path: root/server/src/locale.cppm

module;

#include <boost/locale.hpp>
#include <unicode/localematcher.h>

export module routemon:locale;

import std;
import :util;

export namespace blocale = boost::locale;

export namespace routemon {
using lformat = blocale::format;
using blocale::gettext;
using blocale::translate;
} // namespace routemon

namespace routemon::locale {

struct locale_priority
{
  float weight;
  std::size_t original_index;
};

auto operator<(locale_priority const& lhs, locale_priority const& rhs) -> bool
{
  if (lhs.weight != rhs.weight)
    return lhs.weight > rhs.weight;
  return lhs.original_index < rhs.original_index;
}

struct icu_locale_hash
{
  std::size_t operator()(icu::Locale const& l) const noexcept
  {
    static_assert(sizeof(std::int32_t) < sizeof(std::size_t));
    std::int32_t hash = l.hashCode();
    if (hash < 0)
    {
      return static_cast<std::size_t>(std::numeric_limits<int>::max())
             + static_cast<std::size_t>(-hash) + 1;
    }
    else
    {
      return static_cast<std::size_t>(hash);
    }
  }
};

using icu_locale_priority_map =
  std::unordered_map<icu::Locale, locale_priority, icu_locale_hash>;
using icu_priority_locale = std::pair<icu::Locale, locale_priority>;

auto operator<(icu_priority_locale const& lhs, icu_priority_locale const& rhs)
  -> bool
{
  return lhs.second < rhs.second;
}

class icu_priority_locale_vec_iterator : public icu::Locale::Iterator
{
  std::size_t i_ = 0uz;
  std::vector<icu_priority_locale> ls_;

public:
  explicit icu_priority_locale_vec_iterator(
    std::vector<icu_priority_locale>&& ls)
    : ls_(std::move(ls))
  {
  }

  auto hasNext() const -> UBool override { return i_ < ls_.size(); }

  auto next() -> icu::Locale const& override { return ls_[i_++].first; }

  ~icu_priority_locale_vec_iterator() override = default;
};

export template <class T>
concept locale_input_range =
  std::ranges::input_range<T>
  && std::same_as<std::locale const&, std::ranges::range_const_reference_t<T>>;

// Helps select a locale based on the Accept-Language header in an
// HTTP request.
export class selector
{
  std::locale default_;
  icu::LocaleMatcher matcher_;
  std::shared_ptr<blocale::generator const> lgen_;

  auto make_matcher(
    locale_input_range auto supported_locales, std::locale default_locale)
  {
    auto builder = icu::LocaleMatcher::Builder{};
    for (auto const& supported_locale : supported_locales)
    {
      auto const& supported_locale_info =
        std::use_facet<blocale::info>(supported_locale);
      auto supported_icu_locale =
        icu::Locale{supported_locale_info.name().c_str()};
      if (supported_icu_locale.isBogus())
        throw std::runtime_error{
          "supported locale gives rise to bogus ICU locale"
        };
      builder.addSupportedLocale(supported_icu_locale);
    }
    auto const& default_locale_info =
      std::use_facet<blocale::info>(default_locale);
    auto default_icu_locale = icu::Locale{default_locale_info.name().c_str()};
    if (default_icu_locale.isBogus())
      throw std::runtime_error{"default locale gives rise to bogus ICU locale"};
    builder.setDefaultLocale(&default_icu_locale);
    auto ec = UErrorCode::U_ZERO_ERROR;
    auto matcher = builder.build(ec);
    if (U_FAILURE(ec))
      throw std::runtime_error{"failed to build icu::LocaleMatcher"};
    return matcher;
  }

  // Trimming optional whitespace as defined in RFC 9110, § 12.4.2.
  static auto ltrim_ows(std::string_view s) -> std::string_view
  {
    if (auto i = s.find_first_not_of(" \t"); i != std::string_view::npos)
      s.remove_prefix(i);
    return s;
  }
  static auto rtrim_ows(std::string_view s) -> std::string_view
  {
    if (auto i = s.find_last_not_of(" \t"); i != std::string_view::npos)
      return s.substr(0, i + 1);
    return s;
  }
  static auto trim_ows(std::string_view s) -> std::string_view
  {
    return rtrim_ows(ltrim_ows(s));
  }

