GenerateUnicodeLocale.cpp 45 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109
  1. /*
  2. * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "GeneratorUtil.h"
  7. #include <AK/AllOf.h>
  8. #include <AK/CharacterTypes.h>
  9. #include <AK/Format.h>
  10. #include <AK/HashMap.h>
  11. #include <AK/JsonObject.h>
  12. #include <AK/JsonParser.h>
  13. #include <AK/JsonValue.h>
  14. #include <AK/LexicalPath.h>
  15. #include <AK/QuickSort.h>
  16. #include <AK/SourceGenerator.h>
  17. #include <AK/String.h>
  18. #include <AK/StringBuilder.h>
  19. #include <LibCore/ArgsParser.h>
  20. #include <LibCore/DirIterator.h>
  21. #include <LibCore/File.h>
  22. using StringIndexType = u16;
  23. constexpr auto s_string_index_type = "u16"sv;
  24. struct ListPatterns {
  25. String type;
  26. String style;
  27. StringIndexType start { 0 };
  28. StringIndexType middle { 0 };
  29. StringIndexType end { 0 };
  30. StringIndexType pair { 0 };
  31. };
  32. struct Locale {
  33. String language;
  34. Optional<String> territory;
  35. Optional<String> variant;
  36. HashMap<String, StringIndexType> languages;
  37. HashMap<String, StringIndexType> territories;
  38. HashMap<String, StringIndexType> scripts;
  39. HashMap<String, StringIndexType> currencies;
  40. HashMap<String, StringIndexType> keywords;
  41. Vector<ListPatterns> list_patterns;
  42. };
  43. struct LanguageMapping {
  44. CanonicalLanguageID<StringIndexType> key {};
  45. CanonicalLanguageID<StringIndexType> alias {};
  46. };
  47. struct UnicodeLocaleData {
  48. UniqueStringStorage<StringIndexType> unique_strings;
  49. HashMap<String, Locale> locales;
  50. Vector<String> languages;
  51. Vector<String> territories;
  52. Vector<String> scripts;
  53. Vector<String> variants;
  54. Vector<String> currencies;
  55. Vector<String> keywords;
  56. Vector<String> list_pattern_types;
  57. Vector<String> list_pattern_styles;
  58. HashMap<String, StringIndexType> language_aliases;
  59. HashMap<String, StringIndexType> territory_aliases;
  60. HashMap<String, StringIndexType> script_aliases;
  61. HashMap<String, StringIndexType> variant_aliases;
  62. HashMap<String, StringIndexType> subdivision_aliases;
  63. Vector<LanguageMapping> complex_mappings;
  64. Vector<LanguageMapping> likely_subtags;
  65. size_t max_variant_size { 0 };
  66. };
  67. static Optional<LanguageMapping> parse_language_mapping(UnicodeLocaleData& locale_data, StringView key, StringView alias)
  68. {
  69. auto parsed_key = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, key);
  70. if (!parsed_key.has_value())
  71. return {};
  72. auto parsed_alias = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, alias);
  73. if (!parsed_alias.has_value())
  74. return {};
  75. return LanguageMapping { parsed_key.release_value(), parsed_alias.release_value() };
  76. }
  77. static void parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
  78. {
  79. LexicalPath core_aliases_path(move(core_supplemental_path));
  80. core_aliases_path = core_aliases_path.append("aliases.json"sv);
  81. VERIFY(Core::File::exists(core_aliases_path.string()));
  82. auto core_aliases_file_or_error = Core::File::open(core_aliases_path.string(), Core::OpenMode::ReadOnly);
  83. VERIFY(!core_aliases_file_or_error.is_error());
  84. auto core_aliases = JsonParser(core_aliases_file_or_error.value()->read_all()).parse();
  85. VERIFY(core_aliases.has_value());
  86. auto const& supplemental_object = core_aliases->as_object().get("supplemental"sv);
  87. auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
  88. auto const& alias_object = metadata_object.as_object().get("alias"sv);
  89. auto append_aliases = [&](auto& alias_object, auto& alias_map) {
  90. alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  91. auto alias = value.as_object().get("_replacement"sv).as_string();
  92. if (key.contains('-')) {
  93. auto mapping = parse_language_mapping(locale_data, key, alias);
  94. if (!mapping.has_value())
  95. return;
  96. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  97. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  98. locale_data.complex_mappings.append(mapping.release_value());
  99. } else {
  100. alias_map.set(key, locale_data.unique_strings.ensure(alias));
  101. }
  102. });
  103. };
  104. append_aliases(alias_object.as_object().get("languageAlias"sv), locale_data.language_aliases);
  105. append_aliases(alias_object.as_object().get("territoryAlias"sv), locale_data.territory_aliases);
  106. append_aliases(alias_object.as_object().get("scriptAlias"sv), locale_data.script_aliases);
  107. append_aliases(alias_object.as_object().get("variantAlias"sv), locale_data.variant_aliases);
  108. append_aliases(alias_object.as_object().get("subdivisionAlias"sv), locale_data.subdivision_aliases);
  109. }
  110. static void parse_likely_subtags(String core_supplemental_path, UnicodeLocaleData& locale_data)
  111. {
  112. LexicalPath likely_subtags_path(move(core_supplemental_path));
  113. likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
