GenerateUnicodeLocale.cpp 50 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186
  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. keyword_values.append(calendar_name);
  328. // FIXME: Similar to the calendar aliases defined in GenerateUnicodeDateTimeFormat, this
  329. // should be parsed from BCP47. https://unicode-org.atlassian.net/browse/CLDR-15158
  330. if (calendar_name == "gregorian"sv)
  331. keyword_values.append("gregory"sv);
  332. });
  333. }
  334. StringBuilder builder;
  335. builder.join(',', keyword_values);
  336. auto index = locale_data.unique_strings.ensure(builder.build());
  337. locale.keywords.set(key, index);
  338. if (!locale_data.keywords.contains_slow(key))
  339. locale_data.keywords.append(key);
  340. return {};
  341. }
  342. static ErrorOr<void> parse_default_content_locales(String core_path, UnicodeLocaleData& locale_data)
  343. {
  344. LexicalPath default_content_path(move(core_path));
  345. default_content_path = default_content_path.append("defaultContent.json"sv);
  346. auto default_content_file = TRY(Core::File::open(default_content_path.string(), Core::OpenMode::ReadOnly));
  347. auto default_content = TRY(JsonValue::from_string(default_content_file->read_all()));
  348. auto const& default_content_array = default_content.as_object().get("defaultContent"sv);
  349. default_content_array.as_array().for_each([&](JsonValue const& value) {
  350. auto locale = value.as_string();
  351. StringView default_locale = locale;
  352. while (true) {
  353. if (locale_data.locales.contains(default_locale))
  354. break;
  355. auto pos = default_locale.find_last('-');
  356. if (!pos.has_value())
  357. return;
  358. default_locale = default_locale.substring_view(0, *pos);
  359. }
  360. if (default_locale != locale)
  361. locale_data.locale_aliases.append({ default_locale, move(locale) });
  362. });
  363. return {};
  364. }
  365. static ErrorOr<void> define_aliases_without_scripts(UnicodeLocaleData& locale_data)
  366. {
  367. // From ECMA-402: https://tc39.es/ecma402/#sec-internal-slots
  368. //
  369. // For locales that include a script subtag in addition to language and region, the
  370. // corresponding locale without a script subtag must also be supported.
  371. //
  372. // So we define aliases for locales that contain all three subtags, but we must also take
  373. // care to handle when the locale itself or the locale without a script subtag are an alias
  374. // by way of default-content locales.
  375. auto find_alias = [&](auto const& locale) {
  376. return locale_data.locale_aliases.find_if([&](auto const& alias) { return locale == alias.alias; });
  377. };
  378. auto append_alias_without_script = [&](auto const& locale) -> ErrorOr<void> {
  379. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, locale));
  380. if ((parsed_locale.language == 0) || (parsed_locale.script == 0) || (parsed_locale.region == 0))
  381. return {};
  382. auto locale_without_script = String::formatted("{}-{}",
  383. locale_data.unique_strings.get(parsed_locale.language),
  384. locale_data.unique_strings.get(parsed_locale.region));
  385. if (locale_data.locales.contains(locale_without_script))
  386. return {};
  387. if (find_alias(locale_without_script) != locale_data.locale_aliases.end())
  388. return {};
  389. if (auto it = find_alias(locale); it != locale_data.locale_aliases.end())
  390. locale_data.locale_aliases.append({ it->name, locale_without_script });
  391. else
  392. locale_data.locale_aliases.append({ locale, locale_without_script });
  393. return {};
  394. };
  395. for (auto const& locale : locale_data.locales)
  396. TRY(append_alias_without_script(locale.key));
  397. for (auto const& locale : locale_data.locale_aliases)
  398. TRY(append_alias_without_script(locale.alias));
  399. return {};
  400. }
  401. 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)
  402. {
  403. auto identity_iterator = TRY(path_to_dir_iterator(locale_names_path));
  404. auto locale_names_iterator = TRY(path_to_dir_iterator(move(locale_names_path)));
  405. auto misc_iterator = TRY(path_to_dir_iterator(move(misc_path)));
  406. auto numbers_iterator = TRY(path_to_dir_iterator(move(numbers_path)));
  407. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  408. LexicalPath core_supplemental_path(core_path);
  409. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  410. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  411. TRY(parse_core_aliases(core_supplemental_path.string(), locale_data));
  412. TRY(parse_likely_subtags(core_supplemental_path.string(), locale_data));
  413. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  414. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  415. StringBuilder builder;
