GenerateUnicodeLocale.cpp 62 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513
  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. #include <math.h>
  23. using StringIndexType = u16;
  24. constexpr auto s_string_index_type = "u16"sv;
  25. struct NumberFormat : public Unicode::NumberFormat {
  26. using Base = Unicode::NumberFormat;
  27. static Base::Plurality plurality_from_string(StringView plurality)
  28. {
  29. if (plurality == "other"sv)
  30. return Base::Plurality::Other;
  31. if (plurality == "1"sv)
  32. return Base::Plurality::Single;
  33. if (plurality == "zero"sv)
  34. return Base::Plurality::Zero;
  35. if (plurality == "one"sv)
  36. return Base::Plurality::One;
  37. if (plurality == "two"sv)
  38. return Base::Plurality::Two;
  39. if (plurality == "few"sv)
  40. return Base::Plurality::Few;
  41. if (plurality == "many"sv)
  42. return Base::Plurality::Many;
  43. VERIFY_NOT_REACHED();
  44. }
  45. StringIndexType zero_format_index { 0 };
  46. StringIndexType positive_format_index { 0 };
  47. StringIndexType negative_format_index { 0 };
  48. };
  49. struct NumberSystem {
  50. StringIndexType system { 0 };
  51. HashMap<String, StringIndexType> symbols {};
  52. NumberFormat decimal_format {};
  53. Vector<NumberFormat> decimal_long_formats {};
  54. Vector<NumberFormat> decimal_short_formats {};
  55. NumberFormat currency_format {};
  56. NumberFormat accounting_format {};
  57. Vector<NumberFormat> currency_short_formats {};
  58. NumberFormat percent_format {};
  59. };
  60. struct ListPatterns {
  61. String type;
  62. String style;
  63. StringIndexType start { 0 };
  64. StringIndexType middle { 0 };
  65. StringIndexType end { 0 };
  66. StringIndexType pair { 0 };
  67. };
  68. struct Locale {
  69. String language;
  70. Optional<String> territory;
  71. Optional<String> variant;
  72. HashMap<String, StringIndexType> languages;
  73. HashMap<String, StringIndexType> territories;
  74. HashMap<String, StringIndexType> scripts;
  75. HashMap<String, StringIndexType> currencies;
  76. HashMap<String, StringIndexType> keywords;
  77. HashMap<String, NumberSystem> number_systems;
  78. Vector<ListPatterns> list_patterns;
  79. };
  80. struct LanguageMapping {
  81. CanonicalLanguageID<StringIndexType> key {};
  82. CanonicalLanguageID<StringIndexType> alias {};
  83. };
  84. struct UnicodeLocaleData {
  85. UniqueStringStorage<StringIndexType> unique_strings;
  86. HashMap<String, Locale> locales;
  87. Vector<String> languages;
  88. Vector<String> territories;
  89. Vector<String> scripts;
  90. Vector<String> variants;
  91. Vector<String> currencies;
  92. Vector<String> keywords;
  93. Vector<String> numeric_symbols;
  94. Vector<String> list_pattern_types;
  95. Vector<String> list_pattern_styles;
  96. HashMap<String, StringIndexType> language_aliases;
  97. HashMap<String, StringIndexType> territory_aliases;
  98. HashMap<String, StringIndexType> script_aliases;
  99. HashMap<String, StringIndexType> variant_aliases;
  100. HashMap<String, StringIndexType> subdivision_aliases;
  101. Vector<LanguageMapping> complex_mappings;
  102. Vector<LanguageMapping> likely_subtags;
  103. size_t max_variant_size { 0 };
  104. };
  105. static Optional<LanguageMapping> parse_language_mapping(UnicodeLocaleData& locale_data, StringView key, StringView alias)
  106. {
  107. auto parsed_key = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, key);
  108. if (!parsed_key.has_value())
  109. return {};
  110. auto parsed_alias = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, alias);
  111. if (!parsed_alias.has_value())
  112. return {};
  113. return LanguageMapping { parsed_key.release_value(), parsed_alias.release_value() };
  114. }
  115. static void parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
  116. {
  117. LexicalPath core_aliases_path(move(core_supplemental_path));
  118. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  119. VERIFY(Core::File::exists(core_aliases_path.string()));
  120. auto core_aliases_file_or_error = Core::File::open(core_aliases_path.string(), Core::OpenMode::ReadOnly);
  121. VERIFY(!core_aliases_file_or_error.is_error());
  122. auto core_aliases = JsonParser(core_aliases_file_or_error.value()->read_all()).parse();
  123. VERIFY(core_aliases.has_value());
  124. auto const& supplemental_object = core_aliases->as_object().get("supplemental"sv);
  125. auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
  126. auto const& alias_object = metadata_object.as_object().get("alias"sv);
  127. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  128. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  129. auto alias = value.as_object().get("_replacement"sv).as_string();
  130. if (key.contains('-')) {
  131. auto mapping = parse_language_mapping(locale_data, key, alias);
  132. if (!mapping.has_value())
  133. return;
  134. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  135. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  136. locale_data.complex_mappings.append(mapping.release_value());
  137. } else {
  138. alias_map.set(key, locale_data.unique_strings.ensure(alias));
  139. }
  140. });
  141. };
  142. append_aliases(alias_object.as_object().get("languageAlias"sv), locale_data.language_aliases);
  143. append_aliases(alias_object.as_object().get("territoryAlias"sv), locale_data.territory_aliases);
  144. append_aliases(alias_object.as_object().get("scriptAlias"sv), locale_data.script_aliases);
  145. append_aliases(alias_object.as_object().get("variantAlias"sv), locale_data.variant_aliases);
  146. append_aliases(alias_object.as_object().get("subdivisionAlias"sv), locale_data.subdivision_aliases);
  147. }
  148. static void parse_likely_subtags(String core_supplemental_path, UnicodeLocaleData& locale_data)
  149. {
  150. LexicalPath likely_subtags_path(move(core_supplemental_path));
  151. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  152. VERIFY(Core::File::exists(likely_subtags_path.string()));
  153. auto likely_subtags_file_or_error = Core::File::open(likely_subtags_path.string(), Core::OpenMode::ReadOnly);
  154. VERIFY(!likely_subtags_file_or_error.is_error());
  155. auto likely_subtags = JsonParser(likely_subtags_file_or_error.value()->read_all()).parse();
  156. VERIFY(likely_subtags.has_value());
  157. auto const& supplemental_object = likely_subtags->as_object().get("supplemental"sv);
  158. auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
  159. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  160. auto mapping = parse_language_mapping(locale_data, key, value.as_string());
  161. if (!mapping.has_value())
  162. return;
  163. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  164. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  165. locale_data.likely_subtags.append(mapping.release_value());
  166. });
  167. }
  168. static void parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  169. {
  170. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  171. languages_path = languages_path.append("languages.json"sv);
  172. VERIFY(Core::File::exists(languages_path.string()));