  114. VERIFY(Core::File::exists(likely_subtags_path.string()));
  115. auto likely_subtags_file_or_error = Core::File::open(likely_subtags_path.string(), Core::OpenMode::ReadOnly);
  116. VERIFY(!likely_subtags_file_or_error.is_error());
  117. auto likely_subtags = JsonParser(likely_subtags_file_or_error.value()->read_all()).parse();
  118. VERIFY(likely_subtags.has_value());
  119. auto const& supplemental_object = likely_subtags->as_object().get("supplemental"sv);
  120. auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
  121. likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  122. auto mapping = parse_language_mapping(locale_data, key, value.as_string());
  123. if (!mapping.has_value())
  124. return;
  125. locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
  126. locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
  127. locale_data.likely_subtags.append(mapping.release_value());
  128. });
  129. }
  130. static void parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  131. {
  132. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  133. languages_path = languages_path.append("languages.json"sv);
  134. VERIFY(Core::File::exists(languages_path.string()));
  135. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  136. VERIFY(!languages_file_or_error.is_error());
  137. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  138. VERIFY(languages.has_value());
  139. auto const& main_object = languages->as_object().get("main"sv);
  140. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  141. auto const& identity_object = locale_object.as_object().get("identity"sv);
  142. auto const& language_string = identity_object.as_object().get("language"sv);
  143. auto const& territory_string = identity_object.as_object().get("territory"sv);
  144. auto const& variant_string = identity_object.as_object().get("variant"sv);
  145. locale.language = language_string.as_string();
  146. if (!locale_data.languages.contains_slow(locale.language))
  147. locale_data.languages.append(locale.language);
  148. if (territory_string.is_string()) {
  149. locale.territory = territory_string.as_string();
  150. if (!locale_data.territories.contains_slow(*locale.territory))
  151. locale_data.territories.append(*locale.territory);
  152. }
  153. if (variant_string.is_string()) {
  154. locale.variant = variant_string.as_string();
  155. if (!locale_data.variants.contains_slow(*locale.variant))
  156. locale_data.variants.append(*locale.variant);
  157. }
  158. }
  159. static void parse_locale_languages(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  160. {
  161. LexicalPath languages_path(move(locale_path));
  162. languages_path = languages_path.append("languages.json"sv);
  163. VERIFY(Core::File::exists(languages_path.string()));
  164. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  165. VERIFY(!languages_file_or_error.is_error());
  166. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  167. VERIFY(languages.has_value());
  168. auto const& main_object = languages->as_object().get("main"sv);
  169. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  170. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  171. auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
  172. languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  173. if (!locale_data.languages.contains_slow(key))
  174. return;
  175. auto index = locale_data.unique_strings.ensure(value.as_string());
  176. locale.languages.set(key, index);
  177. });
  178. }
  179. static void parse_locale_territories(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  180. {
  181. LexicalPath territories_path(move(locale_path));
  182. territories_path = territories_path.append("territories.json"sv);
  183. VERIFY(Core::File::exists(territories_path.string()));
  184. auto territories_file_or_error = Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly);
  185. VERIFY(!territories_file_or_error.is_error());
  186. auto territories = JsonParser(territories_file_or_error.value()->read_all()).parse();
  187. VERIFY(territories.has_value());
  188. auto const& main_object = territories->as_object().get("main"sv);
  189. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  190. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  191. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  192. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  193. if (!locale_data.territories.contains_slow(key))
  194. return;
  195. auto index = locale_data.unique_strings.ensure(value.as_string());
  196. locale.territories.set(key, index);
  197. });
  198. }
  199. static void parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  200. {
  201. LexicalPath scripts_path(move(locale_path));
  202. scripts_path = scripts_path.append("scripts.json"sv);
  203. VERIFY(Core::File::exists(scripts_path.string()));
  204. auto scripts_file_or_error = Core::File::open(scripts_path.string(), Core::OpenMode::ReadOnly);
  205. VERIFY(!scripts_file_or_error.is_error());
  206. auto scripts = JsonParser(scripts_file_or_error.value()->read_all()).parse();
