GenerateUnicodeLocale.cpp 27 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702
  1. /*
  2. * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <AK/AllOf.h>
  7. #include <AK/CharacterTypes.h>
  8. #include <AK/Format.h>
  9. #include <AK/HashMap.h>
  10. #include <AK/JsonObject.h>
  11. #include <AK/JsonParser.h>
  12. #include <AK/JsonValue.h>
  13. #include <AK/LexicalPath.h>
  14. #include <AK/QuickSort.h>
  15. #include <AK/SourceGenerator.h>
  16. #include <AK/String.h>
  17. #include <AK/StringBuilder.h>
  18. #include <LibCore/ArgsParser.h>
  19. #include <LibCore/DirIterator.h>
  20. #include <LibCore/File.h>
  21. struct Locale {
  22. String language;
  23. Optional<String> territory;
  24. Optional<String> variant;
  25. HashMap<String, String> languages;
  26. HashMap<String, String> territories;
  27. HashMap<String, String> scripts;
  28. HashMap<String, String> currencies;
  29. };
  30. struct UnicodeLocaleData {
  31. HashMap<String, Locale> locales;
  32. Vector<String> languages;
  33. Vector<String> territories;
  34. Vector<String> scripts;
  35. Vector<String> variants;
  36. Vector<String> currencies;
  37. HashMap<String, String> language_aliases;
  38. HashMap<String, String> territory_aliases;
  39. HashMap<String, String> script_aliases;
  40. HashMap<String, String> variant_aliases;
  41. HashMap<String, String> subdivision_aliases;
  42. HashMap<String, String> complex_mappings;
  43. };
  44. static void write_to_file_if_different(Core::File& file, StringView contents)
  45. {
  46. auto const current_contents = file.read_all();
  47. if (StringView { current_contents.bytes() } == contents)
  48. return;
  49. VERIFY(file.seek(0));
  50. VERIFY(file.truncate(0));
  51. VERIFY(file.write(contents));
  52. }
  53. static void parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
  54. {
  55. LexicalPath core_aliases_path(move(core_supplemental_path));
  56. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  57. VERIFY(Core::File::exists(core_aliases_path.string()));
  58. auto core_aliases_file_or_error = Core::File::open(core_aliases_path.string(), Core::OpenMode::ReadOnly);
  59. VERIFY(!core_aliases_file_or_error.is_error());
  60. auto core_aliases = JsonParser(core_aliases_file_or_error.value()->read_all()).parse();
  61. VERIFY(core_aliases.has_value());
  62. auto const& supplemental_object = core_aliases->as_object().get("supplemental"sv);
  63. auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
  64. auto const& alias_object = metadata_object.as_object().get("alias"sv);
  65. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  66. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  67. auto alias = value.as_object().get("_replacement"sv).as_string();
  68. if (key.contains('-'))
  69. locale_data.complex_mappings.set(key, move(alias));
  70. else
  71. alias_map.set(key, move(alias));
  72. });
  73. };
  74. append_aliases(alias_object.as_object().get("languageAlias"sv), locale_data.language_aliases);
  75. append_aliases(alias_object.as_object().get("territoryAlias"sv), locale_data.territory_aliases);
  76. append_aliases(alias_object.as_object().get("scriptAlias"sv), locale_data.script_aliases);
  77. append_aliases(alias_object.as_object().get("variantAlias"sv), locale_data.variant_aliases);
  78. append_aliases(alias_object.as_object().get("subdivisionAlias"sv), locale_data.subdivision_aliases);
  79. }
  80. static void parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  81. {
  82. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  83. languages_path = languages_path.append("languages.json"sv);
  84. VERIFY(Core::File::exists(languages_path.string()));
  85. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  86. VERIFY(!languages_file_or_error.is_error());
  87. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  88. VERIFY(languages.has_value());
  89. auto const& main_object = languages->as_object().get("main"sv);
  90. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  91. auto const& identity_object = locale_object.as_object().get("identity"sv);
  92. auto const& language_string = identity_object.as_object().get("language"sv);
  93. auto const& territory_string = identity_object.as_object().get("territory"sv);
