GenerateUnicodeLocale.cpp 47 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261
  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. #include <LibUnicode/Locale.h>
  22. struct ListPatterns {
  23. String type;
  24. String style;
  25. size_t start { 0 };
  26. size_t middle { 0 };
  27. size_t end { 0 };
  28. size_t pair { 0 };
  29. };
  30. struct Locale {
  31. String language;
  32. Optional<String> territory;
  33. Optional<String> variant;
  34. HashMap<String, size_t> languages;
  35. HashMap<String, size_t> territories;
  36. HashMap<String, size_t> scripts;
  37. HashMap<String, size_t> currencies;
  38. HashMap<String, size_t> keywords;
  39. Vector<ListPatterns> list_patterns;
  40. };
  41. struct CanonicalLanguageID {
  42. String language {};
  43. String script {};
  44. String region {};
  45. Vector<String> variants {};
  46. };
  47. struct LanguageMapping {
  48. CanonicalLanguageID key {};
  49. CanonicalLanguageID alias {};
  50. };
  51. struct UnicodeLocaleData {
  52. Vector<String> unique_strings;
  53. HashMap<StringView, size_t> unique_string_indices;
  54. HashMap<String, Locale> locales;
  55. Vector<String> languages;
  56. Vector<String> territories;
  57. Vector<String> scripts;
  58. Vector<String> variants;
  59. Vector<String> currencies;
  60. Vector<String> keywords;
  61. Vector<String> list_pattern_types;
  62. Vector<String> list_pattern_styles;
  63. HashMap<String, String> language_aliases;
  64. HashMap<String, String> territory_aliases;
  65. HashMap<String, String> script_aliases;
  66. HashMap<String, String> variant_aliases;
  67. HashMap<String, String> subdivision_aliases;
  68. Vector<LanguageMapping> complex_mappings;
  69. Vector<LanguageMapping> likely_subtags;
  70. size_t max_variant_size { 0 };
  71. };
  72. static size_t ensure_unique_string(UnicodeLocaleData& locale_data, String string)
  73. {
  74. // We maintain a set of unique strings in two structures: a vector which owns the unique string,
  75. // and a hash map which maps that string to its index in the vector. The vector is to ensure the
  76. // strings are generated in an easily known order, and the map is to allow quickly deciding if a
  77. // string is actually unique (otherwise, we'd have to linear-search the vector for each string).
  78. //
  79. // Also note that index 0 will be reserved for the empty string, so the index returned from this
  80. // method is actually the real index in the vector + 1.
  81. if (auto index = locale_data.unique_string_indices.get(string); index.has_value())
  82. return *index;
  83. locale_data.unique_strings.append(move(string));
  84. size_t index = locale_data.unique_strings.size();
  85. locale_data.unique_string_indices.set(locale_data.unique_strings.last(), index);
  86. return index;
  87. }
  88. static Optional<CanonicalLanguageID> parse_language(StringView language)
  89. {
  90. CanonicalLanguageID language_id {};
  91. auto segments = language.split_view('-');
  92. VERIFY(!segments.is_empty());
  93. size_t index = 0;
  94. if (Unicode::is_unicode_language_subtag(segments[index])) {
  95. language_id.language = segments[index];
  96. if (segments.size() == ++index)
  97. return language_id;
  98. } else {
  99. return {};
  100. }
  101. if (Unicode::is_unicode_script_subtag(segments[index])) {
  102. language_id.script = segments[index];
  103. if (segments.size() == ++index)
  104. return language_id;
  105. }
  106. if (Unicode::is_unicode_region_subtag(segments[index])) {
  107. language_id.region = segments[index];
  108. if (segments.size() == ++index)
  109. return language_id;
  110. }
  111. while (index < segments.size()) {
  112. if (!Unicode::is_unicode_variant_subtag(segments[index]))
  113. return {};
  114. language_id.variants.append(segments[index++]);
  115. }
  116. return language_id;
  117. }
  118. static Optional<LanguageMapping> parse_language_mapping(StringView key, StringView alias)
  119. {
  120. auto parsed_key = parse_language(key);
  121. if (!parsed_key.has_value())
  122. return {};
  123. auto parsed_alias = parse_language(alias);
  124. if (!parsed_alias.has_value())
  125. return {};
  126. return LanguageMapping { parsed_key.release_value(), parsed_alias.release_value() };
  127. }
  128. static void parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
  129. {
  130. LexicalPath core_aliases_path(move(core_supplemental_path));
  131. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  132. VERIFY(Core::File::exists(core_aliases_path.string()));
  133. auto core_aliases_file_or_error = Core::File::open(core_aliases_path.string(), Core::OpenMode::ReadOnly);
  134. VERIFY(!core_aliases_file_or_error.is_error());
  135. auto core_aliases = JsonParser(core_aliases_file_or_error.value()->read_all()).parse();
  136. VERIFY(core_aliases.has_value());
  137. auto const& supplemental_object = core_aliases->as_object().get("supplemental"sv);
  138. auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
  139. auto const& alias_object = metadata_object.as_object().get("alias"sv);
  140. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  141. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  142. auto alias = value.as_object().get("_replacement"sv).as_string();
  143. if (key.contains('-')) {
  144. auto mapping = parse_language_mapping(key, alias);
  145. if (!mapping.has_value())
  146. return;
  147. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  148. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  149. locale_data.complex_mappings.append(mapping.release_value());
