GenerateUnicodeLocale.cpp 47 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121
  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 parse_all_locales(String core_path, String locale_names_path, String misc_path, String numbers_path, UnicodeLocaleData& locale_data)
  343. {
  344. auto identity_iterator = path_to_dir_iterator(locale_names_path);
  345. auto locale_names_iterator = path_to_dir_iterator(move(locale_names_path));
  346. auto misc_iterator = path_to_dir_iterator(move(misc_path));
  347. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  348. LexicalPath core_supplemental_path(core_path);
  349. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  350. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  351. parse_core_aliases(core_supplemental_path.string(), locale_data);
  352. parse_likely_subtags(core_supplemental_path.string(), locale_data);
  353. auto remove_variants_from_path = [&](String path) -> Optional<String> {
  354. auto parsed_locale = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path));
  355. if (!parsed_locale.has_value())
  356. return {};
  357. StringBuilder builder;
  358. builder.append(locale_data.unique_strings.get(parsed_locale->language));
  359. if (auto script = locale_data.unique_strings.get(parsed_locale->script); !script.is_empty())
  360. builder.appendff("-{}", script);
  361. if (auto region = locale_data.unique_strings.get(parsed_locale->region); !region.is_empty())
  362. builder.appendff("-{}", region);
  363. return builder.build();
  364. };
  365. while (identity_iterator.has_next()) {
  366. auto locale_path = identity_iterator.next_full_path();
  367. VERIFY(Core::File::is_directory(locale_path));
  368. auto language = remove_variants_from_path(locale_path);
  369. if (!language.has_value())
  370. continue;
  371. auto& locale = locale_data.locales.ensure(*language);
  372. parse_identity(locale_path, locale_data, locale);
  373. }
  374. while (locale_names_iterator.has_next()) {
  375. auto locale_path = locale_names_iterator.next_full_path();
  376. VERIFY(Core::File::is_directory(locale_path));
  377. auto language = remove_variants_from_path(locale_path);
  378. if (!language.has_value())
  379. continue;
  380. auto& locale = locale_data.locales.ensure(*language);
  381. parse_locale_languages(locale_path, locale_data, locale);
  382. parse_locale_territories(locale_path, locale_data, locale);
  383. parse_locale_scripts(locale_path, locale_data, locale);
  384. }
  385. while (misc_iterator.has_next()) {
  386. auto misc_path = misc_iterator.next_full_path();
  387. VERIFY(Core::File::is_directory(misc_path));
  388. auto language = remove_variants_from_path(misc_path);
  389. if (!language.has_value())
  390. continue;
  391. auto& locale = locale_data.locales.ensure(*language);
  392. parse_locale_list_patterns(misc_path, locale_data, locale);
  393. }
  394. while (numbers_iterator.has_next()) {
  395. auto numbers_path = numbers_iterator.next_full_path();
  396. VERIFY(Core::File::is_directory(numbers_path));
  397. auto language = remove_variants_from_path(numbers_path);
  398. if (!language.has_value())
  399. continue;
  400. auto& locale = locale_data.locales.ensure(*language);
  401. parse_locale_currencies(numbers_path, locale_data, locale);
  402. parse_numeric_keywords(numbers_path, locale_data, locale);
  403. }
  404. parse_default_content_locales(move(core_path), locale_data);
  405. }
  406. static String format_identifier(StringView owner, String identifier)
  407. {
  408. identifier = identifier.replace("-"sv, "_"sv, true);
  409. if (all_of(identifier, is_ascii_digit))
  410. return String::formatted("{}_{}", owner[0], identifier);
  411. if (is_ascii_lower_alpha(identifier[0]))
  412. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  413. return identifier;
  414. }
  415. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  416. {
  417. StringBuilder builder;
  418. SourceGenerator generator { builder };
  419. generator.append(R"~~~(
  420. #pragma once
  421. #include <AK/Optional.h>
  422. #include <AK/StringView.h>
  423. #include <AK/Types.h>
  424. #include <AK/Vector.h>
  425. #include <LibUnicode/Forward.h>
  426. namespace Unicode {
  427. )~~~");
  428. auto locales = locale_data.locales.keys();
  429. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales, locale_data.locale_aliases);
  430. generate_enum(generator, format_identifier, "Language"sv, {}, locale_data.languages);
  431. generate_enum(generator, format_identifier, "Territory"sv, {}, locale_data.territories);
  432. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, locale_data.scripts);
  433. generate_enum(generator, format_identifier, "Currency"sv, {}, locale_data.currencies);
  434. generate_enum(generator, format_identifier, "Key"sv, {}, locale_data.keywords);
  435. generate_enum(generator, format_identifier, "Variant"sv, {}, locale_data.variants);
  436. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, locale_data.list_pattern_types);
