GenerateUnicodeNumberFormat.cpp 30 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746
  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. Vector<StringIndexType> identifier_indices {};
  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. size_t max_identifier_count { 0 };
  79. };
  80. static String parse_identifiers(String pattern, StringView replacement, UnicodeLocaleData& locale_data, NumberFormat& format)
  81. {
  82. static Utf8View whitespace { "\u0020\u00a0"sv };
  83. while (true) {
  84. Utf8View utf8_pattern { pattern };
  85. Optional<size_t> start_index;
  86. Optional<size_t> end_index;
  87. bool inside_replacement = false;
  88. for (auto it = utf8_pattern.begin(); it != utf8_pattern.end(); ++it) {
  89. if (*it == '{') {
  90. if (start_index.has_value()) {
  91. end_index = utf8_pattern.byte_offset_of(it);
  92. break;
  93. }
  94. inside_replacement = true;
  95. } else if (*it == '}') {
  96. inside_replacement = false;
  97. } else if (!inside_replacement && !start_index.has_value() && !whitespace.contains(*it)) {
  98. start_index = utf8_pattern.byte_offset_of(it);
  99. }
  100. }
  101. if (!start_index.has_value())
  102. return pattern;
  103. end_index = end_index.value_or(pattern.length());
  104. utf8_pattern = utf8_pattern.substring_view(*start_index, *end_index - *start_index);
  105. utf8_pattern = utf8_pattern.trim(whitespace);
  106. auto identifier = utf8_pattern.as_string().replace("'.'"sv, "."sv);
  107. auto identifier_index = locale_data.unique_strings.ensure(move(identifier));
  108. size_t replacement_index = 0;
  109. if (auto index = format.identifier_indices.find_first_index(identifier_index); index.has_value()) {
  110. replacement_index = *index;
  111. } else {
  112. replacement_index = format.identifier_indices.size();
  113. format.identifier_indices.append(identifier_index);
  114. locale_data.max_identifier_count = max(locale_data.max_identifier_count, format.identifier_indices.size());
  115. }
  116. pattern = String::formatted("{}{{{}:{}}}{}",
  117. *start_index > 0 ? pattern.substring_view(0, *start_index) : ""sv,
  118. replacement,
  119. replacement_index,
  120. pattern.substring_view(*start_index + utf8_pattern.byte_length()));
  121. }
  122. }
  123. static void parse_number_pattern(Vector<String> patterns, UnicodeLocaleData& locale_data, NumberFormatType type, NumberFormat& format, NumberSystem* number_system_for_groupings = nullptr)
  124. {
  125. // https://unicode.org/reports/tr35/tr35-numbers.html#Number_Format_Patterns
  126. // https://cldr.unicode.org/translation/number-currency-formats/number-and-currency-patterns
  127. VERIFY((patterns.size() == 1) || (patterns.size() == 2));
  128. auto replace_patterns = [&](String pattern) {
  129. static HashMap<StringView, StringView> replacements = {
  130. { "{0}"sv, "{number}"sv },
  131. { "{1}"sv, "{currency}"sv },
  132. { "%"sv, "{percentSign}"sv },
  133. { "+"sv, "{plusSign}"sv },
  134. { "-"sv, "{minusSign}"sv },
  135. { "¤"sv, "{currency}"sv }, // U+00A4 Currency Sign
  136. { "E"sv, "{scientificSeparator}"sv },
  137. };
  138. for (auto const& replacement : replacements)
  139. pattern = pattern.replace(replacement.key, replacement.value, true);
  140. if (auto start_number_index = pattern.find_any_of("#0"sv, String::SearchDirection::Forward); start_number_index.has_value()) {
  141. auto end_number_index = *start_number_index + 1;
  142. for (; end_number_index < pattern.length(); ++end_number_index) {
  143. auto ch = pattern[end_number_index];
  144. if ((ch != '#') && (ch != '0') && (ch != ',') && (ch != '.'))
