GenerateUnicodeLocale.cpp 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409
  1. /*
  2. * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <AK/AllOf.h>
  7. #include <AK/CharacterTypes.h>
  8. #include <AK/Format.h>
  9. #include <AK/HashMap.h>
  10. #include <AK/JsonObject.h>
  11. #include <AK/JsonParser.h>
  12. #include <AK/JsonValue.h>
  13. #include <AK/LexicalPath.h>
  14. #include <AK/QuickSort.h>
  15. #include <AK/SourceGenerator.h>
  16. #include <AK/String.h>
  17. #include <AK/StringBuilder.h>
  18. #include <LibCore/ArgsParser.h>
  19. #include <LibCore/DirIterator.h>
  20. #include <LibCore/File.h>
  21. struct Locale {
  22. String language;
  23. Optional<String> territory;
  24. Optional<String> variant;
  25. HashMap<String, String> territories;
  26. };
  27. struct UnicodeLocaleData {
  28. HashMap<String, Locale> locales;
  29. Vector<String> languages;
  30. Vector<String> territories;
  31. Vector<String> variants;
  32. };
  33. static void write_to_file_if_different(Core::File& file, StringView contents)
  34. {
  35. auto const current_contents = file.read_all();
  36. if (StringView { current_contents.bytes() } == contents)
  37. return;
  38. VERIFY(file.seek(0));
  39. VERIFY(file.truncate(0));
  40. VERIFY(file.write(contents));
  41. }
  42. static void parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
  43. {
  44. LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
  45. languages_path = languages_path.append("languages.json"sv);
  46. VERIFY(Core::File::exists(languages_path.string()));
  47. auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
  48. VERIFY(!languages_file_or_error.is_error());
  49. auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
  50. VERIFY(languages.has_value());
  51. auto const& main_object = languages->as_object().get("main"sv);
  52. auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
  53. auto const& identity_object = locale_object.as_object().get("identity"sv);
  54. auto const& language_string = identity_object.as_object().get("language"sv);
  55. auto const& territory_string = identity_object.as_object().get("territory"sv);
  56. auto const& variant_string = identity_object.as_object().get("variant"sv);
  57. locale.language = language_string.as_string();
  58. if (!locale_data.languages.contains_slow(locale.language))
  59. locale_data.languages.append(locale.language);
  60. if (territory_string.is_string()) {
  61. locale.territory = territory_string.as_string();
  62. if (!locale_data.territories.contains_slow(*locale.territory))
  63. locale_data.territories.append(*locale.territory);
  64. }
  65. if (variant_string.is_string()) {
  66. locale.variant = variant_string.as_string();
  67. if (!locale_data.variants.contains_slow(*locale.variant))
  68. locale_data.variants.append(*locale.variant);
  69. }
  70. }
  71. static void parse_locale_territories(String locale_path, Locale& locale)
  72. {
  73. LexicalPath territories_path(move(locale_path));
  74. territories_path = territories_path.append("territories.json"sv);
  75. VERIFY(Core::File::exists(territories_path.string()));
  76. auto territories_file_or_error = Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly);
  77. VERIFY(!territories_file_or_error.is_error());
  78. auto territories = JsonParser(territories_file_or_error.value()->read_all()).parse();
  79. VERIFY(territories.has_value());
  80. auto const& main_object = territories->as_object().get("main"sv);
  81. auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
  82. auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
  83. auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
  84. territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  85. locale.territories.set(key, value.as_string());
  86. });
  87. }
  88. static void parse_all_locales(String locale_names_path, UnicodeLocaleData& locale_data)
  89. {
  90. LexicalPath locale_names(move(locale_names_path));
  91. locale_names = locale_names.append("main"sv);
  92. VERIFY(Core::File::is_directory(locale_names.string()));
  93. Core::DirIterator iterator(locale_names.string(), Core::DirIterator::SkipParentAndBaseDir);
  94. if (iterator.has_error()) {
  95. warnln("{}: {}", locale_names.string(), iterator.error_string());
  96. VERIFY_NOT_REACHED();
  97. }
  98. while (iterator.has_next()) {
  99. auto locale_path = iterator.next_full_path();
  100. VERIFY(Core::File::is_directory(locale_path));