  150. } else {
  151. alias_map.set(key, move(alias));
  152. }
  153. });
  154. };
  155. append_aliases(alias_object.as_object().get("languageAlias"sv), locale_data.language_aliases);
  156. append_aliases(alias_object.as_object().get("territoryAlias"sv), locale_data.territory_aliases);
  157. append_aliases(alias_object.as_object().get("scriptAlias"sv), locale_data.script_aliases);
  158. append_aliases(alias_object.as_object().get("variantAlias"sv), locale_data.variant_aliases);
  159. append_aliases(alias_object.as_object().get("subdivisionAlias"sv), locale_data.subdivision_aliases);
  160. }
  161. static void parse_likely_subtags(String core_supplemental_path, UnicodeLocaleData& locale_data)
  162. {
  163. LexicalPath likely_subtags_path(move(core_supplemental_path));
  164. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  165. VERIFY(Core::File::exists(likely_subtags_path.string()));
  166. auto likely_subtags_file_or_error = Core::File::open(likely_subtags_path.string(), Core::OpenMode::ReadOnly);
  167. VERIFY(!likely_subtags_file_or_error.is_error());
  168. auto likely_subtags = JsonParser(likely_subtags_file_or_error.value()->read_all()).parse();
  169. VERIFY(likely_subtags.has_value());
  170. auto const& supplemental_object = likely_subtags->as_object().get("supplemental"sv);
  171. auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
  172. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  173. auto mapping = parse_language_mapping(key, value.as_string());
  174. if (!mapping.has_value())
  175. return;
  176. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  177. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  178. locale_data.likely_subtags.append(mapping.release_value());
  179. });
  180. }
  181. static void parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  182. {
  183. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  184. languages_path = languages_path.append("languages.json"sv);
  185. VERIFY(Core::File::exists(languages_path.string()));
  186. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  187. VERIFY(!languages_file_or_error.is_error());
  188. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  189. VERIFY(languages.has_value());
  190. auto const& main_object = languages->as_object().get("main"sv);
  191. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  192. auto const& identity_object = locale_object.as_object().get("identity"sv);
  193. auto const& language_string = identity_object.as_object().get("language"sv);
  194. auto const& territory_string = identity_object.as_object().get("territory"sv);
  195. auto const& variant_string = identity_object.as_object().get("variant"sv);
  196. locale.language = language_string.as_string();
  197. if (!locale_data.languages.contains_slow(locale.language))
  198. locale_data.languages.append(locale.language);
  199. if (territory_string.is_string()) {
  200. locale.territory = territory_string.as_string();
  201. if (!locale_data.territories.contains_slow(*locale.territory))
  202. locale_data.territories.append(*locale.territory);
  203. }
  204. if (variant_string.is_string()) {
  205. locale.variant = variant_string.as_string();
  206. if (!locale_data.variants.contains_slow(*locale.variant))
  207. locale_data.variants.append(*locale.variant);
  208. }
  209. }
  210. static void parse_locale_languages(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  211. {
  212. LexicalPath languages_path(move(locale_path));
  213. languages_path = languages_path.append("languages.json"sv);
  214. VERIFY(Core::File::exists(languages_path.string()));
  215. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  216. VERIFY(!languages_file_or_error.is_error());
  217. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  218. VERIFY(languages.has_value());
  219. auto const& main_object = languages->as_object().get("main"sv);
  220. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  221. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  222. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  223. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  224. if (!locale_data.languages.contains_slow(key))
  225. return;
  226. size_t index = ensure_unique_string(locale_data, value.as_string());
  227. locale.languages.set(key, index);
  228. });
  229. }
  230. static void parse_locale_territories(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  231. {
  232. LexicalPath territories_path(move(locale_path));
  233. territories_path = territories_path.append("territories.json"sv);
  234. VERIFY(Core::File::exists(territories_path.string()));
  235. auto territories_file_or_error = Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly);
  236. VERIFY(!territories_file_or_error.is_error());
  237. auto territories = JsonParser(territories_file_or_error.value()->read_all()).parse();
  238. VERIFY(territories.has_value());
  239. auto const& main_object = territories->as_object().get("main"sv);
  240. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  241. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  242. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  243. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  244. if (!locale_data.territories.contains_slow(key))
  245. return;
  246. size_t index = ensure_unique_string(locale_data, value.as_string());
  247. locale.territories.set(key, index);
  248. });
  249. }
  250. static void parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  251. {
  252. LexicalPath scripts_path(move(locale_path));
