GenerateNumberFormatData.cpp 16 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436
  1. /*
  2. * Copyright (c) 2021-2023, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "../LibUnicode/GeneratorUtil.h" // FIXME: Move this somewhere common.
  7. #include <AK/Array.h>
  8. #include <AK/ByteString.h>
  9. #include <AK/Format.h>
  10. #include <AK/HashFunctions.h>
  11. #include <AK/HashMap.h>
  12. #include <AK/JsonObject.h>
  13. #include <AK/JsonParser.h>
  14. #include <AK/JsonValue.h>
  15. #include <AK/LexicalPath.h>
  16. #include <AK/QuickSort.h>
  17. #include <AK/SourceGenerator.h>
  18. #include <AK/StringBuilder.h>
  19. #include <AK/Traits.h>
  20. #include <AK/Utf8View.h>
  21. #include <LibCore/ArgsParser.h>
  22. #include <LibCore/Directory.h>
  23. #include <LibFileSystem/FileSystem.h>
  24. #include <LibLocale/NumberFormat.h>
  25. #include <math.h>
  26. using NumericSymbolList = Vector<size_t>;
  27. struct NumberSystem {
  28. unsigned hash() const { return int_hash(symbols); }
  29. bool operator==(NumberSystem const& other) const { return (symbols == other.symbols); }
  30. size_t symbols { 0 };
  31. };
  32. template<>
  33. struct AK::Formatter<NumberSystem> : Formatter<FormatString> {
  34. ErrorOr<void> format(FormatBuilder& builder, NumberSystem const& system)
  35. {
  36. return Formatter<FormatString>::format(builder, "{{ {} }}"sv, system.symbols);
  37. }
  38. };
  39. template<>
  40. struct AK::Traits<NumberSystem> : public DefaultTraits<NumberSystem> {
  41. static unsigned hash(NumberSystem const& s) { return s.hash(); }
  42. };
  43. struct LocaleData {
  44. Vector<size_t> number_systems;
  45. };
  46. struct CLDR {
  47. UniqueStringStorage unique_strings;
  48. UniqueStorage<NumericSymbolList> unique_symbols;
  49. UniqueStorage<NumberSystem> unique_systems;
  50. HashMap<ByteString, Array<u32, 10>> number_system_digits;
  51. Vector<ByteString> number_systems;
  52. HashMap<ByteString, LocaleData> locales;
  53. };
  54. static ErrorOr<void> parse_number_system_digits(ByteString core_supplemental_path, CLDR& cldr)
  55. {
  56. LexicalPath number_systems_path(move(core_supplemental_path));
  57. number_systems_path = number_systems_path.append("numberingSystems.json"sv);
  58. auto number_systems = TRY(read_json_file(number_systems_path.string()));
  59. auto const& supplemental_object = number_systems.as_object().get_object("supplemental"sv).value();
  60. auto const& number_systems_object = supplemental_object.get_object("numberingSystems"sv).value();
  61. number_systems_object.for_each_member([&](auto const& number_system, auto const& digits_object) {
  62. auto type = digits_object.as_object().get_byte_string("_type"sv).value();
  63. if (type != "numeric"sv)
  64. return;
  65. auto digits = digits_object.as_object().get_byte_string("_digits"sv).value();
  66. Utf8View utf8_digits { digits };
  67. VERIFY(utf8_digits.length() == 10);
  68. auto& number_system_digits = cldr.number_system_digits.ensure(number_system);
  69. size_t index = 0;
  70. for (u32 digit : utf8_digits)
  71. number_system_digits[index++] = digit;
  72. if (!cldr.number_systems.contains_slow(number_system))
  73. cldr.number_systems.append(number_system);
  74. });
  75. return {};
  76. }
