GenerateDateTimeFormatData.cpp 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267
  1. /*
  2. * Copyright (c) 2021-2024, Tim Flynn <trflynn89@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include "../LibUnicode/GeneratorUtil.h" // FIXME: Move this somewhere common.
  7. #include <AK/AllOf.h>
  8. #include <AK/ByteString.h>
  9. #include <AK/CharacterTypes.h>
  10. #include <AK/Error.h>
  11. #include <AK/Format.h>
  12. #include <AK/HashMap.h>
  13. #include <AK/JsonObject.h>
  14. #include <AK/JsonParser.h>
  15. #include <AK/JsonValue.h>
  16. #include <AK/LexicalPath.h>
  17. #include <AK/SourceGenerator.h>
  18. #include <AK/StringBuilder.h>
  19. #include <LibCore/ArgsParser.h>
  20. #include <LibCore/Directory.h>
  21. #include <LibLocale/DateTimeFormat.h>
  22. struct CLDR {
  23. HashMap<ByteString, u8> minimum_days;
  24. Vector<ByteString> minimum_days_regions;
  25. HashMap<ByteString, Locale::Weekday> first_day;
  26. Vector<ByteString> first_day_regions;
  27. HashMap<ByteString, Locale::Weekday> weekend_start;
  28. Vector<ByteString> weekend_start_regions;
  29. HashMap<ByteString, Locale::Weekday> weekend_end;
  30. Vector<ByteString> weekend_end_regions;
  31. };
  32. static ErrorOr<void> parse_week_data(ByteString core_path, CLDR& cldr)
  33. {
  34. // https://unicode.org/reports/tr35/tr35-dates.html#Week_Data
  35. LexicalPath week_data_path(move(core_path));
  36. week_data_path = week_data_path.append("supplemental"sv);
  37. week_data_path = week_data_path.append("weekData.json"sv);
  38. auto week_data = TRY(read_json_file(week_data_path.string()));
  39. auto const& supplemental_object = week_data.as_object().get_object("supplemental"sv).value();
  40. auto const& week_data_object = supplemental_object.get_object("weekData"sv).value();
  41. auto parse_weekday = [](StringView day) -> Locale::Weekday {
  42. if (day == "sun"sv)
  43. return Locale::Weekday::Sunday;
  44. if (day == "mon"sv)
  45. return Locale::Weekday::Monday;
  46. if (day == "tue"sv)
  47. return Locale::Weekday::Tuesday;
  48. if (day == "wed"sv)
  49. return Locale::Weekday::Wednesday;
  50. if (day == "thu"sv)
  51. return Locale::Weekday::Thursday;
  52. if (day == "fri"sv)
  53. return Locale::Weekday::Friday;
  54. if (day == "sat"sv)
  55. return Locale::Weekday::Saturday;
  56. VERIFY_NOT_REACHED();
  57. };
  58. auto parse_regional_weekdays = [&](auto const& region, auto const& weekday, auto& weekdays_map, auto& weekday_regions) {
  59. if (region.ends_with("alt-variant"sv))
  60. return;
  61. weekdays_map.set(region, parse_weekday(weekday));
  62. if (!weekday_regions.contains_slow(region))
  63. weekday_regions.append(region);
  64. };
  65. auto const& minimum_days_object = week_data_object.get_object("minDays"sv).value();
  66. auto const& first_day_object = week_data_object.get_object("firstDay"sv).value();
  67. auto const& weekend_start_object = week_data_object.get_object("weekendStart"sv).value();
  68. auto const& weekend_end_object = week_data_object.get_object("weekendEnd"sv).value();
  69. minimum_days_object.for_each_member([&](auto const& region, auto const& value) {
  70. auto minimum_days = value.as_string().template to_number<u8>();
  71. cldr.minimum_days.set(region, *minimum_days);
  72. if (!cldr.minimum_days_regions.contains_slow(region))
  73. cldr.minimum_days_regions.append(region);
  74. });
  75. first_day_object.for_each_member([&](auto const& region, auto const& value) {
  76. parse_regional_weekdays(region, value.as_string(), cldr.first_day, cldr.first_day_regions);
  77. });
  78. weekend_start_object.for_each_member([&](auto const& region, auto const& value) {
  79. parse_regional_weekdays(region, value.as_string(), cldr.weekend_start, cldr.weekend_start_regions);
  80. });
  81. weekend_end_object.for_each_member([&](auto const& region, auto const& value) {
  82. parse_regional_weekdays(region, value.as_string(), cldr.weekend_end, cldr.weekend_end_regions);
  83. });
  84. return {};
  85. }
  86. static ErrorOr<void> parse_all_locales(ByteString core_path, CLDR& cldr)
  87. {
  88. TRY(parse_week_data(core_path, cldr));
  89. return {};
  90. }
  91. static ByteString format_identifier(StringView owner, ByteString identifier)
  92. {
  93. identifier = identifier.replace("-"sv, "_"sv, ReplaceMode::All);
  94. identifier = identifier.replace("/"sv, "_"sv, ReplaceMode::All);
  95. if (all_of(identifier, is_ascii_digit))
  96. return ByteString::formatted("{}_{}", owner[0], identifier);
  97. if (is_ascii_lower_alpha(identifier[0]))
  98. return ByteString::formatted("{:c}{}", to_ascii_uppercase(identifier[0]), identifier.substring_view(1));
  99. return identifier;
  100. }
  101. static ErrorOr<void> generate_unicode_locale_header(Core::InputBufferedFile& file, CLDR& cldr)
  102. {
  103. StringBuilder builder;
  104. SourceGenerator generator { builder };
  105. generator.append(R"~~~(
  106. #pragma once
  107. #include <AK/Types.h>
  108. namespace Locale {
  109. )~~~");
  110. generate_enum(generator, format_identifier, "MinimumDaysRegion"sv, {}, cldr.minimum_days_regions);
  111. generate_enum(generator, format_identifier, "FirstDayRegion"sv, {}, cldr.first_day_regions);
