GenerateUnicodeLocale.cpp 49 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164
  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. static Optional<LanguageMapping> parse_language_mapping(UnicodeLocaleData& locale_data, StringView key, StringView alias)
  72. {
  73. auto parsed_key = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, key);
  74. if (!parsed_key.has_value())
  75. return {};
  76. auto parsed_alias = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, alias);
  77. if (!parsed_alias.has_value())
  78. return {};
  79. return LanguageMapping { parsed_key.release_value(), parsed_alias.release_value() };
  80. }
  81. static void parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
  82. {
  83. LexicalPath core_aliases_path(move(core_supplemental_path));
  84. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  85. VERIFY(Core::File::exists(core_aliases_path.string()));
  86. auto core_aliases_file_or_error = Core::File::open(core_aliases_path.string(), Core::OpenMode::ReadOnly);
  87. VERIFY(!core_aliases_file_or_error.is_error());
  88. auto core_aliases = JsonValue::from_string(core_aliases_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  89. auto const& supplemental_object = core_aliases.as_object().get("supplemental"sv);
  90. auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
  91. auto const& alias_object = metadata_object.as_object().get("alias"sv);
  92. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  93. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  94. auto alias = value.as_object().get("_replacement"sv).as_string();
  95. if (key.contains('-')) {
  96. auto mapping = parse_language_mapping(locale_data, key, alias);
  97. if (!mapping.has_value())
  98. return;
  99. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  100. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  101. locale_data.complex_mappings.append(mapping.release_value());
  102. } else {
  103. alias_map.set(key, locale_data.unique_strings.ensure(alias));
  104. }
  105. });
  106. };
  107. append_aliases(alias_object.as_object().get("languageAlias"sv), locale_data.language_aliases);
  108. append_aliases(alias_object.as_object().get("territoryAlias"sv), locale_data.territory_aliases);
  109. append_aliases(alias_object.as_object().get("scriptAlias"sv), locale_data.script_aliases);
  110. append_aliases(alias_object.as_object().get("variantAlias"sv), locale_data.variant_aliases);
  111. append_aliases(alias_object.as_object().get("subdivisionAlias"sv), locale_data.subdivision_aliases);
  112. }
  113. static void parse_likely_subtags(String core_supplemental_path, UnicodeLocaleData& locale_data)
  114. {
  115. LexicalPath likely_subtags_path(move(core_supplemental_path));
  116. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  117. VERIFY(Core::File::exists(likely_subtags_path.string()));
  118. auto likely_subtags_file_or_error = Core::File::open(likely_subtags_path.string(), Core::OpenMode::ReadOnly);
  119. VERIFY(!likely_subtags_file_or_error.is_error());
  120. auto likely_subtags = JsonValue::from_string(likely_subtags_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  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 = parse_language_mapping(locale_data, key, value.as_string());
  125. if (!mapping.has_value())
  126. return;
  127. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  128. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  129. locale_data.likely_subtags.append(mapping.release_value());
  130. });
  131. }
  132. static void parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  133. {
  134. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  135. languages_path = languages_path.append("languages.json"sv);
  136. VERIFY(Core::File::exists(languages_path.string()));
  137. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  138. VERIFY(!languages_file_or_error.is_error());
  139. auto languages = JsonValue::from_string(languages_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  140. auto const& main_object = languages.as_object().get("main"sv);
  141. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  142. auto const& identity_object = locale_object.as_object().get("identity"sv);
  143. auto const& language_string = identity_object.as_object().get("language"sv);
  144. auto const& territory_string = identity_object.as_object().get("territory"sv);
  145. auto const& variant_string = identity_object.as_object().get("variant"sv);
  146. locale.language = language_string.as_string();
  147. if (!locale_data.languages.contains_slow(locale.language))
  148. locale_data.languages.append(locale.language);
  149. if (territory_string.is_string()) {
  150. locale.territory = territory_string.as_string();
  151. if (!locale_data.territories.contains_slow(*locale.territory))
  152. locale_data.territories.append(*locale.territory);
  153. }
  154. if (variant_string.is_string()) {
  155. locale.variant = variant_string.as_string();
  156. if (!locale_data.variants.contains_slow(*locale.variant))
  157. locale_data.variants.append(*locale.variant);
  158. }
  159. }
  160. static void parse_locale_languages(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  161. {
  162. LexicalPath languages_path(move(locale_path));
