases_path.string()));
  227. auto const& supplemental_object = core_aliases.as_object().get_deprecated("supplemental"sv);
  228. auto const& metadata_object = supplemental_object.as_object().get_deprecated("metadata"sv);
  229. auto const& alias_object = metadata_object.as_object().get_deprecated("alias"sv);
  230. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  231. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  232. auto alias = value.as_object().get_deprecated("_replacement"sv).as_string();
  233. if (key.contains('-')) {
  234. auto mapping = TRY_OR_DISCARD(parse_language_mapping(cldr, key, alias));
  235. cldr.max_variant_size = max(mapping.key.variants.size(), cldr.max_variant_size);
  236. cldr.max_variant_size = max(mapping.alias.variants.size(), cldr.max_variant_size);
  237. cldr.complex_mappings.append(move(mapping));
  238. } else {
  239. alias_map.set(key, cldr.unique_strings.ensure(alias));
  240. }
  241. });
  242. };
  243. append_aliases(alias_object.as_object().get_deprecated("languageAlias"sv), cldr.language_aliases);
  244. append_aliases(alias_object.as_object().get_deprecated("territoryAlias"sv), cldr.territory_aliases);
  245. append_aliases(alias_object.as_object().get_deprecated("scriptAlias"sv), cldr.script_aliases);
  246. append_aliases(alias_object.as_object().get_deprecated("variantAlias"sv), cldr.variant_aliases);
  247. append_aliases(alias_object.as_object().get_deprecated("subdivisionAlias"sv), cldr.subdivision_aliases);
  248. return {};
  249. }
  250. static ErrorOr<void> parse_likely_subtags(DeprecatedString core_supplemental_path, CLDR& cldr)
  251. {
  252. LexicalPath likely_subtags_path(move(core_supplemental_path));
  253. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  254. auto likely_subtags = TRY(read_json_file(likely_subtags_path.string()));
  255. auto const& supplemental_object = likely_subtags.as_object().get_deprecated("supplemental"sv);
  256. auto const& likely_subtags_object = supplemental_object.as_object().get_deprecated("likelySubtags"sv);
  257. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  258. auto mapping = TRY_OR_DISCARD(parse_language_mapping(cldr, key, value.as_string()));
  259. cldr.max_variant_size = max(mapping.key.variants.size(), cldr.max_variant_size);
  260. cldr.max_variant_size = max(mapping.alias.variants.size(), cldr.max_variant_size);
  261. cldr.likely_subtags.append(move(mapping));
  262. });
  263. return {};
  264. }
  265. static ErrorOr<void> parse_identity(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  266. {
  267. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  268. languages_path = languages_path.append("languages.json"sv);
  269. auto languages = TRY(read_json_file(languages_path.string()));
  270. auto const& main_object = languages.as_object().get_deprecated("main"sv);
  271. auto const& locale_object = main_object.as_object().get_deprecated(languages_path.parent().basename());
  272. auto const& identity_object = locale_object.as_object().get_deprecated("identity"sv);
  273. auto const& language_string = identity_object.as_object().get_deprecated("language"sv);
  274. auto const& territory_string = identity_object.as_object().get_deprecated("territory"sv);
  275. auto const& script_string = identity_object.as_object().get_deprecated("script"sv);
  276. auto const& variant_string = identity_object.as_object().get_deprecated("variant"sv);
  277. locale.language = language_string.as_string();
  278. if (territory_string.is_string()) {
  279. locale.territory = territory_string.as_string();
  280. if (!cldr.territories.contains_slow(*locale.territory))
  281. cldr.territories.append(*locale.territory);
  282. }
  283. if (script_string.is_string()) {
  284. auto script = script_string.as_string();
  285. if (!cldr.scripts.contains_slow(script))
  286. cldr.scripts.append(script);
  287. }
  288. if (variant_string.is_string()) {
  289. locale.variant = variant_string.as_string();
  290. if (!cldr.variants.contains_slow(*locale.variant))
  291. cldr.variants.append(*locale.variant);
  292. }
  293. return {};
  294. }
  295. static ErrorOr<void> parse_locale_display_patterns(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  296. {
  297. LexicalPath locale_display_names_path(move(locale_path));
  298. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  299. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  300. auto const& main_object = locale_display_names.as_object().get_deprecated("main"sv);
  301. auto const& locale_object = main_object.as_object().get_deprecated(locale_display_names_path.parent().basename());
  302. auto const& locale_display_names_object = locale_object.as_object().get_deprecated("localeDisplayNames"sv);
  303. auto const& locale_display_patterns_object = locale_display_names_object.as_object().get_deprecated("localeDisplayPattern"sv);
  304. auto const& locale_pattern = locale_display_patterns_object.as_object().get_deprecated("localePattern"sv);
  305. auto const& locale_separator = locale_display_patterns_object.as_object().get_deprecated("localeSeparator"sv);
  306. DisplayPattern patterns {};
  307. patterns.locale_pattern = cldr.unique_strings.ensure(locale_pattern.as_string());
  308. patterns.locale_separator = cldr.unique_strings.ensure(locale_separator.as_string());
  309. locale.display_patterns = cldr.unique_display_patterns.ensure(move(patterns));
  310. return {};
  311. }
  312. static ErrorOr<void> preprocess_languages(DeprecatedString locale_path, CLDR& cldr)
  313. {
  314. LexicalPath languages_path(move(locale_path));
  315. languages_path = languages_path.append("languages.json"sv);
  316. auto locale_languages = TRY(read_json_file(languages_path.string()));
  317. auto const& main_object = locale_languages.as_object().get_deprecated("main"sv);
  318. auto const& locale_object = main_object.as_object().get_deprecated(languages_path.parent().basename());
  319. auto const& locale_display_names_object = locale_object.as_object().get_deprecated("localeDisplayNames"sv);
  320. auto const& languages_object = locale_display_names_object.as_object().get_deprecated("languages"sv);
  321. languages_object.as_object().for_each_member([&](auto const& key, auto const&) {
  322. if (!key.contains("-alt-"sv) && !cldr.languages.contains_slow(key))
  323. cldr.languages.append(key);
  324. });
  325. return {};
  326. }
  327. static ErrorOr<void> parse_unicode_extension_keywords(DeprecatedString bcp47_path, CLDR& cldr)
  328. {
  329. constexpr auto desired_keywords = Array { "ca"sv, "co"sv, "hc"sv, "kf"sv, "kn"sv, "nu"sv };
  330. auto keywords = TRY(read_json_file(bcp47_path));
  331. auto const& keyword_object = keywords.as_object().get_deprecated("keyword"sv);
  332. auto const& unicode_object = keyword_object.as_object().get_deprecated("u"sv);
  333. if (unicode_object.is_null())
  334. return {};
  335. unicode_object.as_object().for_each_member([&](auto const& key, auto const& value) {
  336. if (!desired_keywords.span().contains_slow(key))
  337. return;
  338. auto const& name = value.as_object().get_deprecated("_alias"sv);
  339. cldr.keyword_names.set(key, name.as_string());
  340. auto& keywords = cldr.keywords.ensure(key);
  341. // FIXME: ECMA-402 requires the list of supported collation types to include "default", but
  342. // that type does not appear in collation.json.
  343. if (key == "co" && !keywords.contains_slow("default"sv))
  344. keywords.append("default"sv);
  345. value.as_object().for_each_member([&](auto const& keyword, auto const& properties) {
  346. if (!properties.is_object())
  347. return;
  348. // Filter out values not permitted by ECMA-402.
