GenerateUnicodeLocale.cpp 50 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191
  1. /*
  2. * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "GeneratorUtil.h"
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/Format.h>
  10. #include <AK/HashMap.h>
  11. #include <AK/JsonObject.h>
  12. #include <AK/JsonParser.h>
  13. #include <AK/JsonValue.h>
  14. #include <AK/LexicalPath.h>
  15. #include <AK/QuickSort.h>
  16. #include <AK/SourceGenerator.h>
  17. #include <AK/String.h>
  18. #include <AK/StringBuilder.h>
  19. #include <LibCore/ArgsParser.h>
  20. #include <LibCore/DirIterator.h>
  21. #include <LibCore/File.h>
  22. using StringIndexType = u16;
  23. constexpr auto s_string_index_type = "u16"sv;
  24. struct ListPatterns {
  25. String type;
  26. String style;
  27. StringIndexType start { 0 };
  28. StringIndexType middle { 0 };
  29. StringIndexType end { 0 };
  30. StringIndexType pair { 0 };
  31. };
  32. struct Locale {
  33. String language;
  34. Optional<String> territory;
  35. Optional<String> variant;
  36. HashMap<String, StringIndexType> languages;
  37. HashMap<String, StringIndexType> territories;
  38. HashMap<String, StringIndexType> scripts;
  39. HashMap<String, StringIndexType> long_currencies;
  40. HashMap<String, StringIndexType> short_currencies;
  41. HashMap<String, StringIndexType> narrow_currencies;
  42. HashMap<String, StringIndexType> numeric_currencies;
  43. HashMap<String, StringIndexType> keywords;
  44. Vector<ListPatterns> list_patterns;
  45. };
  46. struct LanguageMapping {
  47. CanonicalLanguageID<StringIndexType> key {};
  48. CanonicalLanguageID<StringIndexType> alias {};
  49. };
  50. struct UnicodeLocaleData {
  51. UniqueStringStorage<StringIndexType> unique_strings;
  52. HashMap<String, Locale> locales;
  53. Vector<Alias> locale_aliases;
  54. Vector<String> languages;
  55. Vector<String> territories;
  56. Vector<String> scripts;
  57. Vector<String> variants;
  58. Vector<String> currencies;
  59. Vector<String> keywords;
  60. Vector<String> list_pattern_types;
  61. Vector<String> list_pattern_styles;
  62. HashMap<String, StringIndexType> language_aliases;
  63. HashMap<String, StringIndexType> territory_aliases;
  64. HashMap<String, StringIndexType> script_aliases;
  65. HashMap<String, StringIndexType> variant_aliases;
  66. HashMap<String, StringIndexType> subdivision_aliases;
  67. Vector<LanguageMapping> complex_mappings;
  68. Vector<LanguageMapping> likely_subtags;
  69. size_t max_variant_size { 0 };
  70. };
  71. // Some parsing is expected to fail. For example, the CLDR contains language mappings
  72. // with locales such as "en-GB-oed" that are canonically invalid locale IDs.
  73. #define TRY_OR_DISCARD(expression) \
  74. ({ \
  75. auto _temporary_result = (expression); \
  76. if (_temporary_result.is_error()) \
  77. return; \
  78. _temporary_result.release_value(); \
  79. })
  80. static ErrorOr<LanguageMapping> parse_language_mapping(UnicodeLocaleData& locale_data, StringView key, StringView alias)
  81. {
  82. auto parsed_key = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, key));
  83. auto parsed_alias = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, alias));
  84. return LanguageMapping { move(parsed_key), move(parsed_alias) };
  85. }
  86. static ErrorOr<void> parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
  87. {
  88. LexicalPath core_aliases_path(move(core_supplemental_path));
  89. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  90. auto core_aliases_file = TRY(Core::File::open(core_aliases_path.string(), Core::OpenMode::ReadOnly));
  91. auto core_aliases = TRY(JsonValue::from_string(core_aliases_file->read_all()));
  92. auto const& supplemental_object = core_aliases.as_object().get("supplemental"sv);
  93. auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
  94. auto const& alias_object = metadata_object.as_object().get("alias"sv);
  95. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  96. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  97. auto alias = value.as_object().get("_replacement"sv).as_string();
  98. if (key.contains('-')) {
  99. auto mapping = TRY_OR_DISCARD(parse_language_mapping(locale_data, key, alias));
  100. locale_data.max_variant_size = max(mapping.key.variants.size(), locale_data.max_variant_size);
  101. locale_data.max_variant_size = max(mapping.alias.variants.size(), locale_data.max_variant_size);
  102. locale_data.complex_mappings.append(move(mapping));
  103. } else {
  104. alias_map.set(key, locale_data.unique_strings.ensure(alias));
  105. }
  106. });
  107. };
  108. append_aliases(alias_object.as_object().get("languageAlias"sv), locale_data.language_aliases);
  109. append_aliases(alias_object.as_object().get("territoryAlias"sv), locale_data.territory_aliases);
  110. append_aliases(alias_object.as_object().get("scriptAlias"sv), locale_data.script_aliases);
  111. append_aliases(alias_object.as_object().get("variantAlias"sv), locale_data.variant_aliases);
  112. append_aliases(alias_object.as_object().get("subdivisionAlias"sv), locale_data.subdivision_aliases);
  113. return {};
  114. }
  115. static ErrorOr<void> parse_likely_subtags(String core_supplemental_path, UnicodeLocaleData& locale_data)
  116. {
  117. LexicalPath likely_subtags_path(move(core_supplemental_path));
  118. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  119. auto likely_subtags_file = TRY(Core::File::open(likely_subtags_path.string(), Core::OpenMode::ReadOnly));
  120. auto likely_subtags = TRY(JsonValue::from_string(likely_subtags_file->read_all()));
  121. auto const& supplemental_object = likely_subtags.as_object().get("supplemental"sv);
  122. auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
  123. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  124. auto mapping = TRY_OR_DISCARD(parse_language_mapping(locale_data, key, value.as_string()));
