GenerateUnicodeLocale.cpp 46 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093
  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<String> languages;
  54. Vector<String> territories;
  55. Vector<String> scripts;
  56. Vector<String> variants;
  57. Vector<String> currencies;
  58. Vector<String> keywords;
  59. Vector<String> list_pattern_types;
  60. Vector<String> list_pattern_styles;
  61. HashMap<String, StringIndexType> language_aliases;
  62. HashMap<String, StringIndexType> territory_aliases;
  63. HashMap<String, StringIndexType> script_aliases;
  64. HashMap<String, StringIndexType> variant_aliases;
  65. HashMap<String, StringIndexType> subdivision_aliases;
  66. Vector<LanguageMapping> complex_mappings;
  67. Vector<LanguageMapping> likely_subtags;
  68. size_t max_variant_size { 0 };
  69. };
  70. static Optional<LanguageMapping> parse_language_mapping(UnicodeLocaleData& locale_data, StringView key, StringView alias)
  71. {
  72. auto parsed_key = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, key);
  73. if (!parsed_key.has_value())
  74. return {};
  75. auto parsed_alias = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, alias);
  76. if (!parsed_alias.has_value())
  77. return {};
  78. return LanguageMapping { parsed_key.release_value(), parsed_alias.release_value() };
  79. }
  80. static void parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
  81. {
  82. LexicalPath core_aliases_path(move(core_supplemental_path));
  83. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  84. VERIFY(Core::File::exists(core_aliases_path.string()));
  85. auto core_aliases_file_or_error = Core::File::open(core_aliases_path.string(), Core::OpenMode::ReadOnly);
  86. VERIFY(!core_aliases_file_or_error.is_error());
  87. auto core_aliases = JsonParser(core_aliases_file_or_error.value()->read_all()).parse();
  88. VERIFY(core_aliases.has_value());
  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 = JsonParser(likely_subtags_file_or_error.value()->read_all()).parse();
  121. VERIFY(likely_subtags.has_value());
  122. auto const& supplemental_object = likely_subtags->as_object().get("supplemental"sv);
  123. auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
  124. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  125. auto mapping = parse_language_mapping(locale_data, key, value.as_string());
  126. if (!mapping.has_value())
  127. return;
  128. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  129. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  130. locale_data.likely_subtags.append(mapping.release_value());
  131. });
  132. }
  133. static void parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  134. {
  135. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  136. languages_path = languages_path.append("languages.json"sv);
  137. VERIFY(Core::File::exists(languages_path.string()));
  138. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  139. VERIFY(!languages_file_or_error.is_error());
  140. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  141. VERIFY(languages.has_value());
  142. auto const& main_object = languages->as_object().get("main"sv);
  143. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  144. auto const& identity_object = locale_object.as_object().get("identity"sv);
  145. auto const& language_string = identity_object.as_object().get("language"sv);
  146. auto const& territory_string = identity_object.as_object().get("territory"sv);
  147. auto const& variant_string = identity_object.as_object().get("variant"sv);
  148. locale.language = language_string.as_string();
  149. if (!locale_data.languages.contains_slow(locale.language))
  150. locale_data.languages.append(locale.language);
  151. if (territory_string.is_string()) {
  152. locale.territory = territory_string.as_string();
  153. if (!locale_data.territories.contains_slow(*locale.territory))
  154. locale_data.territories.append(*locale.territory);
  155. }
  156. if (variant_string.is_string()) {
  157. locale.variant = variant_string.as_string();
  158. if (!locale_data.variants.contains_slow(*locale.variant))
  159. locale_data.variants.append(*locale.variant);
  160. }
  161. }
  162. static void parse_locale_languages(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  163. {
  164. LexicalPath languages_path(move(locale_path));
  165. languages_path = languages_path.append("languages.json"sv);
  166. VERIFY(Core::File::exists(languages_path.string()));
  167. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  168. VERIFY(!languages_file_or_error.is_error());
  169. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  170. VERIFY(languages.has_value());
  171. auto const& main_object = languages->as_object().get("main"sv);