  349. // https://tc39.es/ecma402/#sec-intl-collator-internal-slots
  350. if (key == "co"sv && keyword.is_one_of("search"sv, "standard"sv))
  351. return;
  352. // https://tc39.es/ecma402/#sec-intl.numberformat-internal-slots
  353. if (key == "nu"sv && keyword.is_one_of("finance"sv, "native"sv, "traditio"sv))
  354. return;
  355. if (auto const& preferred = properties.as_object().get_deprecated("_preferred"sv); preferred.is_string()) {
  356. cldr.keyword_aliases.ensure(key).append({ preferred.as_string(), keyword });
  357. return;
  358. }
  359. if (auto const& alias = properties.as_object().get_deprecated("_alias"sv); alias.is_string())
  360. cldr.keyword_aliases.ensure(key).append({ keyword, alias.as_string() });
  361. keywords.append(keyword);
  362. });
  363. });
  364. return {};
  365. }
  366. static Optional<DeprecatedString> find_keyword_alias(StringView key, StringView calendar, CLDR& cldr)
  367. {
  368. auto it = cldr.keyword_aliases.find(key);
  369. if (it == cldr.keyword_aliases.end())
  370. return {};
  371. auto alias = it->value.find_if([&](auto const& alias) { return calendar == alias.alias; });
  372. if (alias == it->value.end())
  373. return {};
  374. return alias->name;
  375. }
  376. static ErrorOr<void> parse_locale_languages(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  377. {
  378. LexicalPath languages_path(move(locale_path));
  379. languages_path = languages_path.append("languages.json"sv);
  380. auto locale_languages = TRY(read_json_file(languages_path.string()));
  381. auto const& main_object = locale_languages.as_object().get_deprecated("main"sv);
  382. auto const& locale_object = main_object.as_object().get_deprecated(languages_path.parent().basename());
  383. auto const& locale_display_names_object = locale_object.as_object().get_deprecated("localeDisplayNames"sv);
  384. auto const& languages_object = locale_display_names_object.as_object().get_deprecated("languages"sv);
  385. LanguageList languages;
  386. languages.resize(cldr.languages.size());
  387. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  388. if (key.contains("-alt-"sv))
  389. return;
  390. auto index = cldr.languages.find_first_index(key).value();
  391. languages[index] = cldr.unique_strings.ensure(value.as_string());
  392. });
  393. locale.languages = cldr.unique_language_lists.ensure(move(languages));
  394. return {};
  395. }
  396. static ErrorOr<void> parse_locale_territories(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  397. {
  398. LexicalPath territories_path(move(locale_path));
  399. territories_path = territories_path.append("territories.json"sv);
  400. auto locale_territories = TRY(read_json_file(territories_path.string()));
  401. auto const& main_object = locale_territories.as_object().get_deprecated("main"sv);
  402. auto const& locale_object = main_object.as_object().get_deprecated(territories_path.parent().basename());
  403. auto const& locale_display_names_object = locale_object.as_object().get_deprecated("localeDisplayNames"sv);
  404. auto const& territories_object = locale_display_names_object.as_object().get_deprecated("territories"sv);
  405. TerritoryList territories;
  406. territories.resize(cldr.territories.size());
  407. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  408. if (auto index = cldr.territories.find_first_index(key); index.has_value())
  409. territories[*index] = cldr.unique_strings.ensure(value.as_string());
  410. });
  411. locale.territories = cldr.unique_territory_lists.ensure(move(territories));
  412. return {};
  413. }
  414. static ErrorOr<void> parse_locale_scripts(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  415. {
  416. LexicalPath scripts_path(move(locale_path));
  417. scripts_path = scripts_path.append("scripts.json"sv);
  418. auto locale_scripts = TRY(read_json_file(scripts_path.string()));
  419. auto const& main_object = locale_scripts.as_object().get_deprecated("main"sv);
  420. auto const& locale_object = main_object.as_object().get_deprecated(scripts_path.parent().basename());
  421. auto const& locale_display_names_object = locale_object.as_object().get_deprecated("localeDisplayNames"sv);
  422. auto const& scripts_object = locale_display_names_object.as_object().get_deprecated("scripts"sv);
  423. ScriptList scripts;
  424. scripts.resize(cldr.scripts.size());
  425. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  426. if (auto index = cldr.scripts.find_first_index(key); index.has_value())
  427. scripts[*index] = cldr.unique_strings.ensure(value.as_string());
  428. });
  429. locale.scripts = cldr.unique_script_lists.ensure(move(scripts));
  430. return {};
  431. }
  432. static ErrorOr<void> parse_locale_list_patterns(DeprecatedString misc_path, CLDR& cldr, LocaleData& locale)
  433. {
  434. LexicalPath list_patterns_path(move(misc_path));
  435. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  436. auto locale_list_patterns = TRY(read_json_file(list_patterns_path.string()));
  437. auto const& main_object = locale_list_patterns.as_object().get_deprecated("main"sv);
  438. auto const& locale_object = main_object.as_object().get_deprecated(list_patterns_path.parent().basename());
  439. auto const& list_patterns_object = locale_object.as_object().get_deprecated("listPatterns"sv);
  440. auto list_pattern_type = [](StringView key) {
  441. if (key.contains("type-standard"sv))
  442. return "conjunction"sv;
  443. if (key.contains("type-or"sv))
  444. return "disjunction"sv;
  445. if (key.contains("type-unit"sv))
  446. return "unit"sv;
  447. VERIFY_NOT_REACHED();
  448. };
  449. auto list_pattern_style = [](StringView key) {
  450. if (key.contains("short"sv))
  451. return "short"sv;
  452. if (key.contains("narrow"sv))
  453. return "narrow"sv;
  454. return "long"sv;
  455. };
  456. ListPatternList list_patterns;
  457. list_patterns.ensure_capacity(list_patterns_object.as_object().size());
  458. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  459. auto type = list_pattern_type(key);
  460. auto style = list_pattern_style(key);
  461. auto start = cldr.unique_strings.ensure(value.as_object().get_deprecated("start"sv).as_string());
  462. auto middle = cldr.unique_strings.ensure(value.as_object().get_deprecated("middle"sv).as_string());
  463. auto end = cldr.unique_strings.ensure(value.as_object().get_deprecated("end"sv).as_string());
  464. auto pair = cldr.unique_strings.ensure(value.as_object().get_deprecated("2"sv).as_string());
  465. if (!cldr.list_pattern_types.contains_slow(type))
  466. cldr.list_pattern_types.append(type);
  467. ListPatterns list_pattern { type, style, start, middle, end, pair };
  468. list_patterns.append(cldr.unique_list_patterns.ensure(move(list_pattern)));
  469. });
  470. locale.list_patterns = cldr.unique_list_pattern_lists.ensure(move(list_patterns));
  471. return {};
  472. }
  473. static ErrorOr<void> parse_locale_layout(DeprecatedString misc_path, CLDR& cldr, LocaleData& locale)
  474. {
  475. LexicalPath layout_path(move(misc_path));
  476. layout_path = layout_path.append("layout.json"sv);
  477. auto locale_layout = TRY(read_json_file(layout_path.string()));
  478. auto const& main_object = locale_layout.as_object().get_deprecated("main"sv);
  479. auto const& locale_object = main_object.as_object().get_deprecated(layout_path.parent().basename());
  480. auto const& layout_object = locale_object.as_object().get_deprecated("layout"sv);
  481. auto const& orientation_object = layout_object.as_object().get_deprecated("orientation"sv);
  482. auto text_layout_character_order = [](StringView key) {
  483. if (key == "left-to-right"sv)
  484. return "ltr"sv;
  485. if (key == "right-to-left"sv)
  486. return "rtl"sv;
  487. VERIFY_NOT_REACHED();
  488. };
  489. auto const& character_order_string = orientation_object.as_object().get_deprecated("characterOrder"sv);
  490. auto const& character_order = character_order_string.as_string();
  491. TextLayout layout {};
  492. layout.character_order = text_layout_character_order(character_order);
  493. if (!cldr.character_orders.contains_slow(layout.character_order))
  494. cldr.character_orders.append(layout.character_order);
  495. locale.text_layout = cldr.unique_text_layouts.ensure(move(layout));
  496. return {};
  497. }
  498. static ErrorOr<void> parse_locale_currencies(DeprecatedString numbers_path, CLDR& cldr, LocaleData& locale)
  499. {
  500. LexicalPath currencies_path(move(numbers_path));
  501. currencies_path = currencies_path.append("currencies.json"sv);
  502. auto locale_currencies = TRY(read_json_file(currencies_path.string()));
  503. auto const& main_object = locale_currencies.as_object().get_deprecated("main"sv);
  504. auto const& locale_object = main_object.as_object().get_deprecated(currencies_path.parent().basename());
  505. auto const& locale_numbers_object = locale_object.as_object().get_deprecated("numbers"sv);
  506. auto const& currencies_object = locale_numbers_object.as_object().get_deprecated("currencies"sv);
  507. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const&) {
  508. if (!cldr.currencies.contains_slow(key))
  509. cldr.currencies.append(key);
  510. });
  511. CurrencyList long_currencies {};
  512. long_currencies.resize(cldr.currencies.size());
  513. CurrencyList short_currencies {};
  514. short_currencies.resize(cldr.currencies.size());