  163. languages_path = languages_path.append("languages.json"sv);
  164. VERIFY(Core::File::exists(languages_path.string()));
  165. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  166. VERIFY(!languages_file_or_error.is_error());
  167. auto languages = JsonValue::from_string(languages_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  168. auto const& main_object = languages.as_object().get("main"sv);
  169. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  170. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  171. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  172. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  173. if (!locale_data.languages.contains_slow(key))
  174. return;
  175. auto index = locale_data.unique_strings.ensure(value.as_string());
  176. locale.languages.set(key, index);
  177. });
  178. }
  179. static void parse_locale_territories(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  180. {
  181. LexicalPath territories_path(move(locale_path));
  182. territories_path = territories_path.append("territories.json"sv);
  183. VERIFY(Core::File::exists(territories_path.string()));
  184. auto territories_file_or_error = Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly);
  185. VERIFY(!territories_file_or_error.is_error());
  186. auto territories = JsonValue::from_string(territories_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  187. auto const& main_object = territories.as_object().get("main"sv);
  188. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  189. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  190. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  191. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  192. if (!locale_data.territories.contains_slow(key))
  193. return;
  194. auto index = locale_data.unique_strings.ensure(value.as_string());
  195. locale.territories.set(key, index);
  196. });
  197. }
  198. static void parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  199. {
  200. LexicalPath scripts_path(move(locale_path));
  201. scripts_path = scripts_path.append("scripts.json"sv);
  202. VERIFY(Core::File::exists(scripts_path.string()));
  203. auto scripts_file_or_error = Core::File::open(scripts_path.string(), Core::OpenMode::ReadOnly);
  204. VERIFY(!scripts_file_or_error.is_error());
  205. auto scripts = JsonValue::from_string(scripts_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  206. auto const& main_object = scripts.as_object().get("main"sv);
  207. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  208. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  209. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  210. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  211. auto index = locale_data.unique_strings.ensure(value.as_string());
  212. locale.scripts.set(key, index);
  213. if (!locale_data.scripts.contains_slow(key))
  214. locale_data.scripts.append(key);
  215. });
  216. }
  217. static void parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
  218. {
  219. LexicalPath list_patterns_path(move(misc_path));
  220. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  221. VERIFY(Core::File::exists(list_patterns_path.string()));
  222. auto list_patterns_file_or_error = Core::File::open(list_patterns_path.string(), Core::OpenMode::ReadOnly);
  223. VERIFY(!list_patterns_file_or_error.is_error());
  224. auto list_patterns = JsonValue::from_string(list_patterns_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  225. auto const& main_object = list_patterns.as_object().get("main"sv);
  226. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  227. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  228. auto list_pattern_type = [](StringView key) {
  229. if (key.contains("type-standard"sv))
  230. return "conjunction"sv;
  231. if (key.contains("type-or"sv))
  232. return "disjunction"sv;
  233. if (key.contains("type-unit"sv))
  234. return "unit"sv;
  235. VERIFY_NOT_REACHED();
  236. };
  237. auto list_pattern_style = [](StringView key) {
  238. if (key.contains("short"sv))
  239. return "short"sv;
  240. if (key.contains("narrow"sv))
  241. return "narrow"sv;
  242. return "long"sv;
  243. };
  244. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  245. auto type = list_pattern_type(key);
  246. auto style = list_pattern_style(key);
  247. auto start = locale_data.unique_strings.ensure(value.as_object().get("start"sv).as_string());
  248. auto middle = locale_data.unique_strings.ensure(value.as_object().get("middle"sv).as_string());
  249. auto end = locale_data.unique_strings.ensure(value.as_object().get("end"sv).as_string());
  250. auto pair = locale_data.unique_strings.ensure(value.as_object().get("2"sv).as_string());
  251. if (!locale_data.list_pattern_types.contains_slow(type))
  252. locale_data.list_pattern_types.append(type);
  253. if (!locale_data.list_pattern_styles.contains_slow(style))
  254. locale_data.list_pattern_styles.append(style);
  255. locale.list_patterns.append({ move(type), move(style), move(start), move(middle), move(end), move(pair) });
  256. });
  257. }
  258. static void parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  259. {
  260. LexicalPath currencies_path(move(numbers_path));
