123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182 |
- /*
- * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
- *
- * SPDX-License-Identifier: BSD-2-Clause
- */
- #include <AK/AllOf.h>
- #include <AK/CharacterTypes.h>
- #include <AK/Format.h>
- #include <AK/HashMap.h>
- #include <AK/JsonObject.h>
- #include <AK/JsonParser.h>
- #include <AK/JsonValue.h>
- #include <AK/LexicalPath.h>
- #include <AK/QuickSort.h>
- #include <AK/SourceGenerator.h>
- #include <AK/String.h>
- #include <AK/StringBuilder.h>
- #include <LibCore/ArgsParser.h>
- #include <LibCore/DirIterator.h>
- #include <LibCore/File.h>
- #include <LibUnicode/Locale.h>
- struct ListPatterns {
- String type;
- String style;
- String start;
- String middle;
- String end;
- String pair;
- };
- struct Locale {
- String language;
- Optional<String> territory;
- Optional<String> variant;
- HashMap<String, String> languages;
- HashMap<String, String> territories;
- HashMap<String, String> scripts;
- HashMap<String, String> currencies;
- HashMap<String, String> keywords;
- Vector<ListPatterns> list_patterns;
- };
- struct CanonicalLanguageID {
- String language {};
- String script {};
- String region {};
- Vector<String> variants {};
- };
- struct LanguageMapping {
- CanonicalLanguageID key {};
- CanonicalLanguageID alias {};
- };
- struct UnicodeLocaleData {
- HashMap<String, Locale> locales;
- Vector<String> languages;
- Vector<String> territories;
- Vector<String> scripts;
- Vector<String> variants;
- Vector<String> currencies;
- Vector<String> keywords;
- Vector<String> list_pattern_types;
- Vector<String> list_pattern_styles;
- HashMap<String, String> language_aliases;
- HashMap<String, String> territory_aliases;
- HashMap<String, String> script_aliases;
- HashMap<String, String> variant_aliases;
- HashMap<String, String> subdivision_aliases;
- Vector<LanguageMapping> complex_mappings;
- Vector<LanguageMapping> likely_subtags;
- size_t max_variant_size { 0 };
- };
- static Optional<CanonicalLanguageID> parse_language(StringView language)
- {
- CanonicalLanguageID language_id {};
- auto segments = language.split_view('-');
- VERIFY(!segments.is_empty());
- size_t index = 0;
- if (Unicode::is_unicode_language_subtag(segments[index])) {
- language_id.language = segments[index];
- if (segments.size() == ++index)
- return language_id;
- } else {
- return {};
- }
- if (Unicode::is_unicode_script_subtag(segments[index])) {
- language_id.script = segments[index];
- if (segments.size() == ++index)
- return language_id;
- }
- if (Unicode::is_unicode_region_subtag(segments[index])) {
- language_id.region = segments[index];
- if (segments.size() == ++index)
- return language_id;
- }
- while (index < segments.size()) {
- if (!Unicode::is_unicode_variant_subtag(segments[index]))
- return {};
- language_id.variants.append(segments[index++]);
- }
- return language_id;
- }
- static Optional<LanguageMapping> parse_language_mapping(StringView key, StringView alias)
- {
- auto parsed_key = parse_language(key);
- if (!parsed_key.has_value())
- return {};
- auto parsed_alias = parse_language(alias);
- if (!parsed_alias.has_value())
- return {};
- return LanguageMapping { parsed_key.release_value(), parsed_alias.release_value() };
- }
- static void parse_core_aliases(String core_supplemental_path, UnicodeLocaleData& locale_data)
- {
- LexicalPath core_aliases_path(move(core_supplemental_path));
- core_aliases_path = core_aliases_path.append("aliases.json"sv);
- VERIFY(Core::File::exists(core_aliases_path.string()));
- auto core_aliases_file_or_error = Core::File::open(core_aliases_path.string(), Core::OpenMode::ReadOnly);
- VERIFY(!core_aliases_file_or_error.is_error());
- auto core_aliases = JsonParser(core_aliases_file_or_error.value()->read_all()).parse();
- VERIFY(core_aliases.has_value());
- auto const& supplemental_object = core_aliases->as_object().get("supplemental"sv);
- auto const& metadata_object = supplemental_object.as_object().get("metadata"sv);
- auto const& alias_object = metadata_object.as_object().get("alias"sv);
- auto append_aliases = [&](auto& alias_object, auto& alias_map) {
- alias_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
- auto alias = value.as_object().get("_replacement"sv).as_string();
- if (key.contains('-')) {
- auto mapping = parse_language_mapping(key, alias);
- if (!mapping.has_value())
- return;
- locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
- locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
- locale_data.complex_mappings.append(mapping.release_value());
- } else {
- alias_map.set(key, move(alias));
- }
- });
- };
- append_aliases(alias_object.as_object().get("languageAlias"sv), locale_data.language_aliases);
- append_aliases(alias_object.as_object().get("territoryAlias"sv), locale_data.territory_aliases);
- append_aliases(alias_object.as_object().get("scriptAlias"sv), locale_data.script_aliases);
- append_aliases(alias_object.as_object().get("variantAlias"sv), locale_data.variant_aliases);
- append_aliases(alias_object.as_object().get("subdivisionAlias"sv), locale_data.subdivision_aliases);
- }
- static void parse_likely_subtags(String core_supplemental_path, UnicodeLocaleData& locale_data)
- {
- LexicalPath likely_subtags_path(move(core_supplemental_path));
- likely_subtags_path = likely_subtags_path.append("likelySubtags.json"sv);
- VERIFY(Core::File::exists(likely_subtags_path.string()));
- auto likely_subtags_file_or_error = Core::File::open(likely_subtags_path.string(), Core::OpenMode::ReadOnly);
- VERIFY(!likely_subtags_file_or_error.is_error());
- auto likely_subtags = JsonParser(likely_subtags_file_or_error.value()->read_all()).parse();
- VERIFY(likely_subtags.has_value());
- auto const& supplemental_object = likely_subtags->as_object().get("supplemental"sv);
- auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
- likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
- auto mapping = parse_language_mapping(key, value.as_string());
- if (!mapping.has_value())
- return;
- locale_data.max_variant_size = max(mapping->key.variants.size(), locale_data.max_variant_size);
- locale_data.max_variant_size = max(mapping->alias.variants.size(), locale_data.max_variant_size);
- locale_data.likely_subtags.append(mapping.release_value());
- });
- }
- static void parse_identity(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
- {
- LexicalPath languages_path(move(locale_path)); // Note: Every JSON file defines identity data, so we can use any of them.
