GenerateUnicodeNumberFormat.cpp 27 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664
  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. #include <LibUnicode/Locale.h>
  23. #include <math.h>
  24. using StringIndexType = u16;
  25. constexpr auto s_string_index_type = "u16"sv;
  26. struct NumberFormat : public Unicode::NumberFormat {
  27. using Base = Unicode::NumberFormat;
  28. static Base::Plurality plurality_from_string(StringView plurality)
  29. {
  30. if (plurality == "other"sv)
  31. return Base::Plurality::Other;
  32. if (plurality == "1"sv)
  33. return Base::Plurality::Single;
  34. if (plurality == "zero"sv)
  35. return Base::Plurality::Zero;
  36. if (plurality == "one"sv)
  37. return Base::Plurality::One;
  38. if (plurality == "two"sv)
  39. return Base::Plurality::Two;
  40. if (plurality == "few"sv)
  41. return Base::Plurality::Few;
  42. if (plurality == "many"sv)
  43. return Base::Plurality::Many;
  44. VERIFY_NOT_REACHED();
  45. }
  46. StringIndexType zero_format_index { 0 };
  47. StringIndexType positive_format_index { 0 };
  48. StringIndexType negative_format_index { 0 };
  49. };
  50. struct NumberSystem {
  51. StringIndexType system { 0 };
  52. HashMap<String, StringIndexType> symbols {};
  53. u8 primary_grouping_size { 0 };
  54. u8 secondary_grouping_size { 0 };
  55. NumberFormat decimal_format {};
  56. Vector<NumberFormat> decimal_long_formats {};
  57. Vector<NumberFormat> decimal_short_formats {};
  58. NumberFormat currency_format {};
  59. NumberFormat accounting_format {};
  60. Vector<NumberFormat> currency_unit_formats {};
  61. Vector<NumberFormat> currency_short_formats {};
  62. NumberFormat percent_format {};
  63. NumberFormat scientific_format {};
  64. };
  65. struct Locale {
  66. HashMap<String, NumberSystem> number_systems;
  67. };
  68. struct UnicodeLocaleData {
  69. UniqueStringStorage<StringIndexType> unique_strings;
  70. HashMap<String, Locale> locales;
  71. Vector<String> numeric_symbols;
  72. };
  73. static void parse_number_pattern(String pattern, UnicodeLocaleData& locale_data, NumberFormat& format, NumberSystem* number_system_for_groupings = nullptr)
  74. {
  75. // https://unicode.org/reports/tr35/tr35-numbers.html#Number_Format_Patterns
  76. // https://cldr.unicode.org/translation/number-currency-formats/number-and-currency-patterns
  77. auto replace_patterns = [&](String pattern) {
  78. static HashMap<StringView, StringView> replacements = {
  79. { "{0}"sv, "{number}"sv },
  80. { "{1}"sv, "{currency}"sv },
  81. { "%"sv, "{percentSign}"sv },
  82. { "+"sv, "{plusSign}"sv },
  83. { "-"sv, "{minusSign}"sv },
  84. { "¤"sv, "{currency}"sv }, // U+00A4 Currency Sign
  85. { "E"sv, "{scientificSeparator}"sv },
  86. };
  87. for (auto const& replacement : replacements)
  88. pattern = pattern.replace(replacement.key, replacement.value, true);
  89. if (auto start_number_index = pattern.find_any_of("#0"sv, String::SearchDirection::Forward); start_number_index.has_value()) {
  90. auto end_number_index = *start_number_index + 1;
  91. for (; end_number_index < pattern.length(); ++end_number_index) {
  92. auto ch = pattern[end_number_index];
  93. if ((ch != '#') && (ch != '0') && (ch != ',') && (ch != '.'))
