ladybird/Userland/Libraries/LibUnicode/Locale.cpp

/*
 * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
 *
 * SPDX-License-Identifier: BSD-2-Clause
 */

#include <AK/AllOf.h>
#include <AK/GenericLexer.h>
#include <AK/QuickSort.h>
#include <AK/StringBuilder.h>
#include <LibUnicode/Locale.h>

#if ENABLE_UNICODE_DATA
#    include <LibUnicode/UnicodeLocale.h>
#endif

namespace Unicode {

static bool is_key(StringView key)
{
    // key = alphanum alpha
    if (key.length() != 2)
        return false;
    return is_ascii_alphanumeric(key[0]) && is_ascii_alpha(key[1]);
}

static bool is_single_type(StringView type)
{
    // type = alphanum{3,8} (sep alphanum{3,8})*
    // Note: Consecutive types are not handled here, that is left to the caller.
    if ((type.length() < 3) || (type.length() > 8))
        return false;
    return all_of(type, is_ascii_alphanumeric);
}

static bool is_attribute(StringView type)
{
    // attribute = alphanum{3,8}
    if ((type.length() < 3) || (type.length() > 8))
        return false;
    return all_of(type, is_ascii_alphanumeric);
}

static bool is_transformed_key(StringView key)
{
    // tkey = alpha digit
    if (key.length() != 2)
        return false;
    return is_ascii_alpha(key[0]) && is_ascii_digit(key[1]);
}

static bool is_single_transformed_value(StringView value)
{
    // tvalue = (sep alphanum{3,8})+
    // Note: Consecutive values are not handled here, that is left to the caller.
    if ((value.length() < 3) || (value.length() > 8))
        return false;
    return all_of(value, is_ascii_alphanumeric);
}

static Optional<StringView> consume_next_segment(GenericLexer& lexer, bool with_separator = true)
{
    constexpr auto is_separator = is_any_of("-_"sv);

    if (with_separator) {
        if (!lexer.next_is(is_separator))
            return {};
        lexer.ignore();
    }

    auto segment = lexer.consume_until(is_separator);
    if (segment.is_empty()) {
        lexer.retreat(with_separator);
        return {};
    }

    return segment;
}

bool is_type_identifier(StringView identifier)
{
    // type = alphanum{3,8} (sep alphanum{3,8})*
    GenericLexer lexer { identifier };

    while (true) {
        auto type = consume_next_segment(lexer, lexer.tell() > 0);
        if (!type.has_value())
            break;
        if (!is_single_type(*type))
            return false;
    }

    return lexer.is_eof() && (lexer.tell() > 0);
}

static Optional<LanguageID> parse_unicode_language_id(GenericLexer& lexer)
{
    // https://unicode.org/reports/tr35/#Unicode_language_identifier
    //
    // unicode_language_id = "root"
    //     OR
    // unicode_language_id = ((unicode_language_subtag (sep unicode_script_subtag)?) | unicode_script_subtag)
    //                       (sep unicode_region_subtag)?
    //                       (sep unicode_variant_subtag)*
    LanguageID language_id {};

    if (lexer.consume_specific("root"sv)) {
        language_id.is_root = true;
        return language_id;
    }

    enum class ParseState {
        ParsingLanguageOrScript,
        ParsingScript,
        ParsingRegion,
        ParsingVariant,
        Done,
    };

    auto state = ParseState::ParsingLanguageOrScript;

    while (!lexer.is_eof() && (state != ParseState::Done)) {
        auto segment = consume_next_segment(lexer, state != ParseState::ParsingLanguageOrScript);
        if (!segment.has_value())
            return {};

        switch (state) {
        case ParseState::ParsingLanguageOrScript:
            if (is_unicode_language_subtag(*segment)) {
                state = ParseState::ParsingScript;
                language_id.language = *segment;
            } else if (is_unicode_script_subtag(*segment)) {
                state = ParseState::ParsingRegion;
                language_id.script = *segment;
            } else {
                return {};
            }
            break;

        case ParseState::ParsingScript:
            if (is_unicode_script_subtag(*segment)) {
                state = ParseState::ParsingRegion;
                language_id.script = *segment;
                break;
            }