  112. generate_enum(generator, format_identifier, "WeekendStartRegion"sv, {}, cldr.weekend_start_regions);
  113. generate_enum(generator, format_identifier, "WeekendEndRegion"sv, {}, cldr.weekend_end_regions);
  114. generator.append(R"~~~(
  115. }
  116. )~~~");
  117. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  118. return {};
  119. }
  120. static ErrorOr<void> generate_unicode_locale_implementation(Core::InputBufferedFile& file, CLDR& cldr)
  121. {
  122. StringBuilder builder;
  123. SourceGenerator generator { builder };
  124. generator.append(R"~~~(
  125. #include <AK/Array.h>
  126. #include <AK/BinarySearch.h>
  127. #include <AK/Error.h>
  128. #include <AK/Optional.h>
  129. #include <AK/String.h>
  130. #include <AK/StringView.h>
  131. #include <LibLocale/DateTimeFormat.h>
  132. #include <LibLocale/DateTimeFormatData.h>
  133. #include <LibLocale/Locale.h>
  134. #include <LibLocale/LocaleData.h>
  135. namespace Locale {
  136. )~~~");
  137. auto append_mapping = [&](auto const& keys, auto const& map, auto type, auto name, auto mapping_getter) {
  138. generator.set("type", type);
  139. generator.set("name", name);
  140. generator.set("size", ByteString::number(keys.size()));
  141. generator.append(R"~~~(
  142. static constexpr Array<@type@, @size@> @name@ { {)~~~");
  143. bool first = true;
  144. for (auto const& key : keys) {
  145. auto const& value = map.find(key)->value;
  146. auto mapping = mapping_getter(value);
  147. generator.append(first ? " "sv : ", "sv);
  148. generator.append(ByteString::number(mapping));
  149. first = false;
  150. }
  151. generator.append(" } };");
  152. };
  153. append_mapping(cldr.minimum_days_regions, cldr.minimum_days, "u8"sv, "s_minimum_days"sv, [](auto minimum_days) { return minimum_days; });
  154. append_mapping(cldr.first_day_regions, cldr.first_day, "u8"sv, "s_first_day"sv, [](auto first_day) { return to_underlying(first_day); });
  155. append_mapping(cldr.weekend_start_regions, cldr.weekend_start, "u8"sv, "s_weekend_start"sv, [](auto weekend_start) { return to_underlying(weekend_start); });
  156. append_mapping(cldr.weekend_end_regions, cldr.weekend_end, "u8"sv, "s_weekend_end"sv, [](auto weekend_end) { return to_underlying(weekend_end); });
  157. generator.append("\n");
  158. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values, Vector<Alias> const& aliases = {}) -> ErrorOr<void> {
  159. HashValueMap<ByteString> hashes;
  160. TRY(hashes.try_ensure_capacity(values.size()));
  161. for (auto const& value : values)
  162. hashes.set(value.hash(), format_identifier(enum_title, value));
  163. for (auto const& alias : aliases)
  164. hashes.set(alias.alias.hash(), format_identifier(enum_title, alias.alias));
  165. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  166. return {};
  167. };
  168. TRY(append_from_string("MinimumDaysRegion"sv, "minimum_days_region"sv, cldr.minimum_days_regions));
  169. TRY(append_from_string("FirstDayRegion"sv, "first_day_region"sv, cldr.first_day_regions));
  170. TRY(append_from_string("WeekendStartRegion"sv, "weekend_start_region"sv, cldr.weekend_start_regions));
  171. TRY(append_from_string("WeekendEndRegion"sv, "weekend_end_region"sv, cldr.weekend_end_regions));
  172. auto append_regional_lookup = [&](StringView return_type, StringView lookup_type) {
  173. generator.set("return_type", return_type);
  174. generator.set("lookup_type", lookup_type);
  175. generator.append(R"~~~(
  176. Optional<@return_type@> get_regional_@lookup_type@(StringView region)
  177. {
  178. auto region_value = @lookup_type@_region_from_string(region);
  179. if (!region_value.has_value())
  180. return {};
  181. auto region_index = to_underlying(*region_value);
  182. auto @lookup_type@ = s_@lookup_type@.at(region_index);
  183. return static_cast<@return_type@>(@lookup_type@);
  184. }
  185. )~~~");
  186. };
  187. append_regional_lookup("u8"sv, "minimum_days"sv);
  188. append_regional_lookup("Weekday"sv, "first_day"sv);
  189. append_regional_lookup("Weekday"sv, "weekend_start"sv);
  190. append_regional_lookup("Weekday"sv, "weekend_end"sv);
  191. generator.append(R"~~~(
  192. }
  193. )~~~");
  194. TRY(file.write_until_depleted(generator.as_string_view().bytes()));
  195. return {};
  196. }
  197. ErrorOr<int> serenity_main(Main::Arguments arguments)
  198. {
  199. StringView generated_header_path;
  200. StringView generated_implementation_path;
  201. StringView core_path;
  202. Core::ArgsParser args_parser;
  203. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  204. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  205. args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
  206. args_parser.parse(arguments);
  207. auto generated_header_file = TRY(open_file(generated_header_path, Core::File::OpenMode::Write));
  208. auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::File::OpenMode::Write));
  209. CLDR cldr;
  210. TRY(parse_all_locales(core_path, cldr));
  211. TRY(generate_unicode_locale_header(*generated_header_file, cldr));
  212. TRY(generate_unicode_locale_implementation(*generated_implementation_file, cldr));
  213. return 0;
  214. }