  172. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  173. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  174. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  175. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  176. if (!locale_data.languages.contains_slow(key))
  177. return;
  178. auto index = locale_data.unique_strings.ensure(value.as_string());
  179. locale.languages.set(key, index);
  180. });
  181. }
  182. static void parse_locale_territories(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  183. {
  184. LexicalPath territories_path(move(locale_path));
  185. territories_path = territories_path.append("territories.json"sv);
  186. VERIFY(Core::File::exists(territories_path.string()));
  187. auto territories_file_or_error = Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly);
  188. VERIFY(!territories_file_or_error.is_error());
  189. auto territories = JsonParser(territories_file_or_error.value()->read_all()).parse();
  190. VERIFY(territories.has_value());
  191. auto const& main_object = territories->as_object().get("main"sv);
  192. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  193. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  194. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  195. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  196. if (!locale_data.territories.contains_slow(key))
  197. return;
  198. auto index = locale_data.unique_strings.ensure(value.as_string());
  199. locale.territories.set(key, index);
  200. });
  201. }
  202. static void parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  203. {
  204. LexicalPath scripts_path(move(locale_path));
  205. scripts_path = scripts_path.append("scripts.json"sv);
  206. VERIFY(Core::File::exists(scripts_path.string()));
  207. auto scripts_file_or_error = Core::File::open(scripts_path.string(), Core::OpenMode::ReadOnly);
  208. VERIFY(!scripts_file_or_error.is_error());
  209. auto scripts = JsonParser(scripts_file_or_error.value()->read_all()).parse();
  210. VERIFY(scripts.has_value());
  211. auto const& main_object = scripts->as_object().get("main"sv);
  212. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  213. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  214. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  215. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  216. auto index = locale_data.unique_strings.ensure(value.as_string());
  217. locale.scripts.set(key, index);
  218. if (!locale_data.scripts.contains_slow(key))
  219. locale_data.scripts.append(key);
  220. });
  221. }
  222. static void parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
  223. {
  224. LexicalPath list_patterns_path(move(misc_path));
  225. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  226. VERIFY(Core::File::exists(list_patterns_path.string()));
  227. auto list_patterns_file_or_error = Core::File::open(list_patterns_path.string(), Core::OpenMode::ReadOnly);
  228. VERIFY(!list_patterns_file_or_error.is_error());
  229. auto list_patterns = JsonParser(list_patterns_file_or_error.value()->read_all()).parse();
  230. VERIFY(list_patterns.has_value());
  231. auto const& main_object = list_patterns->as_object().get("main"sv);
  232. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  233. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  234. auto list_pattern_type = [](StringView key) {
  235. if (key.contains("type-standard"sv))
  236. return "conjunction"sv;
  237. if (key.contains("type-or"sv))
  238. return "disjunction"sv;
  239. if (key.contains("type-unit"sv))
  240. return "unit"sv;
  241. VERIFY_NOT_REACHED();
  242. };
  243. auto list_pattern_style = [](StringView key) {
  244. if (key.contains("short"sv))
  245. return "short"sv;
  246. if (key.contains("narrow"sv))
  247. return "narrow"sv;
  248. return "long"sv;
  249. };
  250. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  251. auto type = list_pattern_type(key);
  252. auto style = list_pattern_style(key);
  253. auto start = locale_data.unique_strings.ensure(value.as_object().get("start"sv).as_string());
  254. auto middle = locale_data.unique_strings.ensure(value.as_object().get("middle"sv).as_string());
  255. auto end = locale_data.unique_strings.ensure(value.as_object().get("end"sv).as_string());
  256. auto pair = locale_data.unique_strings.ensure(value.as_object().get("2"sv).as_string());
  257. if (!locale_data.list_pattern_types.contains_slow(type))
  258. locale_data.list_pattern_types.append(type);
  259. if (!locale_data.list_pattern_styles.contains_slow(style))
  260. locale_data.list_pattern_styles.append(style);