  437. generate_enum(generator, format_identifier, "ListPatternStyle"sv, {}, locale_data.list_pattern_styles);
  438. generator.append(R"~~~(
  439. namespace Detail {
  440. Optional<Locale> locale_from_string(StringView locale);
  441. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  442. Optional<Language> language_from_string(StringView language);
  443. Optional<StringView> resolve_language_alias(StringView language);
  444. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  445. Optional<Territory> territory_from_string(StringView territory);
  446. Optional<StringView> resolve_territory_alias(StringView territory);
  447. Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
  448. Optional<ScriptTag> script_tag_from_string(StringView script_tag);
  449. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  450. Optional<StringView> get_locale_long_currency_mapping(StringView locale, StringView currency);
  451. Optional<StringView> get_locale_short_currency_mapping(StringView locale, StringView currency);
  452. Optional<StringView> get_locale_narrow_currency_mapping(StringView locale, StringView currency);
  453. Optional<StringView> get_locale_numeric_currency_mapping(StringView locale, StringView currency);
  454. Optional<Currency> currency_from_string(StringView currency);
  455. Optional<StringView> get_locale_key_mapping(StringView locale, StringView key);
  456. Optional<Key> key_from_string(StringView key);
  457. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style);
  458. Optional<ListPatternType> list_pattern_type_from_string(StringView list_pattern_type);
  459. Optional<ListPatternStyle> list_pattern_style_from_string(StringView list_pattern_style);
  460. Optional<StringView> resolve_variant_alias(StringView variant);
  461. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  462. void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
  463. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id);
  464. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
  465. }
  466. }
  467. )~~~");
  468. file.write(generator.as_string_view());
  469. }
  470. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  471. {
  472. StringBuilder builder;
  473. SourceGenerator generator { builder };
  474. generator.set("string_index_type"sv, s_string_index_type);
  475. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  476. generator.set("territories_size", String::number(locale_data.territories.size()));
  477. generator.set("variants_size", String::number(locale_data.max_variant_size));
  478. generator.append(R"~~~(
  479. #include <AK/Array.h>
  480. #include <AK/BinarySearch.h>
  481. #include <AK/Span.h>
  482. #include <LibUnicode/Locale.h>
  483. #include <LibUnicode/UnicodeLocale.h>
  484. namespace Unicode::Detail {
  485. struct Patterns {
  486. ListPatternType type;
  487. ListPatternStyle style;
  488. @string_index_type@ start { 0 };
  489. @string_index_type@ middle { 0 };
  490. @string_index_type@ end { 0 };
  491. @string_index_type@ pair { 0 };
  492. };
  493. )~~~");
  494. locale_data.unique_strings.generate(generator);
  495. auto append_index = [&](auto index) {
  496. generator.append(String::formatted(", {}", index));
  497. };
  498. auto append_list_and_size = [&](auto const& list) {
  499. if (list.is_empty()) {
  500. generator.append(", {}, 0");
  501. return;
  502. }
  503. bool first = true;
  504. generator.append(", {");
  505. for (auto const& item : list) {
  506. generator.append(first ? " " : ", ");
  507. generator.append(String::number(item));
  508. first = false;
  509. }
  510. generator.append(String::formatted(" }}, {}", list.size()));
  511. };
  512. auto append_string_index_list = [&](String name, auto const& keys, auto const& mappings) {
  513. generator.set("name", name);
  514. generator.set("size", String::number(keys.size()));
  515. generator.append(R"~~~(
  516. static constexpr Array<@string_index_type@, @size@> @name@ { {
  517. )~~~");
  518. constexpr size_t max_values_per_row = 30;
  519. size_t values_in_current_row = 0;
  520. for (auto const& key : keys) {
  521. if (values_in_current_row++ > 0)
  522. generator.append(" ");
  523. if (auto it = mappings.find(key); it != mappings.end())
  524. generator.set("mapping"sv, String::number(it->value));
  525. else
  526. generator.set("mapping"sv, "0"sv);
  527. generator.append("@mapping@,");
  528. if (values_in_current_row == max_values_per_row) {
  529. values_in_current_row = 0;
  530. generator.append("\n ");
  531. }
  532. }
  533. generator.append(R"~~~(
  534. } };
  535. )~~~");
  536. };
  537. auto append_list_patterns = [&](StringView name, Vector<ListPatterns> const& list_patterns) {
  538. generator.set("name", name);
  539. generator.set("size", String::number(list_patterns.size()));
  540. generator.append(R"~~~(
  541. static constexpr Array<Patterns, @size@> @name@ { {)~~~");
