GenerateUnicodeLocale.cpp 62 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577
  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 <LibUnicode/Locale.h>
  23. #include <math.h>
  24. using StringIndexType = u16;
  25. constexpr auto s_string_index_type = "u16"sv;
  26. struct NumberFormat : public Unicode::NumberFormat {
  27. using Base = Unicode::NumberFormat;
  28. static Base::Plurality plurality_from_string(StringView plurality)
  29. {
  30. if (plurality == "other"sv)
  31. return Base::Plurality::Other;
  32. if (plurality == "1"sv)
  33. return Base::Plurality::Single;
  34. if (plurality == "zero"sv)
  35. return Base::Plurality::Zero;
  36. if (plurality == "one"sv)
  37. return Base::Plurality::One;
  38. if (plurality == "two"sv)
  39. return Base::Plurality::Two;
  40. if (plurality == "few"sv)
  41. return Base::Plurality::Few;
  42. if (plurality == "many"sv)
  43. return Base::Plurality::Many;
  44. VERIFY_NOT_REACHED();
  45. }
  46. StringIndexType format_index { 0 };
  47. };
  48. struct NumberSystem {
  49. StringIndexType system { 0 };
  50. HashMap<String, StringIndexType> symbols {};
  51. Vector<NumberFormat> decimal_long_formats {};
  52. Vector<NumberFormat> decimal_short_formats {};
  53. Vector<NumberFormat> currency_short_formats {};
  54. };
  55. struct ListPatterns {
  56. String type;
  57. String style;
  58. StringIndexType start { 0 };
  59. StringIndexType middle { 0 };
  60. StringIndexType end { 0 };
  61. StringIndexType pair { 0 };
  62. };
  63. struct Locale {
  64. String language;
  65. Optional<String> territory;
  66. Optional<String> variant;
  67. HashMap<String, StringIndexType> languages;
  68. HashMap<String, StringIndexType> territories;
  69. HashMap<String, StringIndexType> scripts;
  70. HashMap<String, StringIndexType> currencies;
  71. HashMap<String, StringIndexType> keywords;
  72. HashMap<String, NumberSystem> number_systems;
  73. Vector<ListPatterns> list_patterns;
  74. };
  75. struct CanonicalLanguageID {
  76. StringIndexType language { 0 };
  77. StringIndexType script { 0 };
  78. StringIndexType region { 0 };
  79. Vector<StringIndexType> variants {};
  80. };
  81. struct LanguageMapping {
  82. CanonicalLanguageID key {};
  83. CanonicalLanguageID alias {};
  84. };
  85. struct UnicodeLocaleData {
  86. Vector<String> unique_strings;
  87. HashMap<StringView, StringIndexType> unique_string_indices;
  88. HashMap<String, Locale> locales;
  89. Vector<String> languages;
  90. Vector<String> territories;
  91. Vector<String> scripts;
  92. Vector<String> variants;
  93. Vector<String> currencies;
  94. Vector<String> keywords;
  95. Vector<String> numeric_symbols;
  96. Vector<String> list_pattern_types;
  97. Vector<String> list_pattern_styles;
  98. HashMap<String, StringIndexType> language_aliases;
  99. HashMap<String, StringIndexType> territory_aliases;
  100. HashMap<String, StringIndexType> script_aliases;
  101. HashMap<String, StringIndexType> variant_aliases;
  102. HashMap<String, StringIndexType> subdivision_aliases;
  103. Vector<LanguageMapping> complex_mappings;
  104. Vector<LanguageMapping> likely_subtags;
  105. size_t max_variant_size { 0 };
  106. };
  107. static StringIndexType ensure_unique_string(UnicodeLocaleData& locale_data, String string)
  108. {
  109. // We maintain a set of unique strings in two structures: a vector which owns the unique string,
  110. // and a hash map which maps that string to its index in the vector. The vector is to ensure the
  111. // strings are generated in an easily known order, and the map is to allow quickly deciding if a
  112. // string is actually unique (otherwise, we'd have to linear-search the vector for each string).
  113. //
  114. // Also note that index 0 will be reserved for the empty string, so the index returned from this
  115. // method is actually the real index in the vector + 1.
  116. if (auto index = locale_data.unique_string_indices.get(string); index.has_value())
  117. return *index;
  118. locale_data.unique_strings.append(move(string));
  119. size_t index = locale_data.unique_strings.size();
  120. // There are currently on the order of 46K unique strings in UnicodeLocale.cpp.
  121. // If that number reaches 2^16, bump the StringIndexType alias to a u32.
  122. VERIFY(index < NumericLimits<StringIndexType>::max());
  123. auto string_index = static_cast<StringIndexType>(index);
  124. locale_data.unique_string_indices.set(locale_data.unique_strings.last(), string_index);
  125. return string_index;
  126. }
  127. static StringView get_unique_string(UnicodeLocaleData& locale_data, StringIndexType index)
  128. {
  129. if (index == 0)
  130. return {};
  131. VERIFY(index <= locale_data.unique_strings.size());
  132. return locale_data.unique_strings.at(index - 1);
  133. }
  134. static Optional<CanonicalLanguageID> parse_language(UnicodeLocaleData& locale_data, StringView language)
  135. {
  136. CanonicalLanguageID language_id {};
  137. auto segments = language.split_view('-');
  138. VERIFY(!segments.is_empty());
  139. size_t index = 0;
  140. if (Unicode::is_unicode_language_subtag(segments[index])) {
  141. language_id.language = ensure_unique_string(locale_data, segments[index]);
  142. if (segments.size() == ++index)
  143. return language_id;
  144. } else {
  145. return {};
  146. }
  147. if (Unicode::is_unicode_script_subtag(segments[index])) {
  148. language_id.script = ensure_unique_string(locale_data, segments[index]);
  149. if (segments.size() == ++index)
  150. return language_id;
  151. }
  152. if (Unicode::is_unicode_region_subtag(segments[index])) {
  153. language_id.region = ensure_unique_string(locale_data, segments[index]);
  154. if (segments.size() == ++index)
  155. return language_id;
  156. }
  157. while (index < segments.size()) {
  158. if (!Unicode::is_unicode_variant_subtag(segments[index]))
  159. return {};
  160. language_id.variants.append(ensure_unique_string(locale_data, segments[index++]));
  161. }
  162. return language_id;
  163. }
  164. static Optional<LanguageMapping> parse_language_mapping(UnicodeLocaleData& locale_data, StringView key, StringView alias)
  165. {
  166. auto parsed_key = parse_language(locale_data, key);
  167. if (!parsed_key.has_value())
  168. return {};
  169. auto parsed_alias = parse_language(locale_data, alias);
  170. if (!parsed_alias.has_value())
  171. return {};
  172. return LanguageMapping { parsed_key.release_value(), parsed_alias.release_value() };
  173. }
  174. static void parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
  175. {
  176. LexicalPath core_aliases_path(move(core_supplemental_path));
  177. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  178. VERIFY(Core::File::exists(core_aliases_path.string()));
  179. auto core_aliases_file_or_error = Core::File::open(core_aliases_path.string(), Core::OpenMode::ReadOnly);
  180. VERIFY(!core_aliases_file_or_error.is_error());
  181. auto core_aliases = JsonParser(core_aliases_file_or_error.value()->read_all()).parse();
  182. VERIFY(core_aliases.has_value());
  183. auto const& supplemental_object = core_aliases->as_object().get("supplemental"sv);
  184. auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