- languages_path = languages_path.append("languages.json"sv);
- VERIFY(Core::File::exists(languages_path.string()));
- auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
- VERIFY(!languages_file_or_error.is_error());
- auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
- VERIFY(languages.has_value());
- auto const& main_object = languages->as_object().get("main"sv);
- auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
- auto const& identity_object = locale_object.as_object().get("identity"sv);
- auto const& language_string = identity_object.as_object().get("language"sv);
- auto const& territory_string = identity_object.as_object().get("territory"sv);
- auto const& variant_string = identity_object.as_object().get("variant"sv);
- locale.language = language_string.as_string();
- if (!locale_data.languages.contains_slow(locale.language))
- locale_data.languages.append(locale.language);
- if (territory_string.is_string()) {
- locale.territory = territory_string.as_string();
- if (!locale_data.territories.contains_slow(*locale.territory))
- locale_data.territories.append(*locale.territory);
- }
- if (variant_string.is_string()) {
- locale.variant = variant_string.as_string();
- if (!locale_data.variants.contains_slow(*locale.variant))
- locale_data.variants.append(*locale.variant);
- }
- }
- static void parse_locale_languages(String locale_path, Locale& locale)
- {
- LexicalPath languages_path(move(locale_path));
- languages_path = languages_path.append("languages.json"sv);
- VERIFY(Core::File::exists(languages_path.string()));
- auto languages_file_or_error = Core::File::open(languages_path.string(), Core::OpenMode::ReadOnly);
- VERIFY(!languages_file_or_error.is_error());
- auto languages = JsonParser(languages_file_or_error.value()->read_all()).parse();
- VERIFY(languages.has_value());
- auto const& main_object = languages->as_object().get("main"sv);
- auto const& locale_object = main_object.as_object().get(languages_path.parent().basename());
- auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
- auto const& languages_object = locale_display_names_object.as_object().get("languages"sv);
- languages_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
- locale.languages.set(key, value.as_string());
- });
- }
- static void parse_locale_territories(String locale_path, Locale& locale)
- {
- LexicalPath territories_path(move(locale_path));
- territories_path = territories_path.append("territories.json"sv);
- VERIFY(Core::File::exists(territories_path.string()));
- auto territories_file_or_error = Core::File::open(territories_path.string(), Core::OpenMode::ReadOnly);
- VERIFY(!territories_file_or_error.is_error());
- auto territories = JsonParser(territories_file_or_error.value()->read_all()).parse();
- VERIFY(territories.has_value());
- auto const& main_object = territories->as_object().get("main"sv);
- auto const& locale_object = main_object.as_object().get(territories_path.parent().basename());
- auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
- auto const& territories_object = locale_display_names_object.as_object().get("territories"sv);
- territories_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
- locale.territories.set(key, value.as_string());
- });
- }
- static void parse_locale_scripts(String locale_path, UnicodeLocaleData& locale_data, Locale& locale)
- {
- LexicalPath scripts_path(move(locale_path));
- scripts_path = scripts_path.append("scripts.json"sv);
- VERIFY(Core::File::exists(scripts_path.string()));
- auto scripts_file_or_error = Core::File::open(scripts_path.string(), Core::OpenMode::ReadOnly);
- VERIFY(!scripts_file_or_error.is_error());
- auto scripts = JsonParser(scripts_file_or_error.value()->read_all()).parse();
- VERIFY(scripts.has_value());
- auto const& main_object = scripts->as_object().get("main"sv);
- auto const& locale_object = main_object.as_object().get(scripts_path.parent().basename());
- auto const& locale_display_names_object = locale_object.as_object().get("localeDisplayNames"sv);
- auto const& scripts_object = locale_display_names_object.as_object().get("scripts"sv);
- scripts_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
- locale.scripts.set(key, value.as_string());
- if (!locale_data.scripts.contains_slow(key))
- locale_data.scripts.append(key);
- });
- }
- static void parse_locale_list_patterns(String misc_path, UnicodeLocaleData& locale_data, Locale& locale)
- {
- LexicalPath list_patterns_path(move(misc_path));
- list_patterns_path = list_patterns_path.append("listPatterns.json"sv);