  94. auto const& variant_string = identity_object.as_object().get("variant"sv);
  95. locale.language = language_string.as_string();
  96. if (!locale_data.languages.contains_slow(locale.language))
  97. locale_data.languages.append(locale.language);
  98. if (territory_string.is_string()) {
  99. locale.territory = territory_string.as_string();
  100. if (!locale_data.territories.contains_slow(*locale.territory))
  101. locale_data.territories.append(*locale.territory);
  102. }
  103. if (variant_string.is_string()) {
  104. locale.variant = variant_string.as_string();
  105. if (!locale_data.variants.contains_slow(*locale.variant))
  106. locale_data.variants.append(*locale.variant);
  107. }
  108. }
  109. static void parse_locale_languages(String locale_path, Locale& locale)
  110. {
  111. LexicalPath languages_path(move(locale_path));
  112. languages_path = languages_path.append("languages.json"sv);
  113. VERIFY(Core::File::exists(languages_path.string()));
  114. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  115. VERIFY(!languages_file_or_error.is_error());
  116. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  117. VERIFY(languages.has_value());
  118. auto const& main_object = languages->as_object().get("main"sv);
  119. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  120. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  121. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  122. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  123. locale.languages.set(key, value.as_string());
  124. });
  125. }
  126. static void parse_locale_territories(String locale_path, Locale& locale)
  127. {
  128. LexicalPath territories_path(move(locale_path));
  129. territories_path = territories_path.append("territories.json"sv);
  130. VERIFY(Core::File::exists(territories_path.string()));
  131. auto territories_file_or_error = Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly);
  132. VERIFY(!territories_file_or_error.is_error());
  133. auto territories = JsonParser(territories_file_or_error.value()->read_all()).parse();
  134. VERIFY(territories.has_value());
  135. auto const& main_object = territories->as_object().get("main"sv);
  136. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  137. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  138. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  139. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  140. locale.territories.set(key, value.as_string());
  141. });
  142. }
  143. static void parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  144. {
  145. LexicalPath scripts_path(move(locale_path));
  146. scripts_path = scripts_path.append("scripts.json"sv);
  147. VERIFY(Core::File::exists(scripts_path.string()));
  148. auto scripts_file_or_error = Core::File::open(scripts_path.string(), Core::OpenMode::ReadOnly);
  149. VERIFY(!scripts_file_or_error.is_error());
  150. auto scripts = JsonParser(scripts_file_or_error.value()->read_all()).parse();
  151. VERIFY(scripts.has_value());
  152. auto const& main_object = scripts->as_object().get("main"sv);
  153. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  154. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  155. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  156. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  157. locale.scripts.set(key, value.as_string());
  158. if (!locale_data.scripts.contains_slow(key))
  159. locale_data.scripts.append(key);
  160. });
  161. }
  162. static void parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  163. {
  164. LexicalPath currencies_path(move(numbers_path));
  165. currencies_path = currencies_path.append("currencies.json"sv);
  166. VERIFY(Core::File::exists(currencies_path.string()));
  167. auto currencies_file_or_error = Core::File::open(currencies_path.string(), Core::OpenMode::ReadOnly);
  168. VERIFY(!currencies_file_or_error.is_error());
  169. auto currencies = JsonParser(currencies_file_or_error.value()->read_all()).parse();
  170. VERIFY(currencies.has_value());
  171. auto const& main_object = currencies->as_object().get("main"sv);