  145. break;
  146. }
  147. if (number_system_for_groupings) {
  148. auto number_pattern = pattern.substring_view(*start_number_index, end_number_index - *start_number_index);
  149. auto group_separators = number_pattern.find_all(","sv);
  150. VERIFY((group_separators.size() == 1) || (group_separators.size() == 2));
  151. auto decimal = number_pattern.find('.');
  152. VERIFY(decimal.has_value());
  153. if (group_separators.size() == 1) {
  154. number_system_for_groupings->primary_grouping_size = *decimal - group_separators[0] - 1;
  155. number_system_for_groupings->secondary_grouping_size = number_system_for_groupings->primary_grouping_size;
  156. } else {
  157. number_system_for_groupings->primary_grouping_size = *decimal - group_separators[1] - 1;
  158. number_system_for_groupings->secondary_grouping_size = group_separators[1] - group_separators[0] - 1;
  159. }
  160. }
  161. pattern = String::formatted("{}{{number}}{}",
  162. *start_number_index > 0 ? pattern.substring_view(0, *start_number_index) : ""sv,
  163. pattern.substring_view(end_number_index));
  164. // This is specifically handled here rather than in the replacements HashMap above so
  165. // that we do not errantly replace zeroes in number patterns.
  166. if (pattern.contains(*replacements.get("E"sv)))
  167. pattern = pattern.replace("0"sv, "{scientificExponent}"sv);
  168. }
  169. if (type == NumberFormatType::Compact)
  170. return parse_identifiers(move(pattern), "compactIdentifier"sv, locale_data, format);
  171. return pattern;
  172. };
  173. auto zero_format = replace_patterns(move(patterns[0]));
  174. format.positive_format_index = locale_data.unique_strings.ensure(String::formatted("{{plusSign}}{}", zero_format));
  175. if (patterns.size() == 2) {
  176. auto negative_format = replace_patterns(move(patterns[1]));
  177. format.negative_format_index = locale_data.unique_strings.ensure(move(negative_format));
  178. } else {
  179. format.negative_format_index = locale_data.unique_strings.ensure(String::formatted("{{minusSign}}{}", zero_format));
  180. }
  181. format.zero_format_index = locale_data.unique_strings.ensure(move(zero_format));
  182. }
  183. static void parse_number_systems(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
  184. {
  185. LexicalPath numbers_path(move(locale_numbers_path));
  186. numbers_path = numbers_path.append("numbers.json"sv);
  187. VERIFY(Core::File::exists(numbers_path.string()));
  188. auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
  189. VERIFY(!numbers_file_or_error.is_error());
  190. auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
  191. VERIFY(numbers.has_value());
  192. auto const& main_object = numbers->as_object().get("main"sv);
  193. auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
  194. auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
  195. auto ensure_number_system = [&](auto const& system) -> NumberSystem& {
  196. return locale.number_systems.ensure(system, [&]() {
  197. auto system_index = locale_data.unique_strings.ensure(system);
  198. return NumberSystem { .system = system_index };
  199. });
  200. };
  201. auto parse_number_format = [&](auto const& format_object) {
  202. Vector<NumberFormat> result;
  203. result.ensure_capacity(format_object.size());
  204. format_object.for_each_member([&](auto const& key, JsonValue const& value) {
  205. auto split_key = key.split_view('-');
  206. if (split_key.size() != 3)
  207. return;
  208. auto patterns = value.as_string().split(';');
  209. NumberFormat format {};
  210. if (auto type = split_key[0].template to_uint<u64>(); type.has_value()) {
  211. VERIFY(*type % 10 == 0);
  212. format.magnitude = static_cast<u8>(log10(*type));
  213. if (patterns[0] != "0"sv) {
  214. auto number_of_zeroes_in_pattern = patterns[0].count("0"sv);
  215. VERIFY(format.magnitude >= number_of_zeroes_in_pattern);
  216. format.exponent = format.magnitude + 1 - number_of_zeroes_in_pattern;
  217. }
  218. } else {
  219. VERIFY(split_key[0] == "unitPattern"sv);
  220. }
  221. format.plurality = NumberFormat::plurality_from_string(split_key[2]);
  222. parse_number_pattern(move(patterns), locale_data, NumberFormatType::Compact, format);
  223. result.append(move(format));
  224. });
  225. return result;
  226. };
  227. locale_numbers_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
  228. constexpr auto symbols_prefix = "symbols-numberSystem-"sv;
  229. constexpr auto decimal_formats_prefix = "decimalFormats-numberSystem-"sv;