  185. auto const& alias_object = metadata_object.as_object().get("alias"sv);
  186. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  187. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  188. auto alias = value.as_object().get("_replacement"sv).as_string();
  189. if (key.contains('-')) {
  190. auto mapping = parse_language_mapping(locale_data, key, alias);
  191. if (!mapping.has_value())
  192. return;
  193. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  194. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  195. locale_data.complex_mappings.append(mapping.release_value());
  196. } else {
  197. alias_map.set(key, ensure_unique_string(locale_data, alias));
  198. }
  199. });
  200. };
  201. append_aliases(alias_object.as_object().get("languageAlias"sv), locale_data.language_aliases);
  202. append_aliases(alias_object.as_object().get("territoryAlias"sv), locale_data.territory_aliases);
  203. append_aliases(alias_object.as_object().get("scriptAlias"sv), locale_data.script_aliases);
  204. append_aliases(alias_object.as_object().get("variantAlias"sv), locale_data.variant_aliases);
  205. append_aliases(alias_object.as_object().get("subdivisionAlias"sv), locale_data.subdivision_aliases);
  206. }
  207. static void parse_likely_subtags(String core_supplemental_path, UnicodeLocaleData& locale_data)
  208. {
  209. LexicalPath likely_subtags_path(move(core_supplemental_path));
  210. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  211. VERIFY(Core::File::exists(likely_subtags_path.string()));
  212. auto likely_subtags_file_or_error = Core::File::open(likely_subtags_path.string(), Core::OpenMode::ReadOnly);
  213. VERIFY(!likely_subtags_file_or_error.is_error());
  214. auto likely_subtags = JsonParser(likely_subtags_file_or_error.value()->read_all()).parse();
  215. VERIFY(likely_subtags.has_value());
  216. auto const& supplemental_object = likely_subtags->as_object().get("supplemental"sv);
  217. auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
  218. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  219. auto mapping = parse_language_mapping(locale_data, key, value.as_string());
  220. if (!mapping.has_value())
  221. return;
  222. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  223. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  224. locale_data.likely_subtags.append(mapping.release_value());
  225. });
  226. }
  227. static void parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  228. {
  229. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  230. languages_path = languages_path.append("languages.json"sv);
  231. VERIFY(Core::File::exists(languages_path.string()));
  232. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  233. VERIFY(!languages_file_or_error.is_error());
  234. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  235. VERIFY(languages.has_value());
  236. auto const& main_object = languages->as_object().get("main"sv);
  237. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  238. auto const& identity_object = locale_object.as_object().get("identity"sv);
  239. auto const& language_string = identity_object.as_object().get("language"sv);
  240. auto const& territory_string = identity_object.as_object().get("territory"sv);
  241. auto const& variant_string = identity_object.as_object().get("variant"sv);
  242. locale.language = language_string.as_string();
  243. if (!locale_data.languages.contains_slow(locale.language))
  244. locale_data.languages.append(locale.language);
  245. if (territory_string.is_string()) {
  246. locale.territory = territory_string.as_string();
  247. if (!locale_data.territories.contains_slow(*locale.territory))
  248. locale_data.territories.append(*locale.territory);
  249. }
  250. if (variant_string.is_string()) {
  251. locale.variant = variant_string.as_string();
  252. if (!locale_data.variants.contains_slow(*locale.variant))
  253. locale_data.variants.append(*locale.variant);
  254. }
  255. }
  256. static void parse_locale_languages(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  257. {
  258. LexicalPath languages_path(move(locale_path));
  259. languages_path = languages_path.append("languages.json"sv);
  260. VERIFY(Core::File::exists(languages_path.string()));
  261. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  262. VERIFY(!languages_file_or_error.is_error());
  263. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  264. VERIFY(languages.has_value());
  265. auto const& main_object = languages->as_object().get("main"sv);
  266. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  267. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  268. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  269. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  270. if (!locale_data.languages.contains_slow(key))
  271. return;
  272. auto index = ensure_unique_string(locale_data, value.as_string());
  273. locale.languages.set(key, index);
  274. });
  275. }
  276. static void parse_locale_territories(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  277. {
  278. LexicalPath territories_path(move(locale_path));
  279. territories_path = territories_path.append("territories.json"sv);
  280. VERIFY(Core::File::exists(territories_path.string()));
  281. auto territories_file_or_error = Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly);
  282. VERIFY(!territories_file_or_error.is_error());
  283. auto territories = JsonParser(territories_file_or_error.value()->read_all()).parse();
  284. VERIFY(territories.has_value());
  285. auto const& main_object = territories->as_object().get("main"sv);
  286. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  287. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  288. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  289. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  290. if (!locale_data.territories.contains_slow(key))
  291. return;
  292. auto index = ensure_unique_string(locale_data, value.as_string());
  293. locale.territories.set(key, index);
  294. });
  295. }
  296. static void parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  297. {
  298. LexicalPath scripts_path(move(locale_path));
  299. scripts_path = scripts_path.append("scripts.json"sv);
  300. VERIFY(Core::File::exists(scripts_path.string()));
  301. auto scripts_file_or_error = Core::File::open(scripts_path.string(), Core::OpenMode::ReadOnly);
  302. VERIFY(!scripts_file_or_error.is_error());
  303. auto scripts = JsonParser(scripts_file_or_error.value()->read_all()).parse();
  304. VERIFY(scripts.has_value());
  305. auto const& main_object = scripts->as_object().get("main"sv);
  306. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  307. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  308. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  309. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  310. auto index = ensure_unique_string(locale_data, value.as_string());
  311. locale.scripts.set(key, index);
  312. if (!locale_data.scripts.contains_slow(key))
  313. locale_data.scripts.append(key);
  314. });
  315. }
  316. static void parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
  317. {