  173. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  174. VERIFY(!languages_file_or_error.is_error());
  175. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  176. VERIFY(languages.has_value());
  177. auto const& main_object = languages->as_object().get("main"sv);
  178. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  179. auto const& identity_object = locale_object.as_object().get("identity"sv);
  180. auto const& language_string = identity_object.as_object().get("language"sv);
  181. auto const& territory_string = identity_object.as_object().get("territory"sv);
  182. auto const& variant_string = identity_object.as_object().get("variant"sv);
  183. locale.language = language_string.as_string();
  184. if (!locale_data.languages.contains_slow(locale.language))
  185. locale_data.languages.append(locale.language);
  186. if (territory_string.is_string()) {
  187. locale.territory = territory_string.as_string();
  188. if (!locale_data.territories.contains_slow(*locale.territory))
  189. locale_data.territories.append(*locale.territory);
  190. }
  191. if (variant_string.is_string()) {
  192. locale.variant = variant_string.as_string();
  193. if (!locale_data.variants.contains_slow(*locale.variant))
  194. locale_data.variants.append(*locale.variant);
  195. }
  196. }
  197. static void parse_locale_languages(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  198. {
  199. LexicalPath languages_path(move(locale_path));
  200. languages_path = languages_path.append("languages.json"sv);
  201. VERIFY(Core::File::exists(languages_path.string()));
  202. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  203. VERIFY(!languages_file_or_error.is_error());
  204. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  205. VERIFY(languages.has_value());
  206. auto const& main_object = languages->as_object().get("main"sv);
  207. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  208. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  209. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  210. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  211. if (!locale_data.languages.contains_slow(key))
  212. return;
  213. auto index = locale_data.unique_strings.ensure(value.as_string());
  214. locale.languages.set(key, index);
  215. });
  216. }
  217. static void parse_locale_territories(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  218. {
  219. LexicalPath territories_path(move(locale_path));
  220. territories_path = territories_path.append("territories.json"sv);
  221. VERIFY(Core::File::exists(territories_path.string()));
  222. auto territories_file_or_error = Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly);
  223. VERIFY(!territories_file_or_error.is_error());
  224. auto territories = JsonParser(territories_file_or_error.value()->read_all()).parse();
  225. VERIFY(territories.has_value());
  226. auto const& main_object = territories->as_object().get("main"sv);
  227. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  228. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  229. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  230. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  231. if (!locale_data.territories.contains_slow(key))
  232. return;
  233. auto index = locale_data.unique_strings.ensure(value.as_string());
  234. locale.territories.set(key, index);
  235. });
  236. }
  237. static void parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  238. {
  239. LexicalPath scripts_path(move(locale_path));
  240. scripts_path = scripts_path.append("scripts.json"sv);
  241. VERIFY(Core::File::exists(scripts_path.string()));
  242. auto scripts_file_or_error = Core::File::open(scripts_path.string(), Core::OpenMode::ReadOnly);
  243. VERIFY(!scripts_file_or_error.is_error());
  244. auto scripts = JsonParser(scripts_file_or_error.value()->read_all()).parse();
  245. VERIFY(scripts.has_value());
  246. auto const& main_object = scripts->as_object().get("main"sv);
  247. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  248. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  249. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  250. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  251. auto index = locale_data.unique_strings.ensure(value.as_string());
  252. locale.scripts.set(key, index);
  253. if (!locale_data.scripts.contains_slow(key))
  254. locale_data.scripts.append(key);
  255. });
  256. }
  257. static void parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
  258. {
  259. LexicalPath list_patterns_path(move(misc_path));
  260. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  261. VERIFY(Core::File::exists(list_patterns_path.string()));
  262. auto list_patterns_file_or_error = Core::File::open(list_patterns_path.string(), Core::OpenMode::ReadOnly);
  263. VERIFY(!list_patterns_file_or_error.is_error());
  264. auto list_patterns = JsonParser(list_patterns_file_or_error.value()->read_all()).parse();
  265. VERIFY(list_patterns.has_value());
  266. auto const& main_object = list_patterns->as_object().get("main"sv);
  267. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  268. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  269. auto list_pattern_type = [](StringView key) {
  270. if (key.contains("type-standard"sv))
  271. return "conjunction"sv;
  272. if (key.contains("type-or"sv))
  273. return "disjunction"sv;
  274. if (key.contains("type-unit"sv))
  275. return "unit"sv;
  276. VERIFY_NOT_REACHED();
  277. };
  278. auto list_pattern_style = [](StringView key) {
  279. if (key.contains("short"sv))
  280. return "short"sv;
  281. if (key.contains("narrow"sv))
  282. return "narrow"sv;
  283. return "long"sv;
  284. };
  285. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  286. auto type = list_pattern_type(key);
  287. auto style = list_pattern_style(key);
  288. auto start = locale_data.unique_strings.ensure(value.as_object().get("start"sv).as_string());
  289. auto middle = locale_data.unique_strings.ensure(value.as_object().get("middle"sv).as_string());
  290. auto end = locale_data.unique_strings.ensure(value.as_object().get("end"sv).as_string());
  291. auto pair = locale_data.unique_strings.ensure(value.as_object().get("2"sv).as_string());
  292. if (!locale_data.list_pattern_types.contains_slow(type))
  293. locale_data.list_pattern_types.append(type);
  294. if (!locale_data.list_pattern_styles.contains_slow(style))
  295. locale_data.list_pattern_styles.append(style);
  296. locale.list_patterns.append({ move(type), move(style), move(start), move(middle), move(end), move(pair) });
  297. });
  298. }
  299. static void parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  300. {
  301. LexicalPath currencies_path(move(numbers_path));
  302. currencies_path = currencies_path.append("currencies.json"sv);
  303. VERIFY(Core::File::exists(currencies_path.string()));
  304. auto currencies_file_or_error = Core::File::open(currencies_path.string(), Core::OpenMode::ReadOnly);
  305. VERIFY(!currencies_file_or_error.is_error());
  306. auto currencies = JsonParser(currencies_file_or_error.value()->read_all()).parse();
  307. VERIFY(currencies.has_value());