  515. CurrencyList narrow_currencies {};
  516. narrow_currencies.resize(cldr.currencies.size());
  517. CurrencyList numeric_currencies {};
  518. numeric_currencies.resize(cldr.currencies.size());
  519. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  520. auto const& long_name = value.as_object().get_deprecated("displayName"sv);
  521. auto const& short_name = value.as_object().get_deprecated("symbol"sv);
  522. auto const& narrow_name = value.as_object().get_deprecated("symbol-alt-narrow"sv);
  523. auto const& numeric_name = value.as_object().get_deprecated("displayName-count-other"sv);
  524. auto index = cldr.currencies.find_first_index(key).value();
  525. long_currencies[index] = cldr.unique_strings.ensure(long_name.as_string());
  526. short_currencies[index] = cldr.unique_strings.ensure(short_name.as_string());
  527. narrow_currencies[index] = narrow_name.is_null() ? 0 : cldr.unique_strings.ensure(narrow_name.as_string());
  528. numeric_currencies[index] = cldr.unique_strings.ensure(numeric_name.is_null() ? long_name.as_string() : numeric_name.as_string());
  529. });
  530. locale.long_currencies = cldr.unique_currency_lists.ensure(move(long_currencies));
  531. locale.short_currencies = cldr.unique_currency_lists.ensure(move(short_currencies));
  532. locale.narrow_currencies = cldr.unique_currency_lists.ensure(move(narrow_currencies));
  533. locale.numeric_currencies = cldr.unique_currency_lists.ensure(move(numeric_currencies));
  534. return {};
  535. }
  536. static ErrorOr<void> parse_locale_calendars(DeprecatedString locale_path, CLDR& cldr, LocaleData& locale)
  537. {
  538. LexicalPath locale_display_names_path(move(locale_path));
  539. locale_display_names_path = locale_display_names_path.append("localeDisplayNames.json"sv);
  540. auto locale_display_names = TRY(read_json_file(locale_display_names_path.string()));
  541. auto const& main_object = locale_display_names.as_object().get_deprecated("main"sv);
  542. auto const& locale_object = main_object.as_object().get_deprecated(locale_display_names_path.parent().basename());
  543. auto const& locale_display_names_object = locale_object.as_object().get_deprecated("localeDisplayNames"sv);
  544. auto const& types_object = locale_display_names_object.as_object().get_deprecated("types"sv);
  545. auto const& calendar_object = types_object.as_object().get_deprecated("calendar"sv);
  546. auto const& supported_calendars = cldr.keywords.find("ca"sv)->value;
  547. CalendarList calendars;
  548. calendars.resize(supported_calendars.size());
  549. calendar_object.as_object().for_each_member([&](auto const& key, auto const& calendar) {
  550. auto index = supported_calendars.find_first_index(key);
  551. if (!index.has_value()) {
  552. auto alias = find_keyword_alias("ca"sv, key, cldr);
  553. index = supported_calendars.find_first_index(*alias);
  554. }
  555. calendars[*index] = cldr.unique_strings.ensure(calendar.as_string());
  556. });
  557. locale.calendars = cldr.unique_calendar_lists.ensure(move(calendars));
  558. return {};
  559. }
  560. static ErrorOr<void> parse_locale_date_fields(DeprecatedString dates_path, CLDR& cldr, LocaleData& locale)
  561. {
  562. LexicalPath date_fields_path(move(dates_path));
  563. date_fields_path = date_fields_path.append("dateFields.json"sv);
  564. auto locale_date_fields = TRY(read_json_file(date_fields_path.string()));
  565. auto const& main_object = locale_date_fields.as_object().get_deprecated("main"sv);
  566. auto const& locale_object = main_object.as_object().get_deprecated(date_fields_path.parent().basename());
  567. auto const& dates_object = locale_object.as_object().get_deprecated("dates"sv);
  568. auto const& fields_object = dates_object.as_object().get_deprecated("fields"sv);
  569. auto is_sanctioned_field = [](StringView field) {
  570. // This is a copy of the units sanctioned for use within ECMA-402, with names adjusted for the names used by the CLDR.
  571. // https://tc39.es/ecma402/#table-validcodeforDateField
  572. 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);
  573. };
  574. fields_object.as_object().for_each_member([&](auto const& key, JsonValue const&) {
  575. if (!is_sanctioned_field(key))
  576. return;
  577. if (!cldr.date_fields.contains_slow(key))
  578. cldr.date_fields.append(key);
  579. });
  580. quick_sort(cldr.date_fields);
  581. DateFieldList long_date_fields {};
  582. long_date_fields.resize(cldr.date_fields.size());
  583. DateFieldList short_date_fields {};
  584. short_date_fields.resize(cldr.date_fields.size());
  585. DateFieldList narrow_date_fields {};
  586. narrow_date_fields.resize(cldr.date_fields.size());
  587. fields_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  588. if (!is_sanctioned_field(key))
  589. return;
  590. auto const& long_name = value.as_object().get_deprecated("displayName"sv);
  591. auto const& short_name = fields_object.as_object().get_deprecated(DeprecatedString::formatted("{}-short", key)).as_object().get_deprecated("displayName"sv);
  592. auto const& narrow_name = fields_object.as_object().get_deprecated(DeprecatedString::formatted("{}-narrow", key)).as_object().get_deprecated("displayName"sv);
  593. auto index = cldr.date_fields.find_first_index(key).value();
  594. long_date_fields[index] = cldr.unique_strings.ensure(long_name.as_string());
  595. short_date_fields[index] = cldr.unique_strings.ensure(short_name.as_string());
  596. narrow_date_fields[index] = cldr.unique_strings.ensure(narrow_name.as_string());
  597. });
  598. locale.long_date_fields = cldr.unique_date_field_lists.ensure(move(long_date_fields));
  599. locale.short_date_fields = cldr.unique_date_field_lists.ensure(move(short_date_fields));
  600. locale.narrow_date_fields = cldr.unique_date_field_lists.ensure(move(narrow_date_fields));
  601. return {};
  602. }
  603. static ErrorOr<void> parse_number_system_keywords(DeprecatedString locale_numbers_path, CLDR& cldr, LocaleData& locale)
  604. {
  605. LexicalPath numbers_path(move(locale_numbers_path));
  606. numbers_path = numbers_path.append("numbers.json"sv);
  607. auto numbers = TRY(read_json_file(numbers_path.string()));
  608. auto const& main_object = numbers.as_object().get_deprecated("main"sv);
  609. auto const& locale_object = main_object.as_object().get_deprecated(numbers_path.parent().basename());
  610. auto const& locale_numbers_object = locale_object.as_object().get_deprecated("numbers"sv);
  611. auto const& default_numbering_system_object = locale_numbers_object.as_object().get_deprecated("defaultNumberingSystem"sv);
  612. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get_deprecated("otherNumberingSystems"sv);
  613. KeywordList keywords {};
  614. auto append_numbering_system = [&](DeprecatedString system_name) {
  615. if (auto system_alias = find_keyword_alias("nu"sv, system_name, cldr); system_alias.has_value())
  616. system_name = system_alias.release_value();
  617. auto index = cldr.unique_strings.ensure(move(system_name));
  618. if (!keywords.contains_slow(index))
  619. keywords.append(move(index));
  620. };
  621. append_numbering_system(default_numbering_system_object.as_string());
  622. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  623. append_numbering_system(value.as_string());
  624. });
  625. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  626. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  627. return;
  628. append_numbering_system(value.as_string());
  629. });
  630. locale.number_system_keywords = cldr.unique_keyword_lists.ensure(move(keywords));
  631. return {};
  632. }
  633. static ErrorOr<void> parse_calendar_keywords(DeprecatedString locale_dates_path, CLDR& cldr, LocaleData& locale)
  634. {
  635. auto calendars_iterator = TRY(path_to_dir_iterator(locale_dates_path, {}));
  636. KeywordList keywords {};
  637. while (calendars_iterator.has_next()) {
  638. auto locale_calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  639. LexicalPath calendars_path(move(locale_calendars_path));
  640. if (!calendars_path.basename().starts_with("ca-"sv))
  641. continue;
  642. auto calendars = TRY(read_json_file(calendars_path.string()));
  643. auto const& main_object = calendars.as_object().get_deprecated("main"sv);
  644. auto const& locale_object = main_object.as_object().get_deprecated(calendars_path.parent().basename());
  645. auto const& dates_object = locale_object.as_object().get_deprecated("dates"sv);
  646. auto const& calendars_object = dates_object.as_object().get_deprecated("calendars"sv);
  647. calendars_object.as_object().for_each_member([&](auto calendar_name, JsonValue const&) {
  648. // The generic calendar is not a supported Unicode calendar key, so skip it:
  649. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  650. if (calendar_name == "generic"sv)
  651. return;
  652. if (auto calendar_alias = find_keyword_alias("ca"sv, calendar_name, cldr); calendar_alias.has_value())
  653. calendar_name = calendar_alias.release_value();
  654. keywords.append(cldr.unique_strings.ensure(calendar_name));
  655. });
  656. }
  657. locale.calendar_keywords = cldr.unique_keyword_lists.ensure(move(keywords));
  658. return {};
  659. }
  660. static void fill_in_collation_keywords(CLDR& cldr, LocaleData& locale)
  661. {
  662. // FIXME: If collation data becomes available in the CLDR, parse per-locale ordering from there.
