GenerateUnicodeNumberFormat.cpp 29 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717
  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 <AK/Utf8View.h>
  20. #include <LibCore/ArgsParser.h>
  21. #include <LibCore/DirIterator.h>
  22. #include <LibCore/File.h>
  23. #include <LibUnicode/Locale.h>
  24. #include <math.h>
  25. using StringIndexType = u16;
  26. constexpr auto s_string_index_type = "u16"sv;
  27. enum class NumberFormatType {
  28. Standard,
  29. Compact,
  30. };
  31. struct NumberFormat : public Unicode::NumberFormat {
  32. using Base = Unicode::NumberFormat;
  33. static Base::Plurality plurality_from_string(StringView plurality)
  34. {
  35. if (plurality == "other"sv)
  36. return Base::Plurality::Other;
  37. if (plurality == "1"sv)
  38. return Base::Plurality::Single;
  39. if (plurality == "zero"sv)
  40. return Base::Plurality::Zero;
  41. if (plurality == "one"sv)
  42. return Base::Plurality::One;
  43. if (plurality == "two"sv)
  44. return Base::Plurality::Two;
  45. if (plurality == "few"sv)
  46. return Base::Plurality::Few;
  47. if (plurality == "many"sv)
  48. return Base::Plurality::Many;
  49. VERIFY_NOT_REACHED();
  50. }
  51. StringIndexType zero_format_index { 0 };
  52. StringIndexType positive_format_index { 0 };
  53. StringIndexType negative_format_index { 0 };
  54. StringIndexType identifier_index { 0 };
  55. };
  56. struct NumberSystem {
  57. StringIndexType system { 0 };
  58. HashMap<String, StringIndexType> symbols {};
  59. u8 primary_grouping_size { 0 };
  60. u8 secondary_grouping_size { 0 };
  61. NumberFormat decimal_format {};
  62. Vector<NumberFormat> decimal_long_formats {};
  63. Vector<NumberFormat> decimal_short_formats {};
  64. NumberFormat currency_format {};
  65. NumberFormat accounting_format {};
  66. Vector<NumberFormat> currency_unit_formats {};
  67. Vector<NumberFormat> currency_short_formats {};
  68. NumberFormat percent_format {};
  69. NumberFormat scientific_format {};
  70. };
  71. struct Locale {
  72. HashMap<String, NumberSystem> number_systems;
  73. };
  74. struct UnicodeLocaleData {
  75. UniqueStringStorage<StringIndexType> unique_strings;
  76. HashMap<String, Locale> locales;
  77. Vector<String> numeric_symbols;
  78. };
  79. static String parse_identifier(String pattern, StringView replacement, UnicodeLocaleData& locale_data, NumberFormat& format)
  80. {
  81. static Utf8View whitespace { "\u0020\u00a0"sv };
  82. Utf8View utf8_pattern { pattern };
  83. Optional<size_t> start_index;
  84. Optional<size_t> end_index;
  85. bool inside_replacement = false;
  86. for (auto it = utf8_pattern.begin(); it != utf8_pattern.end(); ++it) {
  87. if (*it == '{') {
  88. if (start_index.has_value()) {
  89. end_index = utf8_pattern.byte_offset_of(it);
  90. break;
  91. }
  92. inside_replacement = true;
  93. } else if (*it == '}') {
  94. inside_replacement = false;
  95. } else if (!inside_replacement && !start_index.has_value() && !whitespace.contains(*it)) {
  96. start_index = utf8_pattern.byte_offset_of(it);
  97. }
  98. }
  99. if (!start_index.has_value())
  100. return pattern;