            state = ParseState::ParsingRegion;
            [[fallthrough]];

        case ParseState::ParsingRegion:
            if (is_unicode_region_subtag(*segment)) {
                state = ParseState::ParsingVariant;
                language_id.region = *segment;
                break;
            }

            state = ParseState::ParsingVariant;
            [[fallthrough]];

        case ParseState::ParsingVariant:
            if (is_unicode_variant_subtag(*segment)) {
                language_id.variants.append(*segment);
            } else {
                lexer.retreat(segment->length() + 1);
                state = ParseState::Done;
            }
            break;

        default:
            VERIFY_NOT_REACHED();
        }
    }

    return language_id;
}

static Optional<LocaleExtension> parse_unicode_locale_extension(GenericLexer& lexer)
{
    // https://unicode.org/reports/tr35/#unicode_locale_extensions
    //
    // unicode_locale_extensions = sep [uU] ((sep keyword)+ | (sep attribute)+ (sep keyword)*)
    LocaleExtension locale_extension {};

    enum class ParseState {
        ParsingAttributeOrKeyword,
        ParsingAttribute,
        ParsingKeyword,
        Done,
    };

    auto state = ParseState::ParsingAttributeOrKeyword;

    while (!lexer.is_eof() && (state != ParseState::Done)) {
        auto segment = consume_next_segment(lexer);
        if (!segment.has_value())
            return {};

        if (state == ParseState::ParsingAttributeOrKeyword)
            state = is_key(*segment) ? ParseState::ParsingKeyword : ParseState::ParsingAttribute;

        switch (state) {
        case ParseState::ParsingAttribute:
            if (is_attribute(*segment)) {
                locale_extension.attributes.append(*segment);
                break;
            }

            state = ParseState::ParsingKeyword;
            [[fallthrough]];

        case ParseState::ParsingKeyword: {
            // keyword = key (sep type)?
            Keyword keyword { .key = *segment };

            if (!is_key(*segment)) {
                lexer.retreat(segment->length() + 1);
                state = ParseState::Done;
                break;
            }

            while (true) {
                auto type = consume_next_segment(lexer);

                if (!type.has_value() || !is_single_type(*type)) {
                    if (type.has_value())
                        lexer.retreat(type->length() + 1);
                    break;
                }

                keyword.types.append(*type);
            }

            locale_extension.keywords.append(move(keyword));
            break;
        }

        default:
            VERIFY_NOT_REACHED();
        }
    }

    if (locale_extension.attributes.is_empty() && locale_extension.keywords.is_empty())
        return {};
    return locale_extension;
}

static Optional<TransformedExtension> parse_transformed_extension(GenericLexer& lexer)
{
    // https://unicode.org/reports/tr35/#transformed_extensions
    //
    // transformed_extensions = sep [tT] ((sep tlang (sep tfield)*) | (sep tfield)+)
    TransformedExtension transformed_extension {};

    enum class ParseState {
        ParsingLanguageOrField,
        ParsingLanguage,
        ParsingField,
        Done,
    };

    auto state = ParseState::ParsingLanguageOrField;

    while (!lexer.is_eof() && (state != ParseState::Done)) {
        auto segment = consume_next_segment(lexer);
        if (!segment.has_value())
            return {};

        if (state == ParseState::ParsingLanguageOrField)
            state = is_unicode_language_subtag(*segment) ? ParseState::ParsingLanguage : ParseState::ParsingField;

        switch (state) {
        case ParseState::ParsingLanguage:
            lexer.retreat(segment->length());

            if (auto language_id = parse_unicode_language_id(lexer); language_id.has_value()) {
                transformed_extension.language = language_id.release_value();
                state = ParseState::ParsingField;
                break;
            }

            return {};

        case ParseState::ParsingField: {
            // tfield = tkey tvalue;
            TransformedField field { .key = *segment };

            if (!is_transformed_key(*segment)) {
                lexer.retreat(segment->length() + 1);
                state = ParseState::Done;
                break;
            }

            while (true) {
                auto value = consume_next_segment(lexer);