  663. auto create_list_with_default_first = [&](auto key, auto default_value) {
  664. auto& values = cldr.keywords.find(key)->value;
  665. quick_sort(values, [&](auto const& lhs, auto const& rhs) {
  666. if (lhs == default_value)
  667. return true;
  668. if (rhs == default_value)
  669. return false;
  670. return lhs < rhs;
  671. });
  672. KeywordList keywords;
  673. keywords.ensure_capacity(values.size());
  674. for (auto const& value : values)
  675. keywords.append(cldr.unique_strings.ensure(value));
  676. return cldr.unique_keyword_lists.ensure(move(keywords));
  677. };
  678. static auto kf_index = create_list_with_default_first("kf"sv, "upper"sv);
  679. static auto kn_index = create_list_with_default_first("kn"sv, "true"sv);
  680. locale.collation_case_keywords = kf_index;
  681. locale.collation_numeric_keywords = kn_index;
  682. }
  683. static ErrorOr<void> parse_default_content_locales(DeprecatedString core_path, CLDR& cldr)
  684. {
  685. LexicalPath default_content_path(move(core_path));
  686. default_content_path = default_content_path.append("defaultContent.json"sv);
  687. auto default_content = TRY(read_json_file(default_content_path.string()));
  688. auto const& default_content_array = default_content.as_object().get_deprecated("defaultContent"sv);
  689. default_content_array.as_array().for_each([&](JsonValue const& value) {
  690. auto locale = value.as_string();
  691. StringView default_locale = locale;
  692. while (true) {
  693. if (cldr.locales.contains(default_locale))
  694. break;
  695. auto pos = default_locale.find_last('-');
  696. if (!pos.has_value())
  697. return;
  698. default_locale = default_locale.substring_view(0, *pos);
  699. }
  700. if (default_locale != locale)
  701. cldr.locale_aliases.append({ default_locale, move(locale) });
  702. });
  703. return {};
  704. }
  705. static ErrorOr<void> define_aliases_without_scripts(CLDR& cldr)
  706. {
  707. // From ECMA-402: https://tc39.es/ecma402/#sec-internal-slots
  708. //
  709. // For locales that include a script subtag in addition to language and region, the
  710. // corresponding locale without a script subtag must also be supported.
  711. //
  712. // So we define aliases for locales that contain all three subtags, but we must also take
  713. // care to handle when the locale itself or the locale without a script subtag are an alias
  714. // by way of default-content locales.
  715. auto find_alias = [&](auto const& locale) {
  716. return cldr.locale_aliases.find_if([&](auto const& alias) { return locale == alias.alias; });
  717. };
  718. auto append_alias_without_script = [&](auto const& locale) -> ErrorOr<void> {
  719. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, locale));
  720. if ((parsed_locale.language == 0) || (parsed_locale.script == 0) || (parsed_locale.region == 0))
  721. return {};
  722. auto locale_without_script = DeprecatedString::formatted("{}-{}",
  723. cldr.unique_strings.get(parsed_locale.language),
  724. cldr.unique_strings.get(parsed_locale.region));
  725. if (cldr.locales.contains(locale_without_script))
  726. return {};
  727. if (find_alias(locale_without_script) != cldr.locale_aliases.end())
  728. return {};
  729. if (auto it = find_alias(locale); it != cldr.locale_aliases.end())
  730. cldr.locale_aliases.append({ it->name, locale_without_script });
  731. else
  732. cldr.locale_aliases.append({ locale, locale_without_script });
  733. return {};
  734. };
  735. for (auto const& locale : cldr.locales)
  736. TRY(append_alias_without_script(locale.key));
  737. for (auto const& locale : cldr.locale_aliases)
  738. TRY(append_alias_without_script(locale.alias));
  739. return {};
  740. }
  741. static ErrorOr<void> parse_all_locales(DeprecatedString bcp47_path, DeprecatedString core_path, DeprecatedString locale_names_path, DeprecatedString misc_path, DeprecatedString numbers_path, DeprecatedString dates_path, CLDR& cldr)
  742. {
  743. auto bcp47_iterator = TRY(path_to_dir_iterator(move(bcp47_path), "bcp47"sv));
  744. auto identity_iterator = TRY(path_to_dir_iterator(locale_names_path));
  745. auto preprocess_iterator = TRY(path_to_dir_iterator(locale_names_path));
  746. auto locale_names_iterator = TRY(path_to_dir_iterator(move(locale_names_path)));
  747. auto misc_iterator = TRY(path_to_dir_iterator(move(misc_path)));
  748. auto numbers_iterator = TRY(path_to_dir_iterator(move(numbers_path)));
  749. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  750. LexicalPath core_supplemental_path(core_path);
  751. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  752. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  753. TRY(parse_core_aliases(core_supplemental_path.string(), cldr));
  754. TRY(parse_likely_subtags(core_supplemental_path.string(), cldr));
  755. auto remove_variants_from_path = [&](DeprecatedString path) -> ErrorOr<DeprecatedString> {
  756. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, LexicalPath::basename(path)));
  757. StringBuilder builder;
  758. builder.append(cldr.unique_strings.get(parsed_locale.language));
  759. if (auto script = cldr.unique_strings.get(parsed_locale.script); !script.is_empty())
  760. builder.appendff("-{}", script);
  761. if (auto region = cldr.unique_strings.get(parsed_locale.region); !region.is_empty())
  762. builder.appendff("-{}", region);
  763. return builder.build();
  764. };
  765. while (identity_iterator.has_next()) {
  766. auto locale_path = TRY(next_path_from_dir_iterator(identity_iterator));
  767. auto language = TRY(remove_variants_from_path(locale_path));
  768. auto& locale = cldr.locales.ensure(language);
  769. TRY(parse_identity(locale_path, cldr, locale));
  770. }
  771. while (preprocess_iterator.has_next()) {
  772. auto locale_path = TRY(next_path_from_dir_iterator(preprocess_iterator));
  773. TRY(preprocess_languages(locale_path, cldr));
  774. }
  775. quick_sort(cldr.languages);
  776. quick_sort(cldr.territories);
  777. quick_sort(cldr.scripts);
  778. while (bcp47_iterator.has_next()) {
  779. auto bcp47_path = TRY(next_path_from_dir_iterator(bcp47_iterator));
  780. TRY(parse_unicode_extension_keywords(move(bcp47_path), cldr));
  781. }
  782. while (locale_names_iterator.has_next()) {
  783. auto locale_path = TRY(next_path_from_dir_iterator(locale_names_iterator));
  784. auto language = TRY(remove_variants_from_path(locale_path));