  172. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  173. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  174. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  175. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  176. auto const& display_name = value.as_object().get("displayName"sv);
  177. locale.currencies.set(key, display_name.as_string());
  178. if (!locale_data.currencies.contains_slow(key))
  179. locale_data.currencies.append(key);
  180. });
  181. }
  182. static Core::DirIterator path_to_dir_iterator(String path)
  183. {
  184. LexicalPath lexical_path(move(path));
  185. lexical_path = lexical_path.append("main"sv);
  186. VERIFY(Core::File::is_directory(lexical_path.string()));
  187. Core::DirIterator iterator(lexical_path.string(), Core::DirIterator::SkipParentAndBaseDir);
  188. if (iterator.has_error()) {
  189. warnln("{}: {}", lexical_path.string(), iterator.error_string());
  190. VERIFY_NOT_REACHED();
  191. }
  192. return iterator;
  193. }
  194. static void parse_all_locales(String core_path, String locale_names_path, String numbers_path, UnicodeLocaleData& locale_data)
  195. {
  196. auto locale_names_iterator = path_to_dir_iterator(move(locale_names_path));
  197. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  198. LexicalPath core_supplemental_path(move(core_path));
  199. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  200. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  201. parse_core_aliases(core_supplemental_path.string(), locale_data);
  202. while (locale_names_iterator.has_next()) {
  203. auto locale_path = locale_names_iterator.next_full_path();
  204. VERIFY(Core::File::is_directory(locale_path));
  205. auto& locale = locale_data.locales.ensure(LexicalPath::basename(locale_path));
  206. parse_identity(locale_path, locale_data, locale);
  207. parse_locale_languages(locale_path, locale);
  208. parse_locale_territories(locale_path, locale);
  209. parse_locale_scripts(locale_path, locale_data, locale);
  210. }
  211. while (numbers_iterator.has_next()) {
  212. auto numbers_path = numbers_iterator.next_full_path();
  213. VERIFY(Core::File::is_directory(numbers_path));
  214. auto& locale = locale_data.locales.ensure(LexicalPath::basename(numbers_path));
  215. parse_locale_currencies(numbers_path, locale_data, locale);
  216. }
  217. }
  218. static String format_identifier(StringView owner, String identifier)
  219. {
  220. identifier.replace("-"sv, "_"sv, true);
  221. if (all_of(identifier, is_ascii_digit))
  222. return String::formatted("{}_{}", owner[0], identifier);
  223. return identifier.to_titlecase();
  224. }
  225. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  226. {
  227. StringBuilder builder;
  228. SourceGenerator generator { builder };
  229. auto generate_enum = [&](StringView name, StringView default_, Vector<String>& values) {
  230. quick_sort(values);
  231. generator.set("name", name);
  232. generator.set("underlying", ((values.size() + !default_.is_empty()) < 256) ? "u8"sv : "u16"sv);
  233. generator.append(R"~~~(
  234. enum class @name@ : @underlying@ {)~~~");
  235. if (!default_.is_empty()) {
  236. generator.set("default", default_);
  237. generator.append(R"~~~(
  238. @default@,)~~~");
  239. }
  240. for (auto const& value : values) {
  241. generator.set("value", format_identifier(name, value));
  242. generator.append(R"~~~(
  243. @value@,)~~~");
  244. }
  245. generator.append(R"~~~(
  246. };
  247. )~~~");
  248. };
  249. generator.append(R"~~~(
  250. #pragma once
  251. #include <AK/Optional.h>
  252. #include <AK/StringView.h>
  253. #include <AK/Types.h>
  254. #include <LibUnicode/Forward.h>
  255. namespace Unicode {
  256. )~~~");
  257. auto locales = locale_data.locales.keys();
  258. generate_enum("Locale"sv, "None"sv, locales);
  259. generate_enum("Language"sv, {}, locale_data.languages);
  260. generate_enum("Territory"sv, {}, locale_data.territories);
  261. generate_enum("ScriptTag"sv, {}, locale_data.scripts);
  262. generate_enum("Currency"sv, {}, locale_data.currencies);
  263. generate_enum("Variant"sv, {}, locale_data.variants);
  264. generator.append(R"~~~(
  265. namespace Detail {
  266. Optional<Locale> locale_from_string(StringView const& locale);
  267. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  268. Optional<Language> language_from_string(StringView const& language);
