diff options
Diffstat (limited to 'server/src/locale.cppm')
| -rw-r--r-- | server/src/locale.cppm | 227 |
1 files changed, 11 insertions, 216 deletions
diff --git a/server/src/locale.cppm b/server/src/locale.cppm index da80f76..263de1d 100644 --- a/server/src/locale.cppm +++ b/server/src/locale.cppm | |||
| @@ -11,78 +11,20 @@ import :util; | |||
| 11 | export namespace blocale = boost::locale; | 11 | export namespace blocale = boost::locale; |
| 12 | 12 | ||
| 13 | export namespace routemon { | 13 | export namespace routemon { |
| 14 | |||
| 14 | using lformat = blocale::format; | 15 | using lformat = blocale::format; |
| 15 | using blocale::gettext; | 16 | using blocale::gettext; |
| 16 | using blocale::translate; | 17 | using blocale::translate; |
| 18 | |||
| 17 | } // namespace routemon | 19 | } // namespace routemon |
| 18 | 20 | ||
| 19 | namespace routemon::locale { | 21 | namespace routemon::locale { |
| 20 | 22 | ||
| 21 | struct locale_priority | ||
| 22 | { | ||
| 23 | float weight; | ||
| 24 | std::size_t original_index; | ||
| 25 | }; | ||
| 26 | |||
| 27 | auto operator<(locale_priority const& lhs, locale_priority const& rhs) -> bool | ||
| 28 | { | ||
| 29 | if (lhs.weight != rhs.weight) | ||
| 30 | return lhs.weight > rhs.weight; | ||
| 31 | return lhs.original_index < rhs.original_index; | ||
| 32 | } | ||
| 33 | |||
| 34 | struct icu_locale_hash | ||
| 35 | { | ||
| 36 | std::size_t operator()(icu::Locale const& l) const noexcept | ||
| 37 | { | ||
| 38 | static_assert(sizeof(std::int32_t) < sizeof(std::size_t)); | ||
| 39 | std::int32_t hash = l.hashCode(); | ||
| 40 | if (hash < 0) | ||
| 41 | { | ||
| 42 | return static_cast<std::size_t>(std::numeric_limits<int>::max()) | ||
| 43 | + static_cast<std::size_t>(-hash) + 1; | ||
| 44 | } | ||
| 45 | else | ||
| 46 | { | ||
| 47 | return static_cast<std::size_t>(hash); | ||
| 48 | } | ||
| 49 | } | ||
| 50 | }; | ||
| 51 | |||
| 52 | using icu_locale_priority_map = | ||
| 53 | std::unordered_map<icu::Locale, locale_priority, icu_locale_hash>; | ||
| 54 | using icu_priority_locale = std::pair<icu::Locale, locale_priority>; | ||
| 55 | |||
| 56 | auto operator<(icu_priority_locale const& lhs, icu_priority_locale const& rhs) | ||
| 57 | -> bool | ||
| 58 | { | ||
| 59 | return lhs.second < rhs.second; | ||
| 60 | } | ||
| 61 | |||
| 62 | class icu_priority_locale_vec_iterator : public icu::Locale::Iterator | ||
| 63 | { | ||
| 64 | std::size_t i_ = 0uz; | ||
| 65 | std::vector<icu_priority_locale> ls_; | ||
| 66 | |||
| 67 | public: | ||
| 68 | explicit icu_priority_locale_vec_iterator( | ||
| 69 | std::vector<icu_priority_locale>&& ls) | ||
| 70 | : ls_(std::move(ls)) | ||
| 71 | { | ||
| 72 | } | ||
| 73 | |||
| 74 | auto hasNext() const -> UBool override { return i_ < ls_.size(); } | ||
| 75 | |||
| 76 | auto next() -> icu::Locale const& override { return ls_[i_++].first; } | ||
| 77 | |||
| 78 | ~icu_priority_locale_vec_iterator() override = default; | ||
| 79 | }; | ||
| 80 | |||
| 81 | export template <class T> | 23 | export template <class T> |
| 82 | concept locale_input_range = | 24 | concept locale_input_range = |
| 83 | std::ranges::input_range<T> | 25 | std::ranges::input_range<T> |
| 84 | && std::same_as< | 26 | && std:: |
| 85 | std::locale const&, std::ranges::range_const_reference_t<T>>; | 27 | same_as<std::locale const&, std::ranges::range_const_reference_t<T>>; |
| 86 | 28 | ||
| 87 | // Helps select a locale based on the Accept-Language header in an | 29 | // Helps select a locale based on the Accept-Language header in an |
| 88 | // HTTP request. | 30 | // HTTP request. |
| @@ -122,75 +64,11 @@ export class selector | |||
| 122 | } | 64 | } |
| 123 | 65 | ||
| 124 | // Trimming optional whitespace as defined in RFC 9110, § 12.4.2. | 66 | // Trimming optional whitespace as defined in RFC 9110, § 12.4.2. |
