GenerateUnicodeDateTimeFormat.cpp 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377
  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/Format.h>
  8. #include <AK/HashMap.h>
  9. #include <AK/JsonObject.h>
  10. #include <AK/JsonParser.h>
  11. #include <AK/JsonValue.h>
  12. #include <AK/LexicalPath.h>
  13. #include <AK/SourceGenerator.h>
  14. #include <AK/String.h>
  15. #include <AK/StringBuilder.h>
  16. #include <LibCore/ArgsParser.h>
  17. #include <LibCore/DirIterator.h>
  18. #include <LibCore/File.h>
  19. #include <LibUnicode/DateTimeFormat.h>
  20. using StringIndexType = u16;
  21. constexpr auto s_string_index_type = "u16"sv;
  22. struct CalendarPattern : public Unicode::CalendarPattern {
  23. StringIndexType pattern_index { 0 };
  24. };
  25. struct CalendarFormat {
  26. CalendarPattern full_format {};
  27. CalendarPattern long_format {};
  28. CalendarPattern medium_format {};
  29. CalendarPattern short_format {};
  30. };
  31. struct Calendar {
  32. StringIndexType calendar { 0 };
  33. CalendarFormat date_formats {};
  34. CalendarFormat time_formats {};
  35. CalendarFormat date_time_formats {};
  36. };
  37. struct Locale {
  38. HashMap<String, Calendar> calendars;
  39. };
  40. struct UnicodeLocaleData {
  41. UniqueStringStorage<StringIndexType> unique_strings;
  42. HashMap<String, Locale> locales;
  43. Vector<String> calendars;
  44. };
  45. static void parse_date_time_pattern(CalendarPattern& format, String pattern, UnicodeLocaleData& locale_data)
  46. {
  47. // FIXME: This is very incomplete. Similar to NumberFormat, the pattern string will need to be
  48. // parsed to fill in the CalendarPattern struct, and modified to be useable at runtime.
  49. // For now, this is enough to implement the DateTimeFormat constructor.
  50. //
  51. // https://unicode.org/reports/tr35/tr35-dates.html#Date_Field_Symbol_Table
  52. format.pattern_index = locale_data.unique_strings.ensure(move(pattern));
  53. }
  54. static ErrorOr<void> parse_calendars(String locale_calendars_path, UnicodeLocaleData& locale_data, Locale& locale)
  55. {
  56. LexicalPath calendars_path(move(locale_calendars_path));
  57. if (!calendars_path.basename().starts_with("ca-"sv))
  58. return {};
  59. auto calendars_file = TRY(Core::File::open(calendars_path.string(), Core::OpenMode::ReadOnly));
  60. auto calendars = TRY(JsonValue::from_string(calendars_file->read_all()));
  61. auto const& main_object = calendars.as_object().get("main"sv);
  62. auto const& locale_object = main_object.as_object().get(calendars_path.parent().basename());
  63. auto const& dates_object = locale_object.as_object().get("dates"sv);
  64. auto const& calendars_object = dates_object.as_object().get("calendars"sv);
  65. auto ensure_calendar = [&](auto const& calendar) -> Calendar& {
  66. return locale.calendars.ensure(calendar, [&]() {
  67. auto calendar_index = locale_data.unique_strings.ensure(calendar);
  68. return Calendar { .calendar = calendar_index };
  69. });
  70. };
  71. auto parse_patterns = [&](auto& formats, auto const& patterns_object) {
  72. auto full_format = patterns_object.get("full"sv);
  73. parse_date_time_pattern(formats.full_format, full_format.as_string(), locale_data);
  74. auto long_format = patterns_object.get("long"sv);
  75. parse_date_time_pattern(formats.long_format, long_format.as_string(), locale_data);
  76. auto medium_format = patterns_object.get("medium"sv);
  77. parse_date_time_pattern(formats.medium_format, medium_format.as_string(), locale_data);
  78. auto short_format = patterns_object.get("short"sv);
  79. parse_date_time_pattern(formats.short_format, short_format.as_string(), locale_data);
  80. };
  81. calendars_object.as_object().for_each_member([&](auto const& calendar_name, JsonValue const& value) {
  82. auto& calendar = ensure_calendar(calendar_name);
  83. if (!locale_data.calendars.contains_slow(calendar_name))
  84. locale_data.calendars.append(calendar_name);
  85. auto const& date_formats_object = value.as_object().get("dateFormats"sv);
  86. parse_patterns(calendar.date_formats, date_formats_object.as_object());
  87. auto const& time_formats_object = value.as_object().get("timeFormats"sv);
  88. parse_patterns(calendar.time_formats, time_formats_object.as_object());
  89. auto const& date_time_formats_object = value.as_object().get("dateTimeFormats"sv);
  90. parse_patterns(calendar.date_time_formats, date_time_formats_object.as_object());
  91. });
  92. return {};
  93. }
  94. static ErrorOr<void> parse_all_locales(String dates_path, UnicodeLocaleData& locale_data)