  207. VERIFY(scripts.has_value());
  208. auto const& main_object = scripts->as_object().get("main"sv);
  209. auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
  210. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  211. auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
  212. scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  213. auto index = locale_data.unique_strings.ensure(value.as_string());
  214. locale.scripts.set(key, index);
  215. if (!locale_data.scripts.contains_slow(key))
  216. locale_data.scripts.append(key);
  217. });
  218. }
  219. static void parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
  220. {
  221. LexicalPath list_patterns_path(move(misc_path));
  222. list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
  223. VERIFY(Core::File::exists(list_patterns_path.string()));
  224. auto list_patterns_file_or_error = Core::File::open(list_patterns_path.string(), Core::OpenMode::ReadOnly);
  225. VERIFY(!list_patterns_file_or_error.is_error());
  226. auto list_patterns = JsonParser(list_patterns_file_or_error.value()->read_all()).parse();
  227. VERIFY(list_patterns.has_value());
  228. auto const& main_object = list_patterns->as_object().get("main"sv);
  229. auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
  230. auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
  231. auto list_pattern_type = [](StringView key) {
  232. if (key.contains("type-standard"sv))
  233. return "conjunction"sv;
  234. if (key.contains("type-or"sv))
  235. return "disjunction"sv;
  236. if (key.contains("type-unit"sv))
  237. return "unit"sv;
  238. VERIFY_NOT_REACHED();
  239. };
  240. auto list_pattern_style = [](StringView key) {
  241. if (key.contains("short"sv))
  242. return "short"sv;
  243. if (key.contains("narrow"sv))
  244. return "narrow"sv;
  245. return "long"sv;
  246. };
  247. list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  248. auto type = list_pattern_type(key);
  249. auto style = list_pattern_style(key);
  250. auto start = locale_data.unique_strings.ensure(value.as_object().get("start"sv).as_string());
  251. auto middle = locale_data.unique_strings.ensure(value.as_object().get("middle"sv).as_string());
  252. auto end = locale_data.unique_strings.ensure(value.as_object().get("end"sv).as_string());
  253. auto pair = locale_data.unique_strings.ensure(value.as_object().get("2"sv).as_string());
  254. if (!locale_data.list_pattern_types.contains_slow(type))
  255. locale_data.list_pattern_types.append(type);
  256. if (!locale_data.list_pattern_styles.contains_slow(style))
  257. locale_data.list_pattern_styles.append(style);
  258. locale.list_patterns.append({ move(type), move(style), move(start), move(middle), move(end), move(pair) });
  259. });
  260. }
  261. static void parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  262. {
  263. LexicalPath currencies_path(move(numbers_path));
  264. currencies_path = currencies_path.append("currencies.json"sv);
  265. VERIFY(Core::File::exists(currencies_path.string()));
  266. auto currencies_file_or_error = Core::File::open(currencies_path.string(), Core::OpenMode::ReadOnly);
  267. VERIFY(!currencies_file_or_error.is_error());
  268. auto currencies = JsonParser(currencies_file_or_error.value()->read_all()).parse();
  269. VERIFY(currencies.has_value());
  270. auto const& main_object = currencies->as_object().get("main"sv);
  271. auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
  272. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  273. auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
  274. currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  275. auto const& display_name = value.as_object().get("displayName"sv);
  276. auto index = locale_data.unique_strings.ensure(display_name.as_string());
  277. locale.currencies.set(key, index);
  278. if (!locale_data.currencies.contains_slow(key))
  279. locale_data.currencies.append(key);
  280. });
  281. }
  282. static void parse_numeric_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  283. {
  284. static constexpr StringView key = "nu"sv;
  285. LexicalPath numbers_path(move(locale_numbers_path));
  286. numbers_path = numbers_path.append("numbers.json"sv);
  287. VERIFY(Core::File::exists(numbers_path.string()));
  288. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  289. VERIFY(!numbers_file_or_error.is_error());
  290. auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
  291. VERIFY(numbers.has_value());
  292. auto const& main_object = numbers->as_object().get("main"sv);
  293. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  294. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  295. auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
  296. auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
  297. Vector<String> keyword_values {};
  298. keyword_values.append(default_numbering_system_object.as_string());
  299. other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
  300. auto keyword_value = value.as_string();