  101. auto& locale = locale_data.locales.ensure(LexicalPath::basename(locale_path));
  102. parse_identity(locale_path, locale_data, locale);
  103. parse_locale_territories(locale_path, locale);
  104. }
  105. }
  106. static String format_identifier(StringView owner, String identifier)
  107. {
  108. identifier.replace("-"sv, "_"sv, true);
  109. if (all_of(identifier, is_ascii_digit))
  110. return String::formatted("{}_{}", owner[0], identifier);
  111. return identifier.to_titlecase();
  112. }
  113. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  114. {
  115. StringBuilder builder;
  116. SourceGenerator generator { builder };
  117. auto generate_enum = [&](StringView name, StringView default_, Vector<String>& values) {
  118. quick_sort(values);
  119. generator.set("name", name);
  120. generator.set("underlying", ((values.size() + !default_.is_empty()) < 256) ? "u8"sv : "u16"sv);
  121. generator.append(R"~~~(
  122. enum class @name@ : @underlying@ {)~~~");
  123. if (!default_.is_empty()) {
  124. generator.set("default", default_);
  125. generator.append(R"~~~(
  126. @default@,)~~~");
  127. }
  128. for (auto const& value : values) {
  129. generator.set("value", format_identifier(name, value));
  130. generator.append(R"~~~(
  131. @value@,)~~~");
  132. }
  133. generator.append(R"~~~(
  134. };
  135. )~~~");
  136. };
  137. generator.append(R"~~~(
  138. #pragma once
  139. #include <AK/Optional.h>
  140. #include <AK/StringView.h>
  141. #include <AK/Types.h>
  142. #include <LibUnicode/Forward.h>
  143. namespace Unicode {
  144. )~~~");
  145. auto locales = locale_data.locales.keys();
  146. generate_enum("Locale"sv, "None"sv, locales);
  147. generate_enum("Language"sv, {}, locale_data.languages);
  148. generate_enum("Territory"sv, {}, locale_data.territories);
  149. generate_enum("Variant"sv, {}, locale_data.variants);
  150. generator.append(R"~~~(
  151. namespace Detail {
  152. Optional<Locale> locale_from_string(StringView const& locale);
  153. Optional<Language> language_from_string(StringView const& language);
  154. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  155. Optional<Territory> territory_from_string(StringView const& territory);
  156. }
  157. }
  158. )~~~");
  159. write_to_file_if_different(file, generator.as_string_view());
  160. }
  161. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  162. {
  163. StringBuilder builder;
  164. SourceGenerator generator { builder };
  165. generator.set("locales_size"sv, String::number(locale_data.locales.size()));
  166. generator.set("territories_size", String::number(locale_data.territories.size()));
  167. generator.append(R"~~~(
  168. #include <AK/Array.h>
  169. #include <AK/HashMap.h>
  170. #include <AK/Span.h>
  171. #include <LibUnicode/UnicodeLocale.h>
  172. namespace Unicode {
  173. )~~~");
  174. auto format_mapping_name = [](StringView format, StringView name) {
  175. auto mapping_name = name.to_lowercase_string();
  176. mapping_name.replace("-"sv, "_"sv, true);
  177. return String::formatted(format, mapping_name);
  178. };
  179. auto append_mapping_list = [&](String name, auto const& keys, auto const& mappings) {
  180. generator.set("name", name);
  181. generator.set("size", String::number(keys.size()));
  182. generator.append(R"~~~(
  183. static constexpr Array<StringView, @size@> @name@ { {
  184. )~~~");
  185. constexpr size_t max_values_per_row = 10;
  186. size_t values_in_current_row = 0;
  187. for (auto const& key : keys) {
  188. if (values_in_current_row++ > 0)
  189. generator.append(" ");
  190. if (auto it = mappings.find(key); it != mappings.end())
  191. generator.set("mapping"sv, String::formatted("\"{}\"sv", it->value));
  192. else
  193. generator.set("mapping"sv, "{}"sv);
  194. generator.append("@mapping@,");
  195. if (values_in_current_row == max_values_per_row) {
  196. values_in_current_row = 0;
  197. generator.append("\n ");
  198. }
  199. }
  200. generator.append(R"~~~(
  201. } };
  202. )~~~");
  203. };
  204. auto append_mapping = [&](StringView name, StringView format, auto const& keys, auto get_mapping_callback) {
  205. Vector<String> mapping_names;
  206. for (auto const& locale : locale_data.locales) {
  207. auto mapping_name = format_mapping_name(format, locale.key);
  208. append_mapping_list(mapping_name, keys, get_mapping_callback(locale.value));
  209. mapping_names.append(move(mapping_name));
  210. }
  211. quick_sort(mapping_names);
  212. generator.set("name", name);