  125. locale_data.max_variant_size = max(mapping.key.variants.size(), locale_data.max_variant_size);
  126. locale_data.max_variant_size = max(mapping.alias.variants.size(), locale_data.max_variant_size);
  127. locale_data.likely_subtags.append(move(mapping));
  128. });
  129. return {};
  130. }
  131. static ErrorOr<void> parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  132. {
  133. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  134. languages_path = languages_path.append("languages.json"sv);
  135. auto languages_file = TRY(Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly));
  136. auto languages = TRY(JsonValue::from_string(languages_file->read_all()));
  137. auto const& main_object = languages.as_object().get("main"sv);
  138. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  139. auto const& identity_object = locale_object.as_object().get("identity"sv);
  140. auto const& language_string = identity_object.as_object().get("language"sv);
  141. auto const& territory_string = identity_object.as_object().get("territory"sv);
  142. auto const& variant_string = identity_object.as_object().get("variant"sv);
  143. locale.language = language_string.as_string();
  144. if (!locale_data.languages.contains_slow(locale.language))
  145. locale_data.languages.append(locale.language);
  146. if (territory_string.is_string()) {
  147. locale.territory = territory_string.as_string();
  148. if (!locale_data.territories.contains_slow(*locale.territory))
  149. locale_data.territories.append(*locale.territory);
  150. }
  151. if (variant_string.is_string()) {
  152. locale.variant = variant_string.as_string();
  153. if (!locale_data.variants.contains_slow(*locale.variant))
  154. locale_data.variants.append(*locale.variant);
  155. }
  156. return {};
  157. }
  158. static ErrorOr<void> parse_locale_languages(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  159. {
  160. LexicalPath languages_path(move(locale_path));
  161. languages_path = languages_path.append("languages.json"sv);
  162. auto languages_file = TRY(Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly));
  163. auto languages = TRY(JsonValue::from_string(languages_file->read_all()));
  164. auto const& main_object = languages.as_object().get("main"sv);
  165. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  166. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  167. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  168. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  169. if (!locale_data.languages.contains_slow(key))
  170. return;
  171. auto index = locale_data.unique_strings.ensure(value.as_string());
  172. locale.languages.set(key, index);
  173. });
  174. return {};
  175. }
  176. static ErrorOr<void> parse_locale_territories(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  177. {
  178. LexicalPath territories_path(move(locale_path));
  179. territories_path = territories_path.append("territories.json"sv);
  180. auto territories_file = TRY(Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly));
  181. auto territories = TRY(JsonValue::from_string(territories_file->read_all()));
  182. auto const& main_object = territories.as_object().get("main"sv);
  183. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  184. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  185. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  186. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  187. if (!locale_data.territories.contains_slow(key))
  188. return;
  189. auto index = locale_data.unique_strings.ensure(value.as_string());
  190. locale.territories.set(key, index);
  191. });
  192. return {};
  193. }
  194. static ErrorOr<void> parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  195. {
  196. LexicalPath scripts_path(move(locale_path));
  197. scripts_path = scripts_path.append("scripts.json"sv);
  198. auto scripts_file = TRY(Core::File::open(scripts_path.string(), Core::OpenMode::ReadOnly));
  199. auto scripts = TRY(JsonValue::from_string(scripts_file->read_all()));
  200. auto const& main_object = scripts.as_object().get("main"sv);
  201. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  202. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  203. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  204. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  205. auto index = locale_data.unique_strings.ensure(value.as_string());
  206. locale.scripts.set(key, index);
  207. if (!locale_data.scripts.contains_slow(key))
  208. locale_data.scripts.append(key);
  209. });
  210. return {};
  211. }
  212. static ErrorOr<void> parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
  213. {
  214. LexicalPath list_patterns_path(move(misc_path));
  215. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  216. auto list_patterns_file = TRY(Core::File::open(list_patterns_path.string(), Core::OpenMode::ReadOnly));
  217. auto list_patterns = TRY(JsonValue::from_string(list_patterns_file->read_all()));
  218. auto const& main_object = list_patterns.as_object().get("main"sv);
  219. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  220. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  221. auto list_pattern_type = [](StringView key) {
  222. if (key.contains("type-standard"sv))
  223. return "conjunction"sv;
  224. if (key.contains("type-or"sv))
  225. return "disjunction"sv;
  226. if (key.contains("type-unit"sv))
  227. return "unit"sv;
  228. VERIFY_NOT_REACHED();
  229. };
  230. auto list_pattern_style = [](StringView key) {
  231. if (key.contains("short"sv))
  232. return "short"sv;
  233. if (key.contains("narrow"sv))
  234. return "narrow"sv;
  235. return "long"sv;
  236. };