  301. if (!keyword_values.contains_slow(keyword_value))
  302. keyword_values.append(move(keyword_value));
  303. });
  304. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  305. if (!key.starts_with("defaultNumberingSystem-alt-"sv))
  306. return;
  307. auto keyword_value = value.as_string();
  308. if (!keyword_values.contains_slow(keyword_value))
  309. keyword_values.append(move(keyword_value));
  310. });
  311. StringBuilder builder;
  312. builder.join(',', keyword_values);
  313. auto index = locale_data.unique_strings.ensure(builder.build());
  314. locale.keywords.set(key, index);
  315. if (!locale_data.keywords.contains_slow(key))
  316. locale_data.keywords.append(key);
  317. }
  318. static void parse_default_content_locales(String core_path, UnicodeLocaleData& locale_data)
  319. {
  320. LexicalPath default_content_path(move(core_path));
  321. default_content_path = default_content_path.append("defaultContent.json"sv);
  322. VERIFY(Core::File::exists(default_content_path.string()));
  323. auto default_content_file_or_error = Core::File::open(default_content_path.string(), Core::OpenMode::ReadOnly);
  324. VERIFY(!default_content_file_or_error.is_error());
  325. auto default_content = JsonParser(default_content_file_or_error.value()->read_all()).parse();
  326. VERIFY(default_content.has_value());
  327. auto const& default_content_array = default_content->as_object().get("defaultContent"sv);
  328. default_content_array.as_array().for_each([&](JsonValue const& value) {
  329. auto locale = value.as_string();
  330. StringView default_locale = locale;
  331. while (true) {
  332. if (locale_data.locales.contains(default_locale))
  333. break;
  334. auto pos = default_locale.find_last('-');
  335. if (!pos.has_value())
  336. return;
  337. default_locale = default_locale.substring_view(0, *pos);
  338. }
  339. locale_data.locales.set(locale, locale_data.locales.get(default_locale).value());
  340. });
  341. }
  342. static void parse_all_locales(String core_path, String locale_names_path, String misc_path, String numbers_path, UnicodeLocaleData& locale_data)
  343. {
  344. auto identity_iterator = path_to_dir_iterator(locale_names_path);
  345. auto locale_names_iterator = path_to_dir_iterator(move(locale_names_path));
  346. auto misc_iterator = path_to_dir_iterator(move(misc_path));
  347. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  348. LexicalPath core_supplemental_path(core_path);
  349. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  350. VERIFY(Core::File::is_directory(core_supplemental_path.string()));
  351. parse_core_aliases(core_supplemental_path.string(), locale_data);
  352. parse_likely_subtags(core_supplemental_path.string(), locale_data);
  353. auto remove_variants_from_path = [&](String path) -> Optional<String> {
  354. auto parsed_locale = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path));
  355. if (!parsed_locale.has_value())
  356. return {};
  357. StringBuilder builder;
  358. builder.append(locale_data.unique_strings.get(parsed_locale->language));
  359. if (auto script = locale_data.unique_strings.get(parsed_locale->script); !script.is_empty())
  360. builder.appendff("-{}", script);
  361. if (auto region = locale_data.unique_strings.get(parsed_locale->region); !region.is_empty())
  362. builder.appendff("-{}", region);
  363. return builder.build();
  364. };
  365. while (identity_iterator.has_next()) {
  366. auto locale_path = identity_iterator.next_full_path();
  367. VERIFY(Core::File::is_directory(locale_path));
  368. auto language = remove_variants_from_path(locale_path);
  369. if (!language.has_value())
  370. continue;
  371. auto& locale = locale_data.locales.ensure(*language);
  372. parse_identity(locale_path, locale_data, locale);
  373. }
  374. while (locale_names_iterator.has_next()) {
  375. auto locale_path = locale_names_iterator.next_full_path();
  376. VERIFY(Core::File::is_directory(locale_path));
  377. auto language = remove_variants_from_path(locale_path);
  378. if (!language.has_value())
  379. continue;
  380. auto& locale = locale_data.locales.ensure(*language);
  381. parse_locale_languages(locale_path, locale_data, locale);
  382. parse_locale_territories(locale_path, locale_data, locale);
  383. parse_locale_scripts(locale_path, locale_data, locale);
  384. }
  385. while (misc_iterator.has_next()) {
  386. auto misc_path = misc_iterator.next_full_path();
  387. VERIFY(Core::File::is_directory(misc_path));
  388. auto language = remove_variants_from_path(misc_path);
  389. if (!language.has_value())
  390. continue;
  391. auto& locale = locale_data.locales.ensure(*language);
  392. parse_locale_list_patterns(misc_path, locale_data, locale);
  393. }
  394. while (numbers_iterator.has_next()) {
  395. auto numbers_path = numbers_iterator.next_full_path();
  396. VERIFY(Core::File::is_directory(numbers_path));
  397. auto language = remove_variants_from_path(numbers_path);
  398. if (!language.has_value())
  399. continue;
  400. auto& locale = locale_data.locales.ensure(*language);