  101. end_index = end_index.value_or(pattern.length());
  102. utf8_pattern = utf8_pattern.substring_view(*start_index, *end_index - *start_index);
  103. utf8_pattern = utf8_pattern.trim(whitespace);
  104. auto identifier = utf8_pattern.as_string().replace("'.'"sv, "."sv);
  105. format.identifier_index = locale_data.unique_strings.ensure(move(identifier));
  106. return String::formatted("{}{}{}",
  107. *start_index > 0 ? pattern.substring_view(0, *start_index) : ""sv,
  108. replacement,
  109. pattern.substring_view(*start_index + utf8_pattern.byte_length()));
  110. }
  111. static void parse_number_pattern(Vector<String> patterns, UnicodeLocaleData& locale_data, NumberFormatType type, NumberFormat& format, NumberSystem* number_system_for_groupings = nullptr)
  112. {
  113. // https://unicode.org/reports/tr35/tr35-numbers.html#Number_Format_Patterns
  114. // https://cldr.unicode.org/translation/number-currency-formats/number-and-currency-patterns
  115. VERIFY((patterns.size() == 1) || (patterns.size() == 2));
  116. auto replace_patterns = [&](String pattern) {
  117. static HashMap<StringView, StringView> replacements = {
  118. { "{0}"sv, "{number}"sv },
  119. { "{1}"sv, "{currency}"sv },
  120. { "%"sv, "{percentSign}"sv },
  121. { "+"sv, "{plusSign}"sv },
  122. { "-"sv, "{minusSign}"sv },
  123. { "¤"sv, "{currency}"sv }, // U+00A4 Currency Sign
  124. { "E"sv, "{scientificSeparator}"sv },
  125. };
  126. for (auto const& replacement : replacements)
  127. pattern = pattern.replace(replacement.key, replacement.value, true);
  128. if (auto start_number_index = pattern.find_any_of("#0"sv, String::SearchDirection::Forward); start_number_index.has_value()) {
  129. auto end_number_index = *start_number_index + 1;
  130. for (; end_number_index < pattern.length(); ++end_number_index) {
  131. auto ch = pattern[end_number_index];
  132. if ((ch != '#') && (ch != '0') && (ch != ',') && (ch != '.'))
  133. break;
  134. }
  135. if (number_system_for_groupings) {
  136. auto number_pattern = pattern.substring_view(*start_number_index, end_number_index - *start_number_index);
  137. auto group_separators = number_pattern.find_all(","sv);
  138. VERIFY((group_separators.size() == 1) || (group_separators.size() == 2));
  139. auto decimal = number_pattern.find('.');
  140. VERIFY(decimal.has_value());
  141. if (group_separators.size() == 1) {
  142. number_system_for_groupings->primary_grouping_size = *decimal - group_separators[0] - 1;
  143. number_system_for_groupings->secondary_grouping_size = number_system_for_groupings->primary_grouping_size;
  144. } else {
  145. number_system_for_groupings->primary_grouping_size = *decimal - group_separators[1] - 1;
  146. number_system_for_groupings->secondary_grouping_size = group_separators[1] - group_separators[0] - 1;
  147. }
  148. }
  149. pattern = String::formatted("{}{{number}}{}",
  150. *start_number_index > 0 ? pattern.substring_view(0, *start_number_index) : ""sv,
  151. pattern.substring_view(end_number_index));
  152. // This is specifically handled here rather than in the replacements HashMap above so
  153. // that we do not errantly replace zeroes in number patterns.