  237. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  238. auto type = list_pattern_type(key);
  239. auto style = list_pattern_style(key);
  240. auto start = locale_data.unique_strings.ensure(value.as_object().get("start"sv).as_string());
  241. auto middle = locale_data.unique_strings.ensure(value.as_object().get("middle"sv).as_string());
  242. auto end = locale_data.unique_strings.ensure(value.as_object().get("end"sv).as_string());
  243. auto pair = locale_data.unique_strings.ensure(value.as_object().get("2"sv).as_string());
  244. if (!locale_data.list_pattern_types.contains_slow(type))
  245. locale_data.list_pattern_types.append(type);
  246. if (!locale_data.list_pattern_styles.contains_slow(style))
  247. locale_data.list_pattern_styles.append(style);
  248. locale.list_patterns.append({ move(type), move(style), move(start), move(middle), move(end), move(pair) });
  249. });
  250. return {};
  251. }
  252. static ErrorOr<void> parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  253. {
  254. LexicalPath currencies_path(move(numbers_path));
  255. currencies_path = currencies_path.append("currencies.json"sv);
  256. auto currencies_file = TRY(Core::File::open(currencies_path.string(), Core::OpenMode::ReadOnly));
  257. auto currencies = TRY(JsonValue::from_string(currencies_file->read_all()));
  258. auto const& main_object = currencies.as_object().get("main"sv);
  259. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  260. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  261. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  262. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  263. auto const& long_name = value.as_object().get("displayName"sv);
  264. auto const& short_name = value.as_object().get("symbol"sv);
  265. auto const& narrow_name = value.as_object().get("symbol-alt-narrow"sv);
  266. auto const& numeric_name = value.as_object().get("displayName-count-other"sv);
  267. locale.long_currencies.set(key, locale_data.unique_strings.ensure(long_name.as_string()));
  268. locale.short_currencies.set(key, locale_data.unique_strings.ensure(short_name.as_string()));
  269. locale.narrow_currencies.set(key, narrow_name.is_null() ? 0 : locale_data.unique_strings.ensure(narrow_name.as_string()));
  270. locale.numeric_currencies.set(key, locale_data.unique_strings.ensure(numeric_name.is_null() ? long_name.as_string() : numeric_name.as_string()));
  271. if (!locale_data.currencies.contains_slow(key))
  272. locale_data.currencies.append(key);
  273. });
  274. return {};
  275. }
  276. static ErrorOr<void> parse_numeric_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  277. {
  278. static constexpr StringView key = "nu"sv;
  279. LexicalPath numbers_path(move(locale_numbers_path));
  280. numbers_path = numbers_path.append("numbers.json"sv);
  281. auto numbers_file = TRY(Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly));
  282. auto numbers = TRY(JsonValue::from_string(numbers_file->read_all()));
  283. auto const& main_object = numbers.as_object().get("main"sv);
  284. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  285. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  286. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  287. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  288. Vector<String> keyword_values {};
  289. keyword_values.append(default_numbering_system_object.as_string());
  290. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  291. auto keyword_value = value.as_string();
  292. if (!keyword_values.contains_slow(keyword_value))
  293. keyword_values.append(move(keyword_value));
  294. });
  295. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  296. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  297. return;
  298. auto keyword_value = value.as_string();
  299. if (!keyword_values.contains_slow(keyword_value))
  300. keyword_values.append(move(keyword_value));
  301. });
  302. StringBuilder builder;
  303. builder.join(',', keyword_values);
  304. auto index = locale_data.unique_strings.ensure(builder.build());
  305. locale.keywords.set(key, index);
  306. if (!locale_data.keywords.contains_slow(key))
  307. locale_data.keywords.append(key);
  308. return {};
  309. }
  310. static ErrorOr<void> parse_calendar_keywords(String locale_dates_path, UnicodeLocaleData& locale_data, Locale& locale)
  311. {
  312. static constexpr StringView key = "ca"sv;
  313. auto calendars_iterator = TRY(path_to_dir_iterator(locale_dates_path, {}));
  314. Vector<String> keyword_values {};
  315. while (calendars_iterator.has_next()) {
  316. auto locale_calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  317. LexicalPath calendars_path(move(locale_calendars_path));
  318. if (!calendars_path.basename().starts_with("ca-"sv))
  319. continue;
  320. auto calendars_file = TRY(Core::File::open(calendars_path.string(), Core::OpenMode::ReadOnly));
  321. auto calendars = TRY(JsonValue::from_string(calendars_file->read_all()));
  322. auto const& main_object = calendars.as_object().get("main"sv);
  323. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  324. auto const& dates_object = locale_object.as_object().get("dates"sv);
  325. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  326. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const&) {
  327. // The generic calendar is not a supported Unicode calendar key, so skip it:
  328. // https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Intl/Locale/calendar#unicode_calendar_keys
  329. if (calendar_name == "generic"sv)
  330. return;
  331. // FIXME: Similar to the calendar aliases defined in GenerateUnicodeDateTimeFormat, this
  332. // should be parsed from BCP47. https://unicode-org.atlassian.net/browse/CLDR-15158
  333. if (calendar_name == "gregorian"sv)
  334. keyword_values.append("gregory"sv);
  335. else
  336. keyword_values.append(calendar_name);