  785. auto& locale = cldr.locales.ensure(language);
  786. TRY(parse_locale_display_patterns(locale_path, cldr, locale));
  787. TRY(parse_locale_languages(locale_path, cldr, locale));
  788. TRY(parse_locale_territories(locale_path, cldr, locale));
  789. TRY(parse_locale_scripts(locale_path, cldr, locale));
  790. TRY(parse_locale_calendars(locale_path, cldr, locale));
  791. }
  792. while (misc_iterator.has_next()) {
  793. auto misc_path = TRY(next_path_from_dir_iterator(misc_iterator));
  794. auto language = TRY(remove_variants_from_path(misc_path));
  795. auto& locale = cldr.locales.ensure(language);
  796. TRY(parse_locale_list_patterns(misc_path, cldr, locale));
  797. TRY(parse_locale_layout(misc_path, cldr, locale));
  798. }
  799. while (numbers_iterator.has_next()) {
  800. auto numbers_path = TRY(next_path_from_dir_iterator(numbers_iterator));
  801. auto language = TRY(remove_variants_from_path(numbers_path));
  802. auto& locale = cldr.locales.ensure(language);
  803. TRY(parse_locale_currencies(numbers_path, cldr, locale));
  804. TRY(parse_number_system_keywords(numbers_path, cldr, locale));
  805. fill_in_collation_keywords(cldr, locale);
  806. }
  807. while (dates_iterator.has_next()) {
  808. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  809. auto language = TRY(remove_variants_from_path(dates_path));
  810. auto& locale = cldr.locales.ensure(language);
  811. TRY(parse_locale_date_fields(dates_path, cldr, locale));
  812. TRY(parse_calendar_keywords(dates_path, cldr, locale));
  813. }
  814. TRY(parse_default_content_locales(move(core_path), cldr));
  815. TRY(define_aliases_without_scripts(cldr));
  816. return {};
  817. }
  818. static ErrorOr<void> generate_unicode_locale_header(Core::Stream::BufferedFile& file, CLDR& cldr)
  819. {
  820. StringBuilder builder;
  821. SourceGenerator generator { builder };
  822. generator.append(R"~~~(
  823. #pragma once
  824. #include <AK/Types.h>
  825. namespace Locale {
  826. )~~~");
  827. auto locales = cldr.locales.keys();
  828. auto keywords = cldr.keywords.keys();
  829. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, cldr.locale_aliases);
  830. generate_enum(generator, format_identifier, "Language"sv, {}, cldr.languages);
  831. generate_enum(generator, format_identifier, "Territory"sv, {}, cldr.territories);
  832. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, cldr.scripts);
  833. generate_enum(generator, format_identifier, "Currency"sv, {}, cldr.currencies);
  834. generate_enum(generator, format_identifier, "DateField"sv, {}, cldr.date_fields, cldr.date_field_aliases);
  835. generate_enum(generator, format_identifier, "Variant"sv, {}, cldr.variants);
  836. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, cldr.list_pattern_types);
  837. generate_enum(generator, format_identifier, "CharacterOrder"sv, {}, cldr.character_orders);
  838. generate_enum(generator, format_identifier, "Key"sv, {}, keywords);
  839. for (auto& keyword : cldr.keywords) {
  840. auto const& keyword_name = cldr.keyword_names.find(keyword.key)->value;
  841. auto enum_name = DeprecatedString::formatted("Keyword{}", format_identifier({}, keyword_name));
  842. if (auto aliases = cldr.keyword_aliases.find(keyword.key); aliases != cldr.keyword_aliases.end())
  843. generate_enum(generator, format_identifier, enum_name, {}, keyword.value, aliases->value);
  844. else
  845. generate_enum(generator, format_identifier, enum_name, {}, keyword.value);
  846. }
  847. generator.append(R"~~~(
  848. }
  849. )~~~");
  850. TRY(file.write(generator.as_string_view().bytes()));
  851. return {};
  852. }
  853. static ErrorOr<void> generate_unicode_locale_implementation(Core::Stream::BufferedFile& file, CLDR& cldr)
  854. {
  855. auto string_index_type = cldr.unique_strings.type_that_fits();
  856. StringBuilder builder;
  857. SourceGenerator generator { builder };
  858. generator.set("string_index_type"sv, string_index_type);
  859. generator.set("locales_size"sv, DeprecatedString::number(cldr.locales.size()));
  860. generator.set("territories_size", DeprecatedString::number(cldr.territories.size()));
  861. generator.set("variants_size", DeprecatedString::number(cldr.max_variant_size));
  862. generator.append(R"~~~(
  863. #include <AK/Array.h>
  864. #include <AK/BinarySearch.h>
  865. #include <AK/Optional.h>
  866. #include <AK/Span.h>
  867. #include <AK/String.h>
  868. #include <AK/StringView.h>
  869. #include <AK/Vector.h>
  870. #include <LibLocale/DateTimeFormat.h>
  871. #include <LibLocale/Locale.h>
  872. #include <LibLocale/LocaleData.h>
  873. #include <LibUnicode/CurrencyCode.h>
  874. namespace Locale {
  875. )~~~");
  876. cldr.unique_strings.generate(generator);
  877. generator.append(R"~~~(
  878. struct DisplayPatternImpl {
  879. DisplayPattern to_display_pattern() const
  880. {
  881. DisplayPattern display_patterns {};
  882. display_patterns.locale_pattern = decode_string(locale_pattern);
  883. display_patterns.locale_separator = decode_string(locale_separator);
  884. return display_patterns;
  885. }
  886. @string_index_type@ locale_pattern { 0 };
  887. @string_index_type@ locale_separator { 0 };
  888. };
  889. struct Patterns {
  890. ListPatternType type;
  891. Style style;
  892. @string_index_type@ start { 0 };
  893. @string_index_type@ middle { 0 };
  894. @string_index_type@ end { 0 };
  895. @string_index_type@ pair { 0 };
  896. };
  897. struct TextLayout {
  898. CharacterOrder character_order;
  899. };
  900. )~~~");
  901. generate_available_values(generator, "get_available_calendars"sv, cldr.keywords.find("ca"sv)->value, cldr.keyword_aliases.find("ca"sv)->value,
  902. [](auto calendar) {
  903. // FIXME: Remove this filter when we support all calendars.
  904. return calendar.is_one_of("gregory"sv, "iso8601"sv);
  905. });
  906. generate_available_values(generator, "get_available_collation_case_orderings"sv, cldr.keywords.find("kf"sv)->value, cldr.keyword_aliases.find("kf"sv)->value);
  907. generate_available_values(generator, "get_available_collation_numeric_orderings"sv, cldr.keywords.find("kn"sv)->value, cldr.keyword_aliases.find("kn"sv)->value);
  908. generate_available_values(generator, "get_available_collation_types"sv, cldr.keywords.find("co"sv)->value, cldr.keyword_aliases.find("co"sv)->value,
  909. [](auto collation) {
  910. // FIXME: Remove this filter when we support all collation types.