  416. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  417. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  418. builder.appendff("-{}", script);
  419. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  420. builder.appendff("-{}", region);
  421. return builder.build();
  422. };
  423. while (identity_iterator.has_next()) {
  424. auto locale_path = TRY(next_path_from_dir_iterator(identity_iterator));
  425. auto language = TRY(remove_variants_from_path(locale_path));
  426. auto& locale = locale_data.locales.ensure(language);
  427. TRY(parse_identity(locale_path, locale_data, locale));
  428. }
  429. while (locale_names_iterator.has_next()) {
  430. auto locale_path = TRY(next_path_from_dir_iterator(locale_names_iterator));
  431. auto language = TRY(remove_variants_from_path(locale_path));
  432. auto& locale = locale_data.locales.ensure(language);
  433. TRY(parse_locale_languages(locale_path, locale_data, locale));
  434. TRY(parse_locale_territories(locale_path, locale_data, locale));
  435. TRY(parse_locale_scripts(locale_path, locale_data, locale));
  436. }
  437. while (misc_iterator.has_next()) {
  438. auto misc_path = TRY(next_path_from_dir_iterator(misc_iterator));
  439. auto language = TRY(remove_variants_from_path(misc_path));
  440. auto& locale = locale_data.locales.ensure(language);
  441. TRY(parse_locale_list_patterns(misc_path, locale_data, locale));
  442. }
  443. while (numbers_iterator.has_next()) {
  444. auto numbers_path = TRY(next_path_from_dir_iterator(numbers_iterator));
  445. auto language = TRY(remove_variants_from_path(numbers_path));
  446. auto& locale = locale_data.locales.ensure(language);
  447. TRY(parse_locale_currencies(numbers_path, locale_data, locale));
  448. TRY(parse_numeric_keywords(numbers_path, locale_data, locale));
  449. }
  450. while (dates_iterator.has_next()) {
  451. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  452. auto language = TRY(remove_variants_from_path(dates_path));
  453. auto& locale = locale_data.locales.ensure(language);
  454. TRY(parse_calendar_keywords(dates_path, locale_data, locale));
  455. }
  456. TRY(parse_default_content_locales(move(core_path), locale_data));
  457. TRY(define_aliases_without_scripts(locale_data));
  458. return {};
  459. }
  460. static String format_identifier(StringView owner, String identifier)
  461. {
  462. identifier = identifier.replace("-"sv, "_"sv, true);
  463. if (all_of(identifier, is_ascii_digit))
  464. return String::formatted("{}_{}", owner[0], identifier);
  465. if (is_ascii_lower_alpha(identifier[0]))
  466. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  467. return identifier;
  468. }
  469. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  470. {
  471. StringBuilder builder;
  472. SourceGenerator generator { builder };
  473. generator.append(R"~~~(
  474. #pragma once
  475. #include <AK/Optional.h>
  476. #include <AK/StringView.h>
  477. #include <AK/Types.h>
  478. #include <AK/Vector.h>
  479. #include <LibUnicode/Forward.h>
  480. namespace Unicode {
  481. )~~~");
  482. auto locales = locale_data.locales.keys();
  483. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, locale_data.locale_aliases);
  484. generate_enum(generator, format_identifier, "Language"sv, {}, locale_data.languages);
  485. generate_enum(generator, format_identifier, "Territory"sv, {}, locale_data.territories);
  486. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, locale_data.scripts);
  487. generate_enum(generator, format_identifier, "Currency"sv, {}, locale_data.currencies);
  488. generate_enum(generator, format_identifier, "Key"sv, {}, locale_data.keywords);
  489. generate_enum(generator, format_identifier, "Variant"sv, {}, locale_data.variants);
  490. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, locale_data.list_pattern_types);
  491. generate_enum(generator, format_identifier, "ListPatternStyle"sv, {}, locale_data.list_pattern_styles);
  492. generator.append(R"~~~(
  493. namespace Detail {
  494. Optional<Locale> locale_from_string(StringView locale);
  495. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  496. Optional<Language> language_from_string(StringView language);
  497. Optional<StringView> resolve_language_alias(StringView language);
  498. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  499. Optional<Territory> territory_from_string(StringView territory);
  500. Optional<StringView> resolve_territory_alias(StringView territory);
  501. Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
  502. Optional<ScriptTag> script_tag_from_string(StringView script_tag);
  503. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  504. Optional<StringView> get_locale_long_currency_mapping(StringView locale, StringView currency);