  95. {
  96. auto dates_iterator = TRY(path_to_dir_iterator(move(dates_path)));
  97. auto remove_variants_from_path = [&](String path) -> ErrorOr<String> {
  98. auto parsed_locale = TRY(CanonicalLanguageID<StringIndexType>::parse(locale_data.unique_strings, LexicalPath::basename(path)));
  99. StringBuilder builder;
  100. builder.append(locale_data.unique_strings.get(parsed_locale.language));
  101. if (auto script = locale_data.unique_strings.get(parsed_locale.script); !script.is_empty())
  102. builder.appendff("-{}", script);
  103. if (auto region = locale_data.unique_strings.get(parsed_locale.region); !region.is_empty())
  104. builder.appendff("-{}", region);
  105. return builder.build();
  106. };
  107. while (dates_iterator.has_next()) {
  108. auto dates_path = TRY(next_path_from_dir_iterator(dates_iterator));
  109. auto calendars_iterator = TRY(path_to_dir_iterator(dates_path, {}));
  110. auto language = TRY(remove_variants_from_path(dates_path));
  111. auto& locale = locale_data.locales.ensure(language);
  112. while (calendars_iterator.has_next()) {
  113. auto calendars_path = TRY(next_path_from_dir_iterator(calendars_iterator));
  114. TRY(parse_calendars(move(calendars_path), locale_data, locale));
  115. }
  116. }
  117. return {};
  118. }
  119. static String format_identifier(StringView, StringView identifier)
  120. {
  121. return identifier.to_titlecase_string();
  122. }
  123. static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
  124. {
  125. StringBuilder builder;
  126. SourceGenerator generator { builder };
  127. generator.append(R"~~~(
  128. #pragma once
  129. #include <AK/Optional.h>
  130. #include <AK/StringView.h>
  131. #include <LibUnicode/Forward.h>
  132. namespace Unicode {
  133. )~~~");
  134. generate_enum(generator, format_identifier, "Calendar"sv, {}, locale_data.calendars);
  135. generator.append(R"~~~(
  136. namespace Detail {
  137. Optional<Calendar> calendar_from_string(StringView calendar);
  138. Optional<Unicode::CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar);
  139. Optional<Unicode::CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar);
  140. Optional<Unicode::CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar);
  141. }
  142. }
  143. )~~~");
  144. VERIFY(file.write(generator.as_string_view()));
  145. }
  146. static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
  147. {
  148. StringBuilder builder;
  149. SourceGenerator generator { builder };
  150. generator.set("string_index_type"sv, s_string_index_type);
  151. generator.append(R"~~~(
  152. #include <AK/Array.h>
  153. #include <AK/BinarySearch.h>
  154. #include <LibUnicode/DateTimeFormat.h>
  155. #include <LibUnicode/Locale.h>
  156. #include <LibUnicode/UnicodeDateTimeFormat.h>
  157. namespace Unicode::Detail {
  158. )~~~");
  159. locale_data.unique_strings.generate(generator);
  160. generator.append(R"~~~(
  161. struct CalendarPattern {
  162. Unicode::CalendarPattern to_unicode_calendar_pattern() const {
  163. Unicode::CalendarPattern calendar_pattern {};
  164. calendar_pattern.pattern = s_string_list[pattern];
  165. return calendar_pattern;
  166. }
  167. @string_index_type@ pattern { 0 };
  168. };
  169. struct CalendarFormat {
  170. Unicode::CalendarFormat to_unicode_calendar_format() const {
  171. Unicode::CalendarFormat calendar_format {};
  172. calendar_format.full_format = full_format.to_unicode_calendar_pattern();
  173. calendar_format.long_format = long_format.to_unicode_calendar_pattern();
  174. calendar_format.medium_format = medium_format.to_unicode_calendar_pattern();
  175. calendar_format.short_format = short_format.to_unicode_calendar_pattern();
  176. return calendar_format;
  177. }
  178. CalendarPattern full_format {};
  179. CalendarPattern long_format {};
  180. CalendarPattern medium_format {};
  181. CalendarPattern short_format {};
  182. };
  183. struct CalendarData {
  184. @string_index_type@ calendar { 0 };
  185. CalendarFormat date_formats {};
  186. CalendarFormat time_formats {};
  187. CalendarFormat date_time_formats {};
  188. };
  189. )~~~");
  190. auto append_calendar_pattern = [&](auto const& calendar_pattern) {
  191. generator.set("pattern"sv, String::number(calendar_pattern.pattern_index));
  192. generator.append("{ @pattern@ },");
  193. };
  194. auto append_calendar_format = [&](auto const& calendar_format) {
  195. generator.append("{ ");
  196. append_calendar_pattern(calendar_format.full_format);
  197. generator.append(" ");