  230. constexpr auto currency_formats_prefix = "currencyFormats-numberSystem-"sv;
  231. constexpr auto percent_formats_prefix = "percentFormats-numberSystem-"sv;
  232. constexpr auto scientific_formats_prefix = "scientificFormats-numberSystem-"sv;
  233. if (key.starts_with(symbols_prefix)) {
  234. auto system = key.substring(symbols_prefix.length());
  235. auto& number_system = ensure_number_system(system);
  236. value.as_object().for_each_member([&](auto const& symbol, JsonValue const& localization) {
  237. auto symbol_index = locale_data.unique_strings.ensure(localization.as_string());
  238. number_system.symbols.set(symbol, symbol_index);
  239. if (!locale_data.numeric_symbols.contains_slow(symbol))
  240. locale_data.numeric_symbols.append(symbol);
  241. });
  242. } else if (key.starts_with(decimal_formats_prefix)) {
  243. auto system = key.substring(decimal_formats_prefix.length());
  244. auto& number_system = ensure_number_system(system);
  245. auto format_object = value.as_object().get("standard"sv);
  246. parse_number_pattern(format_object.as_string().split(';'), locale_data, NumberFormatType::Standard, number_system.decimal_format, &number_system);
  247. auto const& long_format = value.as_object().get("long"sv).as_object().get("decimalFormat"sv);
  248. number_system.decimal_long_formats = parse_number_format(long_format.as_object());
  249. auto const& short_format = value.as_object().get("short"sv).as_object().get("decimalFormat"sv);
  250. number_system.decimal_short_formats = parse_number_format(short_format.as_object());
  251. } else if (key.starts_with(currency_formats_prefix)) {
  252. auto system = key.substring(currency_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.currency_format);
  256. format_object = value.as_object().get("accounting"sv);
  257. parse_number_pattern(format_object.as_string().split(';'), locale_data, NumberFormatType::Standard, number_system.accounting_format);
  258. number_system.currency_unit_formats = parse_number_format(value.as_object());
  259. if (value.as_object().has("short"sv)) {
  260. auto const& short_format = value.as_object().get("short"sv).as_object().get("standard"sv);
  261. number_system.currency_short_formats = parse_number_format(short_format.as_object());
  262. }
  263. } else if (key.starts_with(percent_formats_prefix)) {
  264. auto system = key.substring(percent_formats_prefix.length());
  265. auto& number_system = ensure_number_system(system);
  266. auto format_object = value.as_object().get("standard"sv);
  267. parse_number_pattern(format_object.as_string().split(';'), locale_data, NumberFormatType::Standard, number_system.percent_format);
  268. } else if (key.starts_with(scientific_formats_prefix)) {
  269. auto system = key.substring(scientific_formats_prefix.length());
  270. auto& number_system = ensure_number_system(system);
  271. auto format_object = value.as_object().get("standard"sv);
  272. parse_number_pattern(format_object.as_string().split(';'), locale_data, NumberFormatType::Standard, number_system.scientific_format);
  273. }
  274. });
  275. }
  276. static void parse_all_locales(String core_path, String numbers_path, UnicodeLocaleData& locale_data)
  277. {
  278. auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
  279. auto remove_variants_from_path = [&](String path) -> Optional<String> {
  280. auto parsed_locale = CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path));
  281. if (!parsed_locale.has_value())
  282. return {};
  283. StringBuilder builder;
  284. builder.append(locale_data.unique_strings.get(parsed_locale->language));
  285. if (auto script = locale_data.unique_strings.get(parsed_locale->script); !script.is_empty())
  286. builder.appendff("-{}", script);
  287. if (auto region = locale_data.unique_strings.get(parsed_locale->region); !region.is_empty())
  288. builder.appendff("-{}", region);
  289. return builder.build();
  290. };
  291. while (numbers_iterator.has_next()) {
  292. auto numbers_path = numbers_iterator.next_full_path();
  293. VERIFY(Core::File::is_directory(numbers_path));
  294. auto language = remove_variants_from_path(numbers_path);
  295. if (!language.has_value())
  296. continue;
  297. auto& locale = locale_data.locales.ensure(*language);
  298. parse_number_systems(numbers_path, locale_data, locale);
  299. }
  300. parse_default_content_locales(move(core_path), locale_data);
  301. }
  302. static String format_identifier(StringView owner, String identifier)
  303. {
  304. identifier = identifier.replace("-"sv, "_"sv, true);