  401. parse_locale_currencies(numbers_path, locale_data, locale);
  402. parse_numeric_keywords(numbers_path, locale_data, locale);
  403. }
  404. parse_default_content_locales(move(core_path), locale_data);
  405. }
  406. static String format_identifier(StringView owner, String identifier)
  407. {
  408. identifier = identifier.replace("-"sv, "_"sv, true);
  409. if (all_of(identifier, is_ascii_digit))
  410. return String::formatted("{}_{}", owner[0], identifier);
  411. if (is_ascii_lower_alpha(identifier[0]))
  412. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  413. return identifier;
  414. }
  415. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  416. {
  417. StringBuilder builder;
  418. SourceGenerator generator { builder };
  419. generator.append(R"~~~(
  420. #pragma once
  421. #include <AK/Optional.h>
  422. #include <AK/StringView.h>
  423. #include <AK/Types.h>
  424. #include <AK/Vector.h>
  425. #include <LibUnicode/Forward.h>
  426. namespace Unicode {
  427. )~~~");
  428. auto locales = locale_data.locales.keys();
  429. generate_enum(generator, format_identifier, "Locale"sv, "None"sv, locales);
  430. generate_enum(generator, format_identifier, "Language"sv, {}, locale_data.languages);
  431. generate_enum(generator, format_identifier, "Territory"sv, {}, locale_data.territories);
  432. generate_enum(generator, format_identifier, "ScriptTag"sv, {}, locale_data.scripts);
  433. generate_enum(generator, format_identifier, "Currency"sv, {}, locale_data.currencies);
  434. generate_enum(generator, format_identifier, "Key"sv, {}, locale_data.keywords);
  435. generate_enum(generator, format_identifier, "Variant"sv, {}, locale_data.variants);
  436. generate_enum(generator, format_identifier, "ListPatternType"sv, {}, locale_data.list_pattern_types);
  437. generate_enum(generator, format_identifier, "ListPatternStyle"sv, {}, locale_data.list_pattern_styles);
  438. generator.append(R"~~~(
  439. namespace Detail {
  440. Optional<Locale> locale_from_string(StringView locale);
  441. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  442. Optional<Language> language_from_string(StringView language);
  443. Optional<StringView> resolve_language_alias(StringView language);
  444. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  445. Optional<Territory> territory_from_string(StringView territory);
  446. Optional<StringView> resolve_territory_alias(StringView territory);
  447. Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
  448. Optional<ScriptTag> script_tag_from_string(StringView script_tag);
  449. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  450. Optional<StringView> get_locale_currency_mapping(StringView locale, StringView currency);
  451. Optional<Currency> currency_from_string(StringView currency);
  452. Optional<StringView> get_locale_key_mapping(StringView locale, StringView key);
  453. Optional<Key> key_from_string(StringView key);
  454. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style);
  455. Optional<ListPatternType> list_pattern_type_from_string(StringView list_pattern_type);
  456. Optional<ListPatternStyle> list_pattern_style_from_string(StringView list_pattern_style);
  457. Optional<StringView> resolve_variant_alias(StringView variant);
  458. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  459. void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
  460. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id);
  461. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
  462. }
  463. }
  464. )~~~");
  465. file.write(generator.as_string_view());
  466. }
  467. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  468. {
  469. StringBuilder builder;
  470. SourceGenerator generator { builder };
  471. generator.set("string_index_type"sv, s_string_index_type);
  472. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  473. generator.set("territories_size", String::number(locale_data.territories.size()));
  474. generator.set("variants_size", String::number(locale_data.max_variant_size));
  475. generator.append(R"~~~(
  476. #include <AK/Array.h>
  477. #include <AK/BinarySearch.h>
  478. #include <AK/Span.h>
  479. #include <LibUnicode/Locale.h>
  480. #include <LibUnicode/UnicodeLocale.h>
  481. namespace Unicode::Detail {
  482. struct Patterns {
  483. ListPatternType type;
  484. ListPatternStyle style;
  485. @string_index_type@ start { 0 };
  486. @string_index_type@ middle { 0 };
  487. @string_index_type@ end { 0 };
  488. @string_index_type@ pair { 0 };
  489. };
  490. )~~~");
  491. locale_data.unique_strings.generate(generator);
  492. auto append_index = [&](auto index) {
  493. generator.append(String::formatted(", {}", index));
  494. };
  495. auto append_list_and_size = [&](auto const& list) {
  496. if (list.is_empty()) {
  497. generator.append(", {}, 0");
  498. return;
  499. }
  500. bool first = true;
  501. generator.append(", {");
  502. for (auto const& item : list) {
  503. generator.append(first ? " " : ", ");
  504. generator.append(String::number(item));
  505. first = false;
  506. }