  261. locale.list_patterns.append({ move(type), move(style), move(start), move(middle), move(end), move(pair) });
  262. });
  263. }
  264. static void parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  265. {
  266. LexicalPath currencies_path(move(numbers_path));
  267. currencies_path = currencies_path.append("currencies.json"sv);
  268. VERIFY(Core::File::exists(currencies_path.string()));
  269. auto currencies_file_or_error = Core::File::open(currencies_path.string(), Core::OpenMode::ReadOnly);
  270. VERIFY(!currencies_file_or_error.is_error());
  271. auto currencies = JsonParser(currencies_file_or_error.value()->read_all()).parse();
  272. VERIFY(currencies.has_value());
  273. auto const& main_object = currencies->as_object().get("main"sv);
  274. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  275. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  276. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  277. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  278. auto const& long_name = value.as_object().get("displayName"sv);
  279. auto const& short_name = value.as_object().get("symbol"sv);
  280. auto const& narrow_name = value.as_object().get("symbol-alt-narrow"sv);
  281. auto const& numeric_name = value.as_object().get("displayName-count-other"sv);
  282. locale.long_currencies.set(key, locale_data.unique_strings.ensure(long_name.as_string()));
  283. locale.short_currencies.set(key, locale_data.unique_strings.ensure(short_name.as_string()));
  284. locale.narrow_currencies.set(key, narrow_name.is_null() ? 0 : locale_data.unique_strings.ensure(narrow_name.as_string()));
  285. locale.numeric_currencies.set(key, locale_data.unique_strings.ensure(numeric_name.is_null() ? long_name.as_string() : numeric_name.as_string()));
  286. if (!locale_data.currencies.contains_slow(key))
  287. locale_data.currencies.append(key);
  288. });
  289. }
  290. static void parse_numeric_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  291. {
  292. static constexpr StringView key = "nu"sv;
  293. LexicalPath numbers_path(move(locale_numbers_path));
  294. numbers_path = numbers_path.append("numbers.json"sv);
  295. VERIFY(Core::File::exists(numbers_path.string()));
  296. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  297. VERIFY(!numbers_file_or_error.is_error());
  298. auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
  299. VERIFY(numbers.has_value());
  300. auto const& main_object = numbers->as_object().get("main"sv);
  301. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  302. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  303. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  304. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  305. Vector<String> keyword_values {};
  306. keyword_values.append(default_numbering_system_object.as_string());
  307. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  308. auto keyword_value = value.as_string();
  309. if (!keyword_values.contains_slow(keyword_value))
  310. keyword_values.append(move(keyword_value));
  311. });
  312. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  313. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  314. return;
  315. auto keyword_value = value.as_string();
  316. if (!keyword_values.contains_slow(keyword_value))
  317. keyword_values.append(move(keyword_value));
  318. });
  319. StringBuilder builder;
  320. builder.join(',', keyword_values);
  321. auto index = locale_data.unique_strings.ensure(builder.build());
  322. locale.keywords.set(key, index);
  323. if (!locale_data.keywords.contains_slow(key))
  324. locale_data.keywords.append(key);
  325. }
  326. static void parse_all_locales(String core_path, String locale_names_path, String misc_path, String numbers_path, UnicodeLocaleData& locale_data)
  327. {
  328. auto identity_iterator = path_to_dir_iterator(locale_names_path);
  329. auto locale_names_iterator = path_to_dir_iterator(move(locale_names_path));
  330. auto misc_iterator = path_to_dir_iterator(move(misc_path));
  331. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  332. LexicalPath core_supplemental_path(core_path);
  333. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  334. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  335. parse_core_aliases(core_supplemental_path.string(), locale_data);
  336. parse_likely_subtags(core_supplemental_path.string(), locale_data);
  337. auto remove_variants_from_path = [&](String path) -> Optional<String> {
  338. auto parsed_locale = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path));
  339. if (!parsed_locale.has_value())
  340. return {};
  341. StringBuilder builder;
  342. builder.append(locale_data.unique_strings.get(parsed_locale->language));