  94. break;
  95. }
  96. if (number_system_for_groupings) {
  97. auto number_pattern = pattern.substring_view(*start_number_index, end_number_index - *start_number_index);
  98. auto group_separators = number_pattern.find_all(","sv);
  99. VERIFY((group_separators.size() == 1) || (group_separators.size() == 2));
  100. auto decimal = number_pattern.find('.');
  101. VERIFY(decimal.has_value());
  102. if (group_separators.size() == 1) {
  103. number_system_for_groupings->primary_grouping_size = *decimal - group_separators[0] - 1;
  104. number_system_for_groupings->secondary_grouping_size = number_system_for_groupings->primary_grouping_size;
  105. } else {
  106. number_system_for_groupings->primary_grouping_size = *decimal - group_separators[1] - 1;
  107. number_system_for_groupings->secondary_grouping_size = group_separators[1] - group_separators[0] - 1;
  108. }
  109. }
  110. pattern = String::formatted("{}{{number}}{}",
  111. *start_number_index > 0 ? pattern.substring_view(0, *start_number_index) : ""sv,
  112. pattern.substring_view(end_number_index));
  113. // This is specifically handled here rather than in the replacements HashMap above so
  114. // that we do not errantly replace zeroes in number patterns.
  115. if (pattern.contains(*replacements.get("E"sv)))
  116. pattern = pattern.replace("0"sv, "{scientificExponent}"sv);
  117. }
  118. return pattern;
  119. };
  120. auto patterns = pattern.split(';');
  121. VERIFY((patterns.size() == 1) || (patterns.size() == 2));
  122. if (format.magnitude != 0) {
  123. auto number_of_zeroes_in_pattern = patterns[0].count("0"sv);
  124. VERIFY(format.magnitude >= number_of_zeroes_in_pattern);
  125. format.compact_scale = format.magnitude - number_of_zeroes_in_pattern;
  126. }
  127. auto zero_format = replace_patterns(move(patterns[0]));
  128. format.positive_format_index = locale_data.unique_strings.ensure(String::formatted("{{plusSign}}{}", zero_format));
  129. if (patterns.size() == 2) {
  130. auto negative_format = replace_patterns(move(patterns[1]));
  131. format.negative_format_index = locale_data.unique_strings.ensure(move(negative_format));
  132. } else {
  133. format.negative_format_index = locale_data.unique_strings.ensure(String::formatted("{{minusSign}}{}", zero_format));
  134. }
  135. format.zero_format_index = locale_data.unique_strings.ensure(move(zero_format));
  136. }
  137. static void parse_number_systems(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  138. {
  139. LexicalPath numbers_path(move(locale_numbers_path));
  140. numbers_path = numbers_path.append("numbers.json"sv);
  141. VERIFY(Core::File::exists(numbers_path.string()));
  142. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  143. VERIFY(!numbers_file_or_error.is_error());
  144. auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
  145. VERIFY(numbers.has_value());
  146. auto const& main_object = numbers->as_object().get("main"sv);
  147. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  148. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  149. auto ensure_number_system = [&](auto const& system) -> NumberSystem& {
  150. return locale.number_systems.ensure(system, [&]() {
  151. auto system_index = locale_data.unique_strings.ensure(system);
  152. return NumberSystem { .system = system_index };
  153. });
  154. };
  155. auto parse_number_format = [&](auto const& format_object) {
  156. Vector<NumberFormat> result;
  157. result.ensure_capacity(format_object.size());
  158. format_object.for_each_member([&](auto const& key, JsonValue const& value) {
  159. auto split_key = key.split_view('-');
  160. if (split_key.size() != 3)
  161. return;
  162. NumberFormat format {};
  163. if (auto type = split_key[0].template to_uint<u64>(); type.has_value()) {
  164. VERIFY(*type % 10 == 0);
  165. format.magnitude = static_cast<u8>(log10(*type));
  166. } else {
  167. VERIFY(split_key[0] == "unitPattern"sv);
  168. }
  169. format.plurality = NumberFormat::plurality_from_string(split_key[2]);
  170. parse_number_pattern(value.as_string(), locale_data, format);
  171. result.append(move(format));
  172. });
  173. return result;
  174. };
  175. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  176. constexpr auto symbols_prefix = "symbols-numberSystem-"sv;
  177. constexpr auto decimal_formats_prefix = "decimalFormats-numberSystem-"sv;