  507. generator.append(String::formatted(" }}, {}", list.size()));
  508. };
  509. auto append_string_index_list = [&](String name, auto const& keys, auto const& mappings) {
  510. generator.set("name", name);
  511. generator.set("size", String::number(keys.size()));
  512. generator.append(R"~~~(
  513. static constexpr Array<@string_index_type@, @size@> @name@ { {
  514. )~~~");
  515. constexpr size_t max_values_per_row = 30;
  516. size_t values_in_current_row = 0;
  517. for (auto const& key : keys) {
  518. if (values_in_current_row++ > 0)
  519. generator.append(" ");
  520. if (auto it = mappings.find(key); it != mappings.end())
  521. generator.set("mapping"sv, String::number(it->value));
  522. else
  523. generator.set("mapping"sv, "0"sv);
  524. generator.append("@mapping@,");
  525. if (values_in_current_row == max_values_per_row) {
  526. values_in_current_row = 0;
  527. generator.append("\n ");
  528. }
  529. }
  530. generator.append(R"~~~(
  531. } };
  532. )~~~");
  533. };
  534. auto append_list_patterns = [&](StringView name, Vector<ListPatterns> const& list_patterns) {
  535. generator.set("name", name);
  536. generator.set("size", String::number(list_patterns.size()));
  537. generator.append(R"~~~(
  538. static constexpr Array<Patterns, @size@> @name@ { {)~~~");
  539. for (auto const& list_pattern : list_patterns) {
  540. generator.set("type"sv, String::formatted("ListPatternType::{}", format_identifier({}, list_pattern.type)));
  541. generator.set("style"sv, String::formatted("ListPatternStyle::{}", format_identifier({}, list_pattern.style)));
  542. generator.set("start"sv, String::number(list_pattern.start));
  543. generator.set("middle"sv, String::number(list_pattern.middle));
  544. generator.set("end"sv, String::number(list_pattern.end));
  545. generator.set("pair"sv, String::number(list_pattern.pair));
  546. generator.append(R"~~~(
  547. { @type@, @style@, @start@, @middle@, @end@, @pair@ },)~~~");
  548. }
  549. generator.append(R"~~~(
  550. } };
  551. )~~~");
  552. };
  553. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_languages"sv, "s_languages_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.languages, value.languages); });
  554. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_territories"sv, "s_territories_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.territories, value.territories); });
  555. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_scripts"sv, "s_scripts_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.scripts, value.scripts); });
  556. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_currencies"sv, "s_currencies_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.currencies, value.currencies); });
  557. generate_mapping(generator, locale_data.locales, s_string_index_type, "s_keywords"sv, "s_keywords_{}", [&](auto const& name, auto const& value) { append_string_index_list(name, locale_data.keywords, value.keywords); });
  558. generate_mapping(generator, locale_data.locales, "Patterns"sv, "s_list_patterns"sv, "s_list_patterns_{}", [&](auto const& name, auto const& value) { append_list_patterns(name, value.list_patterns); });
  559. generator.append(R"~~~(
  560. struct CanonicalLanguageID {
  561. Unicode::LanguageID to_unicode_language_id() const
  562. {
  563. Unicode::LanguageID language_id {};
  564. language_id.variants.ensure_capacity(variants_size);
  565. language_id.language = s_string_list[language];
  566. if (script != 0)
  567. language_id.script = s_string_list[script];
  568. if (region != 0)
  569. language_id.region = s_string_list[region];
  570. for (size_t i = 0; i < variants_size; ++i)
  571. language_id.variants.append(s_string_list[variants[i]]);
  572. return language_id;
  573. }
  574. bool matches_variants(Vector<String> const& other_variants) const {
  575. if (variants_size == 0)
  576. return true;
  577. if (other_variants.size() != variants_size)
  578. return false;
  579. for (size_t i = 0; i < variants_size; ++i) {
  580. if (s_string_list[variants[i]] != other_variants[i])
  581. return false;
  582. }
  583. return true;
  584. };
  585. @string_index_type@ language { 0 };
  586. @string_index_type@ script { 0 };
  587. @string_index_type@ region { 0 };
  588. Array<@string_index_type@, @variants_size@> variants {};
  589. size_t variants_size { 0 };
  590. };
  591. struct LanguageMapping {
  592. CanonicalLanguageID key;
  593. CanonicalLanguageID alias;
  594. };
  595. )~~~");
  596. auto append_complex_mapping = [&](StringView name, auto& mappings) {
  597. generator.set("size", String::number(mappings.size()));
  598. generator.set("name"sv, name);
  599. generator.append(R"~~~(
  600. static constexpr Array<LanguageMapping, @size@> s_@name@ { {
  601. )~~~");
  602. quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
  603. auto const& lhs_language = locale_data.unique_strings.get(lhs.key.language);
  604. auto const& rhs_language = locale_data.unique_strings.get(rhs.key.language);
  605. // Sort the keys such that "und" language tags are at the end, as those are less specific.