  308. auto const& main_object = currencies->as_object().get("main"sv);
  309. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  310. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  311. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  312. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  313. auto const& display_name = value.as_object().get("displayName"sv);
  314. auto index = locale_data.unique_strings.ensure(display_name.as_string());
  315. locale.currencies.set(key, index);
  316. if (!locale_data.currencies.contains_slow(key))
  317. locale_data.currencies.append(key);
  318. });
  319. }
  320. static void parse_numeric_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  321. {
  322. static constexpr StringView key = "nu"sv;
  323. LexicalPath numbers_path(move(locale_numbers_path));
  324. numbers_path = numbers_path.append("numbers.json"sv);
  325. VERIFY(Core::File::exists(numbers_path.string()));
  326. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  327. VERIFY(!numbers_file_or_error.is_error());
  328. auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
  329. VERIFY(numbers.has_value());
  330. auto const& main_object = numbers->as_object().get("main"sv);
  331. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  332. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  333. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  334. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  335. Vector<String> keyword_values {};
  336. keyword_values.append(default_numbering_system_object.as_string());
  337. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  338. auto keyword_value = value.as_string();
  339. if (!keyword_values.contains_slow(keyword_value))
  340. keyword_values.append(move(keyword_value));
  341. });
  342. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  343. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  344. return;
  345. auto keyword_value = value.as_string();
  346. if (!keyword_values.contains_slow(keyword_value))
  347. keyword_values.append(move(keyword_value));
  348. });
  349. StringBuilder builder;
  350. builder.join(',', keyword_values);
  351. auto index = locale_data.unique_strings.ensure(builder.build());
  352. locale.keywords.set(key, index);
  353. if (!locale_data.keywords.contains_slow(key))
  354. locale_data.keywords.append(key);
  355. }
  356. static void parse_number_pattern(String pattern, UnicodeLocaleData& locale_data, NumberFormat& format)
  357. {
  358. // https://unicode.org/reports/tr35/tr35-numbers.html#Number_Format_Patterns
  359. // https://cldr.unicode.org/translation/number-currency-formats/number-and-currency-patterns
  360. auto replace_patterns = [](String pattern) {
  361. static HashMap<StringView, StringView> replacements = {
  362. { "%"sv, "{percentSign}"sv },
  363. { "+"sv, "{plusSign}"sv },
  364. { "-"sv, "{minusSign}"sv },
  365. { "¤"sv, "{currencyCode}"sv }, // U+00A4 Currency Sign
  366. };
  367. if (auto start_number_index = pattern.find('#'); start_number_index.has_value()) {
  368. auto end_number_index = *start_number_index + 1;
  369. for (; end_number_index < pattern.length(); ++end_number_index) {
  370. auto ch = pattern[end_number_index];
  371. if ((ch != '#') && (ch != '0') && (ch != ',') && (ch != '.'))
  372. break;
  373. }
  374. pattern = String::formatted("{}{{number}}{}",
  375. *start_number_index > 0 ? pattern.substring_view(0, *start_number_index) : ""sv,
  376. pattern.substring_view(end_number_index));
  377. }
  378. for (auto const& replacement : replacements)
  379. pattern = pattern.replace(replacement.key, replacement.value, true);
  380. return pattern;
  381. };
  382. auto patterns = pattern.split(';');
  383. VERIFY((patterns.size() == 1) || (patterns.size() == 2));
  384. if (format.magnitude != 0) {
  385. auto number_of_zeroes_in_pattern = patterns[0].count("0"sv);
  386. VERIFY(format.magnitude >= number_of_zeroes_in_pattern);
  387. format.compact_scale = format.magnitude - number_of_zeroes_in_pattern;
  388. }
  389. auto zero_format = replace_patterns(move(patterns[0]));
  390. format.positive_format_index = locale_data.unique_strings.ensure(String::formatted("{{plusSign}}{}", zero_format));
  391. if (patterns.size() == 2) {
  392. auto negative_format = replace_patterns(move(patterns[1]));
  393. format.negative_format_index = locale_data.unique_strings.ensure(move(negative_format));
  394. } else {
  395. format.negative_format_index = locale_data.unique_strings.ensure(String::formatted("{{minusSign}}{}", zero_format));
  396. }
  397. format.zero_format_index = locale_data.unique_strings.ensure(move(zero_format));
  398. }
  399. static void parse_number_systems(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  400. {
  401. LexicalPath numbers_path(move(locale_numbers_path));
  402. numbers_path = numbers_path.append("numbers.json"sv);
  403. VERIFY(Core::File::exists(numbers_path.string()));
  404. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  405. VERIFY(!numbers_file_or_error.is_error());
  406. auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
  407. VERIFY(numbers.has_value());
  408. auto const& main_object = numbers->as_object().get("main"sv);
  409. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  410. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  411. auto ensure_number_system = [&](auto const& system) -> NumberSystem& {
  412. return locale.number_systems.ensure(system, [&]() {
  413. auto system_index = locale_data.unique_strings.ensure(system);
  414. return NumberSystem { .system = system_index };
  415. });
  416. };
  417. auto parse_number_format = [&](auto const& format_object) {
  418. Vector<NumberFormat> result;
  419. result.ensure_capacity(format_object.size());
  420. format_object.for_each_member([&](auto const& key, JsonValue const& value) {
  421. auto split_key = key.split_view('-');
  422. VERIFY(split_key.size() == 3);
  423. auto type = split_key[0].template to_uint<u64>().value();
  424. VERIFY(type % 10 == 0);
  425. NumberFormat format {};
  426. format.magnitude = static_cast<u8>(log10(type));
  427. format.plurality = NumberFormat::plurality_from_string(split_key[2]);
  428. parse_number_pattern(value.as_string(), locale_data, format);
  429. result.append(move(format));
  430. });
  431. return result;
  432. };
  433. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  434. constexpr auto symbols_prefix = "symbols-numberSystem-"sv;
  435. constexpr auto decimal_formats_prefix = "decimalFormats-numberSystem-"sv;
  436. constexpr auto currency_formats_prefix = "currencyFormats-numberSystem-"sv;
  437. constexpr auto percent_formats_prefix = "percentFormats-numberSystem-"sv;
  438. if (key.starts_with(symbols_prefix)) {
  439. auto system = key.substring(symbols_prefix.length());
  440. auto& number_system = ensure_number_system(system);
  441. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& localization) {
  442. auto symbol_index = locale_data.unique_strings.ensure(localization.as_string());
  443. number_system.symbols.set(symbol, symbol_index);
  444. if (!locale_data.numeric_symbols.contains_slow(symbol))
  445. locale_data.numeric_symbols.append(symbol);