  542. for (auto const& list_pattern : list_patterns) {
  543. generator.set("type"sv, String::formatted("ListPatternType::{}", format_identifier({}, list_pattern.type)));
  544. generator.set("style"sv, String::formatted("ListPatternStyle::{}", format_identifier({}, list_pattern.style)));
  545. generator.set("start"sv, String::number(list_pattern.start));
  546. generator.set("middle"sv, String::number(list_pattern.middle));
  547. generator.set("end"sv, String::number(list_pattern.end));
  548. generator.set("pair"sv, String::number(list_pattern.pair));
  549. generator.append(R"~~~(
  550. { @type@, @style@, @start@, @middle@, @end@, @pair@ },)~~~");
  551. }
  552. generator.append(R"~~~(
  553. } };
  554. )~~~");
  555. };
  556. 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); });
  557. 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); });
  558. 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); });
  559. 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); });
  560. 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); });
  561. 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); });
  562. 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); });
  563. 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); });
  564. 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); });
  565. generator.append(R"~~~(
  566. struct CanonicalLanguageID {
  567. Unicode::LanguageID to_unicode_language_id() const
  568. {
  569. Unicode::LanguageID language_id {};
  570. language_id.variants.ensure_capacity(variants_size);
  571. language_id.language = s_string_list[language];
  572. if (script != 0)
  573. language_id.script = s_string_list[script];
  574. if (region != 0)
  575. language_id.region = s_string_list[region];
  576. for (size_t i = 0; i < variants_size; ++i)
  577. language_id.variants.append(s_string_list[variants[i]]);
  578. return language_id;
  579. }
  580. bool matches_variants(Vector<String> const& other_variants) const {
  581. if (variants_size == 0)
  582. return true;
  583. if (other_variants.size() != variants_size)
  584. return false;
  585. for (size_t i = 0; i < variants_size; ++i) {
  586. if (s_string_list[variants[i]] != other_variants[i])
  587. return false;
  588. }
  589. return true;
  590. };
  591. @string_index_type@ language { 0 };
  592. @string_index_type@ script { 0 };
  593. @string_index_type@ region { 0 };
  594. Array<@string_index_type@, @variants_size@> variants {};
  595. size_t variants_size { 0 };
  596. };
  597. struct LanguageMapping {
  598. CanonicalLanguageID key;
  599. CanonicalLanguageID alias;
  600. };
  601. )~~~");
  602. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  603. generator.set("size", String::number(mappings.size()));
  604. generator.set("name"sv, name);
  605. generator.append(R"~~~(
  606. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  607. )~~~");
  608. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  609. auto const& lhs_language = locale_data.unique_strings.get(lhs.key.language);
  610. auto const& rhs_language = locale_data.unique_strings.get(rhs.key.language);
  611. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  612. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  613. return false;
  614. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  615. return true;
  616. return lhs_language < rhs_language;
  617. });
  618. for (auto const& mapping : mappings) {
  619. generator.set("language"sv, String::number(mapping.key.language));
  620. generator.append(" { { @language@");
  621. append_index(mapping.key.script);
  622. append_index(mapping.key.region);
  623. append_list_and_size(mapping.key.variants);
  624. generator.set("language"sv, String::number(mapping.alias.language));
  625. generator.append(" }, { @language@");
  626. append_index(mapping.alias.script);
  627. append_index(mapping.alias.region);
  628. append_list_and_size(mapping.alias.variants);
  629. generator.append(" } },\n");
  630. }
  631. generator.append("} };\n");
  632. };
  633. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  634. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  635. generator.append(R"~~~(
  636. static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& language_id)
  637. {
  638. // https://unicode.org/reports/tr35/#Likely_Subtags
  639. enum class State {
  640. LanguageScriptRegion,
  641. LanguageRegion,
  642. LanguageScript,
  643. Language,
  644. UndScript,
  645. Done,
  646. };
  647. auto state = State::LanguageScriptRegion;
  648. while (state != State::Done) {
  649. Unicode::LanguageID search_key;
  650. switch (state) {
  651. case State::LanguageScriptRegion:
  652. state = State::LanguageRegion;
  653. if (!language_id.script.has_value() || !language_id.region.has_value())
  654. continue;
  655. search_key.language = *language_id.language;
  656. search_key.script = *language_id.script;
  657. search_key.region = *language_id.region;