  261. currencies_path = currencies_path.append("currencies.json"sv);
  262. VERIFY(Core::File::exists(currencies_path.string()));
  263. auto currencies_file_or_error = Core::File::open(currencies_path.string(), Core::OpenMode::ReadOnly);
  264. VERIFY(!currencies_file_or_error.is_error());
  265. auto currencies = JsonValue::from_string(currencies_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  266. auto const& main_object = currencies.as_object().get("main"sv);
  267. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  268. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  269. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  270. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  271. auto const& long_name = value.as_object().get("displayName"sv);
  272. auto const& short_name = value.as_object().get("symbol"sv);
  273. auto const& narrow_name = value.as_object().get("symbol-alt-narrow"sv);
  274. auto const& numeric_name = value.as_object().get("displayName-count-other"sv);
  275. locale.long_currencies.set(key, locale_data.unique_strings.ensure(long_name.as_string()));
  276. locale.short_currencies.set(key, locale_data.unique_strings.ensure(short_name.as_string()));
  277. locale.narrow_currencies.set(key, narrow_name.is_null() ? 0 : locale_data.unique_strings.ensure(narrow_name.as_string()));
  278. locale.numeric_currencies.set(key, locale_data.unique_strings.ensure(numeric_name.is_null() ? long_name.as_string() : numeric_name.as_string()));
  279. if (!locale_data.currencies.contains_slow(key))
  280. locale_data.currencies.append(key);
  281. });
  282. }
  283. static void parse_numeric_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  284. {
  285. static constexpr StringView key = "nu"sv;
  286. LexicalPath numbers_path(move(locale_numbers_path));
  287. numbers_path = numbers_path.append("numbers.json"sv);
  288. VERIFY(Core::File::exists(numbers_path.string()));
  289. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  290. VERIFY(!numbers_file_or_error.is_error());
  291. auto numbers = JsonValue::from_string(numbers_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  292. auto const& main_object = numbers.as_object().get("main"sv);
  293. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  294. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  295. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  296. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  297. Vector<String> keyword_values {};
  298. keyword_values.append(default_numbering_system_object.as_string());
  299. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  300. auto keyword_value = value.as_string();
  301. if (!keyword_values.contains_slow(keyword_value))
  302. keyword_values.append(move(keyword_value));
  303. });
  304. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  305. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  306. return;
  307. auto keyword_value = value.as_string();
  308. if (!keyword_values.contains_slow(keyword_value))
  309. keyword_values.append(move(keyword_value));
  310. });
  311. StringBuilder builder;
  312. builder.join(',', keyword_values);
  313. auto index = locale_data.unique_strings.ensure(builder.build());
  314. locale.keywords.set(key, index);
  315. if (!locale_data.keywords.contains_slow(key))
  316. locale_data.keywords.append(key);
  317. }
  318. static void parse_default_content_locales(String core_path, UnicodeLocaleData& locale_data)
  319. {
  320. LexicalPath default_content_path(move(core_path));
  321. default_content_path = default_content_path.append("defaultContent.json"sv);
  322. VERIFY(Core::File::exists(default_content_path.string()));
  323. auto default_content_file_or_error = Core::File::open(default_content_path.string(), Core::OpenMode::ReadOnly);
  324. VERIFY(!default_content_file_or_error.is_error());
  325. auto default_content = JsonValue::from_string(default_content_file_or_error.value()->read_all()).release_value_but_fixme_should_propagate_errors();
  326. auto const& default_content_array = default_content.as_object().get("defaultContent"sv);
  327. default_content_array.as_array().for_each([&](JsonValue const& value) {
  328. auto locale = value.as_string();
  329. StringView default_locale = locale;
  330. while (true) {
  331. if (locale_data.locales.contains(default_locale))
  332. break;
  333. auto pos = default_locale.find_last('-');
  334. if (!pos.has_value())
  335. return;
  336. default_locale = default_locale.substring_view(0, *pos);
  337. }
  338. if (default_locale != locale)
  339. locale_data.locale_aliases.append({ default_locale, move(locale) });
  340. });
  341. }
  342. static void define_aliases_without_scripts(UnicodeLocaleData& locale_data)
  343. {
  344. // From ECMA-402: https://tc39.es/ecma402/#sec-internal-slots
  345. //
  346. // For locales that include a script subtag in addition to language and region, the
  347. // corresponding locale without a script subtag must also be supported.
  348. //
  349. // So we define aliases for locales that contain all three subtags, but we must also take
  350. // care to handle when the locale itself or the locale without a script subtag are an alias
  351. // by way of default-content locales.