  446. });
  447. } else if (key.starts_with(decimal_formats_prefix)) {
  448. auto system = key.substring(decimal_formats_prefix.length());
  449. auto& number_system = ensure_number_system(system);
  450. auto format_object = value.as_object().get("standard"sv);
  451. parse_number_pattern(format_object.as_string(), locale_data, number_system.decimal_format);
  452. auto const& long_format = value.as_object().get("long"sv).as_object().get("decimalFormat"sv);
  453. number_system.decimal_long_formats = parse_number_format(long_format.as_object());
  454. auto const& short_format = value.as_object().get("short"sv).as_object().get("decimalFormat"sv);
  455. number_system.decimal_short_formats = parse_number_format(short_format.as_object());
  456. } else if (key.starts_with(currency_formats_prefix)) {
  457. auto system = key.substring(currency_formats_prefix.length());
  458. auto& number_system = ensure_number_system(system);
  459. auto format_object = value.as_object().get("standard"sv);
  460. parse_number_pattern(format_object.as_string(), locale_data, number_system.currency_format);
  461. format_object = value.as_object().get("accounting"sv);
  462. parse_number_pattern(format_object.as_string(), locale_data, number_system.accounting_format);
  463. if (value.as_object().has("short"sv)) {
  464. auto const& short_format = value.as_object().get("short"sv).as_object().get("standard"sv);
  465. number_system.currency_short_formats = parse_number_format(short_format.as_object());
  466. }
  467. } else if (key.starts_with(percent_formats_prefix)) {
  468. auto system = key.substring(decimal_formats_prefix.length());
  469. auto& number_system = ensure_number_system(system);
  470. auto format_object = value.as_object().get("standard"sv);
  471. parse_number_pattern(format_object.as_string(), locale_data, number_system.percent_format);
  472. }
  473. });
  474. }
  475. static void parse_default_content_locales(String core_path, UnicodeLocaleData& locale_data)
  476. {
  477. LexicalPath default_content_path(move(core_path));
  478. default_content_path = default_content_path.append("defaultContent.json"sv);
  479. VERIFY(Core::File::exists(default_content_path.string()));
  480. auto default_content_file_or_error = Core::File::open(default_content_path.string(), Core::OpenMode::ReadOnly);
  481. VERIFY(!default_content_file_or_error.is_error());
  482. auto default_content = JsonParser(default_content_file_or_error.value()->read_all()).parse();
  483. VERIFY(default_content.has_value());
  484. auto const& default_content_array = default_content->as_object().get("defaultContent"sv);
  485. default_content_array.as_array().for_each([&](JsonValue const& value) {
  486. auto locale = value.as_string();
  487. StringView default_locale = locale;
  488. while (true) {
  489. if (locale_data.locales.contains(default_locale))
  490. break;
  491. auto pos = default_locale.find_last('-');
  492. if (!pos.has_value())
  493. return;
  494. default_locale = default_locale.substring_view(0, *pos);
  495. }
  496. locale_data.locales.set(locale, locale_data.locales.get(default_locale).value());
  497. });
  498. }
  499. static void parse_all_locales(String core_path, String locale_names_path, String misc_path, String numbers_path, UnicodeLocaleData& locale_data)
  500. {
  501. auto identity_iterator = path_to_dir_iterator(locale_names_path);
  502. auto locale_names_iterator = path_to_dir_iterator(move(locale_names_path));
  503. auto misc_iterator = path_to_dir_iterator(move(misc_path));
  504. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  505. LexicalPath core_supplemental_path(core_path);
  506. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  507. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  508. parse_core_aliases(core_supplemental_path.string(), locale_data);
  509. parse_likely_subtags(core_supplemental_path.string(), locale_data);
  510. auto remove_variants_from_path = [&](String path) -> Optional<String> {
  511. auto parsed_locale = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path));
  512. if (!parsed_locale.has_value())
  513. return {};
  514. StringBuilder builder;
  515. builder.append(locale_data.unique_strings.get(parsed_locale->language));
  516. if (auto script = locale_data.unique_strings.get(parsed_locale->script); !script.is_empty())
  517. builder.appendff("-{}", script);
  518. if (auto region = locale_data.unique_strings.get(parsed_locale->region); !region.is_empty())
  519. builder.appendff("-{}", region);
  520. return builder.build();
  521. };
  522. while (identity_iterator.has_next()) {
  523. auto locale_path = identity_iterator.next_full_path();
  524. VERIFY(Core::File::is_directory(locale_path));
  525. auto language = remove_variants_from_path(locale_path);
  526. if (!language.has_value())
  527. continue;
  528. auto& locale = locale_data.locales.ensure(*language);
  529. parse_identity(locale_path, locale_data, locale);
  530. }
  531. while (locale_names_iterator.has_next()) {
  532. auto locale_path = locale_names_iterator.next_full_path();
  533. VERIFY(Core::File::is_directory(locale_path));
  534. auto language = remove_variants_from_path(locale_path);
  535. if (!language.has_value())
  536. continue;
  537. auto& locale = locale_data.locales.ensure(*language);
  538. parse_locale_languages(locale_path, locale_data, locale);
  539. parse_locale_territories(locale_path, locale_data, locale);
  540. parse_locale_scripts(locale_path, locale_data, locale);
  541. }
  542. while (misc_iterator.has_next()) {
  543. auto misc_path = misc_iterator.next_full_path();
  544. VERIFY(Core::File::is_directory(misc_path));
  545. auto language = remove_variants_from_path(misc_path);
  546. if (!language.has_value())
  547. continue;
  548. auto& locale = locale_data.locales.ensure(*language);
  549. parse_locale_list_patterns(misc_path, locale_data, locale);
  550. }
  551. while (numbers_iterator.has_next()) {
  552. auto numbers_path = numbers_iterator.next_full_path();
  553. VERIFY(Core::File::is_directory(numbers_path));
  554. auto language = remove_variants_from_path(numbers_path);
  555. if (!language.has_value())
  556. continue;
  557. auto& locale = locale_data.locales.ensure(*language);
  558. parse_locale_currencies(numbers_path, locale_data, locale);
  559. parse_numeric_keywords(numbers_path, locale_data, locale);
  560. parse_number_systems(numbers_path, locale_data, locale);
  561. }
  562. parse_default_content_locales(move(core_path), locale_data);
  563. }
  564. static String format_identifier(StringView owner, String identifier)
  565. {
  566. identifier = identifier.replace("-"sv, "_"sv, true);
  567. if (all_of(identifier, is_ascii_digit))
  568. return String::formatted("{}_{}", owner[0], identifier);
  569. if (is_ascii_lower_alpha(identifier[0]))
  570. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  571. return identifier;
  572. }
  573. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  574. {
  575. StringBuilder builder;
  576. SourceGenerator generator { builder };
  577. generator.append(R"~~~(
  578. #pragma once
  579. #include <AK/Optional.h>