                if (!value.has_value() || !is_single_transformed_value(*value)) {
                    if (value.has_value())
                        lexer.retreat(value->length() + 1);
                    break;
                }

                field.values.append(*value);
            }

            if (field.values.is_empty())
                return {};

            transformed_extension.fields.append(move(field));
            break;
        }

        default:
            VERIFY_NOT_REACHED();
        }
    }

    if (!transformed_extension.language.has_value() && transformed_extension.fields.is_empty())
        return {};
    return transformed_extension;
}

static Optional<OtherExtension> parse_other_extension(char key, GenericLexer& lexer)
{
    // https://unicode.org/reports/tr35/#other_extensions
    //
    // other_extensions = sep [alphanum-[tTuUxX]] (sep alphanum{2,8})+ ;
    OtherExtension other_extension { .key = key };

    if (!is_ascii_alphanumeric(key) || (key == 'x') || (key == 'X'))
        return {};

    while (true) {
        auto segment = consume_next_segment(lexer);
        if (!segment.has_value())
            break;

        if ((segment->length() < 2) || (segment->length() > 8) || !all_of(*segment, is_ascii_alphanumeric)) {
            lexer.retreat(segment->length() + 1);
            break;
        }

        other_extension.values.append(*segment);
    }

    if (other_extension.values.is_empty())
        return {};
    return other_extension;
}

static Optional<Extension> parse_extension(GenericLexer& lexer)
{
    // https://unicode.org/reports/tr35/#extensions
    //
    // extensions = unicode_locale_extensions | transformed_extensions | other_extensions
    size_t starting_position = lexer.tell();

    if (auto header = consume_next_segment(lexer); header.has_value() && (header->length() == 1)) {
        switch (char key = (*header)[0]) {
        case 'u':
        case 'U':
            if (auto extension = parse_unicode_locale_extension(lexer); extension.has_value())
                return Extension { extension.release_value() };
            break;

        case 't':
        case 'T':
            if (auto extension = parse_transformed_extension(lexer); extension.has_value())
                return Extension { extension.release_value() };
            break;

        default:
            if (auto extension = parse_other_extension(key, lexer); extension.has_value())
                return Extension { extension.release_value() };
            break;
        }
    }

    lexer.retreat(lexer.tell() - starting_position);
    return {};
}

static Vector<String> parse_private_use_extensions(GenericLexer& lexer)
{
    // https://unicode.org/reports/tr35/#pu_extensions
    //
    // pu_extensions = = sep [xX] (sep alphanum{1,8})+ ;
    size_t starting_position = lexer.tell();

    auto header = consume_next_segment(lexer);
    if (!header.has_value())
        return {};

    auto parse_values = [&]() -> Vector<String> {
        Vector<String> extensions;

        while (true) {
            auto segment = consume_next_segment(lexer);
            if (!segment.has_value())
                break;

            if ((segment->length() < 1) || (segment->length() > 8) || !all_of(*segment, is_ascii_alphanumeric)) {
                lexer.retreat(segment->length() + 1);
                break;
            }

            extensions.append(*segment);
        }

        return extensions;
    };

    if ((header->length() == 1) && (((*header)[0] == 'x') || ((*header)[0] == 'X'))) {
        if (auto extensions = parse_values(); !extensions.is_empty())
            return extensions;
    }

    lexer.retreat(lexer.tell() - starting_position);
    return {};
}

Optional<LanguageID> parse_unicode_language_id(StringView language)
{
    GenericLexer lexer { language };

    auto language_id = parse_unicode_language_id(lexer);
    if (!lexer.is_eof())
        return {};

    return language_id;
}

Optional<LocaleID> parse_unicode_locale_id(StringView locale)
{
    GenericLexer lexer { locale };

    // https://unicode.org/reports/tr35/#Unicode_locale_identifier
    //
    // unicode_locale_id = unicode_language_id
    //                     extensions*
    //                     pu_extensions?
    auto language_id = parse_unicode_language_id(lexer);
    if (!language_id.has_value())
        return {};