  505. Optional<StringView> get_locale_short_currency_mapping(StringView locale, StringView currency);
  506. Optional<StringView> get_locale_narrow_currency_mapping(StringView locale, StringView currency);
  507. Optional<StringView> get_locale_numeric_currency_mapping(StringView locale, StringView currency);
  508. Optional<Currency> currency_from_string(StringView currency);
  509. Optional<StringView> get_locale_key_mapping(StringView locale, StringView key);
  510. Optional<Key> key_from_string(StringView key);
  511. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style);
  512. Optional<ListPatternType> list_pattern_type_from_string(StringView list_pattern_type);
  513. Optional<ListPatternStyle> list_pattern_style_from_string(StringView list_pattern_style);
  514. Optional<StringView> resolve_variant_alias(StringView variant);
  515. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  516. void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
  517. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id);
  518. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
  519. }
  520. }
  521. )~~~");
  522. VERIFY(file.write(generator.as_string_view()));
  523. }
  524. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  525. {
  526. StringBuilder builder;
  527. SourceGenerator generator { builder };
  528. generator.set("string_index_type"sv, s_string_index_type);
  529. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  530. generator.set("territories_size", String::number(locale_data.territories.size()));
  531. generator.set("variants_size", String::number(locale_data.max_variant_size));
  532. generator.append(R"~~~(
  533. #include <AK/Array.h>
  534. #include <AK/BinarySearch.h>
  535. #include <AK/Span.h>
  536. #include <LibUnicode/Locale.h>
  537. #include <LibUnicode/UnicodeLocale.h>
  538. namespace Unicode::Detail {
  539. struct Patterns {
  540. ListPatternType type;
  541. ListPatternStyle style;
  542. @string_index_type@ start { 0 };
  543. @string_index_type@ middle { 0 };
  544. @string_index_type@ end { 0 };
  545. @string_index_type@ pair { 0 };
  546. };
  547. )~~~");
  548. locale_data.unique_strings.generate(generator);
  549. auto append_index = [&](auto index) {
  550. generator.append(String::formatted(", {}", index));
  551. };
  552. auto append_list_and_size = [&](auto const& list) {
  553. if (list.is_empty()) {
  554. generator.append(", {}, 0");
  555. return;
  556. }
  557. bool first = true;
  558. generator.append(", {");
  559. for (auto const& item : list) {
  560. generator.append(first ? " " : ", ");
  561. generator.append(String::number(item));
  562. first = false;
  563. }
  564. generator.append(String::formatted(" }}, {}", list.size()));
  565. };
  566. auto append_string_index_list = [&](String name, auto const& keys, auto const& mappings) {
  567. generator.set("name", name);
  568. generator.set("size", String::number(keys.size()));
  569. generator.append(R"~~~(
  570. static constexpr Array<@string_index_type@, @size@> @name@ { {
  571. )~~~");
  572. constexpr size_t max_values_per_row = 30;
  573. size_t values_in_current_row = 0;
  574. for (auto const& key : keys) {
  575. if (values_in_current_row++ > 0)
  576. generator.append(" ");
  577. if (auto it = mappings.find(key); it != mappings.end())
  578. generator.set("mapping"sv, String::number(it->value));
  579. else
  580. generator.set("mapping"sv, "0"sv);
  581. generator.append("@mapping@,");
  582. if (values_in_current_row == max_values_per_row) {
  583. values_in_current_row = 0;
  584. generator.append("\n ");
  585. }
  586. }
  587. generator.append(R"~~~(
  588. } };
  589. )~~~");
  590. };
  591. auto append_list_patterns = [&](StringView name, Vector<ListPatterns> const& list_patterns) {
  592. generator.set("name", name);
  593. generator.set("size", String::number(list_patterns.size()));
  594. generator.append(R"~~~(
  595. static constexpr Array<Patterns, @size@> @name@ { {)~~~");
  596. for (auto const& list_pattern : list_patterns) {
  597. generator.set("type"sv, String::formatted("ListPatternType::{}", format_identifier({}, list_pattern.type)));
  598. generator.set("style"sv, String::formatted("ListPatternStyle::{}", format_identifier({}, list_pattern.style)));
  599. generator.set("start"sv, String::number(list_pattern.start));
  600. generator.set("middle"sv, String::number(list_pattern.middle));
  601. generator.set("end"sv, String::number(list_pattern.end));
  602. generator.set("pair"sv, String::number(list_pattern.pair));
  603. generator.append(R"~~~(
  604. { @type@, @style@, @start@, @middle@, @end@, @pair@ },)~~~");
  605. }
  606. generator.append(R"~~~(