  213. generator.set("size", String::number(locale_data.locales.size()));
  214. generator.append(R"~~~(
  215. static constexpr Array<Span<StringView const>, @size@> @name@ { {
  216. )~~~");
  217. constexpr size_t max_values_per_row = 10;
  218. size_t values_in_current_row = 0;
  219. for (auto& mapping_name : mapping_names) {
  220. if (values_in_current_row++ > 0)
  221. generator.append(" ");
  222. generator.set("name", move(mapping_name));
  223. generator.append("@name@.span(),");
  224. if (values_in_current_row == max_values_per_row) {
  225. values_in_current_row = 0;
  226. generator.append("\n ");
  227. }
  228. }
  229. generator.append(R"~~~(
  230. } };
  231. )~~~");
  232. };
  233. append_mapping("s_territories"sv, "s_territories_{}", locale_data.territories, [](auto const& value) { return value.territories; });
  234. generator.append(R"~~~(
  235. namespace Detail {
  236. )~~~");
  237. auto append_mapping_search = [&](StringView enum_title, StringView enum_snake, StringView collection_name) {
  238. generator.set("enum_title", enum_title);
  239. generator.set("enum_snake", enum_snake);
  240. generator.set("collection_name", collection_name);
  241. generator.append(R"~~~(
  242. Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
  243. {
  244. auto locale_value = locale_from_string(locale);
  245. if (!locale_value.has_value())
  246. return {};
  247. auto @enum_snake@_value = @enum_snake@_from_string(@enum_snake@);
  248. if (!@enum_snake@_value.has_value())
  249. return {};
  250. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  251. auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
  252. auto const& mappings = @collection_name@.at(locale_index);
  253. auto @enum_snake@_mapping = mappings.at(@enum_snake@_index);
  254. if (@enum_snake@_mapping.is_empty())
  255. return {};
  256. return @enum_snake@_mapping;
  257. }
  258. )~~~");
  259. };
  260. auto append_from_string = [&](StringView enum_title, StringView enum_snake, Vector<String> const& values) {
  261. generator.set("enum_title", enum_title);
  262. generator.set("enum_snake", enum_snake);
  263. generator.append(R"~~~(
  264. Optional<@enum_title@> @enum_snake@_from_string(StringView const& @enum_snake@)
  265. {
  266. static HashMap<StringView, @enum_title@> @enum_snake@_values { {)~~~");
  267. for (auto const& value : values) {
  268. generator.set("key"sv, value);
  269. generator.set("value"sv, format_identifier(enum_title, value));
  270. generator.append(R"~~~(
  271. { "@key@"sv, @enum_title@::@value@ },)~~~");
  272. }
  273. generator.append(R"~~~(
  274. } };
  275. if (auto value = @enum_snake@_values.get(@enum_snake@); value.has_value())
  276. return value.value();
  277. return {};
  278. }
  279. )~~~");
  280. };
  281. append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys());
  282. append_from_string("Language"sv, "language"sv, locale_data.languages);
  283. append_mapping_search("Territory"sv, "territory"sv, "s_territories"sv);
  284. append_from_string("Territory"sv, "territory"sv, locale_data.territories);
  285. generator.append(R"~~~(
  286. }
  287. }
  288. )~~~");
  289. write_to_file_if_different(file, generator.as_string_view());
  290. }
  291. int main(int argc, char** argv)
  292. {
  293. char const* generated_header_path = nullptr;
  294. char const* generated_implementation_path = nullptr;
  295. char const* locale_names_path = nullptr;
  296. Core::ArgsParser args_parser;
  297. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  298. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  299. args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
  300. args_parser.parse(argc, argv);
  301. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  302. if (path.is_empty()) {
  303. warnln("{} is required", flags);
  304. args_parser.print_usage(stderr, argv[0]);
  305. exit(1);
  306. }
  307. auto file_or_error = Core::File::open(path, mode);
  308. if (file_or_error.is_error()) {
  309. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  310. exit(1);
  311. }
  312. return file_or_error.release_value();
  313. };
  314. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  315. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  316. UnicodeLocaleData locale_data;
  317. parse_all_locales(locale_names_path, locale_data);
  318. generate_unicode_locale_header(generated_header_file, locale_data);
  319. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  320. return 0;
  321. }