- VERIFY(Core::File::exists(list_patterns_path.string()));
- auto list_patterns_file_or_error = Core::File::open(list_patterns_path.string(), Core::OpenMode::ReadOnly);
- VERIFY(!list_patterns_file_or_error.is_error());
- auto list_patterns = JsonParser(list_patterns_file_or_error.value()->read_all()).parse();
- VERIFY(list_patterns.has_value());
- auto const& main_object = list_patterns->as_object().get("main"sv);
- auto const& locale_object = main_object.as_object().get(list_patterns_path.parent().basename());
- auto const& list_patterns_object = locale_object.as_object().get("listPatterns"sv);
- auto list_pattern_type = [](StringView key) {
- if (key.contains("type-standard"sv))
- return "conjunction"sv;
- if (key.contains("type-or"sv))
- return "disjunction"sv;
- if (key.contains("type-unit"sv))
- return "unit"sv;
- VERIFY_NOT_REACHED();
- };
- auto list_pattern_style = [](StringView key) {
- if (key.contains("short"sv))
- return "short"sv;
- if (key.contains("narrow"sv))
- return "narrow"sv;
- return "long"sv;
- };
- list_patterns_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
- auto type = list_pattern_type(key);
- auto style = list_pattern_style(key);
- auto start = value.as_object().get("start"sv).as_string();
- auto middle = value.as_object().get("middle"sv).as_string();
- auto end = value.as_object().get("end"sv).as_string();
- auto pair = value.as_object().get("2"sv).as_string();
- if (!locale_data.list_pattern_types.contains_slow(type))
- locale_data.list_pattern_types.append(type);
- if (!locale_data.list_pattern_styles.contains_slow(style))
- locale_data.list_pattern_styles.append(style);
- locale.list_patterns.append({ move(type), move(style), move(start), move(middle), move(end), move(pair) });
- });
- }
- static void parse_locale_currencies(String numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
- {
- LexicalPath currencies_path(move(numbers_path));
- currencies_path = currencies_path.append("currencies.json"sv);
- VERIFY(Core::File::exists(currencies_path.string()));
- auto currencies_file_or_error = Core::File::open(currencies_path.string(), Core::OpenMode::ReadOnly);
- VERIFY(!currencies_file_or_error.is_error());
- auto currencies = JsonParser(currencies_file_or_error.value()->read_all()).parse();
- VERIFY(currencies.has_value());
- auto const& main_object = currencies->as_object().get("main"sv);
- auto const& locale_object = main_object.as_object().get(currencies_path.parent().basename());
- auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
- auto const& currencies_object = locale_numbers_object.as_object().get("currencies"sv);
- currencies_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
- auto const& display_name = value.as_object().get("displayName"sv);
- locale.currencies.set(key, display_name.as_string());
- if (!locale_data.currencies.contains_slow(key))
- locale_data.currencies.append(key);
- });
- }
- static void parse_numeric_keywords(String locale_numbers_path, UnicodeLocaleData& locale_data, Locale& locale)
- {
- static constexpr StringView key = "nu"sv;
- LexicalPath numbers_path(move(locale_numbers_path));
- numbers_path = numbers_path.append("numbers.json"sv);
- VERIFY(Core::File::exists(numbers_path.string()));
- auto numbers_file_or_error = Core::File::open(numbers_path.string(), Core::OpenMode::ReadOnly);
- VERIFY(!numbers_file_or_error.is_error());
- auto numbers = JsonParser(numbers_file_or_error.value()->read_all()).parse();
- VERIFY(numbers.has_value());
- auto const& main_object = numbers->as_object().get("main"sv);
- auto const& locale_object = main_object.as_object().get(numbers_path.parent().basename());
- auto const& locale_numbers_object = locale_object.as_object().get("numbers"sv);
- auto const& default_numbering_system_object = locale_numbers_object.as_object().get("defaultNumberingSystem"sv);
- auto const& other_numbering_systems_object = locale_numbers_object.as_object().get("otherNumberingSystems"sv);
- Vector<String> keyword_values {};
- keyword_values.append(default_numbering_system_object.as_string());
- other_numbering_systems_object.as_object().for_each_member([&](auto const&, JsonValue const& value) {
- auto keyword_value = value.as_string();
- if (!keyword_values.contains_slow(keyword_value))
- keyword_values.append(move(keyword_value));
- });
- StringBuilder builder;
- builder.join(',', keyword_values);
- locale.keywords.set(key, builder.build());
- if (!locale_data.keywords.contains_slow(key))
- locale_data.keywords.append(key);