  580. #include <AK/StringView.h>
  581. #include <AK/Types.h>
  582. #include <AK/Vector.h>
  583. #include <LibUnicode/Forward.h>
  584. namespace Unicode {
  585. )~~~");
  586. auto locales = locale_data.locales.keys();
  587. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales);
  588. generate_enum(generator, format_identifier, "Language"sv, {}, locale_data.languages);
  589. generate_enum(generator, format_identifier, "Territory"sv, {}, locale_data.territories);
  590. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, locale_data.scripts);
  591. generate_enum(generator, format_identifier, "Currency"sv, {}, locale_data.currencies);
  592. generate_enum(generator, format_identifier, "Key"sv, {}, locale_data.keywords);
  593. generate_enum(generator, format_identifier, "NumericSymbol"sv, {}, locale_data.numeric_symbols);
  594. generate_enum(generator, format_identifier, "Variant"sv, {}, locale_data.variants);
  595. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, locale_data.list_pattern_types);
  596. generate_enum(generator, format_identifier, "ListPatternStyle"sv, {}, locale_data.list_pattern_styles);
  597. generator.append(R"~~~(
  598. namespace Detail {
  599. Optional<Locale> locale_from_string(StringView locale);
  600. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  601. Optional<Language> language_from_string(StringView language);
  602. Optional<StringView> resolve_language_alias(StringView language);
  603. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  604. Optional<Territory> territory_from_string(StringView territory);
  605. Optional<StringView> resolve_territory_alias(StringView territory);
  606. Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
  607. Optional<ScriptTag> script_tag_from_string(StringView script_tag);
  608. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  609. Optional<StringView> get_locale_currency_mapping(StringView locale, StringView currency);
  610. Optional<Currency> currency_from_string(StringView currency);
  611. Optional<StringView> get_locale_key_mapping(StringView locale, StringView key);
  612. Optional<Key> key_from_string(StringView key);
  613. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView numeric_symbol);
  614. Optional<NumberFormat> get_standard_number_system_format(StringView locale, StringView system, StandardNumberFormatType type);
  615. Vector<NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type);
  616. Optional<NumericSymbol> numeric_symbol_from_string(StringView numeric_symbol);
  617. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style);
  618. Optional<ListPatternType> list_pattern_type_from_string(StringView list_pattern_type);
  619. Optional<ListPatternStyle> list_pattern_style_from_string(StringView list_pattern_style);
  620. Optional<StringView> resolve_variant_alias(StringView variant);
  621. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  622. void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
  623. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id);
  624. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
  625. }
  626. }
  627. )~~~");
  628. file.write(generator.as_string_view());
  629. }
  630. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  631. {
  632. StringBuilder builder;
  633. SourceGenerator generator { builder };
  634. generator.set("string_index_type"sv, s_string_index_type);
  635. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  636. generator.set("territories_size", String::number(locale_data.territories.size()));
  637. generator.set("variants_size", String::number(locale_data.max_variant_size));
  638. generator.set("numeric_symbols_size", String::number(locale_data.numeric_symbols.size()));
  639. generator.append(R"~~~(
  640. #include <AK/Array.h>
  641. #include <AK/BinarySearch.h>
  642. #include <AK/Span.h>
  643. #include <LibUnicode/Locale.h>
  644. #include <LibUnicode/UnicodeLocale.h>
  645. namespace Unicode::Detail {
  646. struct Patterns {
  647. ListPatternType type;
  648. ListPatternStyle style;
  649. @string_index_type@ start { 0 };
  650. @string_index_type@ middle { 0 };
  651. @string_index_type@ end { 0 };
  652. @string_index_type@ pair { 0 };
  653. };
  654. )~~~");
  655. locale_data.unique_strings.generate(generator);
  656. generator.append(R"~~~(
  657. struct NumberFormat {
  658. Unicode::NumberFormat to_unicode_number_format() const {
  659. Unicode::NumberFormat number_format {};
  660. number_format.magnitude = magnitude;
  661. number_format.compact_scale = compact_scale;
  662. number_format.plurality = static_cast<Unicode::NumberFormat::Plurality>(plurality);
  663. number_format.zero_format = s_string_list[zero_format];
  664. number_format.positive_format = s_string_list[positive_format];
  665. number_format.negative_format = s_string_list[negative_format];
  666. return number_format;
  667. }
  668. u8 magnitude { 0 };
  669. u8 compact_scale { 0 };
  670. u8 plurality { 0 };
  671. @string_index_type@ zero_format { 0 };
  672. @string_index_type@ positive_format { 0 };
  673. @string_index_type@ negative_format { 0 };
  674. };
  675. struct NumberSystem {
  676. @string_index_type@ system { 0 };
  677. Array<@string_index_type@, @numeric_symbols_size@> symbols {};
  678. NumberFormat decimal_format {};
  679. Span<NumberFormat const> decimal_long_formats {};
  680. Span<NumberFormat const> decimal_short_formats {};
  681. NumberFormat currency_format {};
  682. NumberFormat accounting_format {};
  683. Span<NumberFormat const> currency_short_formats {};
  684. NumberFormat percent_format {};
  685. };
  686. )~~~");
  687. auto append_index = [&](auto index) {
  688. generator.append(String::formatted(", {}", index));
  689. };
  690. auto append_list_and_size = [&](auto const& list) {
  691. if (list.is_empty()) {
  692. generator.append(", {}, 0");
  693. return;
  694. }
  695. bool first = true;
  696. generator.append(", {");
  697. for (auto const& item : list) {
  698. generator.append(first ? " " : ", ");
  699. generator.append(String::number(item));
  700. first = false;
  701. }
  702. generator.append(String::formatted(" }}, {}", list.size()));
  703. };
  704. auto append_string_index_list = [&](String name, auto const& keys, auto const& mappings) {
  705. generator.set("name", name);
  706. generator.set("size", String::number(keys.size()));
  707. generator.append(R"~~~(
  708. static constexpr Array<@string_index_type@, @size@> @name@ { {
  709. )~~~");
  710. constexpr size_t max_values_per_row = 30;
  711. size_t values_in_current_row = 0;
  712. for (auto const& key : keys) {
  713. if (values_in_current_row++ > 0)
  714. generator.append(" ");
  715. if (auto it = mappings.find(key); it != mappings.end())
  716. generator.set("mapping"sv, String::number(it->value));
  717. else
  718. generator.set("mapping"sv, "0"sv);
  719. generator.append("@mapping@,");
  720. if (values_in_current_row == max_values_per_row) {
  721. values_in_current_row = 0;
  722. generator.append("\n ");
  723. }
  724. }
  725. generator.append(R"~~~(
  726. } };
  727. )~~~");
  728. };
  729. auto append_number_format = [&](auto const& number_format) {