  318. LexicalPath list_patterns_path(move(misc_path));
  319. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  320. VERIFY(Core::File::exists(list_patterns_path.string()));
  321. auto list_patterns_file_or_error = Core::File::open(list_patterns_path.string(), Core::OpenMode::ReadOnly);
  322. VERIFY(!list_patterns_file_or_error.is_error());
  323. auto list_patterns = JsonParser(list_patterns_file_or_error.value()->read_all()).parse();
  324. VERIFY(list_patterns.has_value());
  325. auto const& main_object = list_patterns->as_object().get("main"sv);
  326. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  327. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  328. auto list_pattern_type = [](StringView key) {
  329. if (key.contains("type-standard"sv))
  330. return "conjunction"sv;
  331. if (key.contains("type-or"sv))
  332. return "disjunction"sv;
  333. if (key.contains("type-unit"sv))
  334. return "unit"sv;
  335. VERIFY_NOT_REACHED();
  336. };
  337. auto list_pattern_style = [](StringView key) {
  338. if (key.contains("short"sv))
  339. return "short"sv;
  340. if (key.contains("narrow"sv))
  341. return "narrow"sv;
  342. return "long"sv;
  343. };
  344. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  345. auto type = list_pattern_type(key);
  346. auto style = list_pattern_style(key);
  347. auto start = ensure_unique_string(locale_data, value.as_object().get("start"sv).as_string());
  348. auto middle = ensure_unique_string(locale_data, value.as_object().get("middle"sv).as_string());
  349. auto end = ensure_unique_string(locale_data, value.as_object().get("end"sv).as_string());
  350. auto pair = ensure_unique_string(locale_data, value.as_object().get("2"sv).as_string());
  351. if (!locale_data.list_pattern_types.contains_slow(type))
  352. locale_data.list_pattern_types.append(type);
  353. if (!locale_data.list_pattern_styles.contains_slow(style))
  354. locale_data.list_pattern_styles.append(style);
  355. locale.list_patterns.append({ move(type), move(style), move(start), move(middle), move(end), move(pair) });
  356. });
  357. }
  358. static void parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  359. {
  360. LexicalPath currencies_path(move(numbers_path));
  361. currencies_path = currencies_path.append("currencies.json"sv);
  362. VERIFY(Core::File::exists(currencies_path.string()));
  363. auto currencies_file_or_error = Core::File::open(currencies_path.string(), Core::OpenMode::ReadOnly);
  364. VERIFY(!currencies_file_or_error.is_error());
  365. auto currencies = JsonParser(currencies_file_or_error.value()->read_all()).parse();
  366. VERIFY(currencies.has_value());
  367. auto const& main_object = currencies->as_object().get("main"sv);
  368. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  369. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  370. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  371. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  372. auto const& display_name = value.as_object().get("displayName"sv);
  373. auto index = ensure_unique_string(locale_data, display_name.as_string());
  374. locale.currencies.set(key, index);
  375. if (!locale_data.currencies.contains_slow(key))
  376. locale_data.currencies.append(key);
  377. });
  378. }
  379. static void parse_numeric_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  380. {
  381. static constexpr StringView key = "nu"sv;
  382. LexicalPath numbers_path(move(locale_numbers_path));
  383. numbers_path = numbers_path.append("numbers.json"sv);
  384. VERIFY(Core::File::exists(numbers_path.string()));
  385. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  386. VERIFY(!numbers_file_or_error.is_error());
  387. auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
  388. VERIFY(numbers.has_value());
  389. auto const& main_object = numbers->as_object().get("main"sv);
  390. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  391. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  392. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  393. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  394. Vector<String> keyword_values {};
  395. keyword_values.append(default_numbering_system_object.as_string());
  396. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  397. auto keyword_value = value.as_string();
  398. if (!keyword_values.contains_slow(keyword_value))
  399. keyword_values.append(move(keyword_value));
  400. });
  401. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  402. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  403. return;
  404. auto keyword_value = value.as_string();
  405. if (!keyword_values.contains_slow(keyword_value))
  406. keyword_values.append(move(keyword_value));
  407. });
  408. StringBuilder builder;
  409. builder.join(',', keyword_values);
  410. auto index = ensure_unique_string(locale_data, builder.build());
  411. locale.keywords.set(key, index);
  412. if (!locale_data.keywords.contains_slow(key))
  413. locale_data.keywords.append(key);
  414. }
  415. static void parse_number_systems(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  416. {
  417. LexicalPath numbers_path(move(locale_numbers_path));
  418. numbers_path = numbers_path.append("numbers.json"sv);
  419. VERIFY(Core::File::exists(numbers_path.string()));
  420. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  421. VERIFY(!numbers_file_or_error.is_error());
  422. auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
  423. VERIFY(numbers.has_value());
  424. auto const& main_object = numbers->as_object().get("main"sv);
  425. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  426. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  427. auto ensure_number_system = [&](auto const& system) -> NumberSystem& {
  428. return locale.number_systems.ensure(system, [&]() {
  429. auto system_index = ensure_unique_string(locale_data, system);
  430. return NumberSystem { .system = system_index };
  431. });
  432. };
  433. auto parse_number_format = [&](auto const& format_object) {
  434. Vector<NumberFormat> result;
  435. result.ensure_capacity(format_object.size());
  436. format_object.for_each_member([&](auto const& key, JsonValue const& value) {
  437. auto split_key = key.split_view('-');
  438. VERIFY(split_key.size() == 3);
  439. auto type = split_key[0].template to_uint<u64>().value();
  440. VERIFY(type % 10 == 0);
  441. NumberFormat format {};
  442. format.magnitude = static_cast<u8>(log10(type));
  443. format.plurality = NumberFormat::plurality_from_string(split_key[2]);
  444. format.format_index = ensure_unique_string(locale_data, value.as_string());
  445. result.append(move(format));
  446. });
  447. return result;
  448. };
  449. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  450. constexpr auto symbols_prefix = "symbols-numberSystem-"sv;
  451. constexpr auto decimal_formats_prefix = "decimalFormats-numberSystem-"sv;
  452. constexpr auto currency_formats_prefix = "currencyFormats-numberSystem-"sv;
  453. if (key.starts_with(symbols_prefix)) {
  454. auto system = key.substring(symbols_prefix.length());
  455. auto& number_system = ensure_number_system(system);
  456. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& localization) {