  658. break;
  659. case State::LanguageRegion:
  660. state = State::LanguageScript;
  661. if (!language_id.region.has_value())
  662. continue;
  663. search_key.language = *language_id.language;
  664. search_key.region = *language_id.region;
  665. break;
  666. case State::LanguageScript:
  667. state = State::Language;
  668. if (!language_id.script.has_value())
  669. continue;
  670. search_key.language = *language_id.language;
  671. search_key.script = *language_id.script;
  672. break;
  673. case State::Language:
  674. state = State::UndScript;
  675. search_key.language = *language_id.language;
  676. break;
  677. case State::UndScript:
  678. state = State::Done;
  679. if (!language_id.script.has_value())
  680. continue;
  681. search_key.language = "und"sv;
  682. search_key.script = *language_id.script;
  683. break;
  684. default:
  685. VERIFY_NOT_REACHED();
  686. }
  687. for (auto const& map : s_likely_subtags) {
  688. auto const& key_language = s_string_list[map.key.language];
  689. auto const& key_script = s_string_list[map.key.script];
  690. auto const& key_region = s_string_list[map.key.region];
  691. if (key_language != search_key.language)
  692. continue;
  693. if (!key_script.is_empty() || search_key.script.has_value()) {
  694. if (key_script != search_key.script)
  695. continue;
  696. }
  697. if (!key_region.is_empty() || search_key.region.has_value()) {
  698. if (key_region != search_key.region)
  699. continue;
  700. }
  701. return &map;
  702. }
  703. }
  704. return nullptr;
  705. }
  706. )~~~");
  707. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name) {
  708. generator.set("enum_snake", enum_snake);
  709. generator.set("from_string_name", from_string_name);
  710. generator.set("collection_name", collection_name);
  711. generator.append(R"~~~(
  712. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  713. {
  714. auto locale_value = locale_from_string(locale);
  715. if (!locale_value.has_value())
  716. return {};
  717. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  718. if (!@enum_snake@_value.has_value())
  719. return {};
  720. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  721. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  722. auto const& mappings = @collection_name@.at(locale_index);
  723. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  724. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  725. if (@enum_snake@_mapping.is_empty())
  726. return {};
  727. return @enum_snake@_mapping;
  728. }
  729. )~~~");
  730. };
  731. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) {
  732. HashValueMap<String> hashes;
  733. hashes.ensure_capacity(values.size());
  734. for (auto const& value : values)
  735. hashes.set(value.hash(), format_identifier(enum_title, value));
  736. for (auto const& alias : aliases)
  737. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  738. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  739. };
  740. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  741. HashValueMap<StringIndexType> hashes;
  742. hashes.ensure_capacity(aliases.size());
  743. for (auto const& alias : aliases)
  744. hashes.set(alias.key.hash(), alias.value);
  745. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), "StringView"sv, "s_string_list[{}]"sv);
  746. };
  747. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys(), locale_data.locale_aliases);
  748. append_mapping_search("language"sv, "language"sv, "s_languages"sv);
  749. append_from_string("Language"sv, "language"sv, locale_data.languages);
  750. append_alias_search("language"sv, locale_data.language_aliases);
  751. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv);
  752. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  753. append_alias_search("territory"sv, locale_data.territory_aliases);
  754. append_mapping_search("script_tag"sv, "script_tag"sv, "s_scripts"sv);
  755. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  756. append_alias_search("script_tag"sv, locale_data.script_aliases);
  757. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv);
  758. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv);
  759. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv);
  760. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv);
  761. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  762. append_mapping_search("key"sv, "key"sv, "s_keywords"sv);
  763. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  764. append_alias_search("variant"sv, locale_data.variant_aliases);
  765. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  766. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  767. append_from_string("ListPatternStyle"sv, "list_pattern_style"sv, locale_data.list_pattern_styles);
  768. generator.append(R"~~~(
  769. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style)
  770. {
  771. auto locale_value = locale_from_string(locale);