  253. scripts_path = scripts_path.append("scripts.json"sv);
  254. VERIFY(Core::File::exists(scripts_path.string()));
  255. auto scripts_file_or_error = Core::File::open(scripts_path.string(), Core::OpenMode::ReadOnly);
  256. VERIFY(!scripts_file_or_error.is_error());
  257. auto scripts = JsonParser(scripts_file_or_error.value()->read_all()).parse();
  258. VERIFY(scripts.has_value());
  259. auto const& main_object = scripts->as_object().get("main"sv);
  260. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  261. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  262. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  263. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  264. size_t index = ensure_unique_string(locale_data, value.as_string());
  265. locale.scripts.set(key, index);
  266. if (!locale_data.scripts.contains_slow(key))
  267. locale_data.scripts.append(key);
  268. });
  269. }
  270. static void parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
  271. {
  272. LexicalPath list_patterns_path(move(misc_path));
  273. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  274. VERIFY(Core::File::exists(list_patterns_path.string()));
  275. auto list_patterns_file_or_error = Core::File::open(list_patterns_path.string(), Core::OpenMode::ReadOnly);
  276. VERIFY(!list_patterns_file_or_error.is_error());
  277. auto list_patterns = JsonParser(list_patterns_file_or_error.value()->read_all()).parse();
  278. VERIFY(list_patterns.has_value());
  279. auto const& main_object = list_patterns->as_object().get("main"sv);
  280. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  281. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  282. auto list_pattern_type = [](StringView key) {
  283. if (key.contains("type-standard"sv))
  284. return "conjunction"sv;
  285. if (key.contains("type-or"sv))
  286. return "disjunction"sv;
  287. if (key.contains("type-unit"sv))
  288. return "unit"sv;
  289. VERIFY_NOT_REACHED();
  290. };
  291. auto list_pattern_style = [](StringView key) {
  292. if (key.contains("short"sv))
  293. return "short"sv;
  294. if (key.contains("narrow"sv))
  295. return "narrow"sv;
  296. return "long"sv;
  297. };
  298. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  299. auto type = list_pattern_type(key);
  300. auto style = list_pattern_style(key);
  301. auto start = ensure_unique_string(locale_data, value.as_object().get("start"sv).as_string());
  302. auto middle = ensure_unique_string(locale_data, value.as_object().get("middle"sv).as_string());
  303. auto end = ensure_unique_string(locale_data, value.as_object().get("end"sv).as_string());
  304. auto pair = ensure_unique_string(locale_data, value.as_object().get("2"sv).as_string());
  305. if (!locale_data.list_pattern_types.contains_slow(type))
  306. locale_data.list_pattern_types.append(type);
  307. if (!locale_data.list_pattern_styles.contains_slow(style))
  308. locale_data.list_pattern_styles.append(style);
  309. locale.list_patterns.append({ move(type), move(style), move(start), move(middle), move(end), move(pair) });
  310. });
  311. }
  312. static void parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  313. {
  314. LexicalPath currencies_path(move(numbers_path));
  315. currencies_path = currencies_path.append("currencies.json"sv);
  316. VERIFY(Core::File::exists(currencies_path.string()));
  317. auto currencies_file_or_error = Core::File::open(currencies_path.string(), Core::OpenMode::ReadOnly);
  318. VERIFY(!currencies_file_or_error.is_error());
  319. auto currencies = JsonParser(currencies_file_or_error.value()->read_all()).parse();
  320. VERIFY(currencies.has_value());
  321. auto const& main_object = currencies->as_object().get("main"sv);
  322. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  323. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  324. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  325. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  326. auto const& display_name = value.as_object().get("displayName"sv);
  327. size_t index = ensure_unique_string(locale_data, display_name.as_string());
  328. locale.currencies.set(key, index);
  329. if (!locale_data.currencies.contains_slow(key))
  330. locale_data.currencies.append(key);
  331. });
  332. }
  333. static void parse_numeric_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  334. {
  335. static constexpr StringView key = "nu"sv;
  336. LexicalPath numbers_path(move(locale_numbers_path));
  337. numbers_path = numbers_path.append("numbers.json"sv);
  338. VERIFY(Core::File::exists(numbers_path.string()));
  339. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  340. VERIFY(!numbers_file_or_error.is_error());
  341. auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
  342. VERIFY(numbers.has_value());
  343. auto const& main_object = numbers->as_object().get("main"sv);
  344. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  345. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  346. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  347. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  348. Vector<String> keyword_values {};
  349. keyword_values.append(default_numbering_system_object.as_string());
  350. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  351. auto keyword_value = value.as_string();
  352. if (!keyword_values.contains_slow(keyword_value))
  353. keyword_values.append(move(keyword_value));
  354. });
  355. StringBuilder builder;
  356. builder.join(',', keyword_values);
  357. auto index = ensure_unique_string(locale_data, builder.build());