  911. return collation == "default"sv;
  912. });
  913. generate_available_values(generator, "get_available_hour_cycles"sv, cldr.keywords.find("hc"sv)->value);
  914. generate_available_values(generator, "get_available_number_systems"sv, cldr.keywords.find("nu"sv)->value);
  915. generate_available_values(generator, "get_available_currencies"sv, cldr.currencies);
  916. generator.append(R"~~~(
  917. Span<StringView const> get_available_keyword_values(StringView key)
  918. {
  919. auto key_value = key_from_string(key);
  920. if (!key_value.has_value())
  921. return {};
  922. switch (*key_value) {
  923. case Key::Ca:
  924. return get_available_calendars();
  925. case Key::Co:
  926. return get_available_collation_types();
  927. case Key::Hc:
  928. return get_available_hour_cycles();
  929. case Key::Kf:
  930. return get_available_collation_case_orderings();
  931. case Key::Kn:
  932. return get_available_collation_numeric_orderings();
  933. case Key::Nu:
  934. return get_available_number_systems();
  935. }
  936. VERIFY_NOT_REACHED();
  937. }
  938. )~~~");
  939. cldr.unique_display_patterns.generate(generator, "DisplayPatternImpl"sv, "s_display_patterns"sv, 30);
  940. cldr.unique_language_lists.generate(generator, string_index_type, "s_language_lists"sv);
  941. cldr.unique_territory_lists.generate(generator, string_index_type, "s_territory_lists"sv);
  942. cldr.unique_script_lists.generate(generator, string_index_type, "s_script_lists"sv);
  943. cldr.unique_currency_lists.generate(generator, string_index_type, "s_currency_lists"sv);
  944. cldr.unique_calendar_lists.generate(generator, string_index_type, "s_calendar_lists"sv);
  945. cldr.unique_date_field_lists.generate(generator, string_index_type, "s_date_field_lists"sv);
  946. cldr.unique_keyword_lists.generate(generator, string_index_type, "s_keyword_lists"sv);
  947. cldr.unique_list_patterns.generate(generator, "Patterns"sv, "s_list_patterns"sv, 10);
  948. cldr.unique_list_pattern_lists.generate(generator, cldr.unique_list_patterns.type_that_fits(), "s_list_pattern_lists"sv);
  949. cldr.unique_text_layouts.generate(generator, "TextLayout"sv, "s_text_layouts"sv, 30);
  950. auto append_index = [&](auto index) {
  951. generator.append(DeprecatedString::formatted(", {}", index));
  952. };
  953. auto append_list_and_size = [&](auto const& list) {
  954. if (list.is_empty()) {
  955. generator.append(", {}, 0");
  956. return;
  957. }
  958. bool first = true;
  959. generator.append(", {");
  960. for (auto const& item : list) {
  961. generator.append(first ? " "sv : ", "sv);
  962. generator.append(DeprecatedString::number(item));
  963. first = false;
  964. }
  965. generator.append(DeprecatedString::formatted(" }}, {}", list.size()));
  966. };
  967. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  968. generator.set("type", type);
  969. generator.set("name", name);
  970. generator.set("size", DeprecatedString::number(keys.size()));
  971. generator.append(R"~~~(
  972. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  973. bool first = true;
  974. for (auto const& key : keys) {
  975. auto const& value = map.find(key)->value;
  976. auto mapping = mapping_getter(value);
  977. generator.append(first ? " "sv : ", "sv);
  978. generator.append(DeprecatedString::number(mapping));
  979. first = false;
  980. }
  981. generator.append(" } };");
  982. };
  983. auto locales = cldr.locales.keys();
  984. quick_sort(locales);
  985. append_mapping(locales, cldr.locales, cldr.unique_display_patterns.type_that_fits(), "s_locale_display_patterns"sv, [&](auto const& locale) { return locale.display_patterns; });
  986. append_mapping(locales, cldr.locales, cldr.unique_language_lists.type_that_fits(), "s_languages"sv, [&](auto const& locale) { return locale.languages; });
  987. append_mapping(locales, cldr.locales, cldr.unique_territory_lists.type_that_fits(), "s_territories"sv, [&](auto const& locale) { return locale.territories; });
  988. append_mapping(locales, cldr.locales, cldr.unique_script_lists.type_that_fits(), "s_scripts"sv, [&](auto const& locale) { return locale.scripts; });
  989. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_long_currencies"sv, [&](auto const& locale) { return locale.long_currencies; });
  990. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_short_currencies"sv, [&](auto const& locale) { return locale.short_currencies; });
  991. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_narrow_currencies"sv, [&](auto const& locale) { return locale.narrow_currencies; });
  992. append_mapping(locales, cldr.locales, cldr.unique_currency_lists.type_that_fits(), "s_numeric_currencies"sv, [&](auto const& locale) { return locale.numeric_currencies; });
  993. append_mapping(locales, cldr.locales, cldr.unique_calendar_lists.type_that_fits(), "s_calendars"sv, [&](auto const& locale) { return locale.calendars; });
  994. 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; });
  995. 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; });
  996. 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; });
  997. append_mapping(locales, cldr.locales, cldr.unique_keyword_lists.type_that_fits(), "s_calendar_keywords"sv, [&](auto const& locale) { return locale.calendar_keywords; });
  998. 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; });
  999. 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; });
  1000. 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; });
  1001. 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; });
  1002. append_mapping(locales, cldr.locales, cldr.unique_text_layouts.type_that_fits(), "s_locale_text_layouts"sv, [&](auto const& locale) { return locale.text_layout; });
  1003. generator.append(R"~~~(
  1004. struct CanonicalLanguageID
  1005. {
  1006. ErrorOr<LanguageID> to_unicode_language_id() const
  1007. {
  1008. LanguageID language_id {};
  1009. TRY(language_id.variants.try_ensure_capacity(variants_size));
  1010. language_id.language = TRY(String::from_utf8(decode_string(language)));
  1011. if (script != 0)
  1012. language_id.script = TRY(String::from_utf8(decode_string(script)));
  1013. if (region != 0)
  1014. language_id.region = TRY(String::from_utf8(decode_string(region)));
  1015. for (size_t i = 0; i < variants_size; ++i)
  1016. language_id.variants.append(TRY(String::from_utf8(decode_string(variants[i]))));
  1017. return language_id;
  1018. }
  1019. bool matches_variants(Vector<String> const& other_variants) const {
  1020. if (variants_size == 0)
  1021. return true;
  1022. if (other_variants.size() != variants_size)
  1023. return false;
  1024. for (size_t i = 0; i < variants_size; ++i) {
  1025. if (decode_string(variants[i]) != other_variants[i])
  1026. return false;
  1027. }
  1028. return true;
  1029. };
  1030. @string_index_type@ language { 0 };
  1031. @string_index_type@ script { 0 };
  1032. @string_index_type@ region { 0 };
  1033. Array<@string_index_type@, @variants_size@> variants {};
  1034. size_t variants_size { 0 };
  1035. };
  1036. struct LanguageMapping {
  1037. CanonicalLanguageID key;
  1038. CanonicalLanguageID alias;
  1039. };
  1040. )~~~");
  1041. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  1042. generator.set("size", DeprecatedString::number(mappings.size()));
  1043. generator.set("name"sv, name);
  1044. generator.append(R"~~~(
  1045. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  1046. )~~~");
  1047. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  1048. auto const& lhs_language = cldr.unique_strings.get(lhs.key.language);
  1049. auto const& rhs_language = cldr.unique_strings.get(rhs.key.language);
  1050. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  1051. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  1052. return false;
  1053. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  1054. return true;
  1055. return lhs_language < rhs_language;
  1056. });
  1057. for (auto const& mapping : mappings) {
  1058. generator.set("language"sv, DeprecatedString::number(mapping.key.language));
  1059. generator.append(" { { @language@");
  1060. append_index(mapping.key.script);
  1061. append_index(mapping.key.region);
  1062. append_list_and_size(mapping.key.variants);
  1063. generator.set("language"sv, DeprecatedString::number(mapping.alias.language));
  1064. generator.append(" }, { @language@");
  1065. append_index(mapping.alias.script);
  1066. append_index(mapping.alias.region);
  1067. append_list_and_size(mapping.alias.variants);
  1068. generator.append(" } },\n");
  1069. }
  1070. generator.append("} };\n");
  1071. };
  1072. append_complex_mapping("complex_alias"sv, cldr.complex_mappings);
  1073. append_complex_mapping("likely_subtags"sv, cldr.likely_subtags);
  1074. generator.append(R"~~~(
  1075. static LanguageMapping const* resolve_likely_subtag(LanguageID const& language_id)
  1076. {
  1077. // https://unicode.org/reports/tr35/#Likely_Subtags
  1078. enum class State {
  1079. LanguageScriptRegion,
  1080. LanguageRegion,
  1081. LanguageScript,
  1082. Language,
  1083. UndScript,
  1084. Done,