  352. auto find_alias = [&](auto const& locale) {
  353. return locale_data.locale_aliases.find_if([&](auto const& alias) { return locale == alias.alias; });
  354. };
  355. auto append_alias_without_script = [&](auto const& locale) {
  356. auto parsed_locale = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, locale);
  357. VERIFY(parsed_locale.has_value());
  358. if ((parsed_locale->language == 0) || (parsed_locale->script == 0) || (parsed_locale->region == 0))
  359. return;
  360. auto locale_without_script = String::formatted("{}-{}",
  361. locale_data.unique_strings.get(parsed_locale->language),
  362. locale_data.unique_strings.get(parsed_locale->region));
  363. if (locale_data.locales.contains(locale_without_script))
  364. return;
  365. if (find_alias(locale_without_script) != locale_data.locale_aliases.end())
  366. return;
  367. if (auto it = find_alias(locale); it != locale_data.locale_aliases.end())
  368. locale_data.locale_aliases.append({ it->name, locale_without_script });
  369. else
  370. locale_data.locale_aliases.append({ locale, locale_without_script });
  371. };
  372. for (auto const& locale : locale_data.locales)
  373. append_alias_without_script(locale.key);
  374. for (auto const& locale : locale_data.locale_aliases)
  375. append_alias_without_script(locale.alias);
  376. }
  377. static void parse_all_locales(String core_path, String locale_names_path, String misc_path, String numbers_path, UnicodeLocaleData& locale_data)
  378. {
  379. auto identity_iterator = path_to_dir_iterator(locale_names_path);
  380. auto locale_names_iterator = path_to_dir_iterator(move(locale_names_path));
  381. auto misc_iterator = path_to_dir_iterator(move(misc_path));
  382. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  383. LexicalPath core_supplemental_path(core_path);
  384. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  385. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  386. parse_core_aliases(core_supplemental_path.string(), locale_data);
  387. parse_likely_subtags(core_supplemental_path.string(), locale_data);
  388. auto remove_variants_from_path = [&](String path) -> Optional<String> {
  389. auto parsed_locale = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path));
  390. if (!parsed_locale.has_value())
  391. return {};
  392. StringBuilder builder;
  393. builder.append(locale_data.unique_strings.get(parsed_locale->language));
  394. if (auto script = locale_data.unique_strings.get(parsed_locale->script); !script.is_empty())
  395. builder.appendff("-{}", script);
  396. if (auto region = locale_data.unique_strings.get(parsed_locale->region); !region.is_empty())
  397. builder.appendff("-{}", region);
  398. return builder.build();
  399. };
  400. while (identity_iterator.has_next()) {
  401. auto locale_path = identity_iterator.next_full_path();
  402. VERIFY(Core::File::is_directory(locale_path));
  403. auto language = remove_variants_from_path(locale_path);
  404. if (!language.has_value())
  405. continue;
  406. auto& locale = locale_data.locales.ensure(*language);
  407. parse_identity(locale_path, locale_data, locale);
  408. }
  409. while (locale_names_iterator.has_next()) {
  410. auto locale_path = locale_names_iterator.next_full_path();
  411. VERIFY(Core::File::is_directory(locale_path));
  412. auto language = remove_variants_from_path(locale_path);
  413. if (!language.has_value())
  414. continue;
  415. auto& locale = locale_data.locales.ensure(*language);
  416. parse_locale_languages(locale_path, locale_data, locale);
  417. parse_locale_territories(locale_path, locale_data, locale);
  418. parse_locale_scripts(locale_path, locale_data, locale);
  419. }
  420. while (misc_iterator.has_next()) {
  421. auto misc_path = misc_iterator.next_full_path();
  422. VERIFY(Core::File::is_directory(misc_path));
  423. auto language = remove_variants_from_path(misc_path);
  424. if (!language.has_value())
  425. continue;
  426. auto& locale = locale_data.locales.ensure(*language);
  427. parse_locale_list_patterns(misc_path, locale_data, locale);
  428. }
  429. while (numbers_iterator.has_next()) {
  430. auto numbers_path = numbers_iterator.next_full_path();
  431. VERIFY(Core::File::is_directory(numbers_path));
  432. auto language = remove_variants_from_path(numbers_path);
  433. if (!language.has_value())
  434. continue;
  435. auto& locale = locale_data.locales.ensure(*language);
  436. parse_locale_currencies(numbers_path, locale_data, locale);
  437. parse_numeric_keywords(numbers_path, locale_data, locale);
  438. }
  439. parse_default_content_locales(move(core_path), locale_data);
  440. define_aliases_without_scripts(locale_data);
  441. }
  442. static String format_identifier(StringView owner, String identifier)
  443. {
  444. identifier = identifier.replace("-"sv, "_"sv, true);
  445. if (all_of(identifier, is_ascii_digit))
  446. return String::formatted("{}_{}", owner[0], identifier);
  447. if (is_ascii_lower_alpha(identifier[0]))
  448. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  449. return identifier;
  450. }
  451. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  452. {
  453. StringBuilder builder;
  454. SourceGenerator generator { builder };