  772. if (!locale_value.has_value())
  773. return {};
  774. auto type_value = list_pattern_type_from_string(list_pattern_type);
  775. if (!type_value.has_value())
  776. return {};
  777. auto style_value = list_pattern_style_from_string(list_pattern_style);
  778. if (!style_value.has_value())
  779. return {};
  780. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  781. auto const& locale_list_patterns = s_list_patterns.at(locale_index);
  782. for (auto const& list_patterns : locale_list_patterns) {
  783. if ((list_patterns.type == type_value) && (list_patterns.style == style_value)) {
  784. auto const& start = s_string_list[list_patterns.start];
  785. auto const& middle = s_string_list[list_patterns.middle];
  786. auto const& end = s_string_list[list_patterns.end];
  787. auto const& pair = s_string_list[list_patterns.pair];
  788. return ListPatterns { start, middle, end, pair };
  789. }
  790. }
  791. return {};
  792. }
  793. void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
  794. {
  795. for (auto const& map : s_complex_alias) {
  796. auto const& key_language = s_string_list[map.key.language];
  797. auto const& key_script = s_string_list[map.key.script];
  798. auto const& key_region = s_string_list[map.key.region];
  799. if ((key_language != language_id.language) && (key_language != "und"sv))
  800. continue;
  801. if (!key_script.is_empty() && (key_script != language_id.script))
  802. continue;
  803. if (!key_region.is_empty() && (key_region != language_id.region))
  804. continue;
  805. if (!map.key.matches_variants(language_id.variants))
  806. continue;
  807. auto alias = map.alias.to_unicode_language_id();
  808. if (alias.language == "und"sv)
  809. alias.language = move(language_id.language);
  810. if (key_script.is_empty() && !alias.script.has_value())
  811. alias.script = move(language_id.script);
  812. if (key_region.is_empty() && !alias.region.has_value())
  813. alias.region = move(language_id.region);
  814. if (map.key.variants_size == 0 && alias.variants.is_empty())
  815. alias.variants = move(language_id.variants);
  816. language_id = move(alias);
  817. break;
  818. }
  819. }
  820. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id)
  821. {
  822. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  823. auto const* likely_subtag = resolve_likely_subtag(language_id);
  824. if (likely_subtag == nullptr)
  825. return {};
  826. auto maximized = language_id;
  827. auto const& key_script = s_string_list[likely_subtag->key.script];
  828. auto const& key_region = s_string_list[likely_subtag->key.region];
  829. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  830. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  831. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  832. if (maximized.language == "und"sv)
  833. maximized.language = alias_language;
  834. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  835. maximized.script = alias_script;
  836. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  837. maximized.region = alias_region;
  838. return maximized;
  839. }
  840. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
  841. {
  842. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  843. return s_string_list[likely_subtag->alias.region];
  844. return {};
  845. }
  846. }
  847. )~~~");
  848. file.write(generator.as_string_view());
  849. }
  850. int main(int argc, char** argv)
  851. {
  852. char const* generated_header_path = nullptr;
  853. char const* generated_implementation_path = nullptr;
  854. char const* core_path = nullptr;
  855. char const* locale_names_path = nullptr;
  856. char const* misc_path = nullptr;
  857. char const* numbers_path = nullptr;
  858. Core::ArgsParser args_parser;
  859. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  860. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  861. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  862. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  863. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  864. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  865. args_parser.parse(argc, argv);
  866. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  867. if (path.is_empty()) {
  868. warnln("{} is required", flags);
  869. args_parser.print_usage(stderr, argv[0]);
  870. exit(1);
  871. }
  872. auto file_or_error = Core::File::open(path, mode);
  873. if (file_or_error.is_error()) {
  874. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  875. exit(1);
  876. }
  877. return file_or_error.release_value();
  878. };
  879. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  880. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  881. UnicodeLocaleData locale_data;
  882. parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, locale_data);
  883. generate_unicode_locale_header(generated_header_file, locale_data);
  884. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  885. return 0;
  886. }