  606. if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
  607. return false;
  608. if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
  609. return true;
  610. return lhs_language < rhs_language;
  611. });
  612. for (auto const& mapping : mappings) {
  613. generator.set("language"sv, String::number(mapping.key.language));
  614. generator.append(" { { @language@");
  615. append_index(mapping.key.script);
  616. append_index(mapping.key.region);
  617. append_list_and_size(mapping.key.variants);
  618. generator.set("language"sv, String::number(mapping.alias.language));
  619. generator.append(" }, { @language@");
  620. append_index(mapping.alias.script);
  621. append_index(mapping.alias.region);
  622. append_list_and_size(mapping.alias.variants);
  623. generator.append(" } },\n");
  624. }
  625. generator.append("} };\n");
  626. };
  627. append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
  628. append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
  629. generator.append(R"~~~(
  630. static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& language_id)
  631. {
  632. // https://unicode.org/reports/tr35/#Likely_Subtags
  633. enum class State {
  634. LanguageScriptRegion,
  635. LanguageRegion,
  636. LanguageScript,
  637. Language,
  638. UndScript,
  639. Done,
  640. };
  641. auto state = State::LanguageScriptRegion;
  642. while (state != State::Done) {
  643. Unicode::LanguageID search_key;
  644. switch (state) {
  645. case State::LanguageScriptRegion:
  646. state = State::LanguageRegion;
  647. if (!language_id.script.has_value() || !language_id.region.has_value())
  648. continue;
  649. search_key.language = *language_id.language;
  650. search_key.script = *language_id.script;
  651. search_key.region = *language_id.region;
  652. break;
  653. case State::LanguageRegion:
  654. state = State::LanguageScript;
  655. if (!language_id.region.has_value())
  656. continue;
  657. search_key.language = *language_id.language;
  658. search_key.region = *language_id.region;
  659. break;
  660. case State::LanguageScript:
  661. state = State::Language;
  662. if (!language_id.script.has_value())
  663. continue;
  664. search_key.language = *language_id.language;
  665. search_key.script = *language_id.script;
  666. break;
  667. case State::Language:
  668. state = State::UndScript;
  669. search_key.language = *language_id.language;
  670. break;
  671. case State::UndScript:
  672. state = State::Done;
  673. if (!language_id.script.has_value())
  674. continue;
  675. search_key.language = "und"sv;
  676. search_key.script = *language_id.script;
  677. break;
  678. default:
  679. VERIFY_NOT_REACHED();
  680. }
  681. for (auto const& map : s_likely_subtags) {
  682. auto const& key_language = s_string_list[map.key.language];
  683. auto const& key_script = s_string_list[map.key.script];
  684. auto const& key_region = s_string_list[map.key.region];
  685. if (key_language != search_key.language)
  686. continue;
  687. if (!key_script.is_empty() || search_key.script.has_value()) {
  688. if (key_script != search_key.script)
  689. continue;
  690. }
  691. if (!key_region.is_empty() || search_key.region.has_value()) {
  692. if (key_region != search_key.region)
  693. continue;
  694. }
  695. return &map;
  696. }
  697. }
  698. return nullptr;
  699. }
  700. )~~~");
  701. auto append_mapping_search = [&](StringView enum_title, StringView enum_snake, StringView collection_name) {
  702. generator.set("enum_title", enum_title);
  703. generator.set("enum_snake", enum_snake);
  704. generator.set("collection_name", collection_name);
  705. generator.append(R"~~~(
  706. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  707. {
  708. auto locale_value = locale_from_string(locale);
  709. if (!locale_value.has_value())
  710. return {};
  711. auto @enum_snake@_value = @enum_snake@_from_string(@enum_snake@);
  712. if (!@enum_snake@_value.has_value())
  713. return {};
  714. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  715. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  716. auto const& mappings = @collection_name@.at(locale_index);
  717. auto @enum_snake@_string_index = mappings.at(@enum_snake@_index);
  718. auto @enum_snake@_mapping = s_string_list.at(@enum_snake@_string_index);
  719. if (@enum_snake@_mapping.is_empty())
  720. return {};
  721. return @enum_snake@_mapping;
  722. }
  723. )~~~");
  724. };
  725. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values) {
  726. HashValueMap<String> hashes;
  727. hashes.ensure_capacity(values.size());
  728. for (auto const& value : values)
  729. hashes.set(value.hash(), format_identifier(enum_title, value));
  730. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  731. };
  732. auto append_alias_search = [&](StringView enum_snake, auto const& aliases) {
  733. HashValueMap<StringIndexType> hashes;
  734. hashes.ensure_capacity(aliases.size());
  735. for (auto const& alias : aliases)
  736. hashes.set(alias.key.hash(), alias.value);
  737. generate_value_from_string(generator, "resolve_{}_alias"sv, s_string_index_type, enum_snake, move(hashes), "StringView"sv, "s_string_list[{}]"sv);
  738. };
  739. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys());
  740. append_mapping_search("Language"sv, "language"sv, "s_languages"sv);
  741. append_from_string("Language"sv, "language"sv, locale_data.languages);
  742. append_alias_search("language"sv, locale_data.language_aliases);
  743. append_mapping_search("Territory"sv, "territory"sv, "s_territories"sv);
  744. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  745. append_alias_search("territory"sv, locale_data.territory_aliases);
  746. append_mapping_search("ScriptTag"sv, "script_tag"sv, "s_scripts"sv);
  747. append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