  455. generator.append(R"~~~(
  456. #pragma once
  457. #include <AK/Optional.h>
  458. #include <AK/StringView.h>
  459. #include <AK/Types.h>
  460. #include <AK/Vector.h>
  461. #include <LibUnicode/Forward.h>
  462. namespace Unicode {
  463. )~~~");
  464. auto locales = locale_data.locales.keys();
  465. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, locale_data.locale_aliases);
  466. generate_enum(generator, format_identifier, "Language"sv, {}, locale_data.languages);
  467. generate_enum(generator, format_identifier, "Territory"sv, {}, locale_data.territories);
  468. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, locale_data.scripts);
  469. generate_enum(generator, format_identifier, "Currency"sv, {}, locale_data.currencies);
  470. generate_enum(generator, format_identifier, "Key"sv, {}, locale_data.keywords);
  471. generate_enum(generator, format_identifier, "Variant"sv, {}, locale_data.variants);
  472. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, locale_data.list_pattern_types);
  473. generate_enum(generator, format_identifier, "ListPatternStyle"sv, {}, locale_data.list_pattern_styles);
  474. generator.append(R"~~~(
  475. namespace Detail {
  476. Optional<Locale> locale_from_string(StringView locale);
  477. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  478. Optional<Language> language_from_string(StringView language);
  479. Optional<StringView> resolve_language_alias(StringView language);
  480. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  481. Optional<Territory> territory_from_string(StringView territory);
  482. Optional<StringView> resolve_territory_alias(StringView territory);
  483. Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
  484. Optional<ScriptTag> script_tag_from_string(StringView script_tag);
  485. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  486. Optional<StringView> get_locale_long_currency_mapping(StringView locale, StringView currency);
  487. Optional<StringView> get_locale_short_currency_mapping(StringView locale, StringView currency);
  488. Optional<StringView> get_locale_narrow_currency_mapping(StringView locale, StringView currency);
  489. Optional<StringView> get_locale_numeric_currency_mapping(StringView locale, StringView currency);
  490. Optional<Currency> currency_from_string(StringView currency);
  491. Optional<StringView> get_locale_key_mapping(StringView locale, StringView key);
  492. Optional<Key> key_from_string(StringView key);
  493. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style);
  494. Optional<ListPatternType> list_pattern_type_from_string(StringView list_pattern_type);
  495. Optional<ListPatternStyle> list_pattern_style_from_string(StringView list_pattern_style);
  496. Optional<StringView> resolve_variant_alias(StringView variant);
  497. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  498. void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
  499. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id);
  500. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
  501. }
  502. }
  503. )~~~");
  504. file.write(generator.as_string_view());
  505. }
  506. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  507. {
  508. StringBuilder builder;
  509. SourceGenerator generator { builder };
  510. generator.set("string_index_type"sv, s_string_index_type);
  511. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  512. generator.set("territories_size", String::number(locale_data.territories.size()));
  513. generator.set("variants_size", String::number(locale_data.max_variant_size));
  514. generator.append(R"~~~(
  515. #include <AK/Array.h>
  516. #include <AK/BinarySearch.h>
  517. #include <AK/Span.h>
  518. #include <LibUnicode/Locale.h>
  519. #include <LibUnicode/UnicodeLocale.h>
  520. namespace Unicode::Detail {
  521. struct Patterns {
  522. ListPatternType type;
  523. ListPatternStyle style;
  524. @string_index_type@ start { 0 };
  525. @string_index_type@ middle { 0 };
  526. @string_index_type@ end { 0 };
  527. @string_index_type@ pair { 0 };
  528. };
  529. )~~~");
  530. locale_data.unique_strings.generate(generator);
  531. auto append_index = [&](auto index) {
  532. generator.append(String::formatted(", {}", index));
  533. };
  534. auto append_list_and_size = [&](auto const& list) {
  535. if (list.is_empty()) {
  536. generator.append(", {}, 0");
  537. return;
  538. }
  539. bool first = true;
  540. generator.append(", {");
  541. for (auto const& item : list) {
  542. generator.append(first ? " " : ", ");
  543. generator.append(String::number(item));
  544. first = false;
  545. }
  546. generator.append(String::formatted(" }}, {}", list.size()));
  547. };
  548. auto append_string_index_list = [&](String name, auto const& keys, auto const& mappings) {
  549. generator.set("name", name);
  550. generator.set("size", String::number(keys.size()));
  551. generator.append(R"~~~(
  552. static constexpr Array<@string_index_type@, @size@> @name@ { {
  553. )~~~");
  554. constexpr size_t max_values_per_row = 30;
  555. size_t values_in_current_row = 0;
  556. for (auto const& key : keys) {
  557. if (values_in_current_row++ > 0)