  343. if (auto script = locale_data.unique_strings.get(parsed_locale->script); !script.is_empty())
  344. builder.appendff("-{}", script);
  345. if (auto region = locale_data.unique_strings.get(parsed_locale->region); !region.is_empty())
  346. builder.appendff("-{}", region);
  347. return builder.build();
  348. };
  349. while (identity_iterator.has_next()) {
  350. auto locale_path = identity_iterator.next_full_path();
  351. VERIFY(Core::File::is_directory(locale_path));
  352. auto language = remove_variants_from_path(locale_path);
  353. if (!language.has_value())
  354. continue;
  355. auto& locale = locale_data.locales.ensure(*language);
  356. parse_identity(locale_path, locale_data, locale);
  357. }
  358. while (locale_names_iterator.has_next()) {
  359. auto locale_path = locale_names_iterator.next_full_path();
  360. VERIFY(Core::File::is_directory(locale_path));
  361. auto language = remove_variants_from_path(locale_path);
  362. if (!language.has_value())
  363. continue;
  364. auto& locale = locale_data.locales.ensure(*language);
  365. parse_locale_languages(locale_path, locale_data, locale);
  366. parse_locale_territories(locale_path, locale_data, locale);
  367. parse_locale_scripts(locale_path, locale_data, locale);
  368. }
  369. while (misc_iterator.has_next()) {
  370. auto misc_path = misc_iterator.next_full_path();
  371. VERIFY(Core::File::is_directory(misc_path));
  372. auto language = remove_variants_from_path(misc_path);
  373. if (!language.has_value())
  374. continue;
  375. auto& locale = locale_data.locales.ensure(*language);
  376. parse_locale_list_patterns(misc_path, locale_data, locale);
  377. }
  378. while (numbers_iterator.has_next()) {
  379. auto numbers_path = numbers_iterator.next_full_path();
  380. VERIFY(Core::File::is_directory(numbers_path));
  381. auto language = remove_variants_from_path(numbers_path);
  382. if (!language.has_value())
  383. continue;
  384. auto& locale = locale_data.locales.ensure(*language);
  385. parse_locale_currencies(numbers_path, locale_data, locale);
  386. parse_numeric_keywords(numbers_path, locale_data, locale);
  387. }
  388. parse_default_content_locales(move(core_path), locale_data);
  389. }
  390. static String format_identifier(StringView owner, String identifier)
  391. {
  392. identifier = identifier.replace("-"sv, "_"sv, true);
  393. if (all_of(identifier, is_ascii_digit))
  394. return String::formatted("{}_{}", owner[0], identifier);
  395. if (is_ascii_lower_alpha(identifier[0]))
  396. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  397. return identifier;
  398. }
  399. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  400. {
  401. StringBuilder builder;
  402. SourceGenerator generator { builder };
  403. generator.append(R"~~~(
  404. #pragma once
  405. #include <AK/Optional.h>
  406. #include <AK/StringView.h>
  407. #include <AK/Types.h>
  408. #include <AK/Vector.h>
  409. #include <LibUnicode/Forward.h>
  410. namespace Unicode {
  411. )~~~");
  412. auto locales = locale_data.locales.keys();
  413. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales);
  414. generate_enum(generator, format_identifier, "Language"sv, {}, locale_data.languages);
  415. generate_enum(generator, format_identifier, "Territory"sv, {}, locale_data.territories);
  416. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, locale_data.scripts);
  417. generate_enum(generator, format_identifier, "Currency"sv, {}, locale_data.currencies);
  418. generate_enum(generator, format_identifier, "Key"sv, {}, locale_data.keywords);
  419. generate_enum(generator, format_identifier, "Variant"sv, {}, locale_data.variants);
  420. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, locale_data.list_pattern_types);
  421. generate_enum(generator, format_identifier, "ListPatternStyle"sv, {}, locale_data.list_pattern_styles);
  422. generator.append(R"~~~(
  423. namespace Detail {
  424. Optional<Locale> locale_from_string(StringView locale);
  425. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  426. Optional<Language> language_from_string(StringView language);
  427. Optional<StringView> resolve_language_alias(StringView language);
  428. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  429. Optional<Territory> territory_from_string(StringView territory);
  430. Optional<StringView> resolve_territory_alias(StringView territory);
  431. Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
  432. Optional<ScriptTag> script_tag_from_string(StringView script_tag);
  433. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  434. Optional<StringView> get_locale_long_currency_mapping(StringView locale, StringView currency);
  435. Optional<StringView> get_locale_short_currency_mapping(StringView locale, StringView currency);