    LocaleID locale_id { language_id.release_value() };

    while (true) {
        auto extension = parse_extension(lexer);
        if (!extension.has_value())
            break;
        locale_id.extensions.append(extension.release_value());
    }

    locale_id.private_use_extensions = parse_private_use_extensions(lexer);

    if (!lexer.is_eof())
        return {};

    return locale_id;
}

static void perform_hard_coded_key_value_substitutions(String& key, String& value)
{
    // FIXME: In the XML export of CLDR, there are some aliases defined in the following files:
    // https://github.com/unicode-org/cldr-staging/blob/master/production/common/bcp47/calendar.xml
    // https://github.com/unicode-org/cldr-staging/blob/master/production/common/bcp47/collation.xml
    // https://github.com/unicode-org/cldr-staging/blob/master/production/common/bcp47/measure.xml
    // https://github.com/unicode-org/cldr-staging/blob/master/production/common/bcp47/timezone.xml
    // https://github.com/unicode-org/cldr-staging/blob/master/production/common/bcp47/transform.xml
    //
    // There doesn't seem to be a counterpart in the JSON export. Since there aren't many such
    // aliases, until an XML parser is implemented, those aliases are implemented here.
    if ((key == "ca"sv) && (value == "islamicc"sv)) {
        value = "islamic-civil"sv;
    } else if (key.is_one_of("kb"sv, "kc"sv, "kh"sv, "kk"sv, "kn"sv) && (value == "yes"sv)) {
        value = "true"sv;
    } else if (key == "ks"sv) {
        if (value == "primary"sv)
            value = "level1"sv;
        else if (value == "tertiary"sv)
            value = "level3"sv;
        // Note: There are also aliases for "secondary", "quaternary", "quarternary", and "identical",
        // but those are semantically incorrect values (they are too long), so they can be skipped.
    } else if ((key == "m0"sv) && (value == "names"sv)) {
        value = "prprname"sv;
    } else if ((key == "ms"sv) && (value == "imperial"sv)) {
        value = "uksystem"sv;
    } else if (key == "tz"sv) {
        // Formatter disabled because this block is easier to read / check against timezone.xml as one-liners.
        // clang-format off
        if (value == "aqams"sv) value = "nzakl"sv;
        else if (value == "cnckg"sv) value = "cnsha"sv;
        else if (value == "cnhrb"sv) value = "cnsha"sv;
        else if (value == "cnkhg"sv) value = "cnurc"sv;
        else if (value == "cuba"sv) value = "cuhav"sv;
        else if (value == "egypt"sv) value = "egcai"sv;
        else if (value == "eire"sv) value = "iedub"sv;
        else if (value == "est"sv) value = "utcw05"sv;
        else if (value == "gmt0"sv) value = "gmt"sv;
        else if (value == "hongkong"sv) value = "hkhkg"sv;
        else if (value == "hst"sv) value = "utcw10"sv;
        else if (value == "iceland"sv) value = "isrey"sv;
        else if (value == "iran"sv) value = "irthr"sv;
        else if (value == "israel"sv) value = "jeruslm"sv;
        else if (value == "jamaica"sv) value = "jmkin"sv;
        else if (value == "japan"sv) value = "jptyo"sv;
        else if (value == "kwajalein"sv) value = "mhkwa"sv;
        else if (value == "libya"sv) value = "lytip"sv;
        else if (value == "mst"sv) value = "utcw07"sv;
        else if (value == "navajo"sv) value = "usden"sv;
        else if (value == "poland"sv) value = "plwaw"sv;
        else if (value == "portugal"sv) value = "ptlis"sv;
        else if (value == "prc"sv) value = "cnsha"sv;
        else if (value == "roc"sv) value = "twtpe"sv;
        else if (value == "rok"sv) value = "krsel"sv;
        else if (value == "singapore"sv) value = "sgsin"sv;
        else if (value == "turkey"sv) value = "trist"sv;
        else if (value == "uct"sv) value = "utc"sv;
        else if (value == "usnavajo"sv) value = "usden"sv;
        else if (value == "zulu"sv) value = "utc"sv;
        // clang-format on
    }
}