  748. append_alias_search("script_tag"sv, locale_data.script_aliases);
  749. append_mapping_search("Currency"sv, "currency"sv, "s_currencies"sv);
  750. append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
  751. append_mapping_search("Key"sv, "key"sv, "s_keywords"sv);
  752. append_from_string("Key"sv, "key"sv, locale_data.keywords);
  753. append_alias_search("variant"sv, locale_data.variant_aliases);
  754. append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
  755. append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
  756. append_from_string("ListPatternStyle"sv, "list_pattern_style"sv, locale_data.list_pattern_styles);
  757. generator.append(R"~~~(
  758. Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style)
  759. {
  760. auto locale_value = locale_from_string(locale);
  761. if (!locale_value.has_value())
  762. return {};
  763. auto type_value = list_pattern_type_from_string(list_pattern_type);
  764. if (!type_value.has_value())
  765. return {};
  766. auto style_value = list_pattern_style_from_string(list_pattern_style);
  767. if (!style_value.has_value())
  768. return {};
  769. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  770. auto const& locale_list_patterns = s_list_patterns.at(locale_index);
  771. for (auto const& list_patterns : locale_list_patterns) {
  772. if ((list_patterns.type == type_value) && (list_patterns.style == style_value)) {
  773. auto const& start = s_string_list[list_patterns.start];
  774. auto const& middle = s_string_list[list_patterns.middle];
  775. auto const& end = s_string_list[list_patterns.end];
  776. auto const& pair = s_string_list[list_patterns.pair];
  777. return ListPatterns { start, middle, end, pair };
  778. }
  779. }
  780. return {};
  781. }
  782. void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
  783. {
  784. for (auto const& map : s_complex_alias) {
  785. auto const& key_language = s_string_list[map.key.language];
  786. auto const& key_script = s_string_list[map.key.script];
  787. auto const& key_region = s_string_list[map.key.region];
  788. if ((key_language != language_id.language) && (key_language != "und"sv))
  789. continue;
  790. if (!key_script.is_empty() && (key_script != language_id.script))
  791. continue;
  792. if (!key_region.is_empty() && (key_region != language_id.region))
  793. continue;
  794. if (!map.key.matches_variants(language_id.variants))
  795. continue;
  796. auto alias = map.alias.to_unicode_language_id();
  797. if (alias.language == "und"sv)
  798. alias.language = move(language_id.language);
  799. if (key_script.is_empty() && !alias.script.has_value())
  800. alias.script = move(language_id.script);
  801. if (key_region.is_empty() && !alias.region.has_value())
  802. alias.region = move(language_id.region);
  803. if (map.key.variants_size == 0 && alias.variants.is_empty())
  804. alias.variants = move(language_id.variants);
  805. language_id = move(alias);
  806. break;
  807. }
  808. }
  809. Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id)
  810. {
  811. // https://www.unicode.org/reports/tr35/#Likely_Subtags
  812. auto const* likely_subtag = resolve_likely_subtag(language_id);
  813. if (likely_subtag == nullptr)
  814. return {};
  815. auto maximized = language_id;
  816. auto const& key_script = s_string_list[likely_subtag->key.script];
  817. auto const& key_region = s_string_list[likely_subtag->key.region];
  818. auto const& alias_language = s_string_list[likely_subtag->alias.language];
  819. auto const& alias_script = s_string_list[likely_subtag->alias.script];
  820. auto const& alias_region = s_string_list[likely_subtag->alias.region];
  821. if (maximized.language == "und"sv)
  822. maximized.language = alias_language;
  823. if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
  824. maximized.script = alias_script;
  825. if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
  826. maximized.region = alias_region;
  827. return maximized;
  828. }
  829. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
  830. {
  831. if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
  832. return s_string_list[likely_subtag->alias.region];
  833. return {};
  834. }
  835. }
  836. )~~~");
  837. file.write(generator.as_string_view());
  838. }
  839. int main(int argc, char** argv)
  840. {
  841. char const* generated_header_path = nullptr;
  842. char const* generated_implementation_path = nullptr;
  843. char const* core_path = nullptr;
  844. char const* locale_names_path = nullptr;
  845. char const* misc_path = nullptr;
  846. char const* numbers_path = nullptr;
  847. Core::ArgsParser args_parser;
  848. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  849. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  850. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  851. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  852. args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
  853. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  854. args_parser.parse(argc, argv);
  855. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  856. if (path.is_empty()) {
  857. warnln("{} is required", flags);
  858. args_parser.print_usage(stderr, argv[0]);
  859. exit(1);
  860. }
  861. auto file_or_error = Core::File::open(path, mode);
  862. if (file_or_error.is_error()) {
  863. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  864. exit(1);
  865. }
  866. return file_or_error.release_value();
  867. };
  868. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  869. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  870. UnicodeLocaleData locale_data;
  871. parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, locale_data);
  872. generate_unicode_locale_header(generated_header_file, locale_data);
  873. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  874. return 0;
  875. }