  457. auto symbol_index = ensure_unique_string(locale_data, localization.as_string());
  458. number_system.symbols.set(symbol, symbol_index);
  459. if (!locale_data.numeric_symbols.contains_slow(symbol))
  460. locale_data.numeric_symbols.append(symbol);
  461. });
  462. } else if (key.starts_with(decimal_formats_prefix)) {
  463. auto system = key.substring(decimal_formats_prefix.length());
  464. auto& number_system = ensure_number_system(system);
  465. auto const& long_format = value.as_object().get("long"sv).as_object().get("decimalFormat"sv);
  466. number_system.decimal_long_formats = parse_number_format(long_format.as_object());
  467. auto const& short_format = value.as_object().get("short"sv).as_object().get("decimalFormat"sv);
  468. number_system.decimal_short_formats = parse_number_format(short_format.as_object());
  469. } else if (key.starts_with(currency_formats_prefix)) {
  470. if (value.as_object().has("short"sv)) {
  471. auto system = key.substring(currency_formats_prefix.length());
  472. auto& number_system = ensure_number_system(system);
  473. auto const& short_format = value.as_object().get("short"sv).as_object().get("standard"sv);
  474. number_system.currency_short_formats = parse_number_format(short_format.as_object());
  475. }
  476. }
  477. });
  478. }
  479. static void parse_default_content_locales(String core_path, UnicodeLocaleData& locale_data)
  480. {
  481. LexicalPath default_content_path(move(core_path));
  482. default_content_path = default_content_path.append("defaultContent.json"sv);
  483. VERIFY(Core::File::exists(default_content_path.string()));
  484. auto default_content_file_or_error = Core::File::open(default_content_path.string(), Core::OpenMode::ReadOnly);
  485. VERIFY(!default_content_file_or_error.is_error());
  486. auto default_content = JsonParser(default_content_file_or_error.value()->read_all()).parse();
  487. VERIFY(default_content.has_value());
  488. auto const& default_content_array = default_content->as_object().get("defaultContent"sv);
  489. default_content_array.as_array().for_each([&](JsonValue const& value) {
  490. auto locale = value.as_string();
  491. StringView default_locale = locale;
  492. while (true) {
  493. if (locale_data.locales.contains(default_locale))
  494. break;
  495. auto pos = default_locale.find_last('-');
  496. if (!pos.has_value())
  497. return;
  498. default_locale = default_locale.substring_view(0, *pos);
  499. }
  500. locale_data.locales.set(locale, locale_data.locales.get(default_locale).value());
  501. });
  502. }
  503. static Core::DirIterator path_to_dir_iterator(String path)
  504. {
  505. LexicalPath lexical_path(move(path));
  506. lexical_path = lexical_path.append("main"sv);
  507. VERIFY(Core::File::is_directory(lexical_path.string()));
  508. Core::DirIterator iterator(lexical_path.string(), Core::DirIterator::SkipParentAndBaseDir);
  509. if (iterator.has_error()) {
  510. warnln("{}: {}", lexical_path.string(), iterator.error_string());
  511. VERIFY_NOT_REACHED();
  512. }
  513. return iterator;
  514. }
  515. static void parse_all_locales(String core_path, String locale_names_path, String misc_path, String numbers_path, UnicodeLocaleData& locale_data)
  516. {
  517. auto identity_iterator = path_to_dir_iterator(locale_names_path);
  518. auto locale_names_iterator = path_to_dir_iterator(move(locale_names_path));
  519. auto misc_iterator = path_to_dir_iterator(move(misc_path));
  520. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  521. LexicalPath core_supplemental_path(core_path);
  522. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  523. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  524. parse_core_aliases(core_supplemental_path.string(), locale_data);
  525. parse_likely_subtags(core_supplemental_path.string(), locale_data);
  526. auto remove_variants_from_path = [&](String path) -> Optional<String> {
  527. auto parsed_locale = parse_language(locale_data, LexicalPath::basename(path));
  528. if (!parsed_locale.has_value())
  529. return {};
  530. StringBuilder builder;
  531. builder.append(get_unique_string(locale_data, parsed_locale->language));
  532. if (auto script = get_unique_string(locale_data, parsed_locale->script); !script.is_empty())
  533. builder.appendff("-{}", script);
  534. if (auto region = get_unique_string(locale_data, parsed_locale->region); !region.is_empty())
  535. builder.appendff("-{}", region);
  536. return builder.build();
  537. };
  538. while (identity_iterator.has_next()) {
  539. auto locale_path = identity_iterator.next_full_path();
  540. VERIFY(Core::File::is_directory(locale_path));
  541. auto language = remove_variants_from_path(locale_path);
  542. if (!language.has_value())
  543. continue;
  544. auto& locale = locale_data.locales.ensure(*language);
  545. parse_identity(locale_path, locale_data, locale);
  546. }
  547. while (locale_names_iterator.has_next()) {
  548. auto locale_path = locale_names_iterator.next_full_path();
  549. VERIFY(Core::File::is_directory(locale_path));
  550. auto language = remove_variants_from_path(locale_path);
  551. if (!language.has_value())
  552. continue;
  553. auto& locale = locale_data.locales.ensure(*language);
  554. parse_locale_languages(locale_path, locale_data, locale);
  555. parse_locale_territories(locale_path, locale_data, locale);
  556. parse_locale_scripts(locale_path, locale_data, locale);
  557. }
  558. while (misc_iterator.has_next()) {
  559. auto misc_path = misc_iterator.next_full_path();
  560. VERIFY(Core::File::is_directory(misc_path));
  561. auto language = remove_variants_from_path(misc_path);
  562. if (!language.has_value())
  563. continue;
  564. auto& locale = locale_data.locales.ensure(*language);
  565. parse_locale_list_patterns(misc_path, locale_data, locale);
  566. }
  567. while (numbers_iterator.has_next()) {
  568. auto numbers_path = numbers_iterator.next_full_path();
  569. VERIFY(Core::File::is_directory(numbers_path));
  570. auto language = remove_variants_from_path(numbers_path);
  571. if (!language.has_value())
  572. continue;
  573. auto& locale = locale_data.locales.ensure(*language);
  574. parse_locale_currencies(numbers_path, locale_data, locale);
  575. parse_numeric_keywords(numbers_path, locale_data, locale);
  576. parse_number_systems(numbers_path, locale_data, locale);
  577. }
  578. parse_default_content_locales(move(core_path), locale_data);
  579. }
  580. static String format_identifier(StringView owner, String identifier)
  581. {
  582. identifier = identifier.replace("-"sv, "_"sv, true);
  583. if (all_of(identifier, is_ascii_digit))
  584. return String::formatted("{}_{}", owner[0], identifier);
  585. if (is_ascii_lower_alpha(identifier[0]))
  586. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  587. return identifier;
  588. }
  589. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  590. {
  591. StringBuilder builder;
  592. SourceGenerator generator { builder };
  593. auto generate_enum = [&](StringView name, StringView default_, Vector<String>& values) {
  594. quick_sort(values);
  595. generator.set("name", name);
  596. generator.set("underlying", ((values.size() + !default_.is_empty()) < 256) ? "u8"sv : "u16"sv);