  358. locale.keywords.set(key, index);
  359. if (!locale_data.keywords.contains_slow(key))
  360. locale_data.keywords.append(key);
  361. }
  362. static Core::DirIterator path_to_dir_iterator(String path)
  363. {
  364. LexicalPath lexical_path(move(path));
  365. lexical_path = lexical_path.append("main"sv);
  366. VERIFY(Core::File::is_directory(lexical_path.string()));
  367. Core::DirIterator iterator(lexical_path.string(), Core::DirIterator::SkipParentAndBaseDir);
  368. if (iterator.has_error()) {
  369. warnln("{}: {}", lexical_path.string(), iterator.error_string());
  370. VERIFY_NOT_REACHED();
  371. }
  372. return iterator;
  373. }
  374. static void parse_all_locales(String core_path, String locale_names_path, String misc_path, String numbers_path, UnicodeLocaleData& locale_data)
  375. {
  376. auto identity_iterator = path_to_dir_iterator(locale_names_path);
  377. auto locale_names_iterator = path_to_dir_iterator(move(locale_names_path));
  378. auto misc_iterator = path_to_dir_iterator(move(misc_path));
  379. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  380. LexicalPath core_supplemental_path(move(core_path));
  381. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  382. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  383. parse_core_aliases(core_supplemental_path.string(), locale_data);
  384. parse_likely_subtags(core_supplemental_path.string(), locale_data);
  385. auto remove_variants_from_path = [](String path) -> Optional<String> {
  386. auto parsed_locale = parse_language(LexicalPath::basename(path));
  387. if (!parsed_locale.has_value())
  388. return {};
  389. StringBuilder builder;
  390. builder.append(parsed_locale->language);
  391. if (!parsed_locale->script.is_empty())
  392. builder.appendff("-{}", parsed_locale->script);
  393. if (!parsed_locale->region.is_empty())
  394. builder.appendff("-{}", parsed_locale->region);
  395. return builder.build();
  396. };
  397. while (identity_iterator.has_next()) {
  398. auto locale_path = identity_iterator.next_full_path();
  399. VERIFY(Core::File::is_directory(locale_path));
  400. auto language = remove_variants_from_path(locale_path);
  401. if (!language.has_value())
  402. continue;
  403. auto& locale = locale_data.locales.ensure(*language);
  404. parse_identity(locale_path, locale_data, locale);
  405. }
  406. while (locale_names_iterator.has_next()) {
  407. auto locale_path = locale_names_iterator.next_full_path();
  408. VERIFY(Core::File::is_directory(locale_path));
  409. auto language = remove_variants_from_path(locale_path);
  410. if (!language.has_value())
  411. continue;
  412. auto& locale = locale_data.locales.ensure(*language);
  413. parse_locale_languages(locale_path, locale_data, locale);
  414. parse_locale_territories(locale_path, locale_data, locale);
  415. parse_locale_scripts(locale_path, locale_data, locale);
  416. }
  417. while (misc_iterator.has_next()) {
  418. auto misc_path = misc_iterator.next_full_path();
  419. VERIFY(Core::File::is_directory(misc_path));
  420. auto language = remove_variants_from_path(misc_path);
  421. if (!language.has_value())
  422. continue;
  423. auto& locale = locale_data.locales.ensure(*language);
  424. parse_locale_list_patterns(misc_path, locale_data, locale);
  425. }
  426. while (numbers_iterator.has_next()) {
  427. auto numbers_path = numbers_iterator.next_full_path();
  428. VERIFY(Core::File::is_directory(numbers_path));
  429. auto language = remove_variants_from_path(numbers_path);
  430. if (!language.has_value())
  431. continue;
  432. auto& locale = locale_data.locales.ensure(*language);
  433. parse_locale_currencies(numbers_path, locale_data, locale);
  434. parse_numeric_keywords(numbers_path, locale_data, locale);
  435. }
  436. }
  437. static String format_identifier(StringView owner, String identifier)
  438. {
  439. identifier = identifier.replace("-"sv, "_"sv, true);
  440. if (all_of(identifier, is_ascii_digit))
  441. return String::formatted("{}_{}", owner[0], identifier);
  442. return identifier.to_titlecase();
  443. }
  444. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  445. {
  446. StringBuilder builder;
  447. SourceGenerator generator { builder };
  448. auto generate_enum = [&](StringView name, StringView default_, Vector<String>& values) {
  449. quick_sort(values);
  450. generator.set("name", name);
  451. generator.set("underlying", ((values.size() + !default_.is_empty()) < 256) ? "u8"sv : "u16"sv);
  452. generator.append(R"~~~(
  453. enum class @name@ : @underlying@ {)~~~");
  454. if (!default_.is_empty()) {
  455. generator.set("default", default_);
  456. generator.append(R"~~~(
  457. @default@,)~~~");
  458. }
  459. for (auto const& value : values) {
  460. generator.set("value", format_identifier(name, value));
  461. generator.append(R"~~~(
  462. @value@,)~~~");
  463. }
  464. generator.append(R"~~~(
  465. };
  466. )~~~");
  467. };
  468. generator.append(R"~~~(
  469. #pragma once
  470. #include <AK/Optional.h>
  471. #include <AK/StringView.h>
  472. #include <AK/Types.h>
  473. #include <LibUnicode/Forward.h>
  474. namespace Unicode {
  475. )~~~");
  476. auto locales = locale_data.locales.keys();
  477. generate_enum("Locale"sv, "None"sv, locales);
  478. generate_enum("Language"sv, {}, locale_data.languages);
  479. generate_enum("Territory"sv, {}, locale_data.territories);
  480. generate_enum("ScriptTag"sv, {}, locale_data.scripts);
  481. generate_enum("Currency"sv, {}, locale_data.currencies);
  482. generate_enum("Key"sv, {}, locale_data.keywords);