  730. generator.set("magnitude"sv, String::number(number_format.magnitude));
  731. generator.set("compact_scale"sv, String::number(number_format.compact_scale));
  732. generator.set("plurality"sv, String::number(static_cast<u8>(number_format.plurality)));
  733. generator.set("zero_format"sv, String::number(number_format.zero_format_index));
  734. generator.set("positive_format"sv, String::number(number_format.positive_format_index));
  735. generator.set("negative_format"sv, String::number(number_format.negative_format_index));
  736. generator.append("{ @magnitude@, @compact_scale@, @plurality@, @zero_format@, @positive_format@, @negative_format@ },");
  737. };
  738. auto append_number_formats = [&](String name, auto const& number_formats) {
  739. generator.set("name"sv, move(name));
  740. generator.set("size"sv, String::number(number_formats.size()));
  741. generator.append(R"~~~(
  742. static constexpr Array<NumberFormat, @size@> @name@ { {
  743. )~~~");
  744. constexpr size_t max_values_per_row = 10;
  745. size_t values_in_current_row = 0;
  746. for (auto const& number_format : number_formats) {
  747. if (values_in_current_row++ > 0)
  748. generator.append(" ");
  749. append_number_format(number_format);
  750. if (values_in_current_row == max_values_per_row) {
  751. values_in_current_row = 0;
  752. generator.append("\n ");
  753. }
  754. }
  755. generator.append(R"~~~(
  756. } };
  757. )~~~");
  758. };
  759. auto append_number_systems = [&](String name, auto const& number_systems) {
  760. auto format_name = [&](StringView system, StringView format) {
  761. return String::formatted("{}_{}_{}", name, system, format);
  762. };
  763. for (auto const& number_system : number_systems) {
  764. append_number_formats(format_name(number_system.key, "dl"sv), number_system.value.decimal_long_formats);
  765. append_number_formats(format_name(number_system.key, "ds"sv), number_system.value.decimal_short_formats);
  766. append_number_formats(format_name(number_system.key, "cs"sv), number_system.value.currency_short_formats);
  767. }
  768. generator.set("name", name);
  769. generator.set("size", String::number(number_systems.size()));
  770. generator.append(R"~~~(
  771. static constexpr Array<NumberSystem, @size@> @name@ { {)~~~");
  772. for (auto const& number_system : number_systems) {
  773. generator.set("system"sv, String::number(number_system.value.system));
  774. generator.set("decimal_long_formats"sv, format_name(number_system.key, "dl"sv));
  775. generator.set("decimal_short_formats"sv, format_name(number_system.key, "ds"sv));
  776. generator.set("currency_short_formats"sv, format_name(number_system.key, "cs"sv));
  777. generator.append(R"~~~(
  778. { @system@, {)~~~");
  779. for (auto const& symbol : locale_data.numeric_symbols) {
  780. auto index = number_system.value.symbols.get(symbol).value_or(0);
  781. generator.set("index", String::number(index));
  782. generator.append(" @index@,");
  783. }
  784. generator.append(" }, ");
  785. append_number_format(number_system.value.decimal_format);
  786. generator.append(" @decimal_long_formats@.span(), @decimal_short_formats@.span(), ");
  787. append_number_format(number_system.value.currency_format);
  788. generator.append(" ");
  789. append_number_format(number_system.value.accounting_format);
  790. generator.append(" @currency_short_formats@.span(), ");
  791. append_number_format(number_system.value.percent_format);
  792. generator.append(" },");
  793. }
  794. generator.append(R"~~~(
  795. } };
  796. )~~~");
  797. };
  798. auto append_list_patterns = [&](StringView name, Vector<ListPatterns> const& list_patterns) {
  799. generator.set("name", name);
  800. generator.set("size", String::number(list_patterns.size()));
  801. generator.append(R"~~~(
  802. static constexpr Array<Patterns, @size@> @name@ { {)~~~");
  803. for (auto const& list_pattern : list_patterns) {
  804. generator.set("type"sv, String::formatted("ListPatternType::{}", format_identifier({}, list_pattern.type)));
  805. generator.set("style"sv, String::formatted("ListPatternStyle::{}", format_identifier({}, list_pattern.style)));
  806. generator.set("start"sv, String::number(list_pattern.start));
  807. generator.set("middle"sv, String::number(list_pattern.middle));
  808. generator.set("end"sv, String::number(list_pattern.end));
  809. generator.set("pair"sv, String::number(list_pattern.pair));
  810. generator.append(R"~~~(
  811. { @type@, @style@, @start@, @middle@, @end@, @pair@ },)~~~");
  812. }
  813. generator.append(R"~~~(
  814. } };
  815. )~~~");
  816. };
  817. 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); });
  818. 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); });
  819. 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); });
  820. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_currencies"sv, "s_currencies_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.currencies, value.currencies); });
  821. 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); });
  822. generate_mapping(generator, locale_data.locales, "NumberSystem"sv, "s_number_systems"sv, "s_number_systems_{}", [&](auto const& name, auto const& value) { append_number_systems(name, value.number_systems); });
  823. 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); });
  824. generator.append(R"~~~(
  825. struct CanonicalLanguageID {
  826. Unicode::LanguageID to_unicode_language_id() const
  827. {
  828. Unicode::LanguageID language_id {};
  829. language_id.variants.ensure_capacity(variants_size);
  830. language_id.language = s_string_list[language];
  831. if (script != 0)
  832. language_id.script = s_string_list[script];
  833. if (region != 0)
  834. language_id.region = s_string_list[region];
  835. for (size_t i = 0; i < variants_size; ++i)
  836. language_id.variants.append(s_string_list[variants[i]]);
  837. return language_id;
  838. }
  839. bool matches_variants(Vector<String> const& other_variants) const {
  840. if (variants_size == 0)
  841. return true;
  842. if (other_variants.size() != variants_size)
  843. return false;
  844. for (size_t i = 0; i < variants_size; ++i) {
  845. if (s_string_list[variants[i]] != other_variants[i])
  846. return false;
  847. }
  848. return true;
  849. };
  850. @string_index_type@ language { 0 };
  851. @string_index_type@ script { 0 };
  852. @string_index_type@ region { 0 };
  853. Array<@string_index_type@, @variants_size@> variants {};
  854. size_t variants_size { 0 };
  855. };
  856. struct LanguageMapping {
  857. CanonicalLanguageID key;
  858. CanonicalLanguageID alias;
  859. };
  860. )~~~");
  861. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  862. generator.set("size", String::number(mappings.size()));
  863. generator.set("name"sv, name);
  864. generator.append(R"~~~(
  865. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  866. )~~~");
  867. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  868. auto const& lhs_language = locale_data.unique_strings.get(lhs.key.language);
  869. auto const& rhs_language = locale_data.unique_strings.get(rhs.key.language);