  337. });
  338. }
  339. StringBuilder builder;
  340. builder.join(',', keyword_values);
  341. auto index = locale_data.unique_strings.ensure(builder.build());
  342. locale.keywords.set(key, index);
  343. if (!locale_data.keywords.contains_slow(key))
  344. locale_data.keywords.append(key);
  345. return {};
  346. }
  347. static ErrorOr<void> parse_default_content_locales(String core_path, UnicodeLocaleData& locale_data)
  348. {
  349. LexicalPath default_content_path(move(core_path));
  350. default_content_path = default_content_path.append("defaultContent.json"sv);
  351. auto default_content_file = TRY(Core::File::open(default_content_path.string(), Core::OpenMode::ReadOnly));
  352. auto default_content = TRY(JsonValue::from_string(default_content_file->read_all()));
  353. auto const& default_content_array = default_content.as_object().get("defaultContent"sv);
  354. default_content_array.as_array().for_each([&](JsonValue const& value) {
  355. auto locale = value.as_string();
  356. StringView default_locale = locale;
  357. while (true) {
  358. if (locale_data.locales.contains(default_locale))
  359. break;
  360. auto pos = default_locale.find_last('-');
  361. if (!pos.has_value())
  362. return;
  363. default_locale = default_locale.substring_view(0, *pos);
  364. }
  365. if (default_locale != locale)
  366. locale_data.locale_aliases.append({ default_locale, move(locale) });
  367. });
  368. return {};
  369. }
  370. static ErrorOr<void> define_aliases_without_scripts(UnicodeLocaleData& locale_data)
  371. {
  372. // From ECMA-402: https://tc39.es/ecma402/#sec-internal-slots
  373. //
  374. // For locales that include a script subtag in addition to language and region, the
  375. // corresponding locale without a script subtag must also be supported.
  376. //
  377. // So we define aliases for locales that contain all three subtags, but we must also take
  378. // care to handle when the locale itself or the locale without a script subtag are an alias
  379. // by way of default-content locales.
  380. auto find_alias = [&](auto const& locale) {
  381. return locale_data.locale_aliases.find_if([&](auto const& alias) { return locale == alias.alias; });
  382. };
  383. auto append_alias_without_script = [&](auto const& locale) -> ErrorOr<void> {
  384. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, locale));
  385. if ((parsed_locale.language == 0) || (parsed_locale.script == 0) || (parsed_locale.region == 0))
  386. return {};
  387. auto locale_without_script = String::formatted("{}-{}",
  388. locale_data.unique_strings.get(parsed_locale.language),
  389. locale_data.unique_strings.get(parsed_locale.region));
  390. if (locale_data.locales.contains(locale_without_script))
  391. return {};
  392. if (find_alias(locale_without_script) != locale_data.locale_aliases.end())
  393. return {};
  394. if (auto it = find_alias(locale); it != locale_data.locale_aliases.end())
  395. locale_data.locale_aliases.append({ it->name, locale_without_script });
  396. else
  397. locale_data.locale_aliases.append({ locale, locale_without_script });
  398. return {};
  399. };
  400. for (auto const& locale : locale_data.locales)
  401. TRY(append_alias_without_script(locale.key));
  402. for (auto const& locale : locale_data.locale_aliases)
  403. TRY(append_alias_without_script(locale.alias));
  404. return {};
  405. }
  406. static ErrorOr<void> parse_all_locales(String core_path, String locale_names_path, String misc_path, String numbers_path, String dates_path, UnicodeLocaleData& locale_data)
  407. {
  408. auto identity_iterator = TRY(path_to_dir_iterator(locale_names_path));
  409. auto locale_names_iterator = TRY(path_to_dir_iterator(move(locale_names_path)));
  410. auto misc_iterator = TRY(path_to_dir_iterator(move(misc_path)));
  411. auto numbers_iterator = TRY(path_to_dir_iterator(move(numbers_path)));
  412. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  413. LexicalPath core_supplemental_path(core_path);
  414. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  415. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  416. TRY(parse_core_aliases(core_supplemental_path.string(), locale_data));
  417. TRY(parse_likely_subtags(core_supplemental_path.string(), locale_data));
  418. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  419. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  420. StringBuilder builder;
  421. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  422. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  423. builder.appendff("-{}", script);
  424. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  425. builder.appendff("-{}", region);
  426. return builder.build();
  427. };
  428. while (identity_iterator.has_next()) {
  429. auto locale_path = TRY(next_path_from_dir_iterator(identity_iterator));
  430. auto language = TRY(remove_variants_from_path(locale_path));
  431. auto& locale = locale_data.locales.ensure(language);
  432. TRY(parse_identity(locale_path, locale_data, locale));
  433. }
  434. while (locale_names_iterator.has_next()) {
  435. auto locale_path = TRY(next_path_from_dir_iterator(locale_names_iterator));
  436. auto language = TRY(remove_variants_from_path(locale_path));
  437. auto& locale = locale_data.locales.ensure(language);
  438. TRY(parse_locale_languages(locale_path, locale_data, locale));
  439. TRY(parse_locale_territories(locale_path, locale_data, locale));
  440. TRY(parse_locale_scripts(locale_path, locale_data, locale));
  441. }
  442. while (misc_iterator.has_next()) {
  443. auto misc_path = TRY(next_path_from_dir_iterator(misc_iterator));
  444. auto language = TRY(remove_variants_from_path(misc_path));
  445. auto& locale = locale_data.locales.ensure(language);
  446. TRY(parse_locale_list_patterns(misc_path, locale_data, locale));