  558. generator.append(" ");
  559. if (auto it = mappings.find(key); it != mappings.end())
  560. generator.set("mapping"sv, String::number(it->value));
  561. else
  562. generator.set("mapping"sv, "0"sv);
  563. generator.append("@mapping@,");
  564. if (values_in_current_row == max_values_per_row) {
  565. values_in_current_row = 0;
  566. generator.append("\n ");
  567. }
  568. }
  569. generator.append(R"~~~(
  570. } };
  571. )~~~");
  572. };
  573. auto append_list_patterns = [&](StringView name, Vector<ListPatterns> const& list_patterns) {
  574. generator.set("name", name);
  575. generator.set("size", String::number(list_patterns.size()));
  576. generator.append(R"~~~(
  577. static constexpr Array<Patterns, @size@> @name@ { {)~~~");
  578. for (auto const& list_pattern : list_patterns) {
  579. generator.set("type"sv, String::formatted("ListPatternType::{}", format_identifier({}, list_pattern.type)));
  580. generator.set("style"sv, String::formatted("ListPatternStyle::{}", format_identifier({}, list_pattern.style)));
  581. generator.set("start"sv, String::number(list_pattern.start));
  582. generator.set("middle"sv, String::number(list_pattern.middle));
  583. generator.set("end"sv, String::number(list_pattern.end));
  584. generator.set("pair"sv, String::number(list_pattern.pair));
  585. generator.append(R"~~~(
  586. { @type@, @style@, @start@, @middle@, @end@, @pair@ },)~~~");
  587. }
  588. generator.append(R"~~~(
  589. } };
  590. )~~~");
  591. };
  592. 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); });
  593. 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); });
  594. 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); });
  595. 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); });
  596. 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); });
  597. 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); });
  598. 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); });
  599. 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); });
  600. 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); });
  601. generator.append(R"~~~(
  602. struct CanonicalLanguageID {
  603. Unicode::LanguageID to_unicode_language_id() const
  604. {
  605. Unicode::LanguageID language_id {};
  606. language_id.variants.ensure_capacity(variants_size);
  607. language_id.language = s_string_list[language];
  608. if (script != 0)
  609. language_id.script = s_string_list[script];
  610. if (region != 0)
  611. language_id.region = s_string_list[region];
  612. for (size_t i = 0; i < variants_size; ++i)
  613. language_id.variants.append(s_string_list[variants[i]]);
  614. return language_id;
  615. }
  616. bool matches_variants(Vector<String> const& other_variants) const {
  617. if (variants_size == 0)
  618. return true;
  619. if (other_variants.size() != variants_size)
  620. return false;
  621. for (size_t i = 0; i < variants_size; ++i) {
  622. if (s_string_list[variants[i]] != other_variants[i])
  623. return false;
  624. }
  625. return true;
  626. };
  627. @string_index_type@ language { 0 };
  628. @string_index_type@ script { 0 };
  629. @string_index_type@ region { 0 };
  630. Array<@string_index_type@, @variants_size@> variants {};
  631. size_t variants_size { 0 };
  632. };
  633. struct LanguageMapping {
  634. CanonicalLanguageID key;
  635. CanonicalLanguageID alias;
  636. };
  637. )~~~");
  638. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  639. generator.set("size", String::number(mappings.size()));
  640. generator.set("name"sv, name);
  641. generator.append(R"~~~(
  642. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  643. )~~~");
  644. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  645. auto const& lhs_language = locale_data.unique_strings.get(lhs.key.language);
  646. auto const& rhs_language = locale_data.unique_strings.get(rhs.key.language);
  647. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  648. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  649. return false;
  650. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  651. return true;
  652. return lhs_language < rhs_language;
  653. });
  654. for (auto const& mapping : mappings) {
  655. generator.set("language"sv, String::number(mapping.key.language));
  656. generator.append(" { { @language@");
  657. append_index(mapping.key.script);
  658. append_index(mapping.key.region);
  659. append_list_and_size(mapping.key.variants);
  660. generator.set("language"sv, String::number(mapping.alias.language));
  661. generator.append(" }, { @language@");
  662. append_index(mapping.alias.script);
  663. append_index(mapping.alias.region);
  664. append_list_and_size(mapping.alias.variants);
  665. generator.append(" } },\n");
  666. }
  667. generator.append("} };\n");
  668. };
  669. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  670. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  671. generator.append(R"~~~(
  672. static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& language_id)
  673. {
  674. // https://unicode.org/reports/tr35/#Likely_Subtags
  675. enum class State {
  676. LanguageScriptRegion,
  677. LanguageRegion,
  678. LanguageScript,
  679. Language,