  154. if (pattern.contains(*replacements.get("E"sv)))
  155. pattern = pattern.replace("0"sv, "{scientificExponent}"sv);
  156. }
  157. if (type == NumberFormatType::Compact)
  158. return parse_identifier(move(pattern), "{compactIdentifier}"sv, locale_data, format);
  159. return pattern;
  160. };
  161. auto zero_format = replace_patterns(move(patterns[0]));
  162. format.positive_format_index = locale_data.unique_strings.ensure(String::formatted("{{plusSign}}{}", zero_format));
  163. if (patterns.size() == 2) {
  164. auto negative_format = replace_patterns(move(patterns[1]));
  165. format.negative_format_index = locale_data.unique_strings.ensure(move(negative_format));
  166. } else {
  167. format.negative_format_index = locale_data.unique_strings.ensure(String::formatted("{{minusSign}}{}", zero_format));
  168. }
  169. format.zero_format_index = locale_data.unique_strings.ensure(move(zero_format));
  170. }
  171. static void parse_number_systems(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  172. {
  173. LexicalPath numbers_path(move(locale_numbers_path));
  174. numbers_path = numbers_path.append("numbers.json"sv);
  175. VERIFY(Core::File::exists(numbers_path.string()));
  176. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  177. VERIFY(!numbers_file_or_error.is_error());
  178. auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
  179. VERIFY(numbers.has_value());
  180. auto const& main_object = numbers->as_object().get("main"sv);
  181. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  182. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  183. auto ensure_number_system = [&](auto const& system) -> NumberSystem& {
  184. return locale.number_systems.ensure(system, [&]() {
  185. auto system_index = locale_data.unique_strings.ensure(system);
  186. return NumberSystem { .system = system_index };
  187. });
  188. };
  189. auto parse_number_format = [&](auto const& format_object) {
  190. Vector<NumberFormat> result;
  191. result.ensure_capacity(format_object.size());
  192. format_object.for_each_member([&](auto const& key, JsonValue const& value) {
  193. auto split_key = key.split_view('-');
  194. if (split_key.size() != 3)
  195. return;
  196. auto patterns = value.as_string().split(';');
  197. NumberFormat format {};
  198. if (auto type = split_key[0].template to_uint<u64>(); type.has_value()) {
  199. VERIFY(*type % 10 == 0);
  200. format.magnitude = static_cast<u8>(log10(*type));
  201. if (patterns[0] != "0"sv) {
  202. auto number_of_zeroes_in_pattern = patterns[0].count("0"sv);
  203. VERIFY(format.magnitude >= number_of_zeroes_in_pattern);
  204. format.exponent = format.magnitude + 1 - number_of_zeroes_in_pattern;
  205. }
  206. } else {
  207. VERIFY(split_key[0] == "unitPattern"sv);
  208. }
  209. format.plurality = NumberFormat::plurality_from_string(split_key[2]);
  210. parse_number_pattern(move(patterns), locale_data, NumberFormatType::Compact, format);
  211. result.append(move(format));
  212. });
  213. return result;
  214. };
  215. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  216. constexpr auto symbols_prefix = "symbols-numberSystem-"sv;
  217. constexpr auto decimal_formats_prefix = "decimalFormats-numberSystem-"sv;
  218. constexpr auto currency_formats_prefix = "currencyFormats-numberSystem-"sv;
  219. constexpr auto percent_formats_prefix = "percentFormats-numberSystem-"sv;
  220. constexpr auto scientific_formats_prefix = "scientificFormats-numberSystem-"sv;
  221. if (key.starts_with(symbols_prefix)) {
  222. auto system = key.substring(symbols_prefix.length());
  223. auto& number_system = ensure_number_system(system);
  224. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& localization) {
  225. auto symbol_index = locale_data.unique_strings.ensure(localization.as_string());
  226. number_system.symbols.set(symbol, symbol_index);
  227. if (!locale_data.numeric_symbols.contains_slow(symbol))
  228. locale_data.numeric_symbols.append(symbol);
  229. });
  230. } else if (key.starts_with(decimal_formats_prefix)) {
  231. auto system = key.substring(decimal_formats_prefix.length());
  232. auto& number_system = ensure_number_system(system);