  483. generate_enum("Variant"sv, {}, locale_data.variants);
  484. generate_enum("ListPatternType"sv, {}, locale_data.list_pattern_types);
  485. generate_enum("ListPatternStyle"sv, {}, locale_data.list_pattern_styles);
  486. generator.append(R"~~~(
  487. namespace Detail {
  488. Optional<Locale> locale_from_string(StringView const& locale);
  489. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  490. Optional<Language> language_from_string(StringView const& language);
  491. Optional<StringView> resolve_language_alias(StringView const& language);
  492. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  493. Optional<Territory> territory_from_string(StringView const& territory);
  494. Optional<StringView> resolve_territory_alias(StringView const& territory);
  495. Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
  496. Optional<ScriptTag> script_tag_from_string(StringView const& script_tag);
  497. Optional<StringView> resolve_script_tag_alias(StringView const& script_tag);
  498. Optional<StringView> get_locale_currency_mapping(StringView locale, StringView currency);
  499. Optional<Currency> currency_from_string(StringView const& currency);
  500. Optional<StringView> get_locale_key_mapping(StringView locale, StringView key);
  501. Optional<Key> key_from_string(StringView const& key);
  502. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style);
  503. Optional<ListPatternType> list_pattern_type_from_string(StringView const& list_pattern_type);
  504. Optional<ListPatternStyle> list_pattern_style_from_string(StringView const& list_pattern_style);
  505. Optional<StringView> resolve_variant_alias(StringView const& variant);
  506. Optional<StringView> resolve_subdivision_alias(StringView const& subdivision);
  507. void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
  508. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id);
  509. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
  510. }
  511. }
  512. )~~~");
  513. file.write(generator.as_string_view());
  514. }
  515. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  516. {
  517. StringBuilder builder;
  518. SourceGenerator generator { builder };
  519. generator.set("strings_size"sv, String::number(locale_data.unique_strings.size()));
  520. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  521. generator.set("territories_size", String::number(locale_data.territories.size()));
  522. generator.set("variants_size", String::number(locale_data.max_variant_size));
  523. generator.append(R"~~~(
  524. #include <AK/Array.h>
  525. #include <AK/HashMap.h>
  526. #include <AK/Span.h>
  527. #include <LibUnicode/Locale.h>
  528. #include <LibUnicode/UnicodeLocale.h>
  529. namespace Unicode {
  530. struct Patterns {
  531. ListPatternType type;
  532. ListPatternStyle style;
  533. size_t start { 0 };
  534. size_t middle { 0 };
  535. size_t end { 0 };
  536. size_t pair { 0 };
  537. };
  538. )~~~");
  539. generator.append(R"~~~(
  540. static constexpr Array<StringView, @strings_size@ + 1> s_string_list { {
  541. {})~~~");
  542. constexpr size_t max_strings_per_row = 30;
  543. size_t strings_in_current_row = 1;
  544. for (auto const& string : locale_data.unique_strings) {
  545. if (strings_in_current_row++ > 0)
  546. generator.append(", ");
  547. generator.append(String::formatted("\"{}\"sv", string));
  548. if (strings_in_current_row == max_strings_per_row) {
  549. strings_in_current_row = 0;
  550. generator.append(",\n ");
  551. }
  552. }
  553. generator.append(R"~~~(
  554. } };
  555. )~~~");
  556. auto format_mapping_name = [](StringView format, StringView name) {
  557. auto mapping_name = name.to_lowercase_string().replace("-"sv, "_"sv, true);
  558. return String::formatted(format, mapping_name);
  559. };
  560. auto append_string = [&](StringView value) {
  561. if (value.is_empty())
  562. generator.append(", {}"sv);
  563. else
  564. generator.append(String::formatted(", \"{}\"sv", value));
  565. };
  566. auto append_list_and_size = [&](auto const& list) {
  567. if (list.is_empty()) {
  568. generator.append(", {}, 0");
  569. return;
  570. }
  571. bool first = true;
  572. generator.append(", {");
  573. for (auto const& item : list) {
  574. generator.append(first ? " " : ", ");
  575. generator.append(String::formatted("\"{}\"sv", item));
  576. first = false;
  577. }
  578. generator.append(String::formatted(" }}, {}", list.size()));
  579. };
  580. auto append_string_index_list = [&](String name, auto const& keys, auto const& mappings) {
  581. generator.set("name", name);
  582. generator.set("size", String::number(keys.size()));
  583. generator.append(R"~~~(
  584. static constexpr Array<size_t, @size@> @name@ { {
  585. )~~~");
  586. constexpr size_t max_values_per_row = 30;
  587. size_t values_in_current_row = 0;
  588. for (auto const& key : keys) {
  589. if (values_in_current_row++ > 0)
  590. generator.append(" ");
  591. if (auto it = mappings.find(key); it != mappings.end())
  592. generator.set("mapping"sv, String::number(it->value));
  593. else
  594. generator.set("mapping"sv, "0"sv);
  595. generator.append("@mapping@,");
  596. if (values_in_current_row == max_values_per_row) {
  597. values_in_current_row = 0;
  598. generator.append("\n ");
  599. }
  600. }
  601. generator.append(R"~~~(
  602. } };
  603. )~~~");
  604. };
  605. auto append_list_patterns = [&](StringView name, Vector<ListPatterns> const& list_patterns) {
  606. generator.set("name", name);
  607. generator.set("size", String::number(list_patterns.size()));
  608. generator.append(R"~~~(