  597. generator.append(R"~~~(
  598. enum class @name@ : @underlying@ {)~~~");
  599. if (!default_.is_empty()) {
  600. generator.set("default", default_);
  601. generator.append(R"~~~(
  602. @default@,)~~~");
  603. }
  604. for (auto const& value : values) {
  605. generator.set("value", format_identifier(name, value));
  606. generator.append(R"~~~(
  607. @value@,)~~~");
  608. }
  609. generator.append(R"~~~(
  610. };
  611. )~~~");
  612. };
  613. generator.append(R"~~~(
  614. #pragma once
  615. #include <AK/Optional.h>
  616. #include <AK/StringView.h>
  617. #include <AK/Types.h>
  618. #include <AK/Vector.h>
  619. #include <LibUnicode/Forward.h>
  620. namespace Unicode {
  621. )~~~");
  622. auto locales = locale_data.locales.keys();
  623. generate_enum("Locale"sv, "None"sv, locales);
  624. generate_enum("Language"sv, {}, locale_data.languages);
  625. generate_enum("Territory"sv, {}, locale_data.territories);
  626. generate_enum("ScriptTag"sv, {}, locale_data.scripts);
  627. generate_enum("Currency"sv, {}, locale_data.currencies);
  628. generate_enum("Key"sv, {}, locale_data.keywords);
  629. generate_enum("NumericSymbol"sv, {}, locale_data.numeric_symbols);
  630. generate_enum("Variant"sv, {}, locale_data.variants);
  631. generate_enum("ListPatternType"sv, {}, locale_data.list_pattern_types);
  632. generate_enum("ListPatternStyle"sv, {}, locale_data.list_pattern_styles);
  633. generator.append(R"~~~(
  634. namespace Detail {
  635. Optional<Locale> locale_from_string(StringView locale);
  636. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  637. Optional<Language> language_from_string(StringView language);
  638. Optional<StringView> resolve_language_alias(StringView language);
  639. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  640. Optional<Territory> territory_from_string(StringView territory);
  641. Optional<StringView> resolve_territory_alias(StringView territory);
  642. Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
  643. Optional<ScriptTag> script_tag_from_string(StringView script_tag);
  644. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  645. Optional<StringView> get_locale_currency_mapping(StringView locale, StringView currency);
  646. Optional<Currency> currency_from_string(StringView currency);
  647. Optional<StringView> get_locale_key_mapping(StringView locale, StringView key);
  648. Optional<Key> key_from_string(StringView key);
  649. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView numeric_symbol);
  650. Vector<NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type);
  651. Optional<NumericSymbol> numeric_symbol_from_string(StringView numeric_symbol);
  652. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style);
  653. Optional<ListPatternType> list_pattern_type_from_string(StringView list_pattern_type);
  654. Optional<ListPatternStyle> list_pattern_style_from_string(StringView list_pattern_style);
  655. Optional<StringView> resolve_variant_alias(StringView variant);
  656. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  657. void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
  658. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id);
  659. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
  660. }
  661. }
  662. )~~~");
  663. file.write(generator.as_string_view());
  664. }
  665. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  666. {
  667. StringBuilder builder;
  668. SourceGenerator generator { builder };
  669. generator.set("string_index_type"sv, s_string_index_type);
  670. generator.set("strings_size"sv, String::number(locale_data.unique_strings.size()));
  671. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  672. generator.set("territories_size", String::number(locale_data.territories.size()));
  673. generator.set("variants_size", String::number(locale_data.max_variant_size));
  674. generator.set("numeric_symbols_size", String::number(locale_data.numeric_symbols.size()));
  675. generator.append(R"~~~(
  676. #include <AK/Array.h>
  677. #include <AK/BinarySearch.h>
  678. #include <AK/Span.h>
  679. #include <LibUnicode/Locale.h>
  680. #include <LibUnicode/UnicodeLocale.h>
  681. namespace Unicode::Detail {
  682. struct Patterns {
  683. ListPatternType type;
  684. ListPatternStyle style;
  685. @string_index_type@ start { 0 };
  686. @string_index_type@ middle { 0 };
  687. @string_index_type@ end { 0 };
  688. @string_index_type@ pair { 0 };
  689. };
  690. )~~~");
  691. generator.append(R"~~~(
  692. static constexpr Array<StringView, @strings_size@ + 1> s_string_list { {
  693. {})~~~");
  694. constexpr size_t max_strings_per_row = 30;
  695. size_t strings_in_current_row = 1;
  696. for (auto const& string : locale_data.unique_strings) {
  697. if (strings_in_current_row++ > 0)
  698. generator.append(", ");
  699. generator.append(String::formatted("\"{}\"sv", string));
  700. if (strings_in_current_row == max_strings_per_row) {
  701. strings_in_current_row = 0;
  702. generator.append(",\n ");
  703. }
  704. }
  705. generator.append(R"~~~(
  706. } };
  707. struct NumberFormat {
  708. Unicode::NumberFormat to_unicode_number_format() const {
  709. Unicode::NumberFormat number_format {};
  710. number_format.magnitude = magnitude;
  711. number_format.plurality = static_cast<Unicode::NumberFormat::Plurality>(plurality);
  712. number_format.format = s_string_list[format];
  713. return number_format;
  714. }
  715. u8 magnitude { 0 };
  716. u8 plurality { 0 };
  717. @string_index_type@ format { 0 };
  718. };
  719. struct NumberSystem {
  720. @string_index_type@ system { 0 };
  721. Array<@string_index_type@, @numeric_symbols_size@> symbols {};
  722. Span<NumberFormat const> decimal_long_formats {};
  723. Span<NumberFormat const> decimal_short_formats {};
  724. Span<NumberFormat const> currency_short_formats {};
  725. };
  726. )~~~");
  727. auto format_mapping_name = [](StringView format, StringView name) {
  728. auto mapping_name = name.to_lowercase_string().replace("-"sv, "_"sv, true);
  729. return String::formatted(format, mapping_name);
  730. };
  731. auto append_index = [&](auto index) {
  732. generator.append(String::formatted(", {}", index));
  733. };
  734. auto append_list_and_size = [&](auto const& list) {
  735. if (list.is_empty()) {
  736. generator.append(", {}, 0");
  737. return;
  738. }
  739. bool first = true;
  740. generator.append(", {");
  741. for (auto const& item : list) {
  742. generator.append(first ? " " : ", ");
  743. generator.append(String::number(item));
  744. first = false;
  745. }
  746. generator.append(String::formatted(" }}, {}", list.size()));
  747. };
  748. auto append_string_index_list = [&](String name, auto const& keys, auto const& mappings) {
  749. generator.set("name", name);
  750. generator.set("size", String::number(keys.size()));
  751. generator.append(R"~~~(
  752. static constexpr Array<@string_index_type@, @size@> @name@ { {
  753. )~~~");
  754. constexpr size_t max_values_per_row = 30;
  755. size_t values_in_current_row = 0;
  756. for (auto const& key : keys) {
  757. if (values_in_current_row++ > 0)
  758. generator.append(" ");