- }
- static Core::DirIterator path_to_dir_iterator(String path)
- {
- LexicalPath lexical_path(move(path));
- lexical_path = lexical_path.append("main"sv);
- VERIFY(Core::File::is_directory(lexical_path.string()));
- Core::DirIterator iterator(lexical_path.string(), Core::DirIterator::SkipParentAndBaseDir);
- if (iterator.has_error()) {
- warnln("{}: {}", lexical_path.string(), iterator.error_string());
- VERIFY_NOT_REACHED();
- }
- return iterator;
- }
- static void parse_all_locales(String core_path, String locale_names_path, String misc_path, String numbers_path, UnicodeLocaleData& locale_data)
- {
- auto locale_names_iterator = path_to_dir_iterator(move(locale_names_path));
- auto misc_iterator = path_to_dir_iterator(move(misc_path));
- auto numbers_iterator = path_to_dir_iterator(move(numbers_path));
- LexicalPath core_supplemental_path(move(core_path));
- core_supplemental_path = core_supplemental_path.append("supplemental"sv);
- VERIFY(Core::File::is_directory(core_supplemental_path.string()));
- parse_core_aliases(core_supplemental_path.string(), locale_data);
- parse_likely_subtags(core_supplemental_path.string(), locale_data);
- auto remove_variants_from_path = [](String path) -> Optional<String> {
- auto parsed_locale = parse_language(LexicalPath::basename(path));
- if (!parsed_locale.has_value())
- return {};
- StringBuilder builder;
- builder.append(parsed_locale->language);
- if (!parsed_locale->script.is_empty())
- builder.appendff("-{}", parsed_locale->script);
- if (!parsed_locale->region.is_empty())
- builder.appendff("-{}", parsed_locale->region);
- return builder.build();
- };
- while (locale_names_iterator.has_next()) {
- auto locale_path = locale_names_iterator.next_full_path();
- VERIFY(Core::File::is_directory(locale_path));
- auto language = remove_variants_from_path(locale_path);
- if (!language.has_value())
- continue;
- auto& locale = locale_data.locales.ensure(*language);
- parse_identity(locale_path, locale_data, locale);
- parse_locale_languages(locale_path, locale);
- parse_locale_territories(locale_path, locale);
- parse_locale_scripts(locale_path, locale_data, locale);
- }
- while (misc_iterator.has_next()) {
- auto misc_path = misc_iterator.next_full_path();
- VERIFY(Core::File::is_directory(misc_path));
- auto language = remove_variants_from_path(misc_path);
- if (!language.has_value())
- continue;
- auto& locale = locale_data.locales.ensure(*language);
- parse_locale_list_patterns(misc_path, locale_data, locale);
- }
- while (numbers_iterator.has_next()) {
- auto numbers_path = numbers_iterator.next_full_path();
- VERIFY(Core::File::is_directory(numbers_path));
- auto language = remove_variants_from_path(numbers_path);
- if (!language.has_value())
- continue;
- auto& locale = locale_data.locales.ensure(*language);
- parse_locale_currencies(numbers_path, locale_data, locale);
- parse_numeric_keywords(numbers_path, locale_data, locale);
- }
- }
- static String format_identifier(StringView owner, String identifier)
- {
- identifier = identifier.replace("-"sv, "_"sv, true);
- if (all_of(identifier, is_ascii_digit))
- return String::formatted("{}_{}", owner[0], identifier);
- return identifier.to_titlecase();
- }
- static void generate_unicode_locale_header(Core::File& file, UnicodeLocaleData& locale_data)
- {
- StringBuilder builder;
- SourceGenerator generator { builder };
- auto generate_enum = [&](StringView name, StringView default_, Vector<String>& values) {
- quick_sort(values);
- generator.set("name", name);
- generator.set("underlying", ((values.size() + !default_.is_empty()) < 256) ? "u8"sv : "u16"sv);
- generator.append(R"~~~(
- enum class @name@ : @underlying@ {)~~~");
- if (!default_.is_empty()) {
- generator.set("default", default_);
- generator.append(R"~~~(
- @default@,)~~~");
- }
- for (auto const& value : values) {
- generator.set("value", format_identifier(name, value));
- generator.append(R"~~~(
- @value@,)~~~");
- }
- generator.append(R"~~~(
- };
- )~~~");
- };
- generator.append(R"~~~(
- #pragma once
- #include <AK/Optional.h>
- #include <AK/StringView.h>
- #include <AK/Types.h>
- #include <LibUnicode/Forward.h>
- namespace Unicode {
- )~~~");
- auto locales = locale_data.locales.keys();
- generate_enum("Locale"sv, "None"sv, locales);
- generate_enum("Language"sv, {}, locale_data.languages);
- generate_enum("Territory"sv, {}, locale_data.territories);
- generate_enum("ScriptTag"sv, {}, locale_data.scripts);
- generate_enum("Currency"sv, {}, locale_data.currencies);