  233. auto format_object = value.as_object().get("standard"sv);
  234. parse_number_pattern(format_object.as_string().split(';'), locale_data, NumberFormatType::Standard, number_system.decimal_format, &number_system);
  235. auto const& long_format = value.as_object().get("long"sv).as_object().get("decimalFormat"sv);
  236. number_system.decimal_long_formats = parse_number_format(long_format.as_object());
  237. auto const& short_format = value.as_object().get("short"sv).as_object().get("decimalFormat"sv);
  238. number_system.decimal_short_formats = parse_number_format(short_format.as_object());
  239. } else if (key.starts_with(currency_formats_prefix)) {
  240. auto system = key.substring(currency_formats_prefix.length());
  241. auto& number_system = ensure_number_system(system);
  242. auto format_object = value.as_object().get("standard"sv);
  243. parse_number_pattern(format_object.as_string().split(';'), locale_data, NumberFormatType::Standard, number_system.currency_format);
  244. format_object = value.as_object().get("accounting"sv);
  245. parse_number_pattern(format_object.as_string().split(';'), locale_data, NumberFormatType::Standard, number_system.accounting_format);
  246. number_system.currency_unit_formats = parse_number_format(value.as_object());
  247. if (value.as_object().has("short"sv)) {
  248. auto const& short_format = value.as_object().get("short"sv).as_object().get("standard"sv);
  249. number_system.currency_short_formats = parse_number_format(short_format.as_object());
  250. }
  251. } else if (key.starts_with(percent_formats_prefix)) {
  252. auto system = key.substring(percent_formats_prefix.length());
  253. auto& number_system = ensure_number_system(system);
  254. auto format_object = value.as_object().get("standard"sv);
  255. parse_number_pattern(format_object.as_string().split(';'), locale_data, NumberFormatType::Standard, number_system.percent_format);
  256. } else if (key.starts_with(scientific_formats_prefix)) {
  257. auto system = key.substring(scientific_formats_prefix.length());
  258. auto& number_system = ensure_number_system(system);
  259. auto format_object = value.as_object().get("standard"sv);
  260. parse_number_pattern(format_object.as_string().split(';'), locale_data, NumberFormatType::Standard, number_system.scientific_format);
  261. }
  262. });
  263. }
  264. static void parse_all_locales(String core_path, String numbers_path, UnicodeLocaleData& locale_data)
  265. {
  266. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  267. auto remove_variants_from_path = [&](String path) -> Optional<String> {
  268. auto parsed_locale = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path));
  269. if (!parsed_locale.has_value())
  270. return {};
  271. StringBuilder builder;
  272. builder.append(locale_data.unique_strings.get(parsed_locale->language));
  273. if (auto script = locale_data.unique_strings.get(parsed_locale->script); !script.is_empty())
  274. builder.appendff("-{}", script);
  275. if (auto region = locale_data.unique_strings.get(parsed_locale->region); !region.is_empty())
  276. builder.appendff("-{}", region);
  277. return builder.build();
  278. };
  279. while (numbers_iterator.has_next()) {
  280. auto numbers_path = numbers_iterator.next_full_path();
  281. VERIFY(Core::File::is_directory(numbers_path));
  282. auto language = remove_variants_from_path(numbers_path);
  283. if (!language.has_value())
  284. continue;
  285. auto& locale = locale_data.locales.ensure(*language);
  286. parse_number_systems(numbers_path, locale_data, locale);
  287. }
  288. parse_default_content_locales(move(core_path), locale_data);
  289. }
  290. static String format_identifier(StringView owner, String identifier)
  291. {
  292. identifier = identifier.replace("-"sv, "_"sv, true);
  293. if (all_of(identifier, is_ascii_digit))
  294. return String::formatted("{}_{}", owner[0], identifier);
  295. if (is_ascii_lower_alpha(identifier[0]))
  296. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  297. return identifier;
  298. }
  299. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  300. {
  301. StringBuilder builder;
  302. SourceGenerator generator { builder };
  303. generator.append(R"~~~(