  269. Optional<StringView> resolve_language_alias(StringView const& language);
  270. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  271. Optional<Territory> territory_from_string(StringView const& territory);
  272. Optional<StringView> resolve_territory_alias(StringView const& territory);
  273. Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
  274. Optional<ScriptTag> script_tag_from_string(StringView const& script_tag);
  275. Optional<StringView> resolve_script_tag_alias(StringView const& script_tag);
  276. Optional<StringView> get_locale_currency_mapping(StringView locale, StringView currency);
  277. Optional<Currency> currency_from_string(StringView const& currency);
  278. Optional<StringView> resolve_variant_alias(StringView const& variant);
  279. Optional<StringView> resolve_subdivision_alias(StringView const& subdivision);
  280. void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
  281. }
  282. }
  283. )~~~");
  284. write_to_file_if_different(file, generator.as_string_view());
  285. }
  286. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  287. {
  288. StringBuilder builder;
  289. SourceGenerator generator { builder };
  290. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  291. generator.set("territories_size", String::number(locale_data.territories.size()));
  292. generator.append(R"~~~(
  293. #include <AK/Array.h>
  294. #include <AK/HashMap.h>
  295. #include <AK/Span.h>
  296. #include <LibUnicode/Locale.h>
  297. #include <LibUnicode/UnicodeLocale.h>
  298. namespace Unicode {
  299. )~~~");
  300. auto format_mapping_name = [](StringView format, StringView name) {
  301. auto mapping_name = name.to_lowercase_string();
  302. mapping_name.replace("-"sv, "_"sv, true);
  303. return String::formatted(format, mapping_name);
  304. };
  305. auto append_mapping_list = [&](String name, auto const& keys, auto const& mappings) {
  306. generator.set("name", name);
  307. generator.set("size", String::number(keys.size()));
  308. generator.append(R"~~~(
  309. static constexpr Array<StringView, @size@> @name@ { {
  310. )~~~");
  311. constexpr size_t max_values_per_row = 10;
  312. size_t values_in_current_row = 0;
  313. for (auto const& key : keys) {
  314. if (values_in_current_row++ > 0)
  315. generator.append(" ");
  316. if (auto it = mappings.find(key); it != mappings.end())
  317. generator.set("mapping"sv, String::formatted("\"{}\"sv", it->value));
  318. else
  319. generator.set("mapping"sv, "{}"sv);
  320. generator.append("@mapping@,");
  321. if (values_in_current_row == max_values_per_row) {
  322. values_in_current_row = 0;
  323. generator.append("\n ");
  324. }
  325. }
  326. generator.append(R"~~~(
  327. } };
  328. )~~~");
  329. };
  330. auto append_mapping = [&](StringView name, StringView format, auto const& keys, auto get_mapping_callback) {
  331. Vector<String> mapping_names;
  332. for (auto const& locale : locale_data.locales) {
  333. auto mapping_name = format_mapping_name(format, locale.key);
  334. append_mapping_list(mapping_name, keys, get_mapping_callback(locale.value));
  335. mapping_names.append(move(mapping_name));
  336. }
  337. quick_sort(mapping_names);
  338. generator.set("name", name);
  339. generator.set("size", String::number(locale_data.locales.size()));
  340. generator.append(R"~~~(
  341. static constexpr Array<Span<StringView const>, @size@> @name@ { {
  342. )~~~");
  343. constexpr size_t max_values_per_row = 10;
  344. size_t values_in_current_row = 0;
  345. for (auto& mapping_name : mapping_names) {
  346. if (values_in_current_row++ > 0)
  347. generator.append(" ");
  348. generator.set("name", move(mapping_name));
  349. generator.append("@name@.span(),");
  350. if (values_in_current_row == max_values_per_row) {
  351. values_in_current_row = 0;
  352. generator.append("\n ");
  353. }
  354. }
  355. generator.append(R"~~~(
  356. } };
  357. )~~~");
  358. };
  359. append_mapping("s_languages"sv, "s_languages_{}", locale_data.languages, [](auto const& value) { return value.languages; });
  360. append_mapping("s_territories"sv, "s_territories_{}", locale_data.territories, [](auto const& value) { return value.territories; });