  447. }
  448. while (numbers_iterator.has_next()) {
  449. auto numbers_path = TRY(next_path_from_dir_iterator(numbers_iterator));
  450. auto language = TRY(remove_variants_from_path(numbers_path));
  451. auto& locale = locale_data.locales.ensure(language);
  452. TRY(parse_locale_currencies(numbers_path, locale_data, locale));
  453. TRY(parse_numeric_keywords(numbers_path, locale_data, locale));
  454. }
  455. while (dates_iterator.has_next()) {
  456. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  457. auto language = TRY(remove_variants_from_path(dates_path));
  458. auto& locale = locale_data.locales.ensure(language);
  459. TRY(parse_calendar_keywords(dates_path, locale_data, locale));
  460. }
  461. TRY(parse_default_content_locales(move(core_path), locale_data));
  462. TRY(define_aliases_without_scripts(locale_data));
  463. return {};
  464. }
  465. static String format_identifier(StringView owner, String identifier)
  466. {
  467. identifier = identifier.replace("-"sv, "_"sv, true);
  468. if (all_of(identifier, is_ascii_digit))
  469. return String::formatted("{}_{}", owner[0], identifier);
  470. if (is_ascii_lower_alpha(identifier[0]))
  471. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  472. return identifier;
  473. }
  474. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  475. {
  476. StringBuilder builder;
  477. SourceGenerator generator { builder };
  478. generator.append(R"~~~(
  479. #pragma once
  480. #include <AK/Optional.h>
  481. #include <AK/StringView.h>
  482. #include <AK/Types.h>
  483. #include <AK/Vector.h>
  484. #include <LibUnicode/Forward.h>
  485. namespace Unicode {
  486. )~~~");
  487. auto locales = locale_data.locales.keys();
  488. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, locale_data.locale_aliases);
  489. generate_enum(generator, format_identifier, "Language"sv, {}, locale_data.languages);
  490. generate_enum(generator, format_identifier, "Territory"sv, {}, locale_data.territories);
  491. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, locale_data.scripts);
  492. generate_enum(generator, format_identifier, "Currency"sv, {}, locale_data.currencies);
  493. generate_enum(generator, format_identifier, "Key"sv, {}, locale_data.keywords);
  494. generate_enum(generator, format_identifier, "Variant"sv, {}, locale_data.variants);
  495. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, locale_data.list_pattern_types);
  496. generate_enum(generator, format_identifier, "ListPatternStyle"sv, {}, locale_data.list_pattern_styles);
  497. generator.append(R"~~~(
  498. namespace Detail {
  499. Optional<Locale> locale_from_string(StringView locale);
  500. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  501. Optional<Language> language_from_string(StringView language);
  502. Optional<StringView> resolve_language_alias(StringView language);
  503. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  504. Optional<Territory> territory_from_string(StringView territory);
  505. Optional<StringView> resolve_territory_alias(StringView territory);
  506. Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
  507. Optional<ScriptTag> script_tag_from_string(StringView script_tag);
  508. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  509. Optional<StringView> get_locale_long_currency_mapping(StringView locale, StringView currency);
  510. Optional<StringView> get_locale_short_currency_mapping(StringView locale, StringView currency);
  511. Optional<StringView> get_locale_narrow_currency_mapping(StringView locale, StringView currency);
  512. Optional<StringView> get_locale_numeric_currency_mapping(StringView locale, StringView currency);
  513. Optional<Currency> currency_from_string(StringView currency);
  514. Optional<StringView> get_locale_key_mapping(StringView locale, StringView key);
  515. Optional<Key> key_from_string(StringView key);
  516. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style);
  517. Optional<ListPatternType> list_pattern_type_from_string(StringView list_pattern_type);
  518. Optional<ListPatternStyle> list_pattern_style_from_string(StringView list_pattern_style);
  519. Optional<StringView> resolve_variant_alias(StringView variant);
  520. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  521. void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
  522. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id);
  523. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
  524. }
  525. }
  526. )~~~");
  527. VERIFY(file.write(generator.as_string_view()));
  528. }
  529. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  530. {
  531. StringBuilder builder;
  532. SourceGenerator generator { builder };
  533. generator.set("string_index_type"sv, s_string_index_type);
  534. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  535. generator.set("territories_size", String::number(locale_data.territories.size()));
  536. generator.set("variants_size", String::number(locale_data.max_variant_size));
  537. generator.append(R"~~~(
  538. #include <AK/Array.h>
  539. #include <AK/BinarySearch.h>
  540. #include <AK/Span.h>
  541. #include <LibUnicode/Locale.h>
  542. #include <LibUnicode/UnicodeLocale.h>
  543. namespace Unicode::Detail {
  544. struct Patterns {
  545. ListPatternType type;
  546. ListPatternStyle style;
  547. @string_index_type@ start { 0 };
  548. @string_index_type@ middle { 0 };
  549. @string_index_type@ end { 0 };
  550. @string_index_type@ pair { 0 };
  551. };
  552. )~~~");
  553. locale_data.unique_strings.generate(generator);
  554. auto append_index = [&](auto index) {
  555. generator.append(String::formatted(", {}", index));
  556. };
  557. auto append_list_and_size = [&](auto const& list) {
  558. if (list.is_empty()) {
  559. generator.append(", {}, 0");
  560. return;
  561. }