  77. static ErrorOr<void> parse_number_systems(ByteString locale_numbers_path, CLDR& cldr, LocaleData& locale)
  78. {
  79. LexicalPath numbers_path(move(locale_numbers_path));
  80. numbers_path = numbers_path.append("numbers.json"sv);
  81. auto numbers = TRY(read_json_file(numbers_path.string()));
  82. auto const& main_object = numbers.as_object().get_object("main"sv).value();
  83. auto const& locale_object = main_object.get_object(numbers_path.parent().basename()).value();
  84. auto const& locale_numbers_object = locale_object.get_object("numbers"sv).value();
  85. auto const& minimum_grouping_digits = locale_numbers_object.get_byte_string("minimumGroupingDigits"sv).value();
  86. Vector<Optional<NumberSystem>> number_systems;
  87. number_systems.resize(cldr.number_systems.size());
  88. auto ensure_number_system = [&](auto const& system) -> NumberSystem& {
  89. auto system_index = cldr.number_systems.find_first_index(system).value();
  90. VERIFY(system_index < number_systems.size());
  91. auto& number_system = number_systems.at(system_index);
  92. if (!number_system.has_value())
  93. number_system = NumberSystem {};
  94. return number_system.value();
  95. };
  96. auto numeric_symbol_from_string = [&](StringView numeric_symbol) -> Optional<Locale::NumericSymbol> {
  97. if (numeric_symbol == "approximatelySign"sv)
  98. return Locale::NumericSymbol::ApproximatelySign;
  99. if (numeric_symbol == "decimal"sv)
  100. return Locale::NumericSymbol::Decimal;
  101. if (numeric_symbol == "exponential"sv)
  102. return Locale::NumericSymbol::Exponential;
  103. if (numeric_symbol == "group"sv)
  104. return Locale::NumericSymbol::Group;
  105. if (numeric_symbol == "infinity"sv)
  106. return Locale::NumericSymbol::Infinity;
  107. if (numeric_symbol == "minusSign"sv)
  108. return Locale::NumericSymbol::MinusSign;
  109. if (numeric_symbol == "nan"sv)
  110. return Locale::NumericSymbol::NaN;
  111. if (numeric_symbol == "percentSign"sv)
  112. return Locale::NumericSymbol::PercentSign;
  113. if (numeric_symbol == "plusSign"sv)
  114. return Locale::NumericSymbol::PlusSign;
  115. if (numeric_symbol == "timeSeparator"sv)
  116. return Locale::NumericSymbol::TimeSeparator;
  117. return {};
  118. };
  119. locale_numbers_object.for_each_member([&](auto const& key, JsonValue const& value) {
  120. constexpr auto symbols_prefix = "symbols-numberSystem-"sv;
  121. constexpr auto misc_patterns_prefix = "miscPatterns-numberSystem-"sv;
  122. if (key.starts_with(symbols_prefix)) {
  123. auto system = key.substring(symbols_prefix.length());
  124. auto& number_system = ensure_number_system(system);
  125. NumericSymbolList symbols;
  126. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& localization) {
  127. auto numeric_symbol = numeric_symbol_from_string(symbol);
  128. if (!numeric_symbol.has_value())
  129. return;
  130. if (to_underlying(*numeric_symbol) >= symbols.size())
  131. symbols.resize(to_underlying(*numeric_symbol) + 1);
  132. auto symbol_index = cldr.unique_strings.ensure(localization.as_string());
  133. symbols[to_underlying(*numeric_symbol)] = symbol_index;
  134. });
  135. // The range separator does not appear in the symbols list, we have to extract it from
  136. // the range pattern.