  361. append_mapping("s_scripts"sv, "s_scripts_{}", locale_data.scripts, [](auto const& value) { return value.scripts; });
  362. append_mapping("s_currencies"sv, "s_currencies_{}", locale_data.currencies, [](auto const& value) { return value.currencies; });
  363. generator.append(R"~~~(
  364. struct LanguageMapping {
  365. Unicode::LanguageID key;
  366. Unicode::LanguageID alias;
  367. };
  368. )~~~");
  369. auto append_complex_mapping = [&](StringView name, auto const& mappings) {
  370. generator.set("name"sv, name);
  371. generator.append(R"~~~(
  372. static auto const& ensure_@name@_map()
  373. {
  374. static Vector<LanguageMapping> @name@_map;
  375. auto append_mapping = [&](StringView key, StringView alias) {
  376. if (auto key_value = Unicode::parse_unicode_language_id(key); key_value.has_value()) {
  377. if (auto alias_value = Unicode::parse_unicode_language_id(alias); alias_value.has_value())
  378. @name@_map.append({ key_value.release_value(), alias_value.release_value() });
  379. }
  380. };
  381. )~~~");
  382. auto keys = mappings.keys();
  383. quick_sort(keys, [](auto const& lhs, auto const& rhs) {
  384. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  385. if (lhs.starts_with("und"sv) && !rhs.starts_with("und"sv))
  386. return false;
  387. if (!lhs.starts_with("und"sv) && rhs.starts_with("und"sv))
  388. return true;
  389. return lhs < rhs;
  390. });
  391. for (auto const& key : keys) {
  392. generator.set("key"sv, key);
  393. generator.set("alias"sv, mappings.get(key).value());
  394. generator.append(R"~~~(
  395. append_mapping("@key@"sv, "@alias@"sv);)~~~");
  396. }
  397. generator.append(R"~~~(
  398. return @name@_map;
  399. }
  400. )~~~");
  401. };
  402. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  403. generator.append(R"~~~(
  404. namespace Detail {
  405. )~~~");
  406. auto append_mapping_search = [&](StringView enum_title, StringView enum_snake, StringView collection_name) {
  407. generator.set("enum_title", enum_title);
  408. generator.set("enum_snake", enum_snake);
  409. generator.set("collection_name", collection_name);
  410. generator.append(R"~~~(
  411. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  412. {
  413. auto locale_value = locale_from_string(locale);
  414. if (!locale_value.has_value())
  415. return {};
  416. auto @enum_snake@_value = @enum_snake@_from_string(@enum_snake@);
  417. if (!@enum_snake@_value.has_value())
  418. return {};
  419. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  420. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  421. auto const& mappings = @collection_name@.at(locale_index);
  422. auto @enum_snake@_mapping = mappings.at(@enum_snake@_index);
  423. if (@enum_snake@_mapping.is_empty())
  424. return {};
  425. return @enum_snake@_mapping;
  426. }
  427. )~~~");
  428. };
  429. auto append_from_string = [&](StringView enum_title, StringView enum_snake, Vector<String> const& values) {
  430. generator.set("enum_title", enum_title);
  431. generator.set("enum_snake", enum_snake);
  432. generator.append(R"~~~(
  433. Optional<@enum_title@> @enum_snake@_from_string(StringView const& @enum_snake@)
  434. {
  435. static HashMap<StringView, @enum_title@> @enum_snake@_values { {)~~~");
  436. for (auto const& value : values) {
  437. generator.set("key"sv, value);
  438. generator.set("value"sv, format_identifier(enum_title, value));
  439. generator.append(R"~~~(
  440. { "@key@"sv, @enum_title@::@value@ },)~~~");
  441. }
  442. generator.append(R"~~~(
  443. } };
  444. if (auto value = @enum_snake@_values.get(@enum_snake@); value.has_value())
  445. return value.value();
  446. return {};
  447. }
  448. )~~~");
  449. };
  450. auto append_alias_search = [&](StringView enum_snake, HashMap<String, String> const& aliases) {
  451. generator.set("enum_snake", enum_snake);
  452. generator.append(R"~~~(
  453. Optional<StringView> resolve_@enum_snake@_alias(StringView const& @enum_snake@)
  454. {
  455. static HashMap<StringView, StringView> @enum_snake@_aliases { {
  456. )~~~");
  457. constexpr size_t max_values_per_row = 10;
  458. size_t values_in_current_row = 0;
  459. for (auto const& alias : aliases) {
  460. if (values_in_current_row++ > 0)