  607. } };
  608. )~~~");
  609. };
  610. 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); });
  611. 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); });
  612. 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); });
  613. 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); });
  614. 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); });
  615. 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); });
  616. 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); });
  617. 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); });
  618. 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); });
  619. generator.append(R"~~~(
  620. struct CanonicalLanguageID {
  621. Unicode::LanguageID to_unicode_language_id() const
  622. {
  623. Unicode::LanguageID language_id {};
  624. language_id.variants.ensure_capacity(variants_size);
  625. language_id.language = s_string_list[language];
  626. if (script != 0)
  627. language_id.script = s_string_list[script];
  628. if (region != 0)
  629. language_id.region = s_string_list[region];
  630. for (size_t i = 0; i < variants_size; ++i)
  631. language_id.variants.append(s_string_list[variants[i]]);
  632. return language_id;
  633. }
  634. bool matches_variants(Vector<String> const& other_variants) const {
  635. if (variants_size == 0)
  636. return true;
  637. if (other_variants.size() != variants_size)
  638. return false;
  639. for (size_t i = 0; i < variants_size; ++i) {
  640. if (s_string_list[variants[i]] != other_variants[i])
  641. return false;
  642. }
  643. return true;
  644. };
  645. @string_index_type@ language { 0 };
  646. @string_index_type@ script { 0 };
  647. @string_index_type@ region { 0 };
  648. Array<@string_index_type@, @variants_size@> variants {};
  649. size_t variants_size { 0 };
  650. };
  651. struct LanguageMapping {
  652. CanonicalLanguageID key;
  653. CanonicalLanguageID alias;
  654. };
  655. )~~~");
  656. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  657. generator.set("size", String::number(mappings.size()));
  658. generator.set("name"sv, name);
  659. generator.append(R"~~~(
  660. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  661. )~~~");
  662. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  663. auto const& lhs_language = locale_data.unique_strings.get(lhs.key.language);
  664. auto const& rhs_language = locale_data.unique_strings.get(rhs.key.language);
  665. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  666. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  667. return false;
  668. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  669. return true;
  670. return lhs_language < rhs_language;
  671. });
  672. for (auto const& mapping : mappings) {
  673. generator.set("language"sv, String::number(mapping.key.language));
  674. generator.append(" { { @language@");
  675. append_index(mapping.key.script);
  676. append_index(mapping.key.region);
  677. append_list_and_size(mapping.key.variants);
  678. generator.set("language"sv, String::number(mapping.alias.language));
  679. generator.append(" }, { @language@");
  680. append_index(mapping.alias.script);
  681. append_index(mapping.alias.region);
  682. append_list_and_size(mapping.alias.variants);
  683. generator.append(" } },\n");
  684. }
  685. generator.append("} };\n");
  686. };
  687. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  688. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  689. generator.append(R"~~~(
  690. static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& language_id)
  691. {
  692. // https://unicode.org/reports/tr35/#Likely_Subtags
  693. enum class State {
  694. LanguageScriptRegion,
  695. LanguageRegion,
  696. LanguageScript,
  697. Language,
  698. UndScript,
  699. Done,
  700. };
  701. auto state = State::LanguageScriptRegion;
  702. while (state != State::Done) {
  703. Unicode::LanguageID search_key;
  704. switch (state) {
  705. case State::LanguageScriptRegion:
  706. state = State::LanguageRegion;
  707. if (!language_id.script.has_value() || !language_id.region.has_value())
  708. continue;
  709. search_key.language = *language_id.language;
  710. search_key.script = *language_id.script;
  711. search_key.region = *language_id.region;
  712. break;
  713. case State::LanguageRegion:
  714. state = State::LanguageScript;
  715. if (!language_id.region.has_value())
  716. continue;
  717. search_key.language = *language_id.language;
  718. search_key.region = *language_id.region;
  719. break;
  720. case State::LanguageScript:
  721. state = State::Language;
  722. if (!language_id.script.has_value())
  723. continue;
  724. search_key.language = *language_id.language;