  680. UndScript,
  681. Done,
  682. };
  683. auto state = State::LanguageScriptRegion;
  684. while (state != State::Done) {
  685. Unicode::LanguageID search_key;
  686. switch (state) {
  687. case State::LanguageScriptRegion:
  688. state = State::LanguageRegion;
  689. if (!language_id.script.has_value() || !language_id.region.has_value())
  690. continue;
  691. search_key.language = *language_id.language;
  692. search_key.script = *language_id.script;
  693. search_key.region = *language_id.region;
  694. break;
  695. case State::LanguageRegion:
  696. state = State::LanguageScript;
  697. if (!language_id.region.has_value())
  698. continue;
  699. search_key.language = *language_id.language;
  700. search_key.region = *language_id.region;
  701. break;
  702. case State::LanguageScript:
  703. state = State::Language;
  704. if (!language_id.script.has_value())
  705. continue;
  706. search_key.language = *language_id.language;
  707. search_key.script = *language_id.script;
  708. break;
  709. case State::Language:
  710. state = State::UndScript;
  711. search_key.language = *language_id.language;
  712. break;
  713. case State::UndScript:
  714. state = State::Done;
  715. if (!language_id.script.has_value())
  716. continue;
  717. search_key.language = "und"sv;
  718. search_key.script = *language_id.script;
  719. break;
  720. default:
  721. VERIFY_NOT_REACHED();
  722. }
  723. for (auto const& map : s_likely_subtags) {
  724. auto const& key_language = s_string_list[map.key.language];
  725. auto const& key_script = s_string_list[map.key.script];
  726. auto const& key_region = s_string_list[map.key.region];
  727. if (key_language != search_key.language)
  728. continue;
  729. if (!key_script.is_empty() || search_key.script.has_value()) {
  730. if (key_script != search_key.script)
  731. continue;
  732. }
  733. if (!key_region.is_empty() || search_key.region.has_value()) {
  734. if (key_region != search_key.region)
  735. continue;
  736. }
  737. return &map;
  738. }
  739. }
  740. return nullptr;
  741. }
  742. )~~~");
  743. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name) {
  744. generator.set("enum_snake", enum_snake);
  745. generator.set("from_string_name", from_string_name);
  746. generator.set("collection_name", collection_name);
  747. generator.append(R"~~~(
  748. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  749. {
  750. auto locale_value = locale_from_string(locale);
  751. if (!locale_value.has_value())
  752. return {};
  753. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  754. if (!@enum_snake@_value.has_value())
  755. return {};
  756. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  757. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  758. auto const& mappings = @collection_name@.at(locale_index);
  759. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  760. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  761. if (@enum_snake@_mapping.is_empty())
  762. return {};
  763. return @enum_snake@_mapping;
  764. }
  765. )~~~");
  766. };
  767. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  768. HashValueMap<String> hashes;
  769. hashes.ensure_capacity(values.size());
  770. for (auto const& value : values)
  771. hashes.set(value.hash(), format_identifier(enum_title, value));
  772. for (auto const& alias : aliases)
  773. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  774. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  775. };
  776. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  777. HashValueMap<StringIndexType> hashes;
  778. hashes.ensure_capacity(aliases.size());
  779. for (auto const& alias : aliases)
  780. hashes.set(alias.key.hash(), alias.value);
  781. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), "StringView"sv, "s_string_list[{}]"sv);
  782. };
  783. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys(), locale_data.locale_aliases);
  784. append_mapping_search("language"sv, "language"sv, "s_languages"sv);
  785. append_from_string("Language"sv, "language"sv, locale_data.languages);
  786. append_alias_search("language"sv, locale_data.language_aliases);
  787. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv);
  788. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  789. append_alias_search("territory"sv, locale_data.territory_aliases);
  790. append_mapping_search("script_tag"sv, "script_tag"sv, "s_scripts"sv);
  791. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  792. append_alias_search("script_tag"sv, locale_data.script_aliases);
  793. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv);
  794. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv);
  795. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv);
  796. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv);
  797. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  798. append_mapping_search("key"sv, "key"sv, "s_keywords"sv);
  799. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  800. append_alias_search("variant"sv, locale_data.variant_aliases);
  801. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  802. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  803. append_from_string("ListPatternStyle"sv, "list_pattern_style"sv, locale_data.list_pattern_styles);