  870. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  871. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  872. return false;
  873. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  874. return true;
  875. return lhs_language < rhs_language;
  876. });
  877. for (auto const& mapping : mappings) {
  878. generator.set("language"sv, String::number(mapping.key.language));
  879. generator.append(" { { @language@");
  880. append_index(mapping.key.script);
  881. append_index(mapping.key.region);
  882. append_list_and_size(mapping.key.variants);
  883. generator.set("language"sv, String::number(mapping.alias.language));
  884. generator.append(" }, { @language@");
  885. append_index(mapping.alias.script);
  886. append_index(mapping.alias.region);
  887. append_list_and_size(mapping.alias.variants);
  888. generator.append(" } },\n");
  889. }
  890. generator.append("} };\n");
  891. };
  892. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  893. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  894. generator.append(R"~~~(
  895. static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& language_id)
  896. {
  897. // https://unicode.org/reports/tr35/#Likely_Subtags
  898. enum class State {
  899. LanguageScriptRegion,
  900. LanguageRegion,
  901. LanguageScript,
  902. Language,
  903. UndScript,
  904. Done,
  905. };
  906. auto state = State::LanguageScriptRegion;
  907. while (state != State::Done) {
  908. Unicode::LanguageID search_key;
  909. switch (state) {
  910. case State::LanguageScriptRegion:
  911. state = State::LanguageRegion;
  912. if (!language_id.script.has_value() || !language_id.region.has_value())
  913. continue;
  914. search_key.language = *language_id.language;
  915. search_key.script = *language_id.script;
  916. search_key.region = *language_id.region;
  917. break;
  918. case State::LanguageRegion:
  919. state = State::LanguageScript;
  920. if (!language_id.region.has_value())
  921. continue;
  922. search_key.language = *language_id.language;
  923. search_key.region = *language_id.region;
  924. break;
  925. case State::LanguageScript:
  926. state = State::Language;
  927. if (!language_id.script.has_value())
  928. continue;
  929. search_key.language = *language_id.language;
  930. search_key.script = *language_id.script;
  931. break;
  932. case State::Language:
  933. state = State::UndScript;
  934. search_key.language = *language_id.language;
  935. break;
  936. case State::UndScript:
  937. state = State::Done;
  938. if (!language_id.script.has_value())
  939. continue;
  940. search_key.language = "und"sv;
  941. search_key.script = *language_id.script;
  942. break;
  943. default:
  944. VERIFY_NOT_REACHED();
  945. }
  946. for (auto const& map : s_likely_subtags) {
  947. auto const& key_language = s_string_list[map.key.language];
  948. auto const& key_script = s_string_list[map.key.script];
  949. auto const& key_region = s_string_list[map.key.region];
  950. if (key_language != search_key.language)
  951. continue;
  952. if (!key_script.is_empty() || search_key.script.has_value()) {
  953. if (key_script != search_key.script)
  954. continue;
  955. }
  956. if (!key_region.is_empty() || search_key.region.has_value()) {
  957. if (key_region != search_key.region)
  958. continue;
  959. }
  960. return &map;
  961. }
  962. }
  963. return nullptr;
  964. }
  965. )~~~");
  966. auto append_mapping_search = [&](StringView enum_title, StringView enum_snake, StringView collection_name) {
  967. generator.set("enum_title", enum_title);
  968. generator.set("enum_snake", enum_snake);
  969. generator.set("collection_name", collection_name);
  970. generator.append(R"~~~(
  971. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  972. {
  973. auto locale_value = locale_from_string(locale);
  974. if (!locale_value.has_value())
  975. return {};
  976. auto @enum_snake@_value = @enum_snake@_from_string(@enum_snake@);
  977. if (!@enum_snake@_value.has_value())
  978. return {};
  979. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  980. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  981. auto const& mappings = @collection_name@.at(locale_index);
  982. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  983. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  984. if (@enum_snake@_mapping.is_empty())
  985. return {};
  986. return @enum_snake@_mapping;
  987. }
  988. )~~~");
  989. };
  990. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values) {
  991. HashValueMap<String> hashes;
  992. hashes.ensure_capacity(values.size());
  993. for (auto const& value : values)
  994. hashes.set(value.hash(), format_identifier(enum_title, value));
  995. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  996. };
  997. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  998. HashValueMap<StringIndexType> hashes;
  999. hashes.ensure_capacity(aliases.size());
  1000. for (auto const& alias : aliases)
  1001. hashes.set(alias.key.hash(), alias.value);
  1002. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), "StringView"sv, "s_string_list[{}]"sv);
  1003. };
  1004. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys());
  1005. append_mapping_search("Language"sv, "language"sv, "s_languages"sv);
  1006. append_from_string("Language"sv, "language"sv, locale_data.languages);
  1007. append_alias_search("language"sv, locale_data.language_aliases);
  1008. append_mapping_search("Territory"sv, "territory"sv, "s_territories"sv);
  1009. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  1010. append_alias_search("territory"sv, locale_data.territory_aliases);
  1011. append_mapping_search("ScriptTag"sv, "script_tag"sv, "s_scripts"sv);
  1012. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  1013. append_alias_search("script_tag"sv, locale_data.script_aliases);
  1014. append_mapping_search("Currency"sv, "currency"sv, "s_currencies"sv);
  1015. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  1016. append_mapping_search("Key"sv, "key"sv, "s_keywords"sv);
  1017. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  1018. append_alias_search("variant"sv, locale_data.variant_aliases);
  1019. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  1020. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  1021. append_from_string("ListPatternStyle"sv, "list_pattern_style"sv, locale_data.list_pattern_styles);
  1022. append_from_string("NumericSymbol"sv, "numeric_symbol"sv, locale_data.numeric_symbols);
  1023. generator.append(R"~~~(
  1024. static NumberSystem const* find_number_system(StringView locale, StringView system)
  1025. {
  1026. auto locale_value = locale_from_string(locale);
  1027. if (!locale_value.has_value())
  1028. return nullptr;
  1029. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1030. auto const& number_systems = s_number_systems.at(locale_index);
  1031. for (auto const& number_system : number_systems) {
  1032. if (system == s_string_list[number_system.system])
  1033. return &number_system;
  1034. };
  1035. return nullptr;
  1036. }
  1037. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView symbol)
  1038. {
  1039. auto symbol_value = numeric_symbol_from_string(symbol);