  759. if (auto it = mappings.find(key); it != mappings.end())
  760. generator.set("mapping"sv, String::number(it->value));
  761. else
  762. generator.set("mapping"sv, "0"sv);
  763. generator.append("@mapping@,");
  764. if (values_in_current_row == max_values_per_row) {
  765. values_in_current_row = 0;
  766. generator.append("\n ");
  767. }
  768. }
  769. generator.append(R"~~~(
  770. } };
  771. )~~~");
  772. };
  773. auto append_number_formats = [&](String name, auto const& number_formats) {
  774. generator.set("name"sv, move(name));
  775. generator.set("size"sv, String::number(number_formats.size()));
  776. generator.append(R"~~~(
  777. static constexpr Array<NumberFormat, @size@> @name@ { {
  778. )~~~");
  779. constexpr size_t max_values_per_row = 10;
  780. size_t values_in_current_row = 0;
  781. for (auto const& number_format : number_formats) {
  782. if (values_in_current_row++ > 0)
  783. generator.append(" ");
  784. generator.set("magnitude"sv, String::number(number_format.magnitude));
  785. generator.set("plurality"sv, String::number(static_cast<u8>(number_format.plurality)));
  786. generator.set("format"sv, String::number(number_format.format_index));
  787. generator.append("{ @magnitude@, @plurality@, @format@ },");
  788. if (values_in_current_row == max_values_per_row) {
  789. values_in_current_row = 0;
  790. generator.append("\n ");
  791. }
  792. }
  793. generator.append(R"~~~(
  794. } };
  795. )~~~");
  796. };
  797. auto append_number_systems = [&](String name, auto const& number_systems) {
  798. auto format_name = [&](StringView system, StringView format) {
  799. return String::formatted("{}_{}_{}", name, system, format);
  800. };
  801. for (auto const& number_system : number_systems) {
  802. append_number_formats(format_name(number_system.key, "dl"sv), number_system.value.decimal_long_formats);
  803. append_number_formats(format_name(number_system.key, "ds"sv), number_system.value.decimal_short_formats);
  804. append_number_formats(format_name(number_system.key, "cs"sv), number_system.value.currency_short_formats);
  805. }
  806. generator.set("name", name);
  807. generator.set("size", String::number(number_systems.size()));
  808. generator.append(R"~~~(
  809. static constexpr Array<NumberSystem, @size@> @name@ { {)~~~");
  810. for (auto const& number_system : number_systems) {
  811. generator.set("system"sv, String::number(number_system.value.system));
  812. generator.set("decimal_long_formats"sv, format_name(number_system.key, "dl"sv));
  813. generator.set("decimal_short_formats"sv, format_name(number_system.key, "ds"sv));
  814. generator.set("currency_short_formats"sv, format_name(number_system.key, "cs"sv));
  815. generator.append(R"~~~(
  816. { @system@, {)~~~");
  817. for (auto const& symbol : locale_data.numeric_symbols) {
  818. auto index = number_system.value.symbols.get(symbol).value_or(0);
  819. generator.set("index", String::number(index));
  820. generator.append(" @index@,");
  821. }
  822. generator.append(" }, @decimal_long_formats@.span(), @decimal_short_formats@.span(), @currency_short_formats@.span() },");
  823. }
  824. generator.append(R"~~~(
  825. } };
  826. )~~~");
  827. };
  828. auto append_list_patterns = [&](StringView name, Vector<ListPatterns> const& list_patterns) {
  829. generator.set("name", name);
  830. generator.set("size", String::number(list_patterns.size()));
  831. generator.append(R"~~~(
  832. static constexpr Array<Patterns, @size@> @name@ { {)~~~");
  833. for (auto const& list_pattern : list_patterns) {
  834. generator.set("type"sv, String::formatted("ListPatternType::{}", format_identifier({}, list_pattern.type)));
  835. generator.set("style"sv, String::formatted("ListPatternStyle::{}", format_identifier({}, list_pattern.style)));
  836. generator.set("start"sv, String::number(list_pattern.start));
  837. generator.set("middle"sv, String::number(list_pattern.middle));
  838. generator.set("end"sv, String::number(list_pattern.end));
  839. generator.set("pair"sv, String::number(list_pattern.pair));
  840. generator.append(R"~~~(
  841. { @type@, @style@, @start@, @middle@, @end@, @pair@ },)~~~");
  842. }
  843. generator.append(R"~~~(
  844. } };
  845. )~~~");
  846. };
  847. auto append_mapping = [&](StringView type, StringView name, StringView format, auto format_list_callback) {
  848. Vector<String> mapping_names;
  849. for (auto const& locale : locale_data.locales) {
  850. auto mapping_name = format_mapping_name(format, locale.key);
  851. format_list_callback(mapping_name, locale.value);
  852. mapping_names.append(move(mapping_name));
  853. }
  854. quick_sort(mapping_names);
  855. generator.set("type", type);
  856. generator.set("name", name);
  857. generator.set("size", String::number(locale_data.locales.size()));
  858. generator.append(R"~~~(
  859. static constexpr Array<Span<@type@ const>, @size@> @name@ { {
  860. )~~~");
  861. constexpr size_t max_values_per_row = 10;
  862. size_t values_in_current_row = 0;
  863. for (auto& mapping_name : mapping_names) {
  864. if (values_in_current_row++ > 0)
  865. generator.append(" ");
  866. generator.set("name", move(mapping_name));
  867. generator.append("@name@.span(),");
  868. if (values_in_current_row == max_values_per_row) {
  869. values_in_current_row = 0;
  870. generator.append("\n ");
  871. }
  872. }
  873. generator.append(R"~~~(
  874. } };
  875. )~~~");
  876. };
  877. append_mapping(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); });
  878. append_mapping(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); });
  879. append_mapping(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); });
  880. append_mapping(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); });
  881. append_mapping(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); });
  882. append_mapping("NumberSystem"sv, "s_number_systems"sv, "s_number_systems_{}", [&](auto const& name, auto const& value) { append_number_systems(name, value.number_systems); });
  883. append_mapping("Patterns"sv, "s_list_patterns"sv, "s_list_patterns_{}", [&](auto const& name, auto const& value) { append_list_patterns(name, value.list_patterns); });
  884. generator.append(R"~~~(
  885. struct CanonicalLanguageID {
  886. Unicode::LanguageID to_unicode_language_id() const
  887. {
  888. Unicode::LanguageID language_id {};
  889. language_id.variants.ensure_capacity(variants_size);
  890. language_id.language = s_string_list[language];
  891. if (script != 0)
  892. language_id.script = s_string_list[script];
  893. if (region != 0)
  894. language_id.region = s_string_list[region];
  895. for (size_t i = 0; i < variants_size; ++i)
  896. language_id.variants.append(s_string_list[variants[i]]);
  897. return language_id;
  898. }
  899. bool matches_variants(Vector<String> const& other_variants) const {
  900. if (variants_size == 0)
  901. return true;
  902. if (other_variants.size() != variants_size)
  903. return false;
  904. for (size_t i = 0; i < variants_size; ++i) {
  905. if (s_string_list[variants[i]] != other_variants[i])
  906. return false;
  907. }
  908. return true;
  909. };
  910. @string_index_type@ language { 0 };
  911. @string_index_type@ script { 0 };