  804. generator.append(R"~~~(
  805. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style)
  806. {
  807. auto locale_value = locale_from_string(locale);
  808. if (!locale_value.has_value())
  809. return {};
  810. auto type_value = list_pattern_type_from_string(list_pattern_type);
  811. if (!type_value.has_value())
  812. return {};
  813. auto style_value = list_pattern_style_from_string(list_pattern_style);
  814. if (!style_value.has_value())
  815. return {};
  816. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  817. auto const& locale_list_patterns = s_list_patterns.at(locale_index);
  818. for (auto const& list_patterns : locale_list_patterns) {
  819. if ((list_patterns.type == type_value) && (list_patterns.style == style_value)) {
  820. auto const& start = s_string_list[list_patterns.start];
  821. auto const& middle = s_string_list[list_patterns.middle];
  822. auto const& end = s_string_list[list_patterns.end];
  823. auto const& pair = s_string_list[list_patterns.pair];
  824. return ListPatterns { start, middle, end, pair };
  825. }
  826. }
  827. return {};
  828. }
  829. void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
  830. {
  831. for (auto const& map : s_complex_alias) {
  832. auto const& key_language = s_string_list[map.key.language];
  833. auto const& key_script = s_string_list[map.key.script];
  834. auto const& key_region = s_string_list[map.key.region];
  835. if ((key_language != language_id.language) && (key_language != "und"sv))
  836. continue;
  837. if (!key_script.is_empty() && (key_script != language_id.script))
  838. continue;
  839. if (!key_region.is_empty() && (key_region != language_id.region))
  840. continue;
  841. if (!map.key.matches_variants(language_id.variants))
  842. continue;
  843. auto alias = map.alias.to_unicode_language_id();
  844. if (alias.language == "und"sv)
  845. alias.language = move(language_id.language);
  846. if (key_script.is_empty() && !alias.script.has_value())
  847. alias.script = move(language_id.script);
  848. if (key_region.is_empty() && !alias.region.has_value())
  849. alias.region = move(language_id.region);
  850. if (map.key.variants_size == 0 && alias.variants.is_empty())
  851. alias.variants = move(language_id.variants);
  852. language_id = move(alias);
  853. break;
  854. }
  855. }
  856. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id)
  857. {
  858. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  859. auto const* likely_subtag = resolve_likely_subtag(language_id);
  860. if (likely_subtag == nullptr)
  861. return {};
  862. auto maximized = language_id;
  863. auto const& key_script = s_string_list[likely_subtag->key.script];
  864. auto const& key_region = s_string_list[likely_subtag->key.region];
  865. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  866. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  867. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  868. if (maximized.language == "und"sv)
  869. maximized.language = alias_language;
  870. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  871. maximized.script = alias_script;
  872. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  873. maximized.region = alias_region;
  874. return maximized;
  875. }
  876. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
  877. {
  878. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  879. return s_string_list[likely_subtag->alias.region];
  880. return {};
  881. }
  882. }
  883. )~~~");
  884. file.write(generator.as_string_view());
  885. }
  886. int main(int argc, char** argv)
  887. {
  888. char const* generated_header_path = nullptr;
  889. char const* generated_implementation_path = nullptr;
  890. char const* core_path = nullptr;
  891. char const* locale_names_path = nullptr;
  892. char const* misc_path = nullptr;
  893. char const* numbers_path = nullptr;
  894. Core::ArgsParser args_parser;
  895. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  896. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  897. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  898. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  899. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  900. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  901. args_parser.parse(argc, argv);
  902. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  903. if (path.is_empty()) {
  904. warnln("{} is required", flags);
  905. args_parser.print_usage(stderr, argv[0]);
  906. exit(1);
  907. }
  908. auto file_or_error = Core::File::open(path, mode);
  909. if (file_or_error.is_error()) {
  910. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  911. exit(1);
  912. }
  913. return file_or_error.release_value();
  914. };
  915. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  916. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  917. UnicodeLocaleData locale_data;
  918. parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, locale_data);
  919. generate_unicode_locale_header(generated_header_file, locale_data);
  920. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  921. return 0;
  922. }