  auto from_icu_locale(icu::Locale const& l) const -> std::locale
  {
    auto posix_name = std::string{l.getLanguage()};
    if (l.getScript() && std::strlen(l.getScript()) > 0)
    {
      posix_name += "_";
      posix_name += l.getScript();
    }
    if (l.getCountry() && std::strlen(l.getCountry()) > 0)
    {
      posix_name += "_";
      posix_name += l.getCountry();
    }
    posix_name += ".UTF-8";
    auto added_at = false;
    if (l.getVariant() && std::strlen(l.getVariant()) > 0)
    {
      added_at = true;
      posix_name += "@";
      posix_name += l.getVariant();
    }
    auto ec = UErrorCode::U_ZERO_ERROR;
    auto* keywords = l.createKeywords(ec);
    if (U_FAILURE(ec))
      throw std::runtime_error{"failed to create keywords"};
    if (keywords)
    {
      std::int32_t kw_len = 0;
      char const* kw = nullptr;
      while (kw = keywords->next(&kw_len, ec), !U_FAILURE(ec) && kw)
      {
        auto value =
          l.getKeywordValue<std::string>(icu::StringPiece(kw, kw_len), ec);
        if (!added_at)
        {
          posix_name += "@";
          added_at = true;
        }
        else
        {
          posix_name += ";";
        }
        posix_name += kw;
        posix_name += "=";
        posix_name += value;
      }
      if (U_FAILURE(ec))
        throw std::runtime_error{"failed to iterate over keywords"};
      delete keywords;
    }
    return lgen_->generate(posix_name);
  }

public:
  // Note: lgen must live at least as long as the selector constructed here!
  // It is unfortunately not possible to copy/move a blocale::generator.
  explicit selector(
    locale_input_range auto locales, std::locale default_,
    std::shared_ptr<blocale::generator const> lgen)
    : default_{default_}, matcher_{make_matcher(locales, default_)}, lgen_{lgen}
  {
  }

  auto select(std::string_view accept_language) const -> std::locale
  {
    using namespace std::literals::string_view_literals;
    // NOTE: can also contain a *;q=0.1
    // q should have at most 3 digits after period
    auto dlpm = icu_locale_priority_map{};
    for (auto const [i, lang_prio] :
         accept_language | std::views::split(","sv) | std::views::enumerate)
    {
      auto [lang_range_ut, mweight_ut] =
        util::split_on(std::string_view{lang_prio}, ';');
      auto lang_range_str = trim_ows(lang_range_ut);
      auto mweight_str = mweight_ut.transform(trim_ows);
      if (lang_range_str == "*")
        break;

      auto ec = UErrorCode::U_ZERO_ERROR;
      auto icu_locale = icu::Locale::forLanguageTag(lang_range_str, ec);
      if (U_FAILURE(ec) || icu_locale.isBogus())
        continue; // ignore this locale

      auto weight = 1.0f;
      if (mweight_str && mweight_str->starts_with("q="))
      {
        auto weight_str = mweight_str->substr(2, 4);
        if (auto mweight =
              util::parse_float(weight_str, std::chars_format::fixed);
            mweight && 0.0f < *mweight && *mweight < 1.0f)
        {
          weight = *mweight;
        }
      }

      if (weight > 0.0f)
      {
        dlpm[icu_locale] = {
          .weight = weight,
          .original_index = static_cast<std::size_t>(i),
        };
      }
      else
      {
        dlpm.erase(icu_locale);
      }
    }

    auto desired_locales =
      std::vector<icu_priority_locale>{dlpm.begin(), dlpm.end()};
    std::sort(desired_locales.begin(), desired_locales.end());
    auto it = icu_priority_locale_vec_iterator{std::move(desired_locales)};
    auto ec = UErrorCode::U_ZERO_ERROR;
    auto res = matcher_.getBestMatchResult(it, ec);
    if (U_FAILURE(ec))
      return default_;
    auto resolved = res.makeResolvedLocale(ec); // TODO: maybe don't?
    if (U_FAILURE(ec))
      return from_icu_locale(*res.getSupportedLocale());
    return from_icu_locale(resolved);
  }
};

export auto to_bcp47_lang_tag(std::locale locale) -> std::optional<std::string>
{
  auto const& locale_info = std::use_facet<blocale::info>(locale);
  auto ec = UErrorCode::U_ZERO_ERROR;
  auto bcp47_lang_tag =
    icu::Locale{locale_info.name().c_str()}.toLanguageTag<std::string>(ec);
  if (U_FAILURE(ec))
    return std::nullopt;
  return bcp47_lang_tag;
}

#ifdef LOCALEDIR
#define LOCALEDIR_AUX_XSTR(s) LOCALEDIR_AUX_STR(s)
#define LOCALEDIR_AUX_STR(s) #s
constexpr auto messages_path = std::string_view{LOCALEDIR_AUX_XSTR(LOCALEDIR)};
#undef LOCALEDIR_AUX_STR
#undef LOCALEDIR_AUX_XSTR
#else  // ifdef LOCALEDIR
constexpr auto messages_path = std::string_view{"locale/dev"};
#endif // ifdef LOCALEDIR

export auto make_generator() -> std::shared_ptr<blocale::generator const>
{
  auto lgen = std::make_shared<blocale::generator>();
  lgen->add_messages_path(std::string{messages_path});
  lgen->add_messages_domain("routemon");
  return std::static_pointer_cast<blocale::generator const>(lgen);
}

} // namespace routemon::locale