  912. @string_index_type@ region { 0 };
  913. Array<@string_index_type@, @variants_size@> variants {};
  914. size_t variants_size { 0 };
  915. };
  916. struct LanguageMapping {
  917. CanonicalLanguageID key;
  918. CanonicalLanguageID alias;
  919. };
  920. )~~~");
  921. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  922. generator.set("size", String::number(mappings.size()));
  923. generator.set("name"sv, name);
  924. generator.append(R"~~~(
  925. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  926. )~~~");
  927. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  928. auto const& lhs_language = get_unique_string(locale_data, lhs.key.language);
  929. auto const& rhs_language = get_unique_string(locale_data, rhs.key.language);
  930. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  931. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  932. return false;
  933. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  934. return true;
  935. return lhs_language < rhs_language;
  936. });
  937. for (auto const& mapping : mappings) {
  938. generator.set("language"sv, String::number(mapping.key.language));
  939. generator.append(" { { @language@");
  940. append_index(mapping.key.script);
  941. append_index(mapping.key.region);
  942. append_list_and_size(mapping.key.variants);
  943. generator.set("language"sv, String::number(mapping.alias.language));
  944. generator.append(" }, { @language@");
  945. append_index(mapping.alias.script);
  946. append_index(mapping.alias.region);
  947. append_list_and_size(mapping.alias.variants);
  948. generator.append(" } },\n");
  949. }
  950. generator.append("} };\n");
  951. };
  952. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  953. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  954. generator.append(R"~~~(
  955. static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& language_id)
  956. {
  957. // https://unicode.org/reports/tr35/#Likely_Subtags
  958. enum class State {
  959. LanguageScriptRegion,
  960. LanguageRegion,
  961. LanguageScript,
  962. Language,
  963. UndScript,
  964. Done,
  965. };
  966. auto state = State::LanguageScriptRegion;
  967. while (state != State::Done) {
  968. Unicode::LanguageID search_key;
  969. switch (state) {
  970. case State::LanguageScriptRegion:
  971. state = State::LanguageRegion;
  972. if (!language_id.script.has_value() || !language_id.region.has_value())
  973. continue;
  974. search_key.language = *language_id.language;
  975. search_key.script = *language_id.script;
  976. search_key.region = *language_id.region;
  977. break;
  978. case State::LanguageRegion:
  979. state = State::LanguageScript;
  980. if (!language_id.region.has_value())
  981. continue;
  982. search_key.language = *language_id.language;
  983. search_key.region = *language_id.region;
  984. break;
  985. case State::LanguageScript:
  986. state = State::Language;
  987. if (!language_id.script.has_value())
  988. continue;
  989. search_key.language = *language_id.language;
  990. search_key.script = *language_id.script;
  991. break;
  992. case State::Language:
  993. state = State::UndScript;
  994. search_key.language = *language_id.language;
  995. break;
  996. case State::UndScript:
  997. state = State::Done;
  998. if (!language_id.script.has_value())
  999. continue;
  1000. search_key.language = "und"sv;
  1001. search_key.script = *language_id.script;
  1002. break;
  1003. default:
  1004. VERIFY_NOT_REACHED();
  1005. }
  1006. for (auto const& map : s_likely_subtags) {
  1007. auto const& key_language = s_string_list[map.key.language];
  1008. auto const& key_script = s_string_list[map.key.script];
  1009. auto const& key_region = s_string_list[map.key.region];
  1010. if (key_language != search_key.language)
  1011. continue;
  1012. if (!key_script.is_empty() || search_key.script.has_value()) {
  1013. if (key_script != search_key.script)
  1014. continue;
  1015. }
  1016. if (!key_region.is_empty() || search_key.region.has_value()) {
  1017. if (key_region != search_key.region)
  1018. continue;
  1019. }
  1020. return &map;
  1021. }
  1022. }
  1023. return nullptr;
  1024. }
  1025. )~~~");
  1026. auto append_mapping_search = [&](StringView enum_title, StringView enum_snake, StringView collection_name) {
  1027. generator.set("enum_title", enum_title);
  1028. generator.set("enum_snake", enum_snake);
  1029. generator.set("collection_name", collection_name);
  1030. generator.append(R"~~~(
  1031. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  1032. {
  1033. auto locale_value = locale_from_string(locale);
  1034. if (!locale_value.has_value())
  1035. return {};
  1036. auto @enum_snake@_value = @enum_snake@_from_string(@enum_snake@);
  1037. if (!@enum_snake@_value.has_value())
  1038. return {};
  1039. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1040. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  1041. auto const& mappings = @collection_name@.at(locale_index);
  1042. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  1043. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  1044. if (@enum_snake@_mapping.is_empty())
  1045. return {};
  1046. return @enum_snake@_mapping;
  1047. }
  1048. )~~~");
  1049. };
  1050. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values) {
  1051. HashValueMap<String> hashes;
  1052. hashes.ensure_capacity(values.size());
  1053. for (auto const& value : values)
  1054. hashes.set(value.hash(), format_identifier(enum_title, value));
  1055. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  1056. };
  1057. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  1058. HashValueMap<StringIndexType> hashes;
  1059. hashes.ensure_capacity(aliases.size());
  1060. for (auto const& alias : aliases)
  1061. hashes.set(alias.key.hash(), alias.value);
  1062. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), "StringView"sv, "s_string_list[{}]"sv);
  1063. };
  1064. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys());
  1065. append_mapping_search("Language"sv, "language"sv, "s_languages"sv);
  1066. append_from_string("Language"sv, "language"sv, locale_data.languages);
  1067. append_alias_search("language"sv, locale_data.language_aliases);
  1068. append_mapping_search("Territory"sv, "territory"sv, "s_territories"sv);
  1069. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  1070. append_alias_search("territory"sv, locale_data.territory_aliases);
  1071. append_mapping_search("ScriptTag"sv, "script_tag"sv, "s_scripts"sv);
  1072. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  1073. append_alias_search("script_tag"sv, locale_data.script_aliases);
  1074. append_mapping_search("Currency"sv, "currency"sv, "s_currencies"sv);
  1075. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  1076. append_mapping_search("Key"sv, "key"sv, "s_keywords"sv);
  1077. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  1078. append_alias_search("variant"sv, locale_data.variant_aliases);
  1079. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  1080. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  1081. append_from_string("ListPatternStyle"sv, "list_pattern_style"sv, locale_data.list_pattern_styles);