  609. static constexpr Array<Patterns, @size@> @name@ { {)~~~");
  610. for (auto const& list_pattern : list_patterns) {
  611. generator.set("type"sv, String::formatted("ListPatternType::{}", format_identifier({}, list_pattern.type)));
  612. generator.set("style"sv, String::formatted("ListPatternStyle::{}", format_identifier({}, list_pattern.style)));
  613. generator.set("start"sv, String::number(list_pattern.start));
  614. generator.set("middle"sv, String::number(list_pattern.middle));
  615. generator.set("end"sv, String::number(list_pattern.end));
  616. generator.set("pair"sv, String::number(list_pattern.pair));
  617. generator.append(R"~~~(
  618. { @type@, @style@, @start@, @middle@, @end@, @pair@ },)~~~");
  619. }
  620. generator.append(R"~~~(
  621. } };
  622. )~~~");
  623. };
  624. auto append_mapping = [&](StringView type, StringView name, StringView format, auto format_list_callback) {
  625. Vector<String> mapping_names;
  626. for (auto const& locale : locale_data.locales) {
  627. auto mapping_name = format_mapping_name(format, locale.key);
  628. format_list_callback(mapping_name, locale.value);
  629. mapping_names.append(move(mapping_name));
  630. }
  631. quick_sort(mapping_names);
  632. generator.set("type", type);
  633. generator.set("name", name);
  634. generator.set("size", String::number(locale_data.locales.size()));
  635. generator.append(R"~~~(
  636. static constexpr Array<Span<@type@ const>, @size@> @name@ { {
  637. )~~~");
  638. constexpr size_t max_values_per_row = 10;
  639. size_t values_in_current_row = 0;
  640. for (auto& mapping_name : mapping_names) {
  641. if (values_in_current_row++ > 0)
  642. generator.append(" ");
  643. generator.set("name", move(mapping_name));
  644. generator.append("@name@.span(),");
  645. if (values_in_current_row == max_values_per_row) {
  646. values_in_current_row = 0;
  647. generator.append("\n ");
  648. }
  649. }
  650. generator.append(R"~~~(
  651. } };
  652. )~~~");
  653. };
  654. append_mapping("size_t"sv, "s_languages"sv, "s_languages_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.languages, value.languages); });
  655. append_mapping("size_t"sv, "s_territories"sv, "s_territories_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.territories, value.territories); });
  656. append_mapping("size_t"sv, "s_scripts"sv, "s_scripts_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.scripts, value.scripts); });
  657. append_mapping("size_t"sv, "s_currencies"sv, "s_currencies_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.currencies, value.currencies); });
  658. append_mapping("size_t"sv, "s_keywords"sv, "s_keywords_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.keywords, value.keywords); });
  659. append_mapping("Patterns"sv, "s_list_patterns"sv, "s_list_patterns_{}", [&](auto const& name, auto const& value) { append_list_patterns(name, value.list_patterns); });
  660. generator.append(R"~~~(
  661. struct CanonicalLanguageID {
  662. Unicode::LanguageID to_unicode_language_id() const
  663. {
  664. Unicode::LanguageID language_id {};
  665. language_id.variants.ensure_capacity(variants_size);
  666. language_id.language = language.to_string();
  667. if (!script.is_empty())
  668. language_id.script = script.to_string();
  669. if (!region.is_empty())
  670. language_id.region = region.to_string();
  671. for (size_t i = 0; i < variants_size; ++i)
  672. language_id.variants.append(variants[i].to_string());
  673. return language_id;
  674. }
  675. bool matches_variants(Vector<String> const& other_variants) const {
  676. if (variants_size == 0)
  677. return true;
  678. if (other_variants.size() != variants_size)
  679. return false;
  680. for (size_t i = 0; i < variants_size; ++i) {
  681. if (variants[i] != other_variants[i])
  682. return false;
  683. }
  684. return true;
  685. };
  686. StringView language {};
  687. StringView script {};
  688. StringView region {};
  689. Array<StringView, @variants_size@> variants {};
  690. size_t variants_size { 0 };
  691. };
  692. struct LanguageMapping {
  693. CanonicalLanguageID key;
  694. CanonicalLanguageID alias;
  695. };
  696. )~~~");
  697. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  698. generator.set("size", String::number(mappings.size()));
  699. generator.set("name"sv, name);
  700. generator.append(R"~~~(
  701. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  702. )~~~");
  703. quick_sort(mappings, [](auto const& lhs, auto const& rhs) {
  704. auto const& lhs_language = lhs.key.language;
  705. auto const& rhs_language = rhs.key.language;
  706. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  707. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  708. return false;
  709. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  710. return true;
  711. return lhs_language < rhs_language;
  712. });
  713. for (auto const& mapping : mappings) {
  714. generator.set("language"sv, mapping.key.language);
  715. generator.append(" { { \"@language@\"sv");
  716. append_string(mapping.key.script);
  717. append_string(mapping.key.region);
  718. append_list_and_size(mapping.key.variants);
  719. generator.set("language"sv, mapping.alias.language);
  720. generator.append(" }, { \"@language@\"sv");
  721. append_string(mapping.alias.script);
  722. append_string(mapping.alias.region);
  723. append_list_and_size(mapping.alias.variants);
  724. generator.append(" } },\n");
  725. }
  726. generator.append("} };\n");
  727. };