  178. constexpr auto currency_formats_prefix = "currencyFormats-numberSystem-"sv;
  179. constexpr auto percent_formats_prefix = "percentFormats-numberSystem-"sv;
  180. constexpr auto scientific_formats_prefix = "scientificFormats-numberSystem-"sv;
  181. if (key.starts_with(symbols_prefix)) {
  182. auto system = key.substring(symbols_prefix.length());
  183. auto& number_system = ensure_number_system(system);
  184. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& localization) {
  185. auto symbol_index = locale_data.unique_strings.ensure(localization.as_string());
  186. number_system.symbols.set(symbol, symbol_index);
  187. if (!locale_data.numeric_symbols.contains_slow(symbol))
  188. locale_data.numeric_symbols.append(symbol);
  189. });
  190. } else if (key.starts_with(decimal_formats_prefix)) {
  191. auto system = key.substring(decimal_formats_prefix.length());
  192. auto& number_system = ensure_number_system(system);
  193. auto format_object = value.as_object().get("standard"sv);
  194. parse_number_pattern(format_object.as_string(), locale_data, number_system.decimal_format, &number_system);
  195. auto const& long_format = value.as_object().get("long"sv).as_object().get("decimalFormat"sv);
  196. number_system.decimal_long_formats = parse_number_format(long_format.as_object());
  197. auto const& short_format = value.as_object().get("short"sv).as_object().get("decimalFormat"sv);
  198. number_system.decimal_short_formats = parse_number_format(short_format.as_object());
  199. } else if (key.starts_with(currency_formats_prefix)) {
  200. auto system = key.substring(currency_formats_prefix.length());
  201. auto& number_system = ensure_number_system(system);
  202. auto format_object = value.as_object().get("standard"sv);
  203. parse_number_pattern(format_object.as_string(), locale_data, number_system.currency_format);
  204. format_object = value.as_object().get("accounting"sv);
  205. parse_number_pattern(format_object.as_string(), locale_data, number_system.accounting_format);
  206. number_system.currency_unit_formats = parse_number_format(value.as_object());
  207. if (value.as_object().has("short"sv)) {
  208. auto const& short_format = value.as_object().get("short"sv).as_object().get("standard"sv);
  209. number_system.currency_short_formats = parse_number_format(short_format.as_object());
  210. }
  211. } else if (key.starts_with(percent_formats_prefix)) {
  212. auto system = key.substring(percent_formats_prefix.length());
  213. auto& number_system = ensure_number_system(system);
  214. auto format_object = value.as_object().get("standard"sv);
  215. parse_number_pattern(format_object.as_string(), locale_data, number_system.percent_format);
  216. } else if (key.starts_with(scientific_formats_prefix)) {
  217. auto system = key.substring(scientific_formats_prefix.length());
  218. auto& number_system = ensure_number_system(system);
  219. auto format_object = value.as_object().get("standard"sv);
  220. parse_number_pattern(format_object.as_string(), locale_data, number_system.scientific_format);
  221. }
  222. });
  223. }
  224. static void parse_all_locales(String core_path, String numbers_path, UnicodeLocaleData& locale_data)
  225. {
  226. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  227. auto remove_variants_from_path = [&](String path) -> Optional<String> {
  228. auto parsed_locale = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path));
  229. if (!parsed_locale.has_value())
  230. return {};
  231. StringBuilder builder;
  232. builder.append(locale_data.unique_strings.get(parsed_locale->language));
  233. if (auto script = locale_data.unique_strings.get(parsed_locale->script); !script.is_empty())
  234. builder.appendff("-{}", script);
  235. if (auto region = locale_data.unique_strings.get(parsed_locale->region); !region.is_empty())
  236. builder.appendff("-{}", region);
  237. return builder.build();
  238. };
  239. while (numbers_iterator.has_next()) {
  240. auto numbers_path = numbers_iterator.next_full_path();
  241. VERIFY(Core::File::is_directory(numbers_path));
  242. auto language = remove_variants_from_path(numbers_path);
  243. if (!language.has_value())
  244. continue;
  245. auto& locale = locale_data.locales.ensure(*language);
  246. parse_number_systems(numbers_path, locale_data, locale);
  247. }
  248. parse_default_content_locales(move(core_path), locale_data);