  1085. };
  1086. auto state = State::LanguageScriptRegion;
  1087. while (state != State::Done) {
  1088. LanguageID search_key;
  1089. switch (state) {
  1090. case State::LanguageScriptRegion:
  1091. state = State::LanguageRegion;
  1092. if (!language_id.script.has_value() || !language_id.region.has_value())
  1093. continue;
  1094. search_key.language = *language_id.language;
  1095. search_key.script = *language_id.script;
  1096. search_key.region = *language_id.region;
  1097. break;
  1098. case State::LanguageRegion:
  1099. state = State::LanguageScript;
  1100. if (!language_id.region.has_value())
  1101. continue;
  1102. search_key.language = *language_id.language;
  1103. search_key.region = *language_id.region;
  1104. break;
  1105. case State::LanguageScript:
  1106. state = State::Language;
  1107. if (!language_id.script.has_value())
  1108. continue;
  1109. search_key.language = *language_id.language;
  1110. search_key.script = *language_id.script;
  1111. break;
  1112. case State::Language:
  1113. state = State::UndScript;
  1114. search_key.language = *language_id.language;
  1115. break;
  1116. case State::UndScript:
  1117. state = State::Done;
  1118. if (!language_id.script.has_value())
  1119. continue;
  1120. search_key.language = String::from_utf8_short_string("und"sv);
  1121. search_key.script = *language_id.script;
  1122. break;
  1123. default:
  1124. VERIFY_NOT_REACHED();
  1125. }
  1126. for (auto const& map : s_likely_subtags) {
  1127. auto const& key_language = decode_string(map.key.language);
  1128. auto const& key_script = decode_string(map.key.script);
  1129. auto const& key_region = decode_string(map.key.region);
  1130. if (key_language != search_key.language)
  1131. continue;
  1132. if (!key_script.is_empty() || search_key.script.has_value()) {
  1133. if (key_script != search_key.script)
  1134. continue;
  1135. }
  1136. if (!key_region.is_empty() || search_key.region.has_value()) {
  1137. if (key_region != search_key.region)
  1138. continue;
  1139. }
  1140. return &map;
  1141. }
  1142. }
  1143. return nullptr;
  1144. }
  1145. )~~~");
  1146. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name, StringView unique_list) {
  1147. generator.set("enum_snake", enum_snake);
  1148. generator.set("from_string_name", from_string_name);
  1149. generator.set("collection_name", collection_name);
  1150. generator.set("unique_list", unique_list);
  1151. generator.append(R"~~~(
  1152. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  1153. {
  1154. auto locale_value = locale_from_string(locale);
  1155. if (!locale_value.has_value())
  1156. return {};
  1157. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  1158. if (!@enum_snake@_value.has_value())
  1159. return {};
  1160. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1161. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  1162. auto mapping_index = @collection_name@.at(locale_index);
  1163. auto const& mappings = @unique_list@.at(mapping_index);
  1164. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  1165. auto @enum_snake@_mapping = decode_string(@enum_snake@_string_index);
  1166. if (@enum_snake@_mapping.is_empty())
  1167. return {};
  1168. return @enum_snake@_mapping;
  1169. }
  1170. )~~~");
  1171. };
  1172. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) -> ErrorOr<void> {
  1173. HashValueMap<DeprecatedString> hashes;
  1174. TRY(hashes.try_ensure_capacity(values.size()));
  1175. for (auto const& value : values)
  1176. hashes.set(value.hash(), format_identifier(enum_title, value));
  1177. for (auto const& alias : aliases)
  1178. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  1179. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1180. return {};
  1181. };
  1182. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) -> ErrorOr<void> {
  1183. HashValueMap<size_t> hashes;
  1184. TRY(hashes.try_ensure_capacity(aliases.size()));
  1185. for (auto const& alias : aliases)
  1186. hashes.set(alias.key.hash(), alias.value);
  1187. ValueFromStringOptions options {};
  1188. options.return_type = "StringView"sv;
  1189. options.return_format = "decode_string({})"sv;
  1190. generate_value_from_string(generator, "resolve_{}_alias"sv, string_index_type, enum_snake, move(hashes), options);
  1191. return {};
  1192. };
  1193. TRY(append_from_string("Locale"sv, "locale"sv, cldr.locales.keys(), cldr.locale_aliases));
  1194. TRY(append_from_string("Language"sv, "language"sv, cldr.languages));
  1195. append_mapping_search("language"sv, "language"sv, "s_languages"sv, "s_language_lists"sv);
  1196. TRY(append_alias_search("language"sv, cldr.language_aliases));
  1197. TRY(append_from_string("Territory"sv, "territory"sv, cldr.territories));
  1198. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv, "s_territory_lists"sv);
  1199. TRY(append_alias_search("territory"sv, cldr.territory_aliases));
  1200. TRY(append_from_string("ScriptTag"sv, "script_tag"sv, cldr.scripts));
  1201. append_mapping_search("script"sv, "script_tag"sv, "s_scripts"sv, "s_script_lists"sv);
  1202. TRY(append_alias_search("script_tag"sv, cldr.script_aliases));
  1203. TRY(append_from_string("Currency"sv, "currency"sv, cldr.currencies));
  1204. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv, "s_currency_lists"sv);
  1205. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv, "s_currency_lists"sv);
  1206. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv, "s_currency_lists"sv);
  1207. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv, "s_currency_lists"sv);
  1208. TRY(append_from_string("DateField"sv, "date_field"sv, cldr.date_fields, cldr.date_field_aliases));
  1209. append_mapping_search("long_date_field"sv, "date_field"sv, "s_long_date_fields"sv, "s_date_field_lists"sv);
  1210. append_mapping_search("short_date_field"sv, "date_field"sv, "s_short_date_fields"sv, "s_date_field_lists"sv);
  1211. append_mapping_search("narrow_date_field"sv, "date_field"sv, "s_narrow_date_fields"sv, "s_date_field_lists"sv);
  1212. TRY(append_from_string("Key"sv, "key"sv, cldr.keywords.keys()));
  1213. for (auto const& keyword : cldr.keywords) {
  1214. auto const& keyword_name = cldr.keyword_names.find(keyword.key)->value;
  1215. auto enum_name = DeprecatedString::formatted("Keyword{}", format_identifier({}, keyword_name));
  1216. auto enum_snake = DeprecatedString::formatted("keyword_{}", keyword.key);
  1217. if (auto aliases = cldr.keyword_aliases.find(keyword.key); aliases != cldr.keyword_aliases.end())
  1218. TRY(append_from_string(enum_name, enum_snake, keyword.value, aliases->value));
  1219. else
  1220. TRY(append_from_string(enum_name, enum_snake, keyword.value));
  1221. }
  1222. append_mapping_search("calendar"sv, "keyword_ca"sv, "s_calendars"sv, "s_calendar_lists"sv);
  1223. TRY(append_alias_search("variant"sv, cldr.variant_aliases));
  1224. TRY(append_alias_search("subdivision"sv, cldr.subdivision_aliases));
  1225. TRY(append_from_string("ListPatternType"sv, "list_pattern_type"sv, cldr.list_pattern_types));
  1226. TRY(append_from_string("CharacterOrder"sv, "character_order"sv, cldr.character_orders));
  1227. generate_value_to_string(generator, "{}_to_string"sv, "CharacterOrder"sv, "character_order"sv, format_identifier, cldr.character_orders);
  1228. generator.append(R"~~~(
  1229. static Span<@string_index_type@ const> find_keyword_indices(StringView locale, StringView key)
  1230. {
  1231. auto locale_value = locale_from_string(locale);
  1232. if (!locale_value.has_value())
  1233. return {};
  1234. auto key_value = key_from_string(key);
  1235. if (!key_value.has_value())
  1236. return {};
  1237. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1238. size_t keywords_index = 0;
  1239. switch (*key_value) {
  1240. case Key::Ca:
  1241. keywords_index = s_calendar_keywords.at(locale_index);
  1242. break;
  1243. case Key::Kf:
  1244. keywords_index = s_collation_case_keywords.at(locale_index);
  1245. break;
  1246. case Key::Kn:
  1247. keywords_index = s_collation_numeric_keywords.at(locale_index);
  1248. break;
  1249. case Key::Nu:
  1250. keywords_index = s_number_system_keywords.at(locale_index);
  1251. break;
  1252. default:
  1253. VERIFY_NOT_REACHED();
  1254. }
  1255. return s_keyword_lists.at(keywords_index);
  1256. }
  1257. Optional<StringView> get_preferred_keyword_value_for_locale(StringView locale, StringView key)
  1258. {
  1259. // Hour cycle keywords are region-based rather than locale-based, so they need to be handled specially.
  1260. // FIXME: Calendar keywords are also region-based, and will need to be handled here when we support non-Gregorian calendars:
  1261. // https://github.com/unicode-org/cldr-json/blob/main/cldr-json/cldr-core/supplemental/calendarPreferenceData.json
  1262. if (key == "hc"sv) {
  1263. auto hour_cycles = get_locale_hour_cycles(locale);
  1264. if (hour_cycles.is_empty())
  1265. return {};
  1266. return hour_cycle_to_string(hour_cycles[0]);
  1267. }
  1268. // FIXME: Generate locale-preferred collation data when available in the CLDR.