  436. Optional<StringView> get_locale_narrow_currency_mapping(StringView locale, StringView currency);
  437. Optional<StringView> get_locale_numeric_currency_mapping(StringView locale, StringView currency);
  438. Optional<Currency> currency_from_string(StringView currency);
  439. Optional<StringView> get_locale_key_mapping(StringView locale, StringView key);
  440. Optional<Key> key_from_string(StringView key);
  441. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style);
  442. Optional<ListPatternType> list_pattern_type_from_string(StringView list_pattern_type);
  443. Optional<ListPatternStyle> list_pattern_style_from_string(StringView list_pattern_style);
  444. Optional<StringView> resolve_variant_alias(StringView variant);
  445. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  446. void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
  447. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id);
  448. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
  449. }
  450. }
  451. )~~~");
  452. file.write(generator.as_string_view());
  453. }
  454. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  455. {
  456. StringBuilder builder;
  457. SourceGenerator generator { builder };
  458. generator.set("string_index_type"sv, s_string_index_type);
  459. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  460. generator.set("territories_size", String::number(locale_data.territories.size()));
  461. generator.set("variants_size", String::number(locale_data.max_variant_size));
  462. generator.append(R"~~~(
  463. #include <AK/Array.h>
  464. #include <AK/BinarySearch.h>
  465. #include <AK/Span.h>
  466. #include <LibUnicode/Locale.h>
  467. #include <LibUnicode/UnicodeLocale.h>
  468. namespace Unicode::Detail {
  469. struct Patterns {
  470. ListPatternType type;
  471. ListPatternStyle style;
  472. @string_index_type@ start { 0 };
  473. @string_index_type@ middle { 0 };
  474. @string_index_type@ end { 0 };
  475. @string_index_type@ pair { 0 };
  476. };
  477. )~~~");
  478. locale_data.unique_strings.generate(generator);
  479. auto append_index = [&](auto index) {
  480. generator.append(String::formatted(", {}", index));
  481. };
  482. auto append_list_and_size = [&](auto const& list) {
  483. if (list.is_empty()) {
  484. generator.append(", {}, 0");
  485. return;
  486. }
  487. bool first = true;
  488. generator.append(", {");
  489. for (auto const& item : list) {
  490. generator.append(first ? " " : ", ");
  491. generator.append(String::number(item));
  492. first = false;
  493. }
  494. generator.append(String::formatted(" }}, {}", list.size()));
  495. };
  496. auto append_string_index_list = [&](String name, auto const& keys, auto const& mappings) {
  497. generator.set("name", name);
  498. generator.set("size", String::number(keys.size()));
  499. generator.append(R"~~~(
  500. static constexpr Array<@string_index_type@, @size@> @name@ { {
  501. )~~~");
  502. constexpr size_t max_values_per_row = 30;
  503. size_t values_in_current_row = 0;
  504. for (auto const& key : keys) {
  505. if (values_in_current_row++ > 0)
  506. generator.append(" ");
  507. if (auto it = mappings.find(key); it != mappings.end())
  508. generator.set("mapping"sv, String::number(it->value));
  509. else
  510. generator.set("mapping"sv, "0"sv);
  511. generator.append("@mapping@,");
  512. if (values_in_current_row == max_values_per_row) {
  513. values_in_current_row = 0;
  514. generator.append("\n ");
  515. }
  516. }
  517. generator.append(R"~~~(
  518. } };
  519. )~~~");
  520. };
  521. auto append_list_patterns = [&](StringView name, Vector<ListPatterns> const& list_patterns) {
  522. generator.set("name", name);
  523. generator.set("size", String::number(list_patterns.size()));
  524. generator.append(R"~~~(
  525. static constexpr Array<Patterns, @size@> @name@ { {)~~~");
  526. for (auto const& list_pattern : list_patterns) {
  527. generator.set("type"sv, String::formatted("ListPatternType::{}", format_identifier({}, list_pattern.type)));
  528. generator.set("style"sv, String::formatted("ListPatternStyle::{}", format_identifier({}, list_pattern.style)));
  529. generator.set("start"sv, String::number(list_pattern.start));
  530. generator.set("middle"sv, String::number(list_pattern.middle));
  531. generator.set("end"sv, String::number(list_pattern.end));
  532. generator.set("pair"sv, String::number(list_pattern.pair));
  533. generator.append(R"~~~(
  534. { @type@, @style@, @start@, @middle@, @end@, @pair@ },)~~~");
  535. }
  536. generator.append(R"~~~(
  537. } };
  538. )~~~");
  539. };
  540. 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); });