  728. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  729. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  730. generator.append(R"~~~(
  731. static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& language_id)
  732. {
  733. // https://unicode.org/reports/tr35/#Likely_Subtags
  734. enum class State {
  735. LanguageScriptRegion,
  736. LanguageRegion,
  737. LanguageScript,
  738. Language,
  739. UndScript,
  740. Done,
  741. };
  742. auto state = State::LanguageScriptRegion;
  743. while (state != State::Done) {
  744. CanonicalLanguageID search_key;
  745. switch (state) {
  746. case State::LanguageScriptRegion:
  747. state = State::LanguageRegion;
  748. if (!language_id.script.has_value() || !language_id.region.has_value())
  749. continue;
  750. search_key.language = *language_id.language;
  751. search_key.script = *language_id.script;
  752. search_key.region = *language_id.region;
  753. break;
  754. case State::LanguageRegion:
  755. state = State::LanguageScript;
  756. if (!language_id.region.has_value())
  757. continue;
  758. search_key.language = *language_id.language;
  759. search_key.region = *language_id.region;
  760. break;
  761. case State::LanguageScript:
  762. state = State::Language;
  763. if (!language_id.script.has_value())
  764. continue;
  765. search_key.language = *language_id.language;
  766. search_key.script = *language_id.script;
  767. break;
  768. case State::Language:
  769. state = State::UndScript;
  770. search_key.language = *language_id.language;
  771. break;
  772. case State::UndScript:
  773. state = State::Done;
  774. if (!language_id.script.has_value())
  775. continue;
  776. search_key.language = "und"sv;
  777. search_key.script = *language_id.script;
  778. break;
  779. default:
  780. VERIFY_NOT_REACHED();
  781. }
  782. for (auto const& map : s_likely_subtags) {
  783. if (map.key.language != search_key.language)
  784. continue;
  785. if (map.key.script != search_key.script)
  786. continue;
  787. if (map.key.region != search_key.region)
  788. continue;
  789. return &map;
  790. }
  791. }
  792. return nullptr;
  793. }
  794. namespace Detail {
  795. )~~~");
  796. auto append_mapping_search = [&](StringView enum_title, StringView enum_snake, StringView collection_name) {
  797. generator.set("enum_title", enum_title);
  798. generator.set("enum_snake", enum_snake);
  799. generator.set("collection_name", collection_name);
  800. generator.append(R"~~~(
  801. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  802. {
  803. auto locale_value = locale_from_string(locale);
  804. if (!locale_value.has_value())
  805. return {};
  806. auto @enum_snake@_value = @enum_snake@_from_string(@enum_snake@);
  807. if (!@enum_snake@_value.has_value())
  808. return {};
  809. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  810. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  811. auto const& mappings = @collection_name@.at(locale_index);
  812. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  813. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  814. if (@enum_snake@_mapping.is_empty())
  815. return {};
  816. return @enum_snake@_mapping;
  817. }
  818. )~~~");
  819. };
  820. auto append_from_string = [&](StringView enum_title, StringView enum_snake, Vector<String> const& values) {
  821. generator.set("enum_title", enum_title);
  822. generator.set("enum_snake", enum_snake);
  823. generator.append(R"~~~(
  824. Optional<@enum_title@> @enum_snake@_from_string(StringView const& @enum_snake@)
  825. {
  826. static HashMap<StringView, @enum_title@> @enum_snake@_values { {)~~~");
  827. for (auto const& value : values) {
  828. generator.set("key"sv, value);
  829. generator.set("value"sv, format_identifier(enum_title, value));
  830. generator.append(R"~~~(
  831. { "@key@"sv, @enum_title@::@value@ },)~~~");
  832. }
  833. generator.append(R"~~~(
  834. } };
  835. if (auto value = @enum_snake@_values.get(@enum_snake@); value.has_value())
  836. return value.value();
  837. return {};
  838. }
  839. )~~~");
  840. };
  841. auto append_alias_search = [&](StringView enum_snake, HashMap<String, String> const& aliases) {
  842. generator.set("enum_snake", enum_snake);
  843. generator.append(R"~~~(
  844. Optional<StringView> resolve_@enum_snake@_alias(StringView const& @enum_snake@)
  845. {
  846. static HashMap<StringView, StringView> @enum_snake@_aliases { {
  847. )~~~");
  848. constexpr size_t max_values_per_row = 10;
  849. size_t values_in_current_row = 0;
  850. for (auto const& alias : aliases) {
  851. if (values_in_current_row++ > 0)
  852. generator.append(" ");
  853. generator.set("key"sv, alias.key);
  854. generator.set("alias"sv, alias.value);
  855. generator.append("{ \"@key@\"sv, \"@alias@\"sv },");
  856. if (values_in_current_row == max_values_per_row) {
  857. generator.append("\n ");
  858. values_in_current_row = 0;
  859. }
  860. }
  861. generator.append(R"~~~(
  862. } };
  863. if (auto alias = @enum_snake@_aliases.get(@enum_snake@); alias.has_value())
  864. return alias.value();
  865. return {};
  866. }
  867. )~~~");
  868. };
  869. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys());
  870. append_mapping_search("Language"sv, "language"sv, "s_languages"sv);
  871. append_from_string("Language"sv, "language"sv, locale_data.languages);
  872. append_alias_search("language"sv, locale_data.language_aliases);
  873. append_mapping_search("Territory"sv, "territory"sv, "s_territories"sv);
  874. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  875. append_alias_search("territory"sv, locale_data.territory_aliases);