- generate_enum("Key"sv, {}, locale_data.keywords);
- generate_enum("Variant"sv, {}, locale_data.variants);
- generate_enum("ListPatternType"sv, {}, locale_data.list_pattern_types);
- generate_enum("ListPatternStyle"sv, {}, locale_data.list_pattern_styles);
- generator.append(R"~~~(
- namespace Detail {
- Optional<Locale> locale_from_string(StringView const& locale);
- Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
- Optional<Language> language_from_string(StringView const& language);
- Optional<StringView> resolve_language_alias(StringView const& language);
- Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
- Optional<Territory> territory_from_string(StringView const& territory);
- Optional<StringView> resolve_territory_alias(StringView const& territory);
- Optional<StringView> get_locale_script_tag_mapping(StringView locale, StringView script_tag);
- Optional<ScriptTag> script_tag_from_string(StringView const& script_tag);
- Optional<StringView> resolve_script_tag_alias(StringView const& script_tag);
- Optional<StringView> get_locale_currency_mapping(StringView locale, StringView currency);
- Optional<Currency> currency_from_string(StringView const& currency);
- Optional<StringView> get_locale_key_mapping(StringView locale, StringView key);
- Optional<Key> key_from_string(StringView const& key);
- Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style);
- Optional<ListPatternType> list_pattern_type_from_string(StringView const& list_pattern_type);
- Optional<ListPatternStyle> list_pattern_style_from_string(StringView const& list_pattern_style);
- Optional<StringView> resolve_variant_alias(StringView const& variant);
- Optional<StringView> resolve_subdivision_alias(StringView const& subdivision);
- void resolve_complex_language_aliases(Unicode::LanguageID& language_id);
- Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id);
- Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
- }
- }
- )~~~");
- file.write(generator.as_string_view());
- }
- static void generate_unicode_locale_implementation(Core::File& file, UnicodeLocaleData& locale_data)
- {
- StringBuilder builder;
- SourceGenerator generator { builder };
- generator.set("locales_size"sv, String::number(locale_data.locales.size()));
- generator.set("territories_size", String::number(locale_data.territories.size()));
- generator.set("variants_size", String::number(locale_data.max_variant_size));
- generator.append(R"~~~(
- #include <AK/Array.h>
- #include <AK/HashMap.h>
- #include <AK/Span.h>
- #include <LibUnicode/Locale.h>
- #include <LibUnicode/UnicodeLocale.h>
- namespace Unicode {
- struct Patterns {
- ListPatternType type;
- ListPatternStyle style;
- StringView start;
- StringView middle;
- StringView end;
- StringView pair;
- };
- )~~~");
- auto format_mapping_name = [](StringView format, StringView name) {
- auto mapping_name = name.to_lowercase_string().replace("-"sv, "_"sv, true);
- return String::formatted(format, mapping_name);
- };
- auto append_string = [&](StringView value) {
- if (value.is_empty())
- generator.append(", {}"sv);
- else
- generator.append(String::formatted(", \"{}\"sv", value));
- };
- auto append_list_and_size = [&](auto const& list) {
- if (list.is_empty()) {
- generator.append(", {}, 0");
- return;
- }
- bool first = true;
- generator.append(", {");
- for (auto const& item : list) {
- generator.append(first ? " " : ", ");
- generator.append(String::formatted("\"{}\"sv", item));
- first = false;
- }
- generator.append(String::formatted(" }}, {}", list.size()));
- };
- auto append_string_list = [&](String name, auto const& keys, auto const& mappings) {
- generator.set("name", name);
- generator.set("size", String::number(keys.size()));
- generator.append(R"~~~(
- static constexpr Array<StringView, @size@> @name@ { {
- )~~~");
- constexpr size_t max_values_per_row = 10;
- size_t values_in_current_row = 0;
- for (auto const& key : keys) {
- if (values_in_current_row++ > 0)
- generator.append(" ");
- if (auto it = mappings.find(key); it != mappings.end())
- generator.set("mapping"sv, String::formatted("\"{}\"sv", it->value));
- else
- generator.set("mapping"sv, "{}"sv);
- generator.append("@mapping@,");
- if (values_in_current_row == max_values_per_row) {
- values_in_current_row = 0;
- generator.append("\n ");
- }
- }
- generator.append(R"~~~(
- } };
- )~~~");
- };
- auto append_list_patterns = [&](StringView name, Vector<ListPatterns> const& list_patterns) {
- generator.set("name", name);
- generator.set("size", String::number(list_patterns.size()));