  562. bool first = true;
  563. generator.append(", {");
  564. for (auto const& item : list) {
  565. generator.append(first ? " " : ", ");
  566. generator.append(String::number(item));
  567. first = false;
  568. }
  569. generator.append(String::formatted(" }}, {}", list.size()));
  570. };
  571. auto append_string_index_list = [&](String name, auto const& keys, auto const& mappings) {
  572. generator.set("name", name);
  573. generator.set("size", String::number(keys.size()));
  574. generator.append(R"~~~(
  575. static constexpr Array<@string_index_type@, @size@> @name@ { {
  576. )~~~");
  577. constexpr size_t max_values_per_row = 30;
  578. size_t values_in_current_row = 0;
  579. for (auto const& key : keys) {
  580. if (values_in_current_row++ > 0)
  581. generator.append(" ");
  582. if (auto it = mappings.find(key); it != mappings.end())
  583. generator.set("mapping"sv, String::number(it->value));
  584. else
  585. generator.set("mapping"sv, "0"sv);
  586. generator.append("@mapping@,");
  587. if (values_in_current_row == max_values_per_row) {
  588. values_in_current_row = 0;
  589. generator.append("\n ");
  590. }
  591. }
  592. generator.append(R"~~~(
  593. } };
  594. )~~~");
  595. };
  596. auto append_list_patterns = [&](StringView name, Vector<ListPatterns> const& list_patterns) {
  597. generator.set("name", name);
  598. generator.set("size", String::number(list_patterns.size()));
  599. generator.append(R"~~~(
  600. static constexpr Array<Patterns, @size@> @name@ { {)~~~");
  601. for (auto const& list_pattern : list_patterns) {
  602. generator.set("type"sv, String::formatted("ListPatternType::{}", format_identifier({}, list_pattern.type)));
  603. generator.set("style"sv, String::formatted("ListPatternStyle::{}", format_identifier({}, list_pattern.style)));
  604. generator.set("start"sv, String::number(list_pattern.start));
  605. generator.set("middle"sv, String::number(list_pattern.middle));
  606. generator.set("end"sv, String::number(list_pattern.end));
  607. generator.set("pair"sv, String::number(list_pattern.pair));
  608. generator.append(R"~~~(
  609. { @type@, @style@, @start@, @middle@, @end@, @pair@ },)~~~");
  610. }
  611. generator.append(R"~~~(
  612. } };
  613. )~~~");
  614. };
  615. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_languages"sv, "s_languages_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.languages, value.languages); });
  616. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_territories"sv, "s_territories_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.territories, value.territories); });
  617. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_scripts"sv, "s_scripts_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.scripts, value.scripts); });
  618. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_long_currencies"sv, "s_long_currencies_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.currencies, value.long_currencies); });
  619. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_short_currencies"sv, "s_short_currencies_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.currencies, value.short_currencies); });
  620. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_narrow_currencies"sv, "s_narrow_currencies_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.currencies, value.narrow_currencies); });
  621. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_numeric_currencies"sv, "s_numeric_currencies_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.currencies, value.numeric_currencies); });
  622. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_keywords"sv, "s_keywords_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.keywords, value.keywords); });
  623. generate_mapping(generator, locale_data.locales, "Patterns"sv, "s_list_patterns"sv, "s_list_patterns_{}", [&](auto const& name, auto const& value) { append_list_patterns(name, value.list_patterns); });
  624. generator.append(R"~~~(
  625. struct CanonicalLanguageID {
  626. Unicode::LanguageID to_unicode_language_id() const
  627. {
  628. Unicode::LanguageID language_id {};
  629. language_id.variants.ensure_capacity(variants_size);
  630. language_id.language = s_string_list[language];
  631. if (script != 0)
  632. language_id.script = s_string_list[script];
  633. if (region != 0)
  634. language_id.region = s_string_list[region];
  635. for (size_t i = 0; i < variants_size; ++i)
  636. language_id.variants.append(s_string_list[variants[i]]);
  637. return language_id;
  638. }
  639. bool matches_variants(Vector<String> const& other_variants) const {
  640. if (variants_size == 0)
  641. return true;
  642. if (other_variants.size() != variants_size)
  643. return false;
  644. for (size_t i = 0; i < variants_size; ++i) {
  645. if (s_string_list[variants[i]] != other_variants[i])
  646. return false;
  647. }
  648. return true;
  649. };
  650. @string_index_type@ language { 0 };
  651. @string_index_type@ script { 0 };
  652. @string_index_type@ region { 0 };
  653. Array<@string_index_type@, @variants_size@> variants {};
  654. size_t variants_size { 0 };
  655. };
  656. struct LanguageMapping {
  657. CanonicalLanguageID key;
  658. CanonicalLanguageID alias;
  659. };
  660. )~~~");
  661. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  662. generator.set("size", String::number(mappings.size()));
  663. generator.set("name"sv, name);
  664. generator.append(R"~~~(
  665. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  666. )~~~");
  667. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  668. auto const& lhs_language = locale_data.unique_strings.get(lhs.key.language);
  669. auto const& rhs_language = locale_data.unique_strings.get(rhs.key.language);
  670. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  671. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  672. return false;