static void perform_hard_coded_key_multi_value_substitutions(String const& key, Vector<String>& values)
{
    // Similar to perform_hard_coded_key_value_substitutions, some aliases depend on multiple
    // variants being present in the original locale. Those are canonicalized separately here.
    // https://github.com/unicode-org/cldr-staging/blob/master/production/common/bcp47/calendar.xml
    if ((key != "ca"sv) || (values.size() != 3))
        return;

    static Vector<String> ethiopic_amete_alem { "ethiopic"sv, "amete"sv, "alem"sv };

    if (values == ethiopic_amete_alem)
        values = { "ethioaa"sv };
}

static void transform_unicode_locale_id_to_canonical_syntax(LocaleID& locale_id)
{
    auto canonicalize_language = [](LanguageID& language_id, bool force_lowercase) {
        language_id.language = language_id.language->to_lowercase();
        if (language_id.script.has_value())
            language_id.script = language_id.script->to_titlecase();
        if (language_id.region.has_value())
            language_id.region = language_id.region->to_uppercase();
        for (auto& variant : language_id.variants)
            variant = variant.to_lowercase();

#if ENABLE_UNICODE_DATA
        Detail::resolve_complex_language_aliases(language_id);
#endif

        if (auto alias = resolve_language_alias(*language_id.language); alias.has_value()) {
            auto language_alias = parse_unicode_language_id(*alias);
            VERIFY(language_alias.has_value());

            language_id.language = move(language_alias->language);
            if (!language_id.script.has_value() && language_alias->script.has_value())
                language_id.script = move(language_alias->script);
            if (!language_id.region.has_value() && language_alias->region.has_value())
                language_id.region = move(language_alias->region);
            if (language_id.variants.is_empty() && !language_alias->variants.is_empty())
                language_id.variants = move(language_alias->variants);
        }

        if (language_id.script.has_value()) {
            if (auto alias = resolve_script_tag_alias(*language_id.script); alias.has_value())
                language_id.script = move(*alias);
        }

        if (language_id.region.has_value()) {
            if (auto alias = resolve_territory_alias(*language_id.region); alias.has_value())
                language_id.region = resolve_most_likely_territory(language_id, *alias);
        }

        quick_sort(language_id.variants);

        for (auto& variant : language_id.variants) {
            variant = variant.to_lowercase();
            if (auto alias = resolve_variant_alias(variant); alias.has_value())
                variant = move(*alias);
        }

        if (force_lowercase) {
            if (language_id.script.has_value())
                language_id.script = language_id.script->to_lowercase();
            if (language_id.region.has_value())
                language_id.region = language_id.region->to_lowercase();
        }
    };

    auto canonicalize_key_value_list = [&](auto& key, auto& values, bool remove_true_values) {
        key = key.to_lowercase();

        auto raw_values = move(values);

        for (auto& value : raw_values) {
            value = value.to_lowercase();
            perform_hard_coded_key_value_substitutions(key, value);

            // Note: The spec says to remove "true" type and tfield values but that is believed to be a bug in the spec
            // because, for tvalues, that would result in invalid syntax:
            //     https://unicode-org.atlassian.net/browse/CLDR-14318
            // This has also been noted by test262:
            //     https://github.com/tc39/test262/blob/18bb955771669541c56c28748603f6afdb2e25ff/test/intl402/Intl/getCanonicalLocales/transformed-ext-canonical.js
            if (remove_true_values && (value == "true"sv))
                continue;

            if (key.is_one_of("sd"sv, "rg"sv)) {
                if (auto alias = resolve_subdivision_alias(value); alias.has_value()) {
                    auto aliases = alias->split_view(' ');