  461. generator.append(" ");
  462. generator.set("key"sv, alias.key);
  463. generator.set("alias"sv, alias.value);
  464. generator.append("{ \"@key@\"sv, \"@alias@\"sv },");
  465. if (values_in_current_row == max_values_per_row) {
  466. generator.append("\n ");
  467. values_in_current_row = 0;
  468. }
  469. }
  470. generator.append(R"~~~(
  471. } };
  472. if (auto alias = @enum_snake@_aliases.get(@enum_snake@); alias.has_value())
  473. return alias.value();
  474. return {};
  475. }
  476. )~~~");
  477. };
  478. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys());
  479. append_mapping_search("Language"sv, "language"sv, "s_languages"sv);
  480. append_from_string("Language"sv, "language"sv, locale_data.languages);
  481. append_alias_search("language"sv, locale_data.language_aliases);
  482. append_mapping_search("Territory"sv, "territory"sv, "s_territories"sv);
  483. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  484. append_alias_search("territory"sv, locale_data.territory_aliases);
  485. append_mapping_search("ScriptTag"sv, "script_tag"sv, "s_scripts"sv);
  486. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  487. append_alias_search("script_tag"sv, locale_data.script_aliases);
  488. append_mapping_search("Currency"sv, "currency"sv, "s_currencies"sv);
  489. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  490. append_alias_search("variant"sv, locale_data.variant_aliases);
  491. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  492. generator.append(R"~~~(
  493. void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
  494. {
  495. static auto const& complex_alias_map = ensure_complex_alias_map();
  496. for (auto const& map : complex_alias_map) {
  497. if ((map.key.language != language_id.language) && (map.key.language != "und"sv))
  498. continue;
  499. if (map.key.script.has_value() && (map.key.script != language_id.script))
  500. continue;
  501. if (map.key.region.has_value() && (map.key.region != language_id.region))
  502. continue;
  503. if (!map.key.variants.is_empty() && (map.key.variants != language_id.variants))
  504. continue;
  505. auto alias = map.alias;
  506. if (alias.language == "und"sv)
  507. alias.language = move(language_id.language);
  508. if (!map.key.script.has_value() && !alias.script.has_value())
  509. alias.script = move(language_id.script);
  510. if (!map.key.region.has_value() && !alias.region.has_value())
  511. alias.region = move(language_id.region);
  512. if (map.key.variants.is_empty() && alias.variants.is_empty())
  513. alias.variants = move(language_id.variants);
  514. language_id = move(alias);
  515. break;
  516. }
  517. }
  518. }
  519. }
  520. )~~~");
  521. write_to_file_if_different(file, generator.as_string_view());
  522. }
  523. int main(int argc, char** argv)
  524. {
  525. char const* generated_header_path = nullptr;
  526. char const* generated_implementation_path = nullptr;
  527. char const* core_path = nullptr;
  528. char const* locale_names_path = nullptr;
  529. char const* numbers_path = nullptr;
  530. Core::ArgsParser args_parser;
  531. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  532. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  533. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  534. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  535. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  536. args_parser.parse(argc, argv);
  537. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  538. if (path.is_empty()) {
  539. warnln("{} is required", flags);
  540. args_parser.print_usage(stderr, argv[0]);
  541. exit(1);
  542. }
  543. auto file_or_error = Core::File::open(path, mode);
  544. if (file_or_error.is_error()) {
  545. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  546. exit(1);
  547. }
  548. return file_or_error.release_value();
  549. };
  550. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  551. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  552. UnicodeLocaleData locale_data;
  553. parse_all_locales(core_path, locale_names_path, numbers_path, locale_data);
  554. generate_unicode_locale_header(generated_header_file, locale_data);
  555. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  556. return 0;
  557. }