  304. #pragma once
  305. #include <AK/Optional.h>
  306. #include <AK/StringView.h>
  307. #include <AK/Types.h>
  308. #include <AK/Vector.h>
  309. #include <LibUnicode/Forward.h>
  310. namespace Unicode {
  311. )~~~");
  312. generate_enum(generator, format_identifier, "NumericSymbol"sv, {}, locale_data.numeric_symbols);
  313. generator.append(R"~~~(
  314. namespace Detail {
  315. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView numeric_symbol);
  316. Optional<NumberGroupings> get_number_system_groupings(StringView locale, StringView system);
  317. Optional<NumberFormat> get_standard_number_system_format(StringView locale, StringView system, StandardNumberFormatType type);
  318. Vector<NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type);
  319. Optional<NumericSymbol> numeric_symbol_from_string(StringView numeric_symbol);
  320. }
  321. }
  322. )~~~");
  323. file.write(generator.as_string_view());
  324. }
  325. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  326. {
  327. StringBuilder builder;
  328. SourceGenerator generator { builder };
  329. generator.set("string_index_type"sv, s_string_index_type);
  330. generator.set("numeric_symbols_size", String::number(locale_data.numeric_symbols.size()));
  331. generator.append(R"~~~(
  332. #include <AK/Array.h>
  333. #include <AK/BinarySearch.h>
  334. #include <AK/Span.h>
  335. #include <LibUnicode/Locale.h>
  336. #include <LibUnicode/UnicodeNumberFormat.h>
  337. namespace Unicode::Detail {
  338. )~~~");
  339. locale_data.unique_strings.generate(generator);
  340. generator.append(R"~~~(
  341. struct NumberFormat {
  342. Unicode::NumberFormat to_unicode_number_format() const {
  343. Unicode::NumberFormat number_format {};
  344. number_format.magnitude = magnitude;
  345. number_format.exponent = exponent;
  346. number_format.plurality = static_cast<Unicode::NumberFormat::Plurality>(plurality);
  347. number_format.zero_format = s_string_list[zero_format];
  348. number_format.positive_format = s_string_list[positive_format];
  349. number_format.negative_format = s_string_list[negative_format];
  350. number_format.identifier = s_string_list[identifier];
  351. return number_format;
  352. }
  353. u8 magnitude { 0 };
  354. u8 exponent { 0 };
  355. u8 plurality { 0 };
  356. @string_index_type@ zero_format { 0 };
  357. @string_index_type@ positive_format { 0 };
  358. @string_index_type@ negative_format { 0 };
  359. @string_index_type@ identifier { 0 };
  360. };
  361. struct NumberSystem {
  362. @string_index_type@ system { 0 };
  363. Array<@string_index_type@, @numeric_symbols_size@> symbols {};
  364. u8 primary_grouping_size { 0 };
  365. u8 secondary_grouping_size { 0 };
  366. NumberFormat decimal_format {};
  367. Span<NumberFormat const> decimal_long_formats {};
  368. Span<NumberFormat const> decimal_short_formats {};
  369. NumberFormat currency_format {};
  370. NumberFormat accounting_format {};
  371. Span<NumberFormat const> currency_unit_formats {};
  372. Span<NumberFormat const> currency_short_formats {};
  373. NumberFormat percent_format {};
  374. NumberFormat scientific_format {};
  375. };
  376. )~~~");
  377. auto append_number_format = [&](auto const& number_format) {
  378. generator.set("magnitude"sv, String::number(number_format.magnitude));
  379. generator.set("exponent"sv, String::number(number_format.exponent));
  380. generator.set("plurality"sv, String::number(static_cast<u8>(number_format.plurality)));
  381. generator.set("zero_format"sv, String::number(number_format.zero_format_index));
  382. generator.set("positive_format"sv, String::number(number_format.positive_format_index));
  383. generator.set("negative_format"sv, String::number(number_format.negative_format_index));
  384. generator.set("identifier"sv, String::number(number_format.identifier_index));
  385. generator.append("{ @magnitude@, @exponent@, @plurality@, @zero_format@, @positive_format@, @negative_format@, @identifier@ },");
  386. };
  387. auto append_number_formats = [&](String name, auto const& number_formats) {
  388. generator.set("name"sv, move(name));
  389. generator.set("size"sv, String::number(number_formats.size()));