  305. if (all_of(identifier, is_ascii_digit))
  306. return String::formatted("{}_{}", owner[0], identifier);
  307. if (is_ascii_lower_alpha(identifier[0]))
  308. return String::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  309. return identifier;
  310. }
  311. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  312. {
  313. StringBuilder builder;
  314. SourceGenerator generator { builder };
  315. generator.append(R"~~~(
  316. #pragma once
  317. #include <AK/Optional.h>
  318. #include <AK/StringView.h>
  319. #include <AK/Types.h>
  320. #include <AK/Vector.h>
  321. #include <LibUnicode/Forward.h>
  322. namespace Unicode {
  323. )~~~");
  324. generate_enum(generator, format_identifier, "NumericSymbol"sv, {}, locale_data.numeric_symbols);
  325. generator.append(R"~~~(
  326. namespace Detail {
  327. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView numeric_symbol);
  328. Optional<NumberGroupings> get_number_system_groupings(StringView locale, StringView system);
  329. Optional<NumberFormat> get_standard_number_system_format(StringView locale, StringView system, StandardNumberFormatType type);
  330. Vector<NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type);
  331. Optional<NumericSymbol> numeric_symbol_from_string(StringView numeric_symbol);
  332. }
  333. }
  334. )~~~");
  335. file.write(generator.as_string_view());
  336. }
  337. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  338. {
  339. StringBuilder builder;
  340. SourceGenerator generator { builder };
  341. generator.set("string_index_type"sv, s_string_index_type);
  342. generator.set("numeric_symbols_size", String::number(locale_data.numeric_symbols.size()));
  343. generator.set("identifier_count", String::number(locale_data.max_identifier_count));
  344. generator.append(R"~~~(
  345. #include <AK/Array.h>
  346. #include <AK/BinarySearch.h>
  347. #include <AK/Span.h>
  348. #include <LibUnicode/Locale.h>
  349. #include <LibUnicode/UnicodeNumberFormat.h>
  350. namespace Unicode::Detail {
  351. )~~~");
  352. locale_data.unique_strings.generate(generator);
  353. generator.append(R"~~~(
  354. struct NumberFormat {
  355. Unicode::NumberFormat to_unicode_number_format() const {
  356. Unicode::NumberFormat number_format {};
  357. number_format.magnitude = magnitude;
  358. number_format.exponent = exponent;
  359. number_format.plurality = static_cast<Unicode::NumberFormat::Plurality>(plurality);
  360. number_format.zero_format = s_string_list[zero_format];
  361. number_format.positive_format = s_string_list[positive_format];
  362. number_format.negative_format = s_string_list[negative_format];
  363. number_format.identifiers.ensure_capacity(identifiers.size());
  364. for (@string_index_type@ identifier : identifiers)
  365. number_format.identifiers.append(s_string_list[identifier]);
  366. return number_format;
  367. }
  368. u8 magnitude { 0 };
  369. u8 exponent { 0 };
  370. u8 plurality { 0 };
  371. @string_index_type@ zero_format { 0 };
  372. @string_index_type@ positive_format { 0 };
  373. @string_index_type@ negative_format { 0 };
  374. Array<@string_index_type@, @identifier_count@> identifiers {};
  375. };
  376. struct NumberSystem {
  377. @string_index_type@ system { 0 };
  378. Array<@string_index_type@, @numeric_symbols_size@> symbols {};
  379. u8 primary_grouping_size { 0 };
  380. u8 secondary_grouping_size { 0 };
  381. NumberFormat decimal_format {};
  382. Span<NumberFormat const> decimal_long_formats {};
  383. Span<NumberFormat const> decimal_short_formats {};
  384. NumberFormat currency_format {};
  385. NumberFormat accounting_format {};
  386. Span<NumberFormat const> currency_unit_formats {};
  387. Span<NumberFormat const> currency_short_formats {};
  388. NumberFormat percent_format {};
  389. NumberFormat scientific_format {};
  390. };
  391. )~~~");
  392. auto append_number_format = [&](auto const& number_format) {
  393. generator.set("magnitude"sv, String::number(number_format.magnitude));
  394. generator.set("exponent"sv, String::number(number_format.exponent));
  395. generator.set("plurality"sv, String::number(static_cast<u8>(number_format.plurality)));
  396. generator.set("zero_format"sv, String::number(number_format.zero_format_index));
  397. generator.set("positive_format"sv, String::number(number_format.positive_format_index));
  398. generator.set("negative_format"sv, String::number(number_format.negative_format_index));