  1269. if (key == "co"sv) {
  1270. auto collations = get_available_collation_types();
  1271. if (collations.is_empty())
  1272. return {};
  1273. return collations[0];
  1274. }
  1275. auto keyword_indices = find_keyword_indices(locale, key);
  1276. if (keyword_indices.is_empty())
  1277. return {};
  1278. return decode_string(keyword_indices[0]);
  1279. }
  1280. Vector<StringView> get_keywords_for_locale(StringView locale, StringView key)
  1281. {
  1282. // Hour cycle keywords are region-based rather than locale-based, so they need to be handled specially.
  1283. // FIXME: Calendar keywords are also region-based, and will need to be handled here when we support non-Gregorian calendars:
  1284. // https://github.com/unicode-org/cldr-json/blob/main/cldr-json/cldr-core/supplemental/calendarPreferenceData.json
  1285. if (key == "hc"sv) {
  1286. auto hour_cycles = get_locale_hour_cycles(locale);
  1287. Vector<StringView> values;
  1288. values.ensure_capacity(hour_cycles.size());
  1289. for (auto hour_cycle : hour_cycles)
  1290. values.unchecked_append(hour_cycle_to_string(hour_cycle));
  1291. return values;
  1292. }
  1293. // FIXME: Generate locale-preferred collation data when available in the CLDR.
  1294. if (key == "co"sv)
  1295. return Vector<StringView> { get_available_collation_types() };
  1296. auto keyword_indices = find_keyword_indices(locale, key);
  1297. Vector<StringView> keywords;
  1298. keywords.ensure_capacity(keyword_indices.size());
  1299. for (auto keyword : keyword_indices)
  1300. keywords.unchecked_append(decode_string(keyword));
  1301. return keywords;
  1302. }
  1303. Optional<DisplayPattern> get_locale_display_patterns(StringView locale)
  1304. {
  1305. auto locale_value = locale_from_string(locale);
  1306. if (!locale_value.has_value())
  1307. return {};
  1308. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1309. auto display_patterns_index = s_locale_display_patterns.at(locale_index);
  1310. auto const& display_patterns = s_display_patterns.at(display_patterns_index);
  1311. return display_patterns.to_display_pattern();
  1312. }
  1313. Optional<ListPatterns> get_locale_list_patterns(StringView locale, StringView list_pattern_type, Style list_pattern_style)
  1314. {
  1315. auto locale_value = locale_from_string(locale);
  1316. if (!locale_value.has_value())
  1317. return {};
  1318. auto type_value = list_pattern_type_from_string(list_pattern_type);
  1319. if (!type_value.has_value())
  1320. return {};
  1321. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1322. auto list_patterns_list_index = s_locale_list_patterns.at(locale_index);
  1323. auto const& locale_list_patterns = s_list_pattern_lists.at(list_patterns_list_index);
  1324. for (auto list_patterns_index : locale_list_patterns) {
  1325. auto const& list_patterns = s_list_patterns.at(list_patterns_index);
  1326. if ((list_patterns.type == type_value) && (list_patterns.style == list_pattern_style)) {
  1327. auto const& start = decode_string(list_patterns.start);
  1328. auto const& middle = decode_string(list_patterns.middle);
  1329. auto const& end = decode_string(list_patterns.end);
  1330. auto const& pair = decode_string(list_patterns.pair);
  1331. return ListPatterns { start, middle, end, pair };
  1332. }
  1333. }
  1334. return {};
  1335. }
  1336. static Optional<TextLayout> text_layout_for_locale(StringView locale)
  1337. {
  1338. auto locale_value = locale_from_string(locale);
  1339. if (!locale_value.has_value())
  1340. return {};
  1341. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1342. auto text_layouts_index = s_locale_text_layouts.at(locale_index);
  1343. return s_text_layouts.at(text_layouts_index);
  1344. }
  1345. Optional<CharacterOrder> character_order_for_locale(StringView locale)
  1346. {
  1347. if (auto text_layout = text_layout_for_locale(locale); text_layout.has_value())
  1348. return text_layout->character_order;
  1349. return {};
  1350. }
  1351. void resolve_complex_language_aliases(LanguageID& language_id)
  1352. {
  1353. for (auto const& map : s_complex_alias) {
  1354. auto key_language = decode_string(map.key.language);
  1355. auto key_script = decode_string(map.key.script);
  1356. auto key_region = decode_string(map.key.region);
  1357. if ((key_language != language_id.language) && (key_language != "und"sv))
  1358. continue;
  1359. if (!key_script.is_empty() && (key_script != language_id.script))
  1360. continue;
  1361. if (!key_region.is_empty() && (key_region != language_id.region))
  1362. continue;
  1363. if (!map.key.matches_variants(language_id.variants))
  1364. continue;
  1365. auto alias = map.alias.to_unicode_language_id().release_value_but_fixme_should_propagate_errors();
  1366. if (alias.language == "und"sv)
  1367. alias.language = move(language_id.language);
  1368. if (key_script.is_empty() && !alias.script.has_value())
  1369. alias.script = move(language_id.script);
  1370. if (key_region.is_empty() && !alias.region.has_value())
  1371. alias.region = move(language_id.region);
  1372. if (map.key.variants_size == 0 && alias.variants.is_empty())
  1373. alias.variants = move(language_id.variants);
  1374. language_id = move(alias);
  1375. break;
  1376. }
  1377. }
  1378. Optional<LanguageID> add_likely_subtags(LanguageID const& language_id)
  1379. {
  1380. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  1381. auto const* likely_subtag = resolve_likely_subtag(language_id);
  1382. if (likely_subtag == nullptr)
  1383. return {};
  1384. auto maximized = language_id;
  1385. auto key_script = decode_string(likely_subtag->key.script);
  1386. auto key_region = decode_string(likely_subtag->key.region);
  1387. auto alias_language = decode_string(likely_subtag->alias.language);
  1388. auto alias_script = decode_string(likely_subtag->alias.script);
  1389. auto alias_region = decode_string(likely_subtag->alias.region);
  1390. if (maximized.language == "und"sv)
  1391. maximized.language = String::from_utf8(alias_language).release_value_but_fixme_should_propagate_errors();
  1392. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  1393. maximized.script = String::from_utf8(alias_script).release_value_but_fixme_should_propagate_errors();
  1394. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  1395. maximized.region = String::from_utf8(alias_region).release_value_but_fixme_should_propagate_errors();
  1396. return maximized;
  1397. }
  1398. Optional<DeprecatedString> resolve_most_likely_territory(LanguageID const& language_id)
  1399. {
  1400. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  1401. return decode_string(likely_subtag->alias.region);
  1402. return {};
  1403. }
  1404. }
  1405. )~~~");
  1406. TRY(file.write(generator.as_string_view().bytes()));
  1407. return {};
  1408. }
  1409. ErrorOr<int> serenity_main(Main::Arguments arguments)
  1410. {
  1411. StringView generated_header_path;
  1412. StringView generated_implementation_path;
  1413. StringView bcp47_path;
  1414. StringView core_path;
  1415. StringView locale_names_path;
  1416. StringView misc_path;
  1417. StringView numbers_path;
  1418. StringView dates_path;
  1419. Core::ArgsParser args_parser;
  1420. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1421. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1422. args_parser.add_option(bcp47_path, "Path to cldr-bcp47 directory", "bcp47-path", 'b', "bcp47-path");
  1423. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1424. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  1425. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  1426. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  1427. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  1428. args_parser.parse(arguments);
  1429. auto generated_header_file = TRY(open_file(generated_header_path, Core::Stream::OpenMode::Write));
  1430. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::Stream::OpenMode::Write));
  1431. CLDR cldr;
  1432. TRY(parse_all_locales(bcp47_path, core_path, locale_names_path, misc_path, numbers_path, dates_path, cldr));
  1433. TRY(generate_unicode_locale_header(*generated_header_file, cldr));
  1434. TRY(generate_unicode_locale_implementation(*generated_implementation_file, cldr));
  1435. return 0;
  1436. }