  541. 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); });
  542. 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); });
  543. 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); });
  544. 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); });
  545. 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); });
  546. 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); });
  547. 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); });
  548. 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); });
  549. generator.append(R"~~~(
  550. struct CanonicalLanguageID {
  551. Unicode::LanguageID to_unicode_language_id() const
  552. {
  553. Unicode::LanguageID language_id {};
  554. language_id.variants.ensure_capacity(variants_size);
  555. language_id.language = s_string_list[language];
  556. if (script != 0)
  557. language_id.script = s_string_list[script];
  558. if (region != 0)
  559. language_id.region = s_string_list[region];
  560. for (size_t i = 0; i < variants_size; ++i)
  561. language_id.variants.append(s_string_list[variants[i]]);
  562. return language_id;
  563. }
  564. bool matches_variants(Vector<String> const& other_variants) const {
  565. if (variants_size == 0)
  566. return true;
  567. if (other_variants.size() != variants_size)
  568. return false;
  569. for (size_t i = 0; i < variants_size; ++i) {
  570. if (s_string_list[variants[i]] != other_variants[i])
  571. return false;
  572. }
  573. return true;
  574. };
  575. @string_index_type@ language { 0 };
  576. @string_index_type@ script { 0 };
  577. @string_index_type@ region { 0 };
  578. Array<@string_index_type@, @variants_size@> variants {};
  579. size_t variants_size { 0 };
  580. };
  581. struct LanguageMapping {
  582. CanonicalLanguageID key;
  583. CanonicalLanguageID alias;
  584. };
  585. )~~~");
  586. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  587. generator.set("size", String::number(mappings.size()));
  588. generator.set("name"sv, name);
  589. generator.append(R"~~~(
  590. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  591. )~~~");
  592. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  593. auto const& lhs_language = locale_data.unique_strings.get(lhs.key.language);
  594. auto const& rhs_language = locale_data.unique_strings.get(rhs.key.language);
  595. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  596. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  597. return false;
  598. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  599. return true;
  600. return lhs_language < rhs_language;
  601. });
  602. for (auto const& mapping : mappings) {
  603. generator.set("language"sv, String::number(mapping.key.language));
  604. generator.append(" { { @language@");
  605. append_index(mapping.key.script);
  606. append_index(mapping.key.region);
  607. append_list_and_size(mapping.key.variants);
  608. generator.set("language"sv, String::number(mapping.alias.language));
  609. generator.append(" }, { @language@");
  610. append_index(mapping.alias.script);
  611. append_index(mapping.alias.region);
  612. append_list_and_size(mapping.alias.variants);
  613. generator.append(" } },\n");
  614. }
  615. generator.append("} };\n");
  616. };
  617. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  618. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  619. generator.append(R"~~~(
  620. static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& language_id)
  621. {
  622. // https://unicode.org/reports/tr35/#Likely_Subtags
  623. enum class State {
  624. LanguageScriptRegion,
  625. LanguageRegion,
  626. LanguageScript,
  627. Language,
  628. UndScript,
  629. Done,
  630. };
  631. auto state = State::LanguageScriptRegion;
  632. while (state != State::Done) {
  633. Unicode::LanguageID search_key;
  634. switch (state) {
  635. case State::LanguageScriptRegion:
  636. state = State::LanguageRegion;
  637. if (!language_id.script.has_value() || !language_id.region.has_value())
  638. continue;
  639. search_key.language = *language_id.language;
  640. search_key.script = *language_id.script;
  641. search_key.region = *language_id.region;
  642. break;
  643. case State::LanguageRegion:
  644. state = State::LanguageScript;
  645. if (!language_id.region.has_value())
  646. continue;
  647. search_key.language = *language_id.language;
  648. search_key.region = *language_id.region;
  649. break;
  650. case State::LanguageScript:
  651. state = State::Language;
  652. if (!language_id.script.has_value())
  653. continue;
  654. search_key.language = *language_id.language;
  655. search_key.script = *language_id.script;
  656. break;
  657. case State::Language:
  658. state = State::UndScript;