  399. generator.append("{ @magnitude@, @exponent@, @plurality@, @zero_format@, @positive_format@, @negative_format@, { ");
  400. bool first = true;
  401. for (auto identifier_index : number_format.identifier_indices) {
  402. if (!first)
  403. generator.append(", ");
  404. generator.append(String::number(identifier_index));
  405. first = false;
  406. }
  407. generator.append(" } },");
  408. };
  409. auto append_number_formats = [&](String name, auto const& number_formats) {
  410. generator.set("name"sv, move(name));
  411. generator.set("size"sv, String::number(number_formats.size()));
  412. generator.append(R"~~~(
  413. static constexpr Array<NumberFormat, @size@> @name@ { {
  414. )~~~");
  415. constexpr size_t max_values_per_row = 10;
  416. size_t values_in_current_row = 0;
  417. for (auto const& number_format : number_formats) {
  418. if (values_in_current_row++ > 0)
  419. generator.append(" ");
  420. append_number_format(number_format);
  421. if (values_in_current_row == max_values_per_row) {
  422. values_in_current_row = 0;
  423. generator.append("\n ");
  424. }
  425. }
  426. generator.append(R"~~~(
  427. } };
  428. )~~~");
  429. };
  430. auto append_number_systems = [&](String name, auto const& number_systems) {
  431. auto format_name = [&](StringView system, StringView format) {
  432. return String::formatted("{}_{}_{}", name, system, format);
  433. };
  434. for (auto const& number_system : number_systems) {
  435. append_number_formats(format_name(number_system.key, "dl"sv), number_system.value.decimal_long_formats);
  436. append_number_formats(format_name(number_system.key, "ds"sv), number_system.value.decimal_short_formats);
  437. append_number_formats(format_name(number_system.key, "cu"sv), number_system.value.currency_unit_formats);
  438. append_number_formats(format_name(number_system.key, "cs"sv), number_system.value.currency_short_formats);
  439. }
  440. generator.set("name", name);
  441. generator.set("size", String::number(number_systems.size()));
  442. generator.append(R"~~~(
  443. static constexpr Array<NumberSystem, @size@> @name@ { {)~~~");
  444. for (auto const& number_system : number_systems) {
  445. generator.set("system"sv, String::number(number_system.value.system));
  446. generator.set("primary_grouping_size"sv, String::number(number_system.value.primary_grouping_size));
  447. generator.set("secondary_grouping_size"sv, String::number(number_system.value.secondary_grouping_size));
  448. generator.set("decimal_long_formats"sv, format_name(number_system.key, "dl"sv));
  449. generator.set("decimal_short_formats"sv, format_name(number_system.key, "ds"sv));
  450. generator.set("currency_unit_formats"sv, format_name(number_system.key, "cu"sv));
  451. generator.set("currency_short_formats"sv, format_name(number_system.key, "cs"sv));
  452. generator.append(R"~~~(
  453. { @system@, {)~~~");
  454. for (auto const& symbol : locale_data.numeric_symbols) {
  455. auto index = number_system.value.symbols.get(symbol).value_or(0);
  456. generator.set("index", String::number(index));
  457. generator.append(" @index@,");
  458. }
  459. generator.append(" }, @primary_grouping_size@, @secondary_grouping_size@, ");
  460. append_number_format(number_system.value.decimal_format);
  461. generator.append(" @decimal_long_formats@.span(), @decimal_short_formats@.span(), ");
  462. append_number_format(number_system.value.currency_format);
  463. generator.append(" ");
  464. append_number_format(number_system.value.accounting_format);
  465. generator.append(" @currency_unit_formats@.span(), @currency_short_formats@.span(), ");
  466. append_number_format(number_system.value.percent_format);
  467. generator.append(" ");
  468. append_number_format(number_system.value.scientific_format);
  469. generator.append(" },");
  470. }
  471. generator.append(R"~~~(
  472. } };
  473. )~~~");
  474. };
  475. 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); });
  476. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values) {
  477. HashValueMap<String> hashes;
  478. hashes.ensure_capacity(values.size());
  479. for (auto const& value : values)
  480. hashes.set(value.hash(), format_identifier(enum_title, value));
  481. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  482. };
  483. append_from_string("NumericSymbol"sv, "numeric_symbol"sv, locale_data.numeric_symbols);
  484. generator.append(R"~~~(
  485. static NumberSystem const* find_number_system(StringView locale, StringView system)
  486. {
  487. auto locale_value = locale_from_string(locale);