                    // FIXME: Subdivision subtags do not appear in the CLDR likelySubtags.json file.
                    //        Implement the spec's recommendation of using just the first alias for now,
                    //        but we should determine if there's anything else needed here.
                    values.append(aliases[0].to_string());
                    continue;
                }
            }

            values.append(move(value));
        }

        perform_hard_coded_key_multi_value_substitutions(key, values);
    };

    canonicalize_language(locale_id.language_id, false);

    quick_sort(locale_id.extensions, [](auto const& left, auto const& right) {
        auto key = [](auto const& extension) {
            return extension.visit(
                [](LocaleExtension const&) { return 'u'; },
                [](TransformedExtension const&) { return 't'; },
                [](OtherExtension const& ext) { return static_cast<char>(to_ascii_lowercase(ext.key)); });
        };

        return key(left) < key(right);
    });

    for (auto& extension : locale_id.extensions) {
        extension.visit(
            [&](LocaleExtension& ext) {
                for (auto& attribute : ext.attributes)
                    attribute = attribute.to_lowercase();
                for (auto& keyword : ext.keywords)
                    canonicalize_key_value_list(keyword.key, keyword.types, true);

                quick_sort(ext.attributes);
                quick_sort(ext.keywords, [](auto const& a, auto const& b) { return a.key < b.key; });
            },
            [&](TransformedExtension& ext) {
                if (ext.language.has_value())
                    canonicalize_language(*ext.language, true);

                for (auto& field : ext.fields)
                    canonicalize_key_value_list(field.key, field.values, false);

                quick_sort(ext.fields, [](auto const& a, auto const& b) { return a.key < b.key; });
            },
            [&](OtherExtension& ext) {
                ext.key = static_cast<char>(to_ascii_lowercase(ext.key));
                for (auto& value : ext.values)
                    value = value.to_lowercase();
            });
    }

    for (auto& extension : locale_id.private_use_extensions)
        extension = extension.to_lowercase();
}

Optional<String> canonicalize_unicode_locale_id(LocaleID& locale_id)
{
    // https://unicode.org/reports/tr35/#Canonical_Unicode_Locale_Identifiers
    StringBuilder builder;

    auto append_sep_and_string = [&](Optional<String> const& string) {
        if (!string.has_value())
            return;
        builder.appendff("-{}", *string);
    };

    if (!locale_id.language_id.language.has_value())
        return {};

    transform_unicode_locale_id_to_canonical_syntax(locale_id);

    builder.append(locale_id.language_id.language->to_lowercase());
    append_sep_and_string(locale_id.language_id.script);
    append_sep_and_string(locale_id.language_id.region);
    for (auto const& variant : locale_id.language_id.variants)
        append_sep_and_string(variant);

    auto append_key_value_list = [&](auto const& key, auto& values) {
        append_sep_and_string(key);

        for (auto& value : values)
            append_sep_and_string(value);
    };

    for (auto const& extension : locale_id.extensions) {
        extension.visit(
            [&](LocaleExtension const& ext) {
                builder.append("-u"sv);

                for (auto const& attribute : ext.attributes)
                    append_sep_and_string(attribute);
                for (auto const& keyword : ext.keywords)
                    append_key_value_list(keyword.key, keyword.types);
            },
            [&](TransformedExtension const& ext) {
                builder.append("-t"sv);

                if (ext.language.has_value()) {
                    append_sep_and_string(ext.language->language);
                    append_sep_and_string(ext.language->script);
                    append_sep_and_string(ext.language->region);
                    for (auto const& variant : ext.language->variants)
                        append_sep_and_string(variant);
                }

                for (auto const& field : ext.fields)
                    append_key_value_list(field.key, field.values);
            },
            [&](OtherExtension const& ext) {
                builder.appendff("-{:c}", to_ascii_lowercase(ext.key));
                for (auto const& value : ext.values)
                    append_sep_and_string(value);
            });
    }

    if (!locale_id.private_use_extensions.is_empty()) {
        builder.append("-x"sv);
        for (auto const& extension : locale_id.private_use_extensions)
            append_sep_and_string(extension);
    }

    return builder.build();
}

String const& default_locale()
{
    static String locale = "en"sv;
    return locale;
}

bool is_locale_available([[maybe_unused]] StringView locale)
{
#if ENABLE_UNICODE_DATA
    return Detail::locale_from_string(locale).has_value();
#else
    return false;
#endif
}

Optional<StringView> get_locale_language_mapping([[maybe_unused]] StringView locale, [[maybe_unused]] StringView language)
{
#if ENABLE_UNICODE_DATA
    return Detail::get_locale_language_mapping(locale, language);
#else
    return {};
#endif
}