  249. }
  250. static String format_identifier(StringView owner, String identifier)
  251. {
  252. identifier = identifier.replace("-"sv, "_"sv, true);
  253. if (all_of(identifier, is_ascii_digit))
  254. return String::formatted("{}_{}", owner[0], identifier);
  255. if (is_ascii_lower_alpha(identifier[0]))
  256. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  257. return identifier;
  258. }
  259. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  260. {
  261. StringBuilder builder;
  262. SourceGenerator generator { builder };
  263. generator.append(R"~~~(
  264. #pragma once
  265. #include <AK/Optional.h>
  266. #include <AK/StringView.h>
  267. #include <AK/Types.h>
  268. #include <AK/Vector.h>
  269. #include <LibUnicode/Forward.h>
  270. namespace Unicode {
  271. )~~~");
  272. generate_enum(generator, format_identifier, "NumericSymbol"sv, {}, locale_data.numeric_symbols);
  273. generator.append(R"~~~(
  274. namespace Detail {
  275. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView numeric_symbol);
  276. Optional<NumberGroupings> get_number_system_groupings(StringView locale, StringView system);
  277. Optional<NumberFormat> get_standard_number_system_format(StringView locale, StringView system, StandardNumberFormatType type);
  278. Vector<NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type);
  279. Optional<NumericSymbol> numeric_symbol_from_string(StringView numeric_symbol);
  280. }
  281. }
  282. )~~~");
  283. file.write(generator.as_string_view());
  284. }
  285. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  286. {
  287. StringBuilder builder;
  288. SourceGenerator generator { builder };
  289. generator.set("string_index_type"sv, s_string_index_type);
  290. generator.set("numeric_symbols_size", String::number(locale_data.numeric_symbols.size()));
  291. generator.append(R"~~~(
  292. #include <AK/Array.h>
  293. #include <AK/BinarySearch.h>
  294. #include <AK/Span.h>
  295. #include <LibUnicode/Locale.h>
  296. #include <LibUnicode/UnicodeNumberFormat.h>
  297. namespace Unicode::Detail {
  298. )~~~");
  299. locale_data.unique_strings.generate(generator);
  300. generator.append(R"~~~(
  301. struct NumberFormat {
  302. Unicode::NumberFormat to_unicode_number_format() const {
  303. Unicode::NumberFormat number_format {};
  304. number_format.magnitude = magnitude;
  305. number_format.compact_scale = compact_scale;
  306. number_format.plurality = static_cast<Unicode::NumberFormat::Plurality>(plurality);
  307. number_format.zero_format = s_string_list[zero_format];
  308. number_format.positive_format = s_string_list[positive_format];
  309. number_format.negative_format = s_string_list[negative_format];
  310. return number_format;
  311. }
  312. u8 magnitude { 0 };
  313. u8 compact_scale { 0 };
  314. u8 plurality { 0 };
  315. @string_index_type@ zero_format { 0 };
  316. @string_index_type@ positive_format { 0 };
  317. @string_index_type@ negative_format { 0 };
  318. };
  319. struct NumberSystem {
  320. @string_index_type@ system { 0 };
  321. Array<@string_index_type@, @numeric_symbols_size@> symbols {};
  322. u8 primary_grouping_size { 0 };
  323. u8 secondary_grouping_size { 0 };
  324. NumberFormat decimal_format {};
  325. Span<NumberFormat const> decimal_long_formats {};
  326. Span<NumberFormat const> decimal_short_formats {};
  327. NumberFormat currency_format {};
  328. NumberFormat accounting_format {};
  329. Span<NumberFormat const> currency_unit_formats {};
  330. Span<NumberFormat const> currency_short_formats {};
  331. NumberFormat percent_format {};
  332. NumberFormat scientific_format {};
  333. };
  334. )~~~");
  335. auto append_number_format = [&](auto const& number_format) {
  336. generator.set("magnitude"sv, String::number(number_format.magnitude));
  337. generator.set("compact_scale"sv, String::number(number_format.compact_scale));
  338. generator.set("plurality"sv, String::number(static_cast<u8>(number_format.plurality)));
  339. generator.set("zero_format"sv, String::number(number_format.zero_format_index));
  340. generator.set("positive_format"sv, String::number(number_format.positive_format_index));
  341. generator.set("negative_format"sv, String::number(number_format.negative_format_index));
  342. generator.append("{ @magnitude@, @compact_scale@, @plurality@, @zero_format@, @positive_format@, @negative_format@ },");