  659. search_key.language = *language_id.language;
  660. break;
  661. case State::UndScript:
  662. state = State::Done;
  663. if (!language_id.script.has_value())
  664. continue;
  665. search_key.language = "und"sv;
  666. search_key.script = *language_id.script;
  667. break;
  668. default:
  669. VERIFY_NOT_REACHED();
  670. }
  671. for (auto const& map : s_likely_subtags) {
  672. auto const& key_language = s_string_list[map.key.language];
  673. auto const& key_script = s_string_list[map.key.script];
  674. auto const& key_region = s_string_list[map.key.region];
  675. if (key_language != search_key.language)
  676. continue;
  677. if (!key_script.is_empty() || search_key.script.has_value()) {
  678. if (key_script != search_key.script)
  679. continue;
  680. }
  681. if (!key_region.is_empty() || search_key.region.has_value()) {
  682. if (key_region != search_key.region)
  683. continue;
  684. }
  685. return &map;
  686. }
  687. }
  688. return nullptr;
  689. }
  690. )~~~");
  691. auto append_mapping_search = [&](StringView enum_snake, StringView from_string_name, StringView collection_name) {
  692. generator.set("enum_snake", enum_snake);
  693. generator.set("from_string_name", from_string_name);
  694. generator.set("collection_name", collection_name);
  695. generator.append(R"~~~(
  696. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  697. {
  698. auto locale_value = locale_from_string(locale);
  699. if (!locale_value.has_value())
  700. return {};
  701. auto @enum_snake@_value = @from_string_name@_from_string(@enum_snake@);
  702. if (!@enum_snake@_value.has_value())
  703. return {};
  704. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  705. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  706. auto const& mappings = @collection_name@.at(locale_index);
  707. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  708. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  709. if (@enum_snake@_mapping.is_empty())
  710. return {};
  711. return @enum_snake@_mapping;
  712. }
  713. )~~~");
  714. };
  715. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values) {
  716. HashValueMap<String> hashes;
  717. hashes.ensure_capacity(values.size());
  718. for (auto const& value : values)
  719. hashes.set(value.hash(), format_identifier(enum_title, value));
  720. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  721. };
  722. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  723. HashValueMap<StringIndexType> hashes;
  724. hashes.ensure_capacity(aliases.size());
  725. for (auto const& alias : aliases)
  726. hashes.set(alias.key.hash(), alias.value);
  727. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), "StringView"sv, "s_string_list[{}]"sv);
  728. };
  729. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys());
  730. append_mapping_search("language"sv, "language"sv, "s_languages"sv);
  731. append_from_string("Language"sv, "language"sv, locale_data.languages);
  732. append_alias_search("language"sv, locale_data.language_aliases);
  733. append_mapping_search("territory"sv, "territory"sv, "s_territories"sv);
  734. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  735. append_alias_search("territory"sv, locale_data.territory_aliases);
  736. append_mapping_search("script_tag"sv, "script_tag"sv, "s_scripts"sv);
  737. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  738. append_alias_search("script_tag"sv, locale_data.script_aliases);
  739. append_mapping_search("long_currency"sv, "currency"sv, "s_long_currencies"sv);
  740. append_mapping_search("short_currency"sv, "currency"sv, "s_short_currencies"sv);
  741. append_mapping_search("narrow_currency"sv, "currency"sv, "s_narrow_currencies"sv);
  742. append_mapping_search("numeric_currency"sv, "currency"sv, "s_numeric_currencies"sv);
  743. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  744. append_mapping_search("key"sv, "key"sv, "s_keywords"sv);
  745. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  746. append_alias_search("variant"sv, locale_data.variant_aliases);
  747. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  748. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  749. append_from_string("ListPatternStyle"sv, "list_pattern_style"sv, locale_data.list_pattern_styles);
  750. generator.append(R"~~~(
  751. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style)
  752. {
  753. auto locale_value = locale_from_string(locale);
  754. if (!locale_value.has_value())
  755. return {};
  756. auto type_value = list_pattern_type_from_string(list_pattern_type);
  757. if (!type_value.has_value())
  758. return {};
  759. auto style_value = list_pattern_style_from_string(list_pattern_style);