- generator.append(R"~~~(
- static constexpr Array<Patterns, @size@> @name@ { {)~~~");
- for (auto const& list_pattern : list_patterns) {
- generator.set("type"sv, String::formatted("ListPatternType::{}", format_identifier({}, list_pattern.type)));
- generator.set("style"sv, String::formatted("ListPatternStyle::{}", format_identifier({}, list_pattern.style)));
- generator.set("start"sv, String::formatted("\"{}\"sv", list_pattern.start));
- generator.set("middle"sv, String::formatted("\"{}\"sv", list_pattern.middle));
- generator.set("end"sv, String::formatted("\"{}\"sv", list_pattern.end));
- generator.set("pair"sv, String::formatted("\"{}\"sv", list_pattern.pair));
- generator.append(R"~~~(
- { @type@, @style@, @start@, @middle@, @end@, @pair@ },)~~~");
- }
- generator.append(R"~~~(
- } };
- )~~~");
- };
- auto append_mapping = [&](StringView type, StringView name, StringView format, auto format_list_callback) {
- Vector<String> mapping_names;
- for (auto const& locale : locale_data.locales) {
- auto mapping_name = format_mapping_name(format, locale.key);
- format_list_callback(mapping_name, locale.value);
- mapping_names.append(move(mapping_name));
- }
- quick_sort(mapping_names);
- generator.set("type", type);
- generator.set("name", name);
- generator.set("size", String::number(locale_data.locales.size()));
- generator.append(R"~~~(
- static constexpr Array<Span<@type@ const>, @size@> @name@ { {
- )~~~");
- constexpr size_t max_values_per_row = 10;
- size_t values_in_current_row = 0;
- for (auto& mapping_name : mapping_names) {
- if (values_in_current_row++ > 0)
- generator.append(" ");
- generator.set("name", move(mapping_name));
- generator.append("@name@.span(),");
- if (values_in_current_row == max_values_per_row) {
- values_in_current_row = 0;
- generator.append("\n ");
- }
- }
- generator.append(R"~~~(
- } };
- )~~~");
- };
- append_mapping("StringView"sv, "s_languages"sv, "s_languages_{}", [&](auto const& name, auto const& value) { append_string_list(name, locale_data.languages, value.languages); });
- append_mapping("StringView"sv, "s_territories"sv, "s_territories_{}", [&](auto const& name, auto const& value) { append_string_list(name, locale_data.territories, value.territories); });
- append_mapping("StringView"sv, "s_scripts"sv, "s_scripts_{}", [&](auto const& name, auto const& value) { append_string_list(name, locale_data.scripts, value.scripts); });
- append_mapping("StringView"sv, "s_currencies"sv, "s_currencies_{}", [&](auto const& name, auto const& value) { append_string_list(name, locale_data.currencies, value.currencies); });
- append_mapping("StringView"sv, "s_keywords"sv, "s_keywords_{}", [&](auto const& name, auto const& value) { append_string_list(name, locale_data.keywords, value.keywords); });
- append_mapping("Patterns"sv, "s_list_patterns"sv, "s_list_patterns_{}", [&](auto const& name, auto const& value) { append_list_patterns(name, value.list_patterns); });
- generator.append(R"~~~(
- struct CanonicalLanguageID {
- Unicode::LanguageID to_unicode_language_id() const
- {
- Unicode::LanguageID language_id {};
- language_id.variants.ensure_capacity(variants_size);
- language_id.language = language.to_string();
- if (!script.is_empty())
- language_id.script = script.to_string();
- if (!region.is_empty())
- language_id.region = region.to_string();
- for (size_t i = 0; i < variants_size; ++i)
- language_id.variants.append(variants[i].to_string());
- return language_id;
- }
- bool matches_variants(Vector<String> const& other_variants) const {
- if (variants_size == 0)
- return true;
- if (other_variants.size() != variants_size)
- return false;
- for (size_t i = 0; i < variants_size; ++i) {
- if (variants[i] != other_variants[i])
- return false;
- }
- return true;
- };
- StringView language {};
- StringView script {};
- StringView region {};
- Array<StringView, @variants_size@> variants {};
- size_t variants_size { 0 };
- };
- struct LanguageMapping {
- CanonicalLanguageID key;
- CanonicalLanguageID alias;
- };
- )~~~");
- auto append_complex_mapping = [&](StringView name, auto& mappings) {
- generator.set("size", String::number(mappings.size()));
- generator.set("name"sv, name);
- generator.append(R"~~~(
- static constexpr Array<LanguageMapping, @size@> s_@name@ { {
- )~~~");
- quick_sort(mappings, [](auto const& lhs, auto const& rhs) {
- auto const& lhs_language = lhs.key.language;
- auto const& rhs_language = rhs.key.language;
- // Sort the keys such that "und" language tags are at the end, as those are less specific.