  198. append_calendar_pattern(calendar_format.long_format);
  199. generator.append(" ");
  200. append_calendar_pattern(calendar_format.medium_format);
  201. generator.append(" ");
  202. append_calendar_pattern(calendar_format.short_format);
  203. generator.append(" },");
  204. };
  205. auto append_calendars = [&](String name, auto const& calendars) {
  206. generator.set("name", name);
  207. generator.set("size", String::number(calendars.size()));
  208. generator.append(R"~~~(
  209. static constexpr Array<CalendarData, @size@> @name@ { {)~~~");
  210. for (auto const& calendar_key : locale_data.calendars) {
  211. auto const& calendar = calendars.find(calendar_key)->value;
  212. generator.set("calendar"sv, String::number(calendar.calendar));
  213. generator.append(R"~~~(
  214. { @calendar@, )~~~");
  215. append_calendar_format(calendar.date_formats);
  216. generator.append(" ");
  217. append_calendar_format(calendar.time_formats);
  218. generator.append(" ");
  219. append_calendar_format(calendar.date_time_formats);
  220. generator.append(" },");
  221. }
  222. generator.append(R"~~~(
  223. } };
  224. )~~~");
  225. };
  226. generate_mapping(generator, locale_data.locales, "CalendarData"sv, "s_calendars"sv, "s_calendars_{}", [&](auto const& name, auto const& value) { append_calendars(name, value.calendars); });
  227. auto append_from_string = [&](StringView enum_title, StringView enum_snake, auto const& values) {
  228. HashValueMap<String> hashes;
  229. hashes.ensure_capacity(values.size());
  230. for (auto const& value : values)
  231. hashes.set(value.hash(), format_identifier(enum_title, value));
  232. generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
  233. };
  234. append_from_string("Calendar"sv, "calendar"sv, locale_data.calendars);
  235. generator.append(R"~~~(
  236. static CalendarData const* find_calendar_data(StringView locale, StringView calendar)
  237. {
  238. auto locale_value = locale_from_string(locale);
  239. if (!locale_value.has_value())
  240. return nullptr;
  241. auto calendar_value = calendar_from_string(calendar);
  242. if (!calendar_value.has_value())
  243. return nullptr;
  244. auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
  245. auto calendar_index = to_underlying(*calendar_value);
  246. auto const& calendars = s_calendars.at(locale_index);
  247. return &calendars[calendar_index];
  248. }
  249. Optional<Unicode::CalendarFormat> get_calendar_date_format(StringView locale, StringView calendar)
  250. {
  251. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr)
  252. return data->date_formats.to_unicode_calendar_format();
  253. return {};
  254. }
  255. Optional<Unicode::CalendarFormat> get_calendar_time_format(StringView locale, StringView calendar)
  256. {
  257. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr)
  258. return data->time_formats.to_unicode_calendar_format();
  259. return {};
  260. }
  261. Optional<Unicode::CalendarFormat> get_calendar_date_time_format(StringView locale, StringView calendar)
  262. {
  263. if (auto const* data = find_calendar_data(locale, calendar); data != nullptr)
  264. return data->date_time_formats.to_unicode_calendar_format();
  265. return {};
  266. }
  267. }
  268. )~~~");
  269. VERIFY(file.write(generator.as_string_view()));
  270. }
  271. ErrorOr<int> serenity_main(Main::Arguments arguments)
  272. {
  273. StringView generated_header_path;
  274. StringView generated_implementation_path;
  275. StringView dates_path;
  276. Core::ArgsParser args_parser;
  277. args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
  278. args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
  279. args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
  280. args_parser.parse(arguments);
  281. auto open_file = [&](StringView path) -> ErrorOr<NonnullRefPtr<Core::File>> {
  282. if (path.is_empty()) {
  283. args_parser.print_usage(stderr, arguments.argv[0]);
  284. return Error::from_string_literal("Must provide all command line options"sv);
  285. }
  286. return Core::File::open(path, Core::OpenMode::ReadWrite);
  287. };
  288. auto generated_header_file = TRY(open_file(generated_header_path));
  289. auto generated_implementation_file = TRY(open_file(generated_implementation_path));
  290. UnicodeLocaleData locale_data;
  291. TRY(parse_all_locales(dates_path, locale_data));
  292. generate_unicode_locale_header(generated_header_file, locale_data);
  293. generate_unicode_locale_implementation(generated_implementation_file, locale_data);
  294. return 0;
  295. }