  673. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  674. return true;
  675. return lhs_language < rhs_language;
  676. });
  677. for (auto const& mapping : mappings) {
  678. generator.set("language"sv, String::number(mapping.key.language));
  679. generator.append(" { { @language@");
  680. append_index(mapping.key.script);
  681. append_index(mapping.key.region);
  682. append_list_and_size(mapping.key.variants);
  683. generator.set("language"sv, String::number(mapping.alias.language));
  684. generator.append(" }, { @language@");
  685. append_index(mapping.alias.script);
  686. append_index(mapping.alias.region);
  687. append_list_and_size(mapping.alias.variants);
  688. generator.append(" } },\n");
  689. }
  690. generator.append("} };\n");
  691. };
  692. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  693. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  694. generator.append(R"~~~(
  695. static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& language_id)
  696. {
  697. // https://unicode.org/reports/tr35/#Likely_Subtags
  698. enum class State {
  699. LanguageScriptRegion,
  700. LanguageRegion,
  701. LanguageScript,
  702. Language,
  703. UndScript,
  704. Done,
  705. };
  706. auto state = State::LanguageScriptRegion;
  707. while (state != State::Done) {
  708. Unicode::LanguageID search_key;
  709. switch (state) {
  710. case State::LanguageScriptRegion:
  711. state = State::LanguageRegion;
  712. if (!language_id.script.has_value() || !language_id.region.has_value())
  713. continue;
  714. search_key.language = *language_id.language;
  715. search_key.script = *language_id.script;
  716. search_key.region = *language_id.region;
  717. break;
  718. case State::LanguageRegion:
  719. state = State::LanguageScript;
  720. if (!language_id.region.has_value())
  721. continue;
  722. search_key.language = *language_id.language;
  723. search_key.region = *language_id.region;
  724. break;
  725. case State::LanguageScript:
  726. state = State::Language;
  727. if (!language_id.script.has_value())
  728. continue;
  729. search_key.language = *language_id.language;
  730. search_key.script = *language_id.script;
  731. break;
  732. case State::Language:
  733. state = State::UndScript;
  734. search_key.language = *language_id.language;
  735. break;
  736. case State::UndScript:
  737. state = State::Done;
  738. if (!language_id.script.has_value())
  739. continue;
  740. search_key.language = "und"sv;
  741. search_key.script = *language_id.script;
  742. break;
  743. default:
  744. VERIFY_NOT_REACHED();
  745. }
  746. for (auto const& map : s_likely_subtags) {
  747. auto const& key_language = s_string_list[map.key.language];
  748. auto const& key_script = s_string_list[map.key.script];
  749. auto const& key_region = s_string_list[map.key.region];
  750. if (key_language != search_key.language)
  751. continue;
  752. if (!key_script.is_empty() || search_key.script.has_value()) {
  753. if (key_script != search_key.script)
  754. continue;
  755. }
  756. if (!key_region.is_empty() || search_key.region.has_value()) {
  757. if (key_region != search_key.region)
  758. continue;
  759. }
  760. return &map;
  761. }
  762. }
  763. return nullptr;
  764. }
  765. )~~~");
  766. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name) {
  767. generator.set("enum_snake", enum_snake);
  768. generator.set("from_string_name", from_string_name);
  769. generator.set("collection_name", collection_name);
  770. generator.append(R"~~~(
  771. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  772. {
  773. auto locale_value = locale_from_string(locale);
  774. if (!locale_value.has_value())
  775. return {};
  776. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  777. if (!@enum_snake@_value.has_value())
  778. return {};
  779. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  780. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  781. auto const& mappings = @collection_name@.at(locale_index);
  782. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  783. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  784. if (@enum_snake@_mapping.is_empty())
  785. return {};
  786. return @enum_snake@_mapping;
  787. }
  788. )~~~");
  789. };
  790. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  791. HashValueMap<String> hashes;
  792. hashes.ensure_capacity(values.size());
  793. for (auto const& value : values)
  794. hashes.set(value.hash(), format_identifier(enum_title, value));
  795. for (auto const& alias : aliases)
  796. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  797. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  798. };
  799. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  800. HashValueMap<StringIndexType> hashes;
  801. hashes.ensure_capacity(aliases.size());
  802. for (auto const& alias : aliases)
  803. hashes.set(alias.key.hash(), alias.value);
  804. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), "StringView"sv, "s_string_list[{}]"sv);
  805. };
  806. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys(), locale_data.locale_aliases);
  807. append_mapping_search("language"sv, "language"sv, "s_languages"sv);
  808. append_from_string("Language"sv, "language"sv, locale_data.languages);
  809. append_alias_search("language"sv, locale_data.language_aliases);
  810. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv);
  811. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  812. append_alias_search("territory"sv, locale_data.territory_aliases);
  813. append_mapping_search("script_tag"sv, "script_tag"sv, "s_scripts"sv);
  814. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  815. append_alias_search("script_tag"sv, locale_data.script_aliases);
  816. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv);