  725. search_key.script = *language_id.script;
  726. break;
  727. case State::Language:
  728. state = State::UndScript;
  729. search_key.language = *language_id.language;
  730. break;
  731. case State::UndScript:
  732. state = State::Done;
  733. if (!language_id.script.has_value())
  734. continue;
  735. search_key.language = "und"sv;
  736. search_key.script = *language_id.script;
  737. break;
  738. default:
  739. VERIFY_NOT_REACHED();
  740. }
  741. for (auto const& map : s_likely_subtags) {
  742. auto const& key_language = s_string_list[map.key.language];
  743. auto const& key_script = s_string_list[map.key.script];
  744. auto const& key_region = s_string_list[map.key.region];
  745. if (key_language != search_key.language)
  746. continue;
  747. if (!key_script.is_empty() || search_key.script.has_value()) {
  748. if (key_script != search_key.script)
  749. continue;
  750. }
  751. if (!key_region.is_empty() || search_key.region.has_value()) {
  752. if (key_region != search_key.region)
  753. continue;
  754. }
  755. return &map;
  756. }
  757. }
  758. return nullptr;
  759. }
  760. )~~~");
  761. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name) {
  762. generator.set("enum_snake", enum_snake);
  763. generator.set("from_string_name", from_string_name);
  764. generator.set("collection_name", collection_name);
  765. generator.append(R"~~~(
  766. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  767. {
  768. auto locale_value = locale_from_string(locale);
  769. if (!locale_value.has_value())
  770. return {};
  771. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  772. if (!@enum_snake@_value.has_value())
  773. return {};
  774. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  775. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  776. auto const& mappings = @collection_name@.at(locale_index);
  777. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  778. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  779. if (@enum_snake@_mapping.is_empty())
  780. return {};
  781. return @enum_snake@_mapping;
  782. }
  783. )~~~");
  784. };
  785. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  786. HashValueMap<String> hashes;
  787. hashes.ensure_capacity(values.size());
  788. for (auto const& value : values)
  789. hashes.set(value.hash(), format_identifier(enum_title, value));
  790. for (auto const& alias : aliases)
  791. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  792. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  793. };
  794. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  795. HashValueMap<StringIndexType> hashes;
  796. hashes.ensure_capacity(aliases.size());
  797. for (auto const& alias : aliases)
  798. hashes.set(alias.key.hash(), alias.value);
  799. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), "StringView"sv, "s_string_list[{}]"sv);
  800. };
  801. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys(), locale_data.locale_aliases);
  802. append_mapping_search("language"sv, "language"sv, "s_languages"sv);
  803. append_from_string("Language"sv, "language"sv, locale_data.languages);
  804. append_alias_search("language"sv, locale_data.language_aliases);
  805. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv);
  806. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  807. append_alias_search("territory"sv, locale_data.territory_aliases);
  808. append_mapping_search("script_tag"sv, "script_tag"sv, "s_scripts"sv);
  809. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  810. append_alias_search("script_tag"sv, locale_data.script_aliases);
  811. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv);
  812. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv);
  813. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv);
  814. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv);
  815. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  816. append_mapping_search("key"sv, "key"sv, "s_keywords"sv);
  817. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  818. append_alias_search("variant"sv, locale_data.variant_aliases);
  819. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  820. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  821. append_from_string("ListPatternStyle"sv, "list_pattern_style"sv, locale_data.list_pattern_styles);
  822. generator.append(R"~~~(
  823. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style)
  824. {
  825. auto locale_value = locale_from_string(locale);
  826. if (!locale_value.has_value())
  827. return {};
  828. auto type_value = list_pattern_type_from_string(list_pattern_type);
  829. if (!type_value.has_value())
  830. return {};
  831. auto style_value = list_pattern_style_from_string(list_pattern_style);