- if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
- return false;
- if (!lhs_language.starts_with("und"sv) && rhs_language.starts_with("und"sv))
- return true;
- return lhs_language < rhs_language;
- });
- for (auto const& mapping : mappings) {
- generator.set("language"sv, mapping.key.language);
- generator.append(" { { \"@language@\"sv");
- append_string(mapping.key.script);
- append_string(mapping.key.region);
- append_list_and_size(mapping.key.variants);
- generator.set("language"sv, mapping.alias.language);
- generator.append(" }, { \"@language@\"sv");
- append_string(mapping.alias.script);
- append_string(mapping.alias.region);
- append_list_and_size(mapping.alias.variants);
- generator.append(" } },\n");
- }
- generator.append("} };\n");
- };
- append_complex_mapping("complex_alias"sv, locale_data.complex_mappings);
- append_complex_mapping("likely_subtags"sv, locale_data.likely_subtags);
- generator.append(R"~~~(
- static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& language_id)
- {
- // https://unicode.org/reports/tr35/#Likely_Subtags
- enum class State {
- LanguageScriptRegion,
- LanguageRegion,
- LanguageScript,
- Language,
- UndScript,
- Done,
- };
- auto state = State::LanguageScriptRegion;
- while (state != State::Done) {
- CanonicalLanguageID search_key;
- switch (state) {
- case State::LanguageScriptRegion:
- state = State::LanguageRegion;
- if (!language_id.script.has_value() || !language_id.region.has_value())
- continue;
- search_key.language = *language_id.language;
- search_key.script = *language_id.script;
- search_key.region = *language_id.region;
- break;
- case State::LanguageRegion:
- state = State::LanguageScript;
- if (!language_id.region.has_value())
- continue;
- search_key.language = *language_id.language;
- search_key.region = *language_id.region;
- break;
- case State::LanguageScript:
- state = State::Language;
- if (!language_id.script.has_value())
- continue;
- search_key.language = *language_id.language;
- search_key.script = *language_id.script;
- break;
- case State::Language:
- state = State::UndScript;
- search_key.language = *language_id.language;
- break;
- case State::UndScript:
- state = State::Done;
- if (!language_id.script.has_value())
- continue;
- search_key.language = "und"sv;
- search_key.script = *language_id.script;
- break;
- default:
- VERIFY_NOT_REACHED();
- }
- for (auto const& map : s_likely_subtags) {
- if (map.key.language != search_key.language)
- continue;
- if (map.key.script != search_key.script)
- continue;
- if (map.key.region != search_key.region)
- continue;
- return ↦
- }
- }
- return nullptr;
- }
- namespace Detail {
- )~~~");
- auto append_mapping_search = [&](StringView enum_title, StringView enum_snake, StringView collection_name) {
- generator.set("enum_title", enum_title);
- generator.set("enum_snake", enum_snake);
- generator.set("collection_name", collection_name);
- generator.append(R"~~~(
- Optional<StringView> get_locale_@enum_snake@_mapping(StringView locale, StringView @enum_snake@)
- {
- auto locale_value = locale_from_string(locale);
- if (!locale_value.has_value())
- return {};
- auto @enum_snake@_value = @enum_snake@_from_string(@enum_snake@);
- if (!@enum_snake@_value.has_value())
- return {};
- auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
- auto @enum_snake@_index = to_underlying(*@enum_snake@_value);
- auto const& mappings = @collection_name@.at(locale_index);
- auto @enum_snake@_mapping = mappings.at(@enum_snake@_index);
- if (@enum_snake@_mapping.is_empty())
- return {};
- return @enum_snake@_mapping;
- }
- )~~~");
- };
- auto append_from_string = [&](StringView enum_title, StringView enum_snake, Vector<String> const& values) {
- generator.set("enum_title", enum_title);
- generator.set("enum_snake", enum_snake);
- generator.append(R"~~~(
- Optional<@enum_title@> @enum_snake@_from_string(StringView const& @enum_snake@)
- {
- static HashMap<StringView, @enum_title@> @enum_snake@_values { {)~~~");
- for (auto const& value : values) {
- generator.set("key"sv, value);
- generator.set("value"sv, format_identifier(enum_title, value));
- generator.append(R"~~~(
- { "@key@"sv, @enum_title@::@value@ },)~~~");
- }
- generator.append(R"~~~(
- } };
- if (auto value = @enum_snake@_values.get(@enum_snake@); value.has_value())
- return value.value();
- return {};
- }
- )~~~");
- };
- auto append_alias_search = [&](StringView enum_snake, HashMap<String, String> const& aliases) {
- generator.set("enum_snake", enum_snake);
- generator.append(R"~~~(
- Optional<StringView> resolve_@enum_snake@_alias(StringView const& @enum_snake@)
- {
- static HashMap<StringView, StringView> @enum_snake@_aliases { {
- )~~~");
- constexpr size_t max_values_per_row = 10;
- size_t values_in_current_row = 0;
- for (auto const& alias : aliases) {
- if (values_in_current_row++ > 0)
- generator.append(" ");
- generator.set("key"sv, alias.key);
- generator.set("alias"sv, alias.value);
- generator.append("{ \"@key@\"sv, \"@alias@\"sv },");
- if (values_in_current_row == max_values_per_row) {
- generator.append("\n ");
- values_in_current_row = 0;
- }
- }
- generator.append(R"~~~(
- } };
- if (auto alias = @enum_snake@_aliases.get(@enum_snake@); alias.has_value())
- return alias.value();
- return {};
- }
- )~~~");
- };
- append_from_string("Locale"sv, "locale"sv, locale_data.locales.keys());
- append_mapping_search("Language"sv, "language"sv, "s_languages"sv);
- append_from_string("Language"sv, "language"sv, locale_data.languages);
- append_alias_search("language"sv, locale_data.language_aliases);
- append_mapping_search("Territory"sv, "territory"sv, "s_territories"sv);