  1082. append_from_string("NumericSymbol"sv, "numeric_symbol"sv, locale_data.numeric_symbols);
  1083. generator.append(R"~~~(
  1084. static NumberSystem const* find_number_system(StringView locale, StringView system)
  1085. {
  1086. auto locale_value = locale_from_string(locale);
  1087. if (!locale_value.has_value())
  1088. return nullptr;
  1089. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1090. auto const& number_systems = s_number_systems.at(locale_index);
  1091. for (auto const& number_system : number_systems) {
  1092. if (system == s_string_list[number_system.system])
  1093. return &number_system;
  1094. };
  1095. return nullptr;
  1096. }
  1097. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView symbol)
  1098. {
  1099. auto symbol_value = numeric_symbol_from_string(symbol);
  1100. if (!symbol_value.has_value())
  1101. return {};
  1102. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  1103. auto symbol_index = to_underlying(*symbol_value);
  1104. return s_string_list[number_system->symbols[symbol_index]];
  1105. }
  1106. return {};
  1107. }
  1108. Vector<Unicode::NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type)
  1109. {
  1110. Vector<Unicode::NumberFormat> formats;
  1111. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  1112. Span<NumberFormat const> number_formats;
  1113. switch (type) {
  1114. case CompactNumberFormatType::DecimalLong:
  1115. number_formats = number_system->decimal_long_formats;
  1116. break;
  1117. case CompactNumberFormatType::DecimalShort:
  1118. number_formats = number_system->decimal_short_formats;
  1119. break;
  1120. case CompactNumberFormatType::CurrencyShort:
  1121. number_formats = number_system->currency_short_formats;
  1122. break;
  1123. }
  1124. formats.ensure_capacity(number_formats.size());
  1125. for (auto const& number_format : number_formats)
  1126. formats.append(number_format.to_unicode_number_format());
  1127. }
  1128. return formats;
  1129. }
  1130. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style)
  1131. {
  1132. auto locale_value = locale_from_string(locale);
  1133. if (!locale_value.has_value())
  1134. return {};
  1135. auto type_value = list_pattern_type_from_string(list_pattern_type);
  1136. if (!type_value.has_value())
  1137. return {};
  1138. auto style_value = list_pattern_style_from_string(list_pattern_style);
  1139. if (!style_value.has_value())
  1140. return {};
  1141. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  1142. auto const& locale_list_patterns = s_list_patterns.at(locale_index);
  1143. for (auto const& list_patterns : locale_list_patterns) {
  1144. if ((list_patterns.type == type_value) && (list_patterns.style == style_value)) {
  1145. auto const& start = s_string_list[list_patterns.start];
  1146. auto const& middle = s_string_list[list_patterns.middle];
  1147. auto const& end = s_string_list[list_patterns.end];
  1148. auto const& pair = s_string_list[list_patterns.pair];
  1149. return ListPatterns { start, middle, end, pair };
  1150. }
  1151. }
  1152. return {};
  1153. }
  1154. void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
  1155. {
  1156. for (auto const& map : s_complex_alias) {
  1157. auto const& key_language = s_string_list[map.key.language];
  1158. auto const& key_script = s_string_list[map.key.script];
  1159. auto const& key_region = s_string_list[map.key.region];
  1160. if ((key_language != language_id.language) && (key_language != "und"sv))
  1161. continue;
  1162. if (!key_script.is_empty() && (key_script != language_id.script))
  1163. continue;
  1164. if (!key_region.is_empty() && (key_region != language_id.region))
  1165. continue;
  1166. if (!map.key.matches_variants(language_id.variants))
  1167. continue;
  1168. auto alias = map.alias.to_unicode_language_id();
  1169. if (alias.language == "und"sv)
  1170. alias.language = move(language_id.language);
  1171. if (key_script.is_empty() && !alias.script.has_value())
  1172. alias.script = move(language_id.script);
  1173. if (key_region.is_empty() && !alias.region.has_value())
  1174. alias.region = move(language_id.region);
  1175. if (map.key.variants_size == 0 && alias.variants.is_empty())
  1176. alias.variants = move(language_id.variants);
  1177. language_id = move(alias);
  1178. break;
  1179. }
  1180. }
  1181. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id)
  1182. {
  1183. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  1184. auto const* likely_subtag = resolve_likely_subtag(language_id);
  1185. if (likely_subtag == nullptr)
  1186. return {};
  1187. auto maximized = language_id;
  1188. auto const& key_script = s_string_list[likely_subtag->key.script];
  1189. auto const& key_region = s_string_list[likely_subtag->key.region];
  1190. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  1191. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  1192. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  1193. if (maximized.language == "und"sv)
  1194. maximized.language = alias_language;
  1195. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  1196. maximized.script = alias_script;
  1197. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  1198. maximized.region = alias_region;
  1199. return maximized;
  1200. }
  1201. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
  1202. {
  1203. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  1204. return s_string_list[likely_subtag->alias.region];
  1205. return {};
  1206. }
  1207. }
  1208. )~~~");
  1209. file.write(generator.as_string_view());
  1210. }
  1211. int main(int argc, char** argv)
  1212. {
  1213. char const* generated_header_path = nullptr;
  1214. char const* generated_implementation_path = nullptr;
  1215. char const* core_path = nullptr;
  1216. char const* locale_names_path = nullptr;
  1217. char const* misc_path = nullptr;
  1218. char const* numbers_path = nullptr;
  1219. Core::ArgsParser args_parser;
  1220. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  1221. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  1222. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  1223. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  1224. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  1225. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  1226. args_parser.parse(argc, argv);
  1227. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  1228. if (path.is_empty()) {
  1229. warnln("{} is required", flags);
  1230. args_parser.print_usage(stderr, argv[0]);
  1231. exit(1);
  1232. }
  1233. auto file_or_error = Core::File::open(path, mode);
  1234. if (file_or_error.is_error()) {
  1235. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  1236. exit(1);
  1237. }
  1238. return file_or_error.release_value();
  1239. };
  1240. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  1241. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  1242. UnicodeLocaleData locale_data;
  1243. parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, locale_data);
  1244. generate_unicode_locale_header(generated_header_file, locale_data);
  1245. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  1246. return 0;
  1247. }