  137. auto misc_patterns_key = ByteString::formatted("{}{}", misc_patterns_prefix, system);
  138. auto misc_patterns = locale_numbers_object.get_object(misc_patterns_key).value();
  139. auto range_separator = misc_patterns.get_byte_string("range"sv).value();
  140. auto begin_index = range_separator.find("{0}"sv).value() + "{0}"sv.length();
  141. auto end_index = range_separator.find("{1}"sv).value();
  142. range_separator = range_separator.substring(begin_index, end_index - begin_index);
  143. if (to_underlying(Locale::NumericSymbol::RangeSeparator) >= symbols.size())
  144. symbols.resize(to_underlying(Locale::NumericSymbol::RangeSeparator) + 1);
  145. auto symbol_index = cldr.unique_strings.ensure(move(range_separator));
  146. symbols[to_underlying(Locale::NumericSymbol::RangeSeparator)] = symbol_index;
  147. number_system.symbols = cldr.unique_symbols.ensure(move(symbols));
  148. }
  149. });
  150. locale.number_systems.ensure_capacity(number_systems.size());
  151. for (auto& number_system : number_systems) {
  152. size_t system_index = 0;
  153. if (number_system.has_value())
  154. system_index = cldr.unique_systems.ensure(number_system.release_value());
  155. locale.number_systems.append(system_index);
  156. }
  157. // locale.minimum_grouping_digits = minimum_grouping_digits.template to_number<u8>().value();
  158. return {};
  159. }
  160. static ErrorOr<void> parse_all_locales(ByteString core_path, ByteString numbers_path, CLDR& cldr)
  161. {
  162. LexicalPath core_supplemental_path(move(core_path));
  163. core_supplemental_path = core_supplemental_path.append("supplemental"sv);
  164. VERIFY(FileSystem::is_directory(core_supplemental_path.string()));
  165. TRY(parse_number_system_digits(core_supplemental_path.string(), cldr));
  166. auto remove_variants_from_path = [&](ByteString path) -> ErrorOr<ByteString> {
  167. auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, LexicalPath::basename(path)));
  168. StringBuilder builder;
  169. builder.append(cldr.unique_strings.get(parsed_locale.language));
  170. if (auto script = cldr.unique_strings.get(parsed_locale.script); !script.is_empty())
  171. builder.appendff("-{}", script);
  172. if (auto region = cldr.unique_strings.get(parsed_locale.region); !region.is_empty())
  173. builder.appendff("-{}", region);
  174. return builder.to_byte_string();
  175. };
  176. TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", numbers_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
  177. auto numbers_path = LexicalPath::join(directory.path().string(), entry.name).string();
  178. auto language = TRY(remove_variants_from_path(numbers_path));
  179. auto& locale = cldr.locales.ensure(language);
  180. TRY(parse_number_systems(numbers_path, cldr, locale));
  181. return IterationDecision::Continue;
  182. }));
  183. return {};
  184. }
  185. static ByteString format_identifier(StringView, ByteString const& identifier)
  186. {
  187. return identifier.to_titlecase();
  188. }
  189. static ErrorOr<void> generate_unicode_locale_header(Core::InputBufferedFile& file, CLDR& cldr)
  190. {
  191. StringBuilder builder;
  192. SourceGenerator generator { builder };
  193. generator.append(R"~~~(
  194. #pragma once
  195. #include <AK/Types.h>
  196. namespace Locale {
  197. )~~~");
  198. generate_enum(generator, format_identifier, "NumberSystem"sv, {}, cldr.number_systems);
  199. generator.append(R"~~~(
  200. }
  201. )~~~");
  202. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  203. return {};
  204. }
  205. static ErrorOr<void> generate_unicode_locale_implementation(Core::InputBufferedFile& file, CLDR& cldr)
  206. {
  207. StringBuilder builder;
  208. SourceGenerator generator { builder };
  209. generator.set("string_index_type"sv, cldr.unique_strings.type_that_fits());
  210. generator.set("numeric_symbol_list_index_type"sv, cldr.unique_symbols.type_that_fits());
  211. generator.append(R"~~~(
  212. #include <AK/Array.h>
  213. #include <AK/Optional.h>
  214. #include <AK/Span.h>
  215. #include <AK/StringView.h>
  216. #include <LibLocale/Locale.h>
  217. #include <LibLocale/LocaleData.h>
  218. #include <LibLocale/NumberFormat.h>
  219. #include <LibLocale/NumberFormatData.h>
  220. namespace Locale {
  221. )~~~");
  222. cldr.unique_strings.generate(generator);
  223. generator.append(R"~~~(
  224. struct NumberSystemData {
  225. @numeric_symbol_list_index_type@ symbols { 0 };
  226. };
  227. )~~~");
  228. cldr.unique_symbols.generate(generator, cldr.unique_strings.type_that_fits(), "s_numeric_symbol_lists"sv);
  229. cldr.unique_systems.generate(generator, "NumberSystemData"sv, "s_number_systems"sv, 10);
  230. auto locales = cldr.locales.keys();
  231. quick_sort(locales);
  232. auto append_map = [&](ByteString name, auto type, auto const& map) {
  233. generator.set("name", move(name));