  1040. if (!symbol_value.has_value())
  1041. return {};
  1042. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  1043. auto symbol_index = to_underlying(*symbol_value);
  1044. return s_string_list[number_system->symbols[symbol_index]];
  1045. }
  1046. return {};
  1047. }
  1048. Optional<Unicode::NumberFormat> get_standard_number_system_format(StringView locale, StringView system, StandardNumberFormatType type)
  1049. {
  1050. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  1051. switch (type) {
  1052. case StandardNumberFormatType::Decimal:
  1053. return number_system->decimal_format.to_unicode_number_format();
  1054. case StandardNumberFormatType::Currency:
  1055. return number_system->currency_format.to_unicode_number_format();
  1056. case StandardNumberFormatType::Accounting:
  1057. return number_system->accounting_format.to_unicode_number_format();
  1058. case StandardNumberFormatType::Percent:
  1059. return number_system->percent_format.to_unicode_number_format();
  1060. }
  1061. }
  1062. return {};
  1063. }
  1064. Vector<Unicode::NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type)
  1065. {
  1066. Vector<Unicode::NumberFormat> formats;
  1067. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  1068. Span<NumberFormat const> number_formats;
  1069. switch (type) {
  1070. case CompactNumberFormatType::DecimalLong:
  1071. number_formats = number_system->decimal_long_formats;
  1072. break;
  1073. case CompactNumberFormatType::DecimalShort:
  1074. number_formats = number_system->decimal_short_formats;
  1075. break;
  1076. case CompactNumberFormatType::CurrencyShort:
  1077. number_formats = number_system->currency_short_formats;
  1078. break;
  1079. }
  1080. formats.ensure_capacity(number_formats.size());
  1081. for (auto const& number_format : number_formats)
  1082. formats.append(number_format.to_unicode_number_format());
  1083. }
  1084. return formats;
  1085. }
  1086. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style)
  1087. {
  1088. auto locale_value = locale_from_string(locale);
  1089. if (!locale_value.has_value())
  1090. return {};
  1091. auto type_value = list_pattern_type_from_string(list_pattern_type);
  1092. if (!type_value.has_value())
  1093. return {};
  1094. auto style_value = list_pattern_style_from_string(list_pattern_style);
  1095. if (!style_value.has_value())
  1096. return {};
  1097. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1098. auto const& locale_list_patterns = s_list_patterns.at(locale_index);
  1099. for (auto const& list_patterns : locale_list_patterns) {
  1100. if ((list_patterns.type == type_value) && (list_patterns.style == style_value)) {
  1101. auto const& start = s_string_list[list_patterns.start];
  1102. auto const& middle = s_string_list[list_patterns.middle];
  1103. auto const& end = s_string_list[list_patterns.end];
  1104. auto const& pair = s_string_list[list_patterns.pair];
  1105. return ListPatterns { start, middle, end, pair };
  1106. }
  1107. }
  1108. return {};
  1109. }
  1110. void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
  1111. {
  1112. for (auto const& map : s_complex_alias) {
  1113. auto const& key_language = s_string_list[map.key.language];
  1114. auto const& key_script = s_string_list[map.key.script];
  1115. auto const& key_region = s_string_list[map.key.region];
  1116. if ((key_language != language_id.language) && (key_language != "und"sv))
  1117. continue;
  1118. if (!key_script.is_empty() && (key_script != language_id.script))
  1119. continue;
  1120. if (!key_region.is_empty() && (key_region != language_id.region))
  1121. continue;
  1122. if (!map.key.matches_variants(language_id.variants))
  1123. continue;
  1124. auto alias = map.alias.to_unicode_language_id();
  1125. if (alias.language == "und"sv)
  1126. alias.language = move(language_id.language);
  1127. if (key_script.is_empty() && !alias.script.has_value())
  1128. alias.script = move(language_id.script);
  1129. if (key_region.is_empty() && !alias.region.has_value())
  1130. alias.region = move(language_id.region);
  1131. if (map.key.variants_size == 0 && alias.variants.is_empty())
  1132. alias.variants = move(language_id.variants);
  1133. language_id = move(alias);
  1134. break;
  1135. }
  1136. }
  1137. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id)
  1138. {
  1139. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  1140. auto const* likely_subtag = resolve_likely_subtag(language_id);
  1141. if (likely_subtag == nullptr)
  1142. return {};
  1143. auto maximized = language_id;
  1144. auto const& key_script = s_string_list[likely_subtag->key.script];
  1145. auto const& key_region = s_string_list[likely_subtag->key.region];
  1146. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  1147. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  1148. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  1149. if (maximized.language == "und"sv)
  1150. maximized.language = alias_language;
  1151. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  1152. maximized.script = alias_script;
  1153. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  1154. maximized.region = alias_region;
  1155. return maximized;
  1156. }
  1157. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
  1158. {
  1159. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  1160. return s_string_list[likely_subtag->alias.region];
  1161. return {};
  1162. }
  1163. }
  1164. )~~~");
  1165. file.write(generator.as_string_view());
  1166. }
  1167. int main(int argc, char** argv)
  1168. {
  1169. char const* generated_header_path = nullptr;
  1170. char const* generated_implementation_path = nullptr;
  1171. char const* core_path = nullptr;
  1172. char const* locale_names_path = nullptr;
  1173. char const* misc_path = nullptr;
  1174. char const* numbers_path = nullptr;
  1175. Core::ArgsParser args_parser;
  1176. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1177. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1178. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1179. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  1180. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  1181. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  1182. args_parser.parse(argc, argv);
  1183. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  1184. if (path.is_empty()) {
  1185. warnln("{} is required", flags);
  1186. args_parser.print_usage(stderr, argv[0]);
  1187. exit(1);
  1188. }
  1189. auto file_or_error = Core::File::open(path, mode);
  1190. if (file_or_error.is_error()) {
  1191. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  1192. exit(1);
  1193. }
  1194. return file_or_error.release_value();
  1195. };
  1196. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  1197. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  1198. UnicodeLocaleData locale_data;
  1199. parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, locale_data);
  1200. generate_unicode_locale_header(generated_header_file, locale_data);
  1201. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  1202. return 0;
  1203. }