  817. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv);
  818. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv);
  819. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv);
  820. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  821. append_mapping_search("key"sv, "key"sv, "s_keywords"sv);
  822. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  823. append_alias_search("variant"sv, locale_data.variant_aliases);
  824. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  825. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  826. append_from_string("ListPatternStyle"sv, "list_pattern_style"sv, locale_data.list_pattern_styles);
  827. generator.append(R"~~~(
  828. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style)
  829. {
  830. auto locale_value = locale_from_string(locale);
  831. if (!locale_value.has_value())
  832. return {};
  833. auto type_value = list_pattern_type_from_string(list_pattern_type);
  834. if (!type_value.has_value())
  835. return {};
  836. auto style_value = list_pattern_style_from_string(list_pattern_style);
  837. if (!style_value.has_value())
  838. return {};
  839. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  840. auto const& locale_list_patterns = s_list_patterns.at(locale_index);
  841. for (auto const& list_patterns : locale_list_patterns) {
  842. if ((list_patterns.type == type_value) && (list_patterns.style == style_value)) {
  843. auto const& start = s_string_list[list_patterns.start];
  844. auto const& middle = s_string_list[list_patterns.middle];
  845. auto const& end = s_string_list[list_patterns.end];
  846. auto const& pair = s_string_list[list_patterns.pair];
  847. return ListPatterns { start, middle, end, pair };
  848. }
  849. }
  850. return {};
  851. }
  852. void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
  853. {
  854. for (auto const& map : s_complex_alias) {
  855. auto const& key_language = s_string_list[map.key.language];
  856. auto const& key_script = s_string_list[map.key.script];
  857. auto const& key_region = s_string_list[map.key.region];
  858. if ((key_language != language_id.language) && (key_language != "und"sv))
  859. continue;
  860. if (!key_script.is_empty() && (key_script != language_id.script))
  861. continue;
  862. if (!key_region.is_empty() && (key_region != language_id.region))
  863. continue;
  864. if (!map.key.matches_variants(language_id.variants))
  865. continue;
  866. auto alias = map.alias.to_unicode_language_id();
  867. if (alias.language == "und"sv)
  868. alias.language = move(language_id.language);
  869. if (key_script.is_empty() && !alias.script.has_value())
  870. alias.script = move(language_id.script);
  871. if (key_region.is_empty() && !alias.region.has_value())
  872. alias.region = move(language_id.region);
  873. if (map.key.variants_size == 0 && alias.variants.is_empty())
  874. alias.variants = move(language_id.variants);
  875. language_id = move(alias);
  876. break;
  877. }
  878. }
  879. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id)
  880. {
  881. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  882. auto const* likely_subtag = resolve_likely_subtag(language_id);
  883. if (likely_subtag == nullptr)
  884. return {};
  885. auto maximized = language_id;
  886. auto const& key_script = s_string_list[likely_subtag->key.script];
  887. auto const& key_region = s_string_list[likely_subtag->key.region];
  888. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  889. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  890. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  891. if (maximized.language == "und"sv)
  892. maximized.language = alias_language;
  893. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  894. maximized.script = alias_script;
  895. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  896. maximized.region = alias_region;
  897. return maximized;
  898. }
  899. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
  900. {
  901. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  902. return s_string_list[likely_subtag->alias.region];
  903. return {};
  904. }
  905. }
  906. )~~~");
  907. VERIFY(file.write(generator.as_string_view()));
  908. }
  909. ErrorOr<int> serenity_main(Main::Arguments arguments)
  910. {
  911. StringView generated_header_path;
  912. StringView generated_implementation_path;
  913. StringView core_path;
  914. StringView locale_names_path;
  915. StringView misc_path;
  916. StringView numbers_path;
  917. StringView dates_path;
  918. Core::ArgsParser args_parser;
  919. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  920. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  921. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  922. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  923. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  924. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  925. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  926. args_parser.parse(arguments);
  927. auto open_file = [&](StringView path) -> ErrorOr<NonnullRefPtr<Core::File>> {
  928. if (path.is_empty()) {
  929. args_parser.print_usage(stderr, arguments.argv[0]);
  930. return Error::from_string_literal("Must provide all command line options"sv);
  931. }
  932. return Core::File::open(path, Core::OpenMode::ReadWrite);
  933. };
  934. auto generated_header_file = TRY(open_file(generated_header_path));
  935. auto generated_implementation_file = TRY(open_file(generated_implementation_path));
  936. UnicodeLocaleData locale_data;
  937. TRY(parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, dates_path, locale_data));
  938. generate_unicode_locale_header(generated_header_file, locale_data);
  939. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  940. return 0;
  941. }