  234. generator.set("type", type);
  235. generator.set("size", ByteString::number(map.size()));
  236. generator.append(R"~~~(
  237. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  238. bool first = true;
  239. for (auto const& item : map) {
  240. generator.append(first ? " "sv : ", "sv);
  241. if constexpr (requires { item.value; })
  242. generator.append(ByteString::number(item.value));
  243. else
  244. generator.append(ByteString::number(item));
  245. first = false;
  246. }
  247. generator.append(" } };");
  248. };
  249. generate_mapping(generator, cldr.number_system_digits, "u32"sv, "s_number_systems_digits"sv, "s_number_systems_digits_{}"sv, nullptr, [&](auto const& name, auto const& value) { append_map(name, "u32"sv, value); });
  250. generate_mapping(generator, cldr.locales, cldr.unique_systems.type_that_fits(), "s_locale_number_systems"sv, "s_number_systems_{}"sv, nullptr, [&](auto const& name, auto const& value) { append_map(name, cldr.unique_systems.type_that_fits(), value.number_systems); });
  251. generator.append(R"~~~(
  252. static Optional<NumberSystem> keyword_to_number_system(KeywordNumbers keyword)
  253. {
  254. switch (keyword) {)~~~");
  255. for (auto const& number_system : cldr.number_systems) {
  256. generator.set("name"sv, format_identifier({}, number_system));
  257. generator.append(R"~~~(
  258. case KeywordNumbers::@name@:
  259. return NumberSystem::@name@;)~~~");
  260. }
  261. generator.append(R"~~~(
  262. default:
  263. return {};
  264. }
  265. }
  266. Optional<ReadonlySpan<u32>> get_digits_for_number_system(StringView system)
  267. {
  268. auto number_system_keyword = keyword_nu_from_string(system);
  269. if (!number_system_keyword.has_value())
  270. return {};
  271. auto number_system_value = keyword_to_number_system(*number_system_keyword);
  272. if (!number_system_value.has_value())
  273. return {};
  274. auto number_system_index = to_underlying(*number_system_value);
  275. return s_number_systems_digits[number_system_index];
  276. }
  277. static NumberSystemData const* find_number_system(StringView locale, StringView system)
  278. {
  279. auto locale_value = locale_from_string(locale);
  280. if (!locale_value.has_value())
  281. return nullptr;
  282. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  283. auto const& number_systems = s_locale_number_systems.at(locale_index);
  284. auto lookup_number_system = [&](auto number_system) -> NumberSystemData const* {
  285. auto number_system_keyword = keyword_nu_from_string(number_system);
  286. if (!number_system_keyword.has_value())
  287. return nullptr;
  288. auto number_system_value = keyword_to_number_system(*number_system_keyword);
  289. if (!number_system_value.has_value())
  290. return nullptr;
  291. auto number_system_index = to_underlying(*number_system_value);
  292. number_system_index = number_systems.at(number_system_index);
  293. if (number_system_index == 0)
  294. return nullptr;
  295. return &s_number_systems.at(number_system_index);
  296. };
  297. if (auto const* number_system = lookup_number_system(system))
  298. return number_system;
  299. auto default_number_system = get_preferred_keyword_value_for_locale(locale, "nu"sv);
  300. if (!default_number_system.has_value())
  301. return nullptr;
  302. return lookup_number_system(*default_number_system);
  303. }
  304. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, NumericSymbol symbol)
  305. {
  306. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  307. auto symbols = s_numeric_symbol_lists.at(number_system->symbols);
  308. auto symbol_index = to_underlying(symbol);
  309. if (symbol_index >= symbols.size())
  310. return {};
  311. return decode_string(symbols[symbol_index]);
  312. }
  313. return {};
  314. }
  315. }
  316. )~~~");
  317. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  318. return {};
  319. }
  320. ErrorOr<int> serenity_main(Main::Arguments arguments)
  321. {
  322. StringView generated_header_path;
  323. StringView generated_implementation_path;
  324. StringView core_path;
  325. StringView numbers_path;
  326. Core::ArgsParser args_parser;
  327. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  328. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  329. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  330. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  331. args_parser.parse(arguments);
  332. auto generated_header_file = TRY(open_file(generated_header_path, Core::File::OpenMode::Write));
  333. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::File::OpenMode::Write));
  334. CLDR cldr;
  335. TRY(parse_all_locales(core_path, numbers_path, cldr));
  336. TRY(generate_unicode_locale_header(*generated_header_file, cldr));
  337. TRY(generate_unicode_locale_implementation(*generated_implementation_file, cldr));
  338. return 0;
  339. }