  832. if (!style_value.has_value())
  833. return {};
  834. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  835. auto const& locale_list_patterns = s_list_patterns.at(locale_index);
  836. for (auto const& list_patterns : locale_list_patterns) {
  837. if ((list_patterns.type == type_value) && (list_patterns.style == style_value)) {
  838. auto const& start = s_string_list[list_patterns.start];
  839. auto const& middle = s_string_list[list_patterns.middle];
  840. auto const& end = s_string_list[list_patterns.end];
  841. auto const& pair = s_string_list[list_patterns.pair];
  842. return ListPatterns { start, middle, end, pair };
  843. }
  844. }
  845. return {};
  846. }
  847. void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
  848. {
  849. for (auto const& map : s_complex_alias) {
  850. auto const& key_language = s_string_list[map.key.language];
  851. auto const& key_script = s_string_list[map.key.script];
  852. auto const& key_region = s_string_list[map.key.region];
  853. if ((key_language != language_id.language) && (key_language != "und"sv))
  854. continue;
  855. if (!key_script.is_empty() && (key_script != language_id.script))
  856. continue;
  857. if (!key_region.is_empty() && (key_region != language_id.region))
  858. continue;
  859. if (!map.key.matches_variants(language_id.variants))
  860. continue;
  861. auto alias = map.alias.to_unicode_language_id();
  862. if (alias.language == "und"sv)
  863. alias.language = move(language_id.language);
  864. if (key_script.is_empty() && !alias.script.has_value())
  865. alias.script = move(language_id.script);
  866. if (key_region.is_empty() && !alias.region.has_value())
  867. alias.region = move(language_id.region);
  868. if (map.key.variants_size == 0 && alias.variants.is_empty())
  869. alias.variants = move(language_id.variants);
  870. language_id = move(alias);
  871. break;
  872. }
  873. }
  874. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id)
  875. {
  876. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  877. auto const* likely_subtag = resolve_likely_subtag(language_id);
  878. if (likely_subtag == nullptr)
  879. return {};
  880. auto maximized = language_id;
  881. auto const& key_script = s_string_list[likely_subtag->key.script];
  882. auto const& key_region = s_string_list[likely_subtag->key.region];
  883. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  884. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  885. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  886. if (maximized.language == "und"sv)
  887. maximized.language = alias_language;
  888. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  889. maximized.script = alias_script;
  890. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  891. maximized.region = alias_region;
  892. return maximized;
  893. }
  894. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
  895. {
  896. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  897. return s_string_list[likely_subtag->alias.region];
  898. return {};
  899. }
  900. }
  901. )~~~");
  902. VERIFY(file.write(generator.as_string_view()));
  903. }
  904. ErrorOr<int> serenity_main(Main::Arguments arguments)
  905. {
  906. StringView generated_header_path;
  907. StringView generated_implementation_path;
  908. StringView core_path;
  909. StringView locale_names_path;
  910. StringView misc_path;
  911. StringView numbers_path;
  912. StringView dates_path;
  913. Core::ArgsParser args_parser;
  914. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  915. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  916. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  917. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  918. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  919. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  920. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  921. args_parser.parse(arguments);
  922. auto open_file = [&](StringView path) -> ErrorOr<NonnullRefPtr<Core::File>> {
  923. if (path.is_empty()) {
  924. args_parser.print_usage(stderr, arguments.argv[0]);
  925. return Error::from_string_literal("Must provide all command line options"sv);
  926. }
  927. return Core::File::open(path, Core::OpenMode::ReadWrite);
  928. };
  929. auto generated_header_file = TRY(open_file(generated_header_path));
  930. auto generated_implementation_file = TRY(open_file(generated_implementation_path));
  931. UnicodeLocaleData locale_data;
  932. TRY(parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, dates_path, locale_data));
  933. generate_unicode_locale_header(generated_header_file, locale_data);
  934. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  935. return 0;
  936. }