  390. generator.append(R"~~~(
  391. static constexpr Array<NumberFormat, @size@> @name@ { {
  392. )~~~");
  393. constexpr size_t max_values_per_row = 10;
  394. size_t values_in_current_row = 0;
  395. for (auto const& number_format : number_formats) {
  396. if (values_in_current_row++ > 0)
  397. generator.append(" ");
  398. append_number_format(number_format);
  399. if (values_in_current_row == max_values_per_row) {
  400. values_in_current_row = 0;
  401. generator.append("\n ");
  402. }
  403. }
  404. generator.append(R"~~~(
  405. } };
  406. )~~~");
  407. };
  408. auto append_number_systems = [&](String name, auto const& number_systems) {
  409. auto format_name = [&](StringView system, StringView format) {
  410. return String::formatted("{}_{}_{}", name, system, format);
  411. };
  412. for (auto const& number_system : number_systems) {
  413. append_number_formats(format_name(number_system.key, "dl"sv), number_system.value.decimal_long_formats);
  414. append_number_formats(format_name(number_system.key, "ds"sv), number_system.value.decimal_short_formats);
  415. append_number_formats(format_name(number_system.key, "cu"sv), number_system.value.currency_unit_formats);
  416. append_number_formats(format_name(number_system.key, "cs"sv), number_system.value.currency_short_formats);
  417. }
  418. generator.set("name", name);
  419. generator.set("size", String::number(number_systems.size()));
  420. generator.append(R"~~~(
  421. static constexpr Array<NumberSystem, @size@> @name@ { {)~~~");
  422. for (auto const& number_system : number_systems) {
  423. generator.set("system"sv, String::number(number_system.value.system));
  424. generator.set("primary_grouping_size"sv, String::number(number_system.value.primary_grouping_size));
  425. generator.set("secondary_grouping_size"sv, String::number(number_system.value.secondary_grouping_size));
  426. generator.set("decimal_long_formats"sv, format_name(number_system.key, "dl"sv));
  427. generator.set("decimal_short_formats"sv, format_name(number_system.key, "ds"sv));
  428. generator.set("currency_unit_formats"sv, format_name(number_system.key, "cu"sv));
  429. generator.set("currency_short_formats"sv, format_name(number_system.key, "cs"sv));
  430. generator.append(R"~~~(
  431. { @system@, {)~~~");
  432. for (auto const& symbol : locale_data.numeric_symbols) {
  433. auto index = number_system.value.symbols.get(symbol).value_or(0);
  434. generator.set("index", String::number(index));
  435. generator.append(" @index@,");
  436. }
  437. generator.append(" }, @primary_grouping_size@, @secondary_grouping_size@, ");
  438. append_number_format(number_system.value.decimal_format);
  439. generator.append(" @decimal_long_formats@.span(), @decimal_short_formats@.span(), ");
  440. append_number_format(number_system.value.currency_format);
  441. generator.append(" ");
  442. append_number_format(number_system.value.accounting_format);
  443. generator.append(" @currency_unit_formats@.span(), @currency_short_formats@.span(), ");
  444. append_number_format(number_system.value.percent_format);
  445. generator.append(" ");
  446. append_number_format(number_system.value.scientific_format);
  447. generator.append(" },");
  448. }
  449. generator.append(R"~~~(
  450. } };
  451. )~~~");
  452. };
  453. 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); });
  454. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values) {
  455. HashValueMap<String> hashes;
  456. hashes.ensure_capacity(values.size());
  457. for (auto const& value : values)
  458. hashes.set(value.hash(), format_identifier(enum_title, value));
  459. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  460. };
  461. append_from_string("NumericSymbol"sv, "numeric_symbol"sv, locale_data.numeric_symbols);
  462. generator.append(R"~~~(
  463. static NumberSystem const* find_number_system(StringView locale, StringView system)
  464. {
  465. auto locale_value = locale_from_string(locale);
  466. if (!locale_value.has_value())
  467. return nullptr;
  468. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  469. auto const& number_systems = s_number_systems.at(locale_index);
  470. for (auto const& number_system : number_systems) {