- append_from_string("Territory"sv, "territory"sv, locale_data.territories);
- append_alias_search("territory"sv, locale_data.territory_aliases);
- append_mapping_search("ScriptTag"sv, "script_tag"sv, "s_scripts"sv);
- append_from_string("ScriptTag"sv, "script_tag"sv, locale_data.scripts);
- append_alias_search("script_tag"sv, locale_data.script_aliases);
- append_mapping_search("Currency"sv, "currency"sv, "s_currencies"sv);
- append_from_string("Currency"sv, "currency"sv, locale_data.currencies);
- append_mapping_search("Key"sv, "key"sv, "s_keywords"sv);
- append_from_string("Key"sv, "key"sv, locale_data.keywords);
- append_alias_search("variant"sv, locale_data.variant_aliases);
- append_alias_search("subdivision"sv, locale_data.subdivision_aliases);
- append_from_string("ListPatternType"sv, "list_pattern_type"sv, locale_data.list_pattern_types);
- append_from_string("ListPatternStyle"sv, "list_pattern_style"sv, locale_data.list_pattern_styles);
- generator.append(R"~~~(
- Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, StringView list_pattern_type, StringView list_pattern_style)
- {
- auto locale_value = locale_from_string(locale);
- if (!locale_value.has_value())
- return {};
- auto type_value = list_pattern_type_from_string(list_pattern_type);
- if (!type_value.has_value())
- return {};
- auto style_value = list_pattern_style_from_string(list_pattern_style);
- if (!style_value.has_value())
- return {};
- auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
- auto const& locale_list_patterns = s_list_patterns.at(locale_index);
- for (auto const& list_patterns : locale_list_patterns) {
- if ((list_patterns.type == type_value) && (list_patterns.style == style_value))
- return ListPatterns { list_patterns.start, list_patterns.middle, list_patterns.end, list_patterns.pair };
- }
- return {};
- }
- void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
- {
- for (auto const& map : s_complex_alias) {
- if ((map.key.language != language_id.language) && (map.key.language != "und"sv))
- continue;
- if (!map.key.script.is_empty() && (map.key.script != language_id.script))
- continue;
- if (!map.key.region.is_empty() && (map.key.region != language_id.region))
- continue;
- if (!map.key.matches_variants(language_id.variants))
- continue;
- auto alias = map.alias.to_unicode_language_id();
- if (alias.language == "und"sv)
- alias.language = move(language_id.language);
- if (map.key.script.is_empty() && !alias.script.has_value())
- alias.script = move(language_id.script);
- if (map.key.region.is_empty() && !alias.region.has_value())
- alias.region = move(language_id.region);
- if (map.key.variants_size == 0 && alias.variants.is_empty())
- alias.variants = move(language_id.variants);
- language_id = move(alias);
- break;
- }
- }
- Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& language_id)
- {
- // https://www.unicode.org/reports/tr35/#Likely_Subtags
- auto const* likely_subtag = resolve_likely_subtag(language_id);
- if (likely_subtag == nullptr)
- return {};
- auto maximized = language_id;
- auto const& key = likely_subtag->key;
- auto const& alias = likely_subtag->alias;
- if (maximized.language == "und"sv)
- maximized.language = alias.language;
- if (!maximized.script.has_value() || (!key.script.is_empty() && !alias.script.is_empty()))
- maximized.script = alias.script;
- if (!maximized.region.has_value() || (!key.region.is_empty() && !alias.region.is_empty()))
- maximized.region = alias.region;
- return maximized;
- }
- Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
- {
- if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
- return likely_subtag->alias.region;
- return {};
- }
- }
- }
- )~~~");
- file.write(generator.as_string_view());
- }
- int main(int argc, char** argv)
- {
- char const* generated_header_path = nullptr;
- char const* generated_implementation_path = nullptr;
- char const* core_path = nullptr;
- char const* locale_names_path = nullptr;
- char const* misc_path = nullptr;
- char const* numbers_path = nullptr;
- Core::ArgsParser args_parser;
- args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
- args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
- args_parser.add_option(core_path, "Path to cldr-core directory", "core-path", 'r', "core-path");
- args_parser.add_option(locale_names_path, "Path to cldr-localenames directory", "locale-names-path", 'l', "locale-names-path");
- args_parser.add_option(misc_path, "Path to cldr-misc directory", "misc-path", 'm', "misc-path");
- args_parser.add_option(numbers_path, "Path to cldr-numbers directory", "numbers-path", 'n', "numbers-path");
- args_parser.parse(argc, argv);
- auto open_file = [&](StringView path, StringView flags, Core::OpenMode mode = Core::OpenMode::ReadOnly) {
- if (path.is_empty()) {
- warnln("{} is required", flags);
- args_parser.print_usage(stderr, argv[0]);
- exit(1);
- }
- auto file_or_error = Core::File::open(path, mode);
- if (file_or_error.is_error()) {
- warnln("Failed to open {}: {}", path, file_or_error.release_error());
- exit(1);
- }
- return file_or_error.release_value();
- };
- auto generated_header_file = open_file(generated_header_path, "-h/--generated-header-path", Core::OpenMode::ReadWrite);
- auto generated_implementation_file = open_file(generated_implementation_path, "-c/--generated-implementation-path", Core::OpenMode::ReadWrite);
- UnicodeLocaleData locale_data;
- parse_all_locales(core_path, locale_names_path, misc_path, numbers_path, locale_data);
- generate_unicode_locale_header(generated_header_file, locale_data);
- generate_unicode_locale_implementation(generated_implementation_file, locale_data);
- return 0;
- }
|