  343. };
  344. auto append_number_formats = [&](String name, auto const& number_formats) {
  345. generator.set("name"sv, move(name));
  346. generator.set("size"sv, String::number(number_formats.size()));
  347. generator.append(R"~~~(
  348. static constexpr Array<NumberFormat, @size@> @name@ { {
  349. )~~~");
  350. constexpr size_t max_values_per_row = 10;
  351. size_t values_in_current_row = 0;
  352. for (auto const& number_format : number_formats) {
  353. if (values_in_current_row++ > 0)
  354. generator.append(" ");
  355. append_number_format(number_format);
  356. if (values_in_current_row == max_values_per_row) {
  357. values_in_current_row = 0;
  358. generator.append("\n ");
  359. }
  360. }
  361. generator.append(R"~~~(
  362. } };
  363. )~~~");
  364. };
  365. auto append_number_systems = [&](String name, auto const& number_systems) {
  366. auto format_name = [&](StringView system, StringView format) {
  367. return String::formatted("{}_{}_{}", name, system, format);
  368. };
  369. for (auto const& number_system : number_systems) {
  370. append_number_formats(format_name(number_system.key, "dl"sv), number_system.value.decimal_long_formats);
  371. append_number_formats(format_name(number_system.key, "ds"sv), number_system.value.decimal_short_formats);
  372. append_number_formats(format_name(number_system.key, "cu"sv), number_system.value.currency_unit_formats);
  373. append_number_formats(format_name(number_system.key, "cs"sv), number_system.value.currency_short_formats);
  374. }
  375. generator.set("name", name);
  376. generator.set("size", String::number(number_systems.size()));
  377. generator.append(R"~~~(
  378. static constexpr Array<NumberSystem, @size@> @name@ { {)~~~");
  379. for (auto const& number_system : number_systems) {
  380. generator.set("system"sv, String::number(number_system.value.system));
  381. generator.set("primary_grouping_size"sv, String::number(number_system.value.primary_grouping_size));
  382. generator.set("secondary_grouping_size"sv, String::number(number_system.value.secondary_grouping_size));
  383. generator.set("decimal_long_formats"sv, format_name(number_system.key, "dl"sv));
  384. generator.set("decimal_short_formats"sv, format_name(number_system.key, "ds"sv));
  385. generator.set("currency_unit_formats"sv, format_name(number_system.key, "cu"sv));
  386. generator.set("currency_short_formats"sv, format_name(number_system.key, "cs"sv));
  387. generator.append(R"~~~(
  388. { @system@, {)~~~");
  389. for (auto const& symbol : locale_data.numeric_symbols) {
  390. auto index = number_system.value.symbols.get(symbol).value_or(0);
  391. generator.set("index", String::number(index));
  392. generator.append(" @index@,");
  393. }
  394. generator.append(" }, @primary_grouping_size@, @secondary_grouping_size@, ");
  395. append_number_format(number_system.value.decimal_format);
  396. generator.append(" @decimal_long_formats@.span(), @decimal_short_formats@.span(), ");
  397. append_number_format(number_system.value.currency_format);
  398. generator.append(" ");
  399. append_number_format(number_system.value.accounting_format);
  400. generator.append(" @currency_unit_formats@.span(), @currency_short_formats@.span(), ");
  401. append_number_format(number_system.value.percent_format);
  402. generator.append(" ");
  403. append_number_format(number_system.value.scientific_format);
  404. generator.append(" },");
  405. }
  406. generator.append(R"~~~(
  407. } };
  408. )~~~");
  409. };
  410. generate_mapping(generator, locale_data.locales, "NumberSystem"sv, "s_number_systems"sv, "s_number_systems_{}", [&](auto const& name, auto const& value) { append_number_systems(name, value.number_systems); });
  411. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values) {
  412. HashValueMap<String> hashes;
  413. hashes.ensure_capacity(values.size());
  414. for (auto const& value : values)
  415. hashes.set(value.hash(), format_identifier(enum_title, value));
  416. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  417. };
  418. append_from_string("NumericSymbol"sv, "numeric_symbol"sv, locale_data.numeric_symbols);
  419. generator.append(R"~~~(
  420. static NumberSystem const* find_number_system(StringView locale, StringView system)
  421. {
  422. auto locale_value = locale_from_string(locale);
  423. if (!locale_value.has_value())