  488. if (!locale_value.has_value())
  489. return nullptr;
  490. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  491. auto const& number_systems = s_number_systems.at(locale_index);
  492. for (auto const& number_system : number_systems) {
  493. if (system == s_string_list[number_system.system])
  494. return &number_system;
  495. };
  496. return nullptr;
  497. }
  498. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView symbol)
  499. {
  500. auto symbol_value = numeric_symbol_from_string(symbol);
  501. if (!symbol_value.has_value())
  502. return {};
  503. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  504. auto symbol_index = to_underlying(*symbol_value);
  505. return s_string_list[number_system->symbols[symbol_index]];
  506. }
  507. return {};
  508. }
  509. Optional<NumberGroupings> get_number_system_groupings(StringView locale, StringView system)
  510. {
  511. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr)
  512. return NumberGroupings { number_system->primary_grouping_size, number_system->secondary_grouping_size };
  513. return {};
  514. }
  515. Optional<Unicode::NumberFormat> get_standard_number_system_format(StringView locale, StringView system, StandardNumberFormatType type)
  516. {
  517. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  518. switch (type) {
  519. case StandardNumberFormatType::Decimal:
  520. return number_system->decimal_format.to_unicode_number_format();
  521. case StandardNumberFormatType::Currency:
  522. return number_system->currency_format.to_unicode_number_format();
  523. case StandardNumberFormatType::Accounting:
  524. return number_system->accounting_format.to_unicode_number_format();
  525. case StandardNumberFormatType::Percent:
  526. return number_system->percent_format.to_unicode_number_format();
  527. case StandardNumberFormatType::Scientific:
  528. return number_system->scientific_format.to_unicode_number_format();
  529. }
  530. }
  531. return {};
  532. }
  533. Vector<Unicode::NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type)
  534. {
  535. Vector<Unicode::NumberFormat> formats;
  536. if (auto const* number_system = find_number_system(locale, system); number_system != nullptr) {
  537. Span<NumberFormat const> number_formats;
  538. switch (type) {
  539. case CompactNumberFormatType::DecimalLong:
  540. number_formats = number_system->decimal_long_formats;
  541. break;
  542. case CompactNumberFormatType::DecimalShort:
  543. number_formats = number_system->decimal_short_formats;
  544. break;
  545. case CompactNumberFormatType::CurrencyUnit:
  546. number_formats = number_system->currency_unit_formats;
  547. break;
  548. case CompactNumberFormatType::CurrencyShort:
  549. number_formats = number_system->currency_short_formats;
  550. break;
  551. }
  552. formats.ensure_capacity(number_formats.size());
  553. for (auto const& number_format : number_formats)
  554. formats.append(number_format.to_unicode_number_format());
  555. }
  556. return formats;
  557. }
  558. }
  559. )~~~");
  560. file.write(generator.as_string_view());
  561. }
  562. int main(int argc, char** argv)
  563. {
  564. char const* generated_header_path = nullptr;
  565. char const* generated_implementation_path = nullptr;
  566. char const* core_path = nullptr;
  567. char const* numbers_path = nullptr;
  568. Core::ArgsParser args_parser;
  569. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  570. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  571. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  572. args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
  573. args_parser.parse(argc, argv);
  574. auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
  575. if (path.is_empty()) {
  576. warnln("{} is required", flags);
  577. args_parser.print_usage(stderr, argv[0]);
  578. exit(1);
  579. }
  580. auto file_or_error = Core::File::open(path, mode);
  581. if (file_or_error.is_error()) {
  582. warnln("Failed to open {}: {}", path, file_or_error.release_error());
  583. exit(1);
  584. }
  585. return file_or_error.release_value();
  586. };
  587. auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
  588. auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
  589. UnicodeLocaleData locale_data;
  590. parse_all_locales(core_path, numbers_path, locale_data);
  591. generate_unicode_locale_header(generated_header_file, locale_data);
  592. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  593. return 0;
  594. }