Optional<StringView> get_locale_territory_mapping([[maybe_unused]] StringView locale, [[maybe_unused]] StringView territory)
{
#if ENABLE_UNICODE_DATA
    return Detail::get_locale_territory_mapping(locale, territory);
#else
    return {};
#endif
}

Optional<StringView> get_locale_script_mapping([[maybe_unused]] StringView locale, [[maybe_unused]] StringView script)
{
#if ENABLE_UNICODE_DATA
    return Detail::get_locale_script_tag_mapping(locale, script);
#else
    return {};
#endif
}

Optional<StringView> get_locale_currency_mapping([[maybe_unused]] StringView locale, [[maybe_unused]] StringView currency)
{
#if ENABLE_UNICODE_DATA
    return Detail::get_locale_currency_mapping(locale, currency);
#else
    return {};
#endif
}

Optional<StringView> resolve_language_alias(StringView language)
{
#if ENABLE_UNICODE_DATA
    return Detail::resolve_language_alias(language);
#else
    return language;
#endif
}

Optional<StringView> resolve_territory_alias(StringView territory)
{
#if ENABLE_UNICODE_DATA
    return Detail::resolve_territory_alias(territory);
#else
    return territory;
#endif
}

Optional<StringView> resolve_script_tag_alias(StringView script_tag)
{
#if ENABLE_UNICODE_DATA
    return Detail::resolve_script_tag_alias(script_tag);
#else
    return script_tag;
#endif
}

Optional<StringView> resolve_variant_alias(StringView variant)
{
#if ENABLE_UNICODE_DATA
    return Detail::resolve_variant_alias(variant);
#else
    return variant;
#endif
}

Optional<StringView> resolve_subdivision_alias(StringView subdivision)
{
#if ENABLE_UNICODE_DATA
    return Detail::resolve_subdivision_alias(subdivision);
#else
    return subdivision;
#endif
}

String resolve_most_likely_territory([[maybe_unused]] LanguageID const& language_id, StringView territory_alias)
{
    auto aliases = territory_alias.split_view(' ');

#if ENABLE_UNICODE_DATA
    if (aliases.size() > 1) {
        auto territory = Detail::resolve_most_likely_territory(language_id);
        if (territory.has_value() && aliases.contains_slow(*territory))
            return territory.release_value();
    }
#endif

    return aliases[0].to_string();
}

String LanguageID::to_string() const
{
    StringBuilder builder;

    auto append_segment = [&](Optional<String> const& segment) {
        if (!segment.has_value())
            return;
        if (!builder.is_empty())
            builder.append('-');
        builder.append(*segment);
    };

    append_segment(language);
    append_segment(script);
    append_segment(region);
    for (auto const& variant : variants)
        append_segment(variant);

    return builder.build();
}

String LocaleID::to_string() const
{
    StringBuilder builder;

    auto append_segment = [&](Optional<String> const& segment) {
        if (!segment.has_value())
            return;
        if (!builder.is_empty())
            builder.append('-');
        builder.append(*segment);
    };

    auto append_key_value_list = [&](auto const& key, auto const& values) {
        append_segment(key);
        for (auto const& value : values)
            append_segment(value);
    };

    append_segment(language_id.to_string());

    for (auto const& extension : extensions) {
        extension.visit(
            [&](LocaleExtension const& ext) {
                builder.append("-u"sv);
                for (auto const& attribute : ext.attributes)
                    append_segment(attribute);
                for (auto const& keyword : ext.keywords)
                    append_key_value_list(keyword.key, keyword.types);
            },
            [&](TransformedExtension const& ext) {
                builder.append("-t"sv);
                if (ext.language.has_value())
                    append_segment(ext.language->to_string());
                for (auto const& field : ext.fields)
                    append_key_value_list(field.key, field.values);
            },
            [&](OtherExtension const& ext) {
                builder.appendff("-{}", ext.key);
                for (auto const& value : ext.values)
                    append_segment(value);
            });
    }

    if (!private_use_extensions.is_empty()) {
        builder.append("-x"sv);
        for (auto const& extension : private_use_extensions)
            append_segment(extension);
    }

    return builder.build();
}

}