  876. append_mapping_search("ScriptTag"sv, "script_tag"sv, "s_scripts"sv);
  877. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  878. append_alias_search("script_tag"sv, locale_data.script_aliases);
  879. append_mapping_search("Currency"sv, "currency"sv, "s_currencies"sv);
  880. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  881. append_mapping_search("Key"sv, "key"sv, "s_keywords"sv);
  882. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  883. append_alias_search("variant"sv, locale_data.variant_aliases);
  884. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  885. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  886. append_from_string("ListPatternStyle"sv, "list_pattern_style"sv, locale_data.list_pattern_styles);
  887. generator.append(R"~~~(
  888. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style)
  889. {
  890. auto locale_value = locale_from_string(locale);
  891. if (!locale_value.has_value())
  892. return {};
  893. auto type_value = list_pattern_type_from_string(list_pattern_type);
  894. if (!type_value.has_value())
  895. return {};
  896. auto style_value = list_pattern_style_from_string(list_pattern_style);
  897. if (!style_value.has_value())
  898. return {};
  899. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  900. auto const& locale_list_patterns = s_list_patterns.at(locale_index);
  901. for (auto const& list_patterns : locale_list_patterns) {
  902. if ((list_patterns.type == type_value) && (list_patterns.style == style_value)) {
  903. auto const& start = s_string_list[list_patterns.start];
  904. auto const& middle = s_string_list[list_patterns.middle];
  905. auto const& end = s_string_list[list_patterns.end];
  906. auto const& pair = s_string_list[list_patterns.pair];
  907. return ListPatterns { start, middle, end, pair };
  908. }
  909. }
  910. return {};
  911. }
  912. void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
  913. {
  914. for (auto const& map : s_complex_alias) {
  915. if ((map.key.language != language_id.language) && (map.key.language != "und"sv))
  916. continue;
  917. if (!map.key.script.is_empty() && (map.key.script != language_id.script))
  918. continue;
  919. if (!map.key.region.is_empty() && (map.key.region != language_id.region))
  920. continue;
  921. if (!map.key.matches_variants(language_id.variants))
  922. continue;
  923. auto alias = map.alias.to_unicode_language_id();
  924. if (alias.language == "und"sv)
  925. alias.language = move(language_id.language);
  926. if (map.key.script.is_empty() && !alias.script.has_value())
  927. alias.script = move(language_id.script);
  928. if (map.key.region.is_empty() && !alias.region.has_value())
  929. alias.region = move(language_id.region);
  930. if (map.key.variants_size == 0 && alias.variants.is_empty())
  931. alias.variants = move(language_id.variants);
  932. language_id = move(alias);
  933. break;
  934. }
  935. }
  936. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id)
  937. {
  938. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  939. auto const* likely_subtag = resolve_likely_subtag(language_id);
  940. if (likely_subtag == nullptr)
  941. return {};
  942. auto maximized = language_id;
  943. auto const& key = likely_subtag->key;
  944. auto const& alias = likely_subtag->alias;
  945. if (maximized.language == "und"sv)
  946. maximized.language = alias.language;
  947. if (!maximized.script.has_value() || (!key.script.is_empty() && !alias.script.is_empty()))
  948. maximized.script = alias.script;
  949. if (!maximized.region.has_value() || (!key.region.is_empty() && !alias.region.is_empty()))
  950. maximized.region = alias.region;
  951. return maximized;
  952. }
  953. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
  954. {
  955. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  956. return likely_subtag->alias.region;
  957. return {};
  958. }
  959. }
  960. }
  961. )~~~");
  962. file.write(generator.as_string_view());
  963. }
  964. int main(int argc, char** argv)
  965. {
  966. char const* generated_header_path = nullptr;
  967. char const* generated_implementation_path = nullptr;
  968. char const* core_path = nullptr;
  969. char const* locale_names_path = nullptr;
  970. char const* misc_path = nullptr;
  971. char const* numbers_path = nullptr;
  972. Core::ArgsParser args_parser;
  973. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  974. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  975. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  976. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  977. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  978. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  979. args_parser.parse(argc, argv);
  980. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  981. if (path.is_empty()) {
  982. warnln("{} is required", flags);
  983. args_parser.print_usage(stderr, argv[0]);
  984. exit(1);
  985. }
  986. auto file_or_error = Core::File::open(path, mode);
  987. if (file_or_error.is_error()) {
  988. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  989. exit(1);
  990. }
  991. return file_or_error.release_value();
  992. };
  993. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  994. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  995. UnicodeLocaleData locale_data;
  996. parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, locale_data);
  997. generate_unicode_locale_header(generated_header_file, locale_data);
  998. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  999. return 0;
  1000. }