  760. if (!style_value.has_value())
  761. return {};
  762. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  763. auto const& locale_list_patterns = s_list_patterns.at(locale_index);
  764. for (auto const& list_patterns : locale_list_patterns) {
  765. if ((list_patterns.type == type_value) && (list_patterns.style == style_value)) {
  766. auto const& start = s_string_list[list_patterns.start];
  767. auto const& middle = s_string_list[list_patterns.middle];
  768. auto const& end = s_string_list[list_patterns.end];
  769. auto const& pair = s_string_list[list_patterns.pair];
  770. return ListPatterns { start, middle, end, pair };
  771. }
  772. }
  773. return {};
  774. }
  775. void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
  776. {
  777. for (auto const& map : s_complex_alias) {
  778. auto const& key_language = s_string_list[map.key.language];
  779. auto const& key_script = s_string_list[map.key.script];
  780. auto const& key_region = s_string_list[map.key.region];
  781. if ((key_language != language_id.language) && (key_language != "und"sv))
  782. continue;
  783. if (!key_script.is_empty() && (key_script != language_id.script))
  784. continue;
  785. if (!key_region.is_empty() && (key_region != language_id.region))
  786. continue;
  787. if (!map.key.matches_variants(language_id.variants))
  788. continue;
  789. auto alias = map.alias.to_unicode_language_id();
  790. if (alias.language == "und"sv)
  791. alias.language = move(language_id.language);
  792. if (key_script.is_empty() && !alias.script.has_value())
  793. alias.script = move(language_id.script);
  794. if (key_region.is_empty() && !alias.region.has_value())
  795. alias.region = move(language_id.region);
  796. if (map.key.variants_size == 0 && alias.variants.is_empty())
  797. alias.variants = move(language_id.variants);
  798. language_id = move(alias);
  799. break;
  800. }
  801. }
  802. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id)
  803. {
  804. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  805. auto const* likely_subtag = resolve_likely_subtag(language_id);
  806. if (likely_subtag == nullptr)
  807. return {};
  808. auto maximized = language_id;
  809. auto const& key_script = s_string_list[likely_subtag->key.script];
  810. auto const& key_region = s_string_list[likely_subtag->key.region];
  811. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  812. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  813. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  814. if (maximized.language == "und"sv)
  815. maximized.language = alias_language;
  816. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  817. maximized.script = alias_script;
  818. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  819. maximized.region = alias_region;
  820. return maximized;
  821. }
  822. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
  823. {
  824. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  825. return s_string_list[likely_subtag->alias.region];
  826. return {};
  827. }
  828. }
  829. )~~~");
  830. file.write(generator.as_string_view());
  831. }
  832. int main(int argc, char** argv)
  833. {
  834. char const* generated_header_path = nullptr;
  835. char const* generated_implementation_path = nullptr;
  836. char const* core_path = nullptr;
  837. char const* locale_names_path = nullptr;
  838. char const* misc_path = nullptr;
  839. char const* numbers_path = nullptr;
  840. Core::ArgsParser args_parser;
  841. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  842. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  843. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  844. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  845. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  846. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  847. args_parser.parse(argc, argv);
  848. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  849. if (path.is_empty()) {
  850. warnln("{} is required", flags);
  851. args_parser.print_usage(stderr, argv[0]);
  852. exit(1);
  853. }
  854. auto file_or_error = Core::File::open(path, mode);
  855. if (file_or_error.is_error()) {
  856. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  857. exit(1);
  858. }
  859. return file_or_error.release_value();
  860. };
  861. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  862. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  863. UnicodeLocaleData locale_data;
  864. parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, locale_data);
  865. generate_unicode_locale_header(generated_header_file, locale_data);
  866. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  867. return 0;
  868. }