  471. if (system == s_string_list[number_system.system])
  472. return &number_system;
  473. };
  474. return nullptr;
  475. }
  476. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView symbol)
  477. {
  478. auto symbol_value = numeric_symbol_from_string(symbol);
  479. if (!symbol_value.has_value())
  480. return {};
  481. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  482. auto symbol_index = to_underlying(*symbol_value);
  483. return s_string_list[number_system->symbols[symbol_index]];
  484. }
  485. return {};
  486. }
  487. Optional<NumberGroupings> get_number_system_groupings(StringView locale, StringView system)
  488. {
  489. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr)
  490. return NumberGroupings { number_system->primary_grouping_size, number_system->secondary_grouping_size };
  491. return {};
  492. }
  493. Optional<Unicode::NumberFormat> get_standard_number_system_format(StringView locale, StringView system, StandardNumberFormatType type)
  494. {
  495. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  496. switch (type) {
  497. case StandardNumberFormatType::Decimal:
  498. return number_system->decimal_format.to_unicode_number_format();
  499. case StandardNumberFormatType::Currency:
  500. return number_system->currency_format.to_unicode_number_format();
  501. case StandardNumberFormatType::Accounting:
  502. return number_system->accounting_format.to_unicode_number_format();
  503. case StandardNumberFormatType::Percent:
  504. return number_system->percent_format.to_unicode_number_format();
  505. case StandardNumberFormatType::Scientific:
  506. return number_system->scientific_format.to_unicode_number_format();
  507. }
  508. }
  509. return {};
  510. }
  511. Vector<Unicode::NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type)
  512. {
  513. Vector<Unicode::NumberFormat> formats;
  514. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  515. Span<NumberFormat const> number_formats;
  516. switch (type) {
  517. case CompactNumberFormatType::DecimalLong:
  518. number_formats = number_system->decimal_long_formats;
  519. break;
  520. case CompactNumberFormatType::DecimalShort:
  521. number_formats = number_system->decimal_short_formats;
  522. break;
  523. case CompactNumberFormatType::CurrencyUnit:
  524. number_formats = number_system->currency_unit_formats;
  525. break;
  526. case CompactNumberFormatType::CurrencyShort:
  527. number_formats = number_system->currency_short_formats;
  528. break;
  529. }
  530. formats.ensure_capacity(number_formats.size());
  531. for (auto const& number_format : number_formats)
  532. formats.append(number_format.to_unicode_number_format());
  533. }
  534. return formats;
  535. }
  536. }
  537. )~~~");
  538. file.write(generator.as_string_view());
  539. }
  540. int main(int argc, char** argv)
  541. {
  542. char const* generated_header_path = nullptr;
  543. char const* generated_implementation_path = nullptr;
  544. char const* core_path = nullptr;
  545. char const* numbers_path = nullptr;
  546. Core::ArgsParser args_parser;
  547. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  548. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  549. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  550. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  551. args_parser.parse(argc, argv);
  552. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  553. if (path.is_empty()) {
  554. warnln("{} is required", flags);
  555. args_parser.print_usage(stderr, argv[0]);
  556. exit(1);
  557. }
  558. auto file_or_error = Core::File::open(path, mode);
  559. if (file_or_error.is_error()) {
  560. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  561. exit(1);
  562. }
  563. return file_or_error.release_value();
  564. };
  565. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  566. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  567. UnicodeLocaleData locale_data;
  568. parse_all_locales(core_path, numbers_path, locale_data);
  569. generate_unicode_locale_header(generated_header_file, locale_data);
  570. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  571. return 0;
  572. }