  424. return nullptr;
  425. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  426. auto const& number_systems = s_number_systems.at(locale_index);
  427. for (auto const& number_system : number_systems) {
  428. if (system == s_string_list[number_system.system])
  429. return &number_system;
  430. };
  431. return nullptr;
  432. }
  433. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView symbol)
  434. {
  435. auto symbol_value = numeric_symbol_from_string(symbol);
  436. if (!symbol_value.has_value())
  437. return {};
  438. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  439. auto symbol_index = to_underlying(*symbol_value);
  440. return s_string_list[number_system->symbols[symbol_index]];
  441. }
  442. return {};
  443. }
  444. Optional<NumberGroupings> get_number_system_groupings(StringView locale, StringView system)
  445. {
  446. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr)
  447. return NumberGroupings { number_system->primary_grouping_size, number_system->secondary_grouping_size };
  448. return {};
  449. }
  450. Optional<Unicode::NumberFormat> get_standard_number_system_format(StringView locale, StringView system, StandardNumberFormatType type)
  451. {
  452. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  453. switch (type) {
  454. case StandardNumberFormatType::Decimal:
  455. return number_system->decimal_format.to_unicode_number_format();
  456. case StandardNumberFormatType::Currency:
  457. return number_system->currency_format.to_unicode_number_format();
  458. case StandardNumberFormatType::Accounting:
  459. return number_system->accounting_format.to_unicode_number_format();
  460. case StandardNumberFormatType::Percent:
  461. return number_system->percent_format.to_unicode_number_format();
  462. case StandardNumberFormatType::Scientific:
  463. return number_system->scientific_format.to_unicode_number_format();
  464. }
  465. }
  466. return {};
  467. }
  468. Vector<Unicode::NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type)
  469. {
  470. Vector<Unicode::NumberFormat> formats;
  471. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  472. Span<NumberFormat const> number_formats;
  473. switch (type) {
  474. case CompactNumberFormatType::DecimalLong:
  475. number_formats = number_system->decimal_long_formats;
  476. break;
  477. case CompactNumberFormatType::DecimalShort:
  478. number_formats = number_system->decimal_short_formats;
  479. break;
  480. case CompactNumberFormatType::CurrencyUnit:
  481. number_formats = number_system->currency_unit_formats;
  482. break;
  483. case CompactNumberFormatType::CurrencyShort:
  484. number_formats = number_system->currency_short_formats;
  485. break;
  486. }
  487. formats.ensure_capacity(number_formats.size());
  488. for (auto const& number_format : number_formats)
  489. formats.append(number_format.to_unicode_number_format());
  490. }
  491. return formats;
  492. }
  493. }
  494. )~~~");
  495. file.write(generator.as_string_view());
  496. }
  497. int main(int argc, char** argv)
  498. {
  499. char const* generated_header_path = nullptr;
  500. char const* generated_implementation_path = nullptr;
  501. char const* core_path = nullptr;
  502. char const* numbers_path = nullptr;
  503. Core::ArgsParser args_parser;
  504. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  505. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  506. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  507. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  508. args_parser.parse(argc, argv);
  509. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  510. if (path.is_empty()) {
  511. warnln("{} is required", flags);
  512. args_parser.print_usage(stderr, argv[0]);
  513. exit(1);
  514. }
  515. auto file_or_error = Core::File::open(path, mode);
  516. if (file_or_error.is_error()) {
  517. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  518. exit(1);
  519. }
  520. return file_or_error.release_value();
  521. };
  522. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  523. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  524. UnicodeLocaleData locale_data;
  525. parse_all_locales(core_path, numbers_path, locale_data);
  526. generate_unicode_locale_header(generated_header_file, locale_data);
  527. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  528. return 0;
  529. }