| 125 | static auto ltrim_ows(std::string_view s) -> std::string_view | 67 | static auto ltrim_ows(std::string_view s) -> std::string_view; |
| 126 | { | 68 | static auto rtrim_ows(std::string_view s) -> std::string_view; |
| 127 | if (auto i = s.find_first_not_of(" \t"); i != std::string_view::npos) | 69 | static auto trim_ows(std::string_view s) -> std::string_view; |
| 128 | s.remove_prefix(i); | ||
| 129 | return s; | ||
| 130 | } | ||
| 131 | static auto rtrim_ows(std::string_view s) -> std::string_view | ||
| 132 | { | ||
| 133 | if (auto i = s.find_last_not_of(" \t"); i != std::string_view::npos) | ||
| 134 | return s.substr(0, i + 1); | ||
| 135 | return s; | ||
| 136 | } | ||
| 137 | static auto trim_ows(std::string_view s) -> std::string_view | ||
| 138 | { | ||
| 139 | return rtrim_ows(ltrim_ows(s)); | ||
| 140 | } | ||
| 141 | 70 | ||
| 142 | auto from_icu_locale(icu::Locale const& l) const -> std::locale | 71 | auto from_icu_locale(icu::Locale const& l) const -> std::locale; |
| 143 | { | ||
| 144 | auto posix_name = std::string{l.getLanguage()}; | ||
| 145 | if (l.getScript() && std::strlen(l.getScript()) > 0) | ||
| 146 | { | ||
| 147 | posix_name += "_"; | ||
| 148 | posix_name += l.getScript(); | ||
| 149 | } | ||
| 150 | if (l.getCountry() && std::strlen(l.getCountry()) > 0) | ||
| 151 | { | ||
| 152 | posix_name += "_"; | ||
| 153 | posix_name += l.getCountry(); | ||
| 154 | } | ||
| 155 | posix_name += ".UTF-8"; | ||
| 156 | auto added_at = false; | ||
| 157 | if (l.getVariant() && std::strlen(l.getVariant()) > 0) | ||
| 158 | { | ||
| 159 | added_at = true; | ||
| 160 | posix_name += "@"; | ||
| 161 | posix_name += l.getVariant(); | ||
| 162 | } | ||
| 163 | auto ec = UErrorCode::U_ZERO_ERROR; | ||
| 164 | auto* keywords = l.createKeywords(ec); | ||
| 165 | if (U_FAILURE(ec)) | ||
| 166 | throw std::runtime_error{"failed to create keywords"}; | ||
| 167 | if (keywords) | ||
| 168 | { | ||
| 169 | std::int32_t kw_len = 0; | ||
| 170 | char const* kw = nullptr; | ||
| 171 | while (kw = keywords->next(&kw_len, ec), !U_FAILURE(ec) && kw) | ||
| 172 | { | ||
| 173 | auto value = | ||
| 174 | l.getKeywordValue<std::string>(icu::StringPiece(kw, kw_len), ec); | ||
| 175 | if (!added_at) | ||
| 176 | { | ||
| 177 | posix_name += "@"; | ||
| 178 | added_at = true; | ||
| 179 | } | ||
| 180 | else | ||
| 181 | { | ||
| 182 | posix_name += ";"; | ||
| 183 | } | ||
| 184 | posix_name += kw; | ||
| 185 | posix_name += "="; | ||
| 186 | posix_name += value; | ||
| 187 | } | ||
| 188 | if (U_FAILURE(ec)) | ||
| 189 | throw std::runtime_error{"failed to iterate over keywords"}; | ||
| 190 | delete keywords; | ||
| 191 | } | ||
| 192 | return lgen_->generate(posix_name); | ||
| 193 | } | ||
| 194 | 72 | ||
| 195 | public: | 73 | public: |
| 196 | // Note: lgen must live at least as long as the selector constructed here! | 74 | // Note: lgen must live at least as long as the selector constructed here! |
| @@ -202,94 +80,11 @@ public: | |||
| 202 | { | 80 | { |
| 203 | } | 81 | } |
| 204 | 82 | ||
| 205 | auto select(std::string_view accept_language) const -> std::locale | 83 | auto select(std::string_view accept_language) const -> std::locale; |
| 206 | { | ||
| 207 | using namespace std::literals::string_view_literals; | ||
| 208 | // NOTE: can also contain a *;q=0.1 | ||
| 209 | // q should have at most 3 digits after period | ||
| 210 | auto dlpm = icu_locale_priority_map{}; | ||
| 211 | for (auto const [i, lang_prio] : | ||
| 212 | accept_language | std::views::split(","sv) | std::views::enumerate) | ||
| 213 | { | ||
| 214 | auto [lang_range_ut, mweight_ut] = | ||
| 215 | util::split_on(std::string_view{lang_prio}, ';'); | ||
| 216 | auto lang_range_str = trim_ows(lang_range_ut); | ||
| 217 | auto mweight_str = mweight_ut.transform(trim_ows); | ||
| 218 | if (lang_range_str == "*") | ||
| 219 | break; | ||
| 220 | |||
| 221 | auto ec = UErrorCode::U_ZERO_ERROR; | ||
| 222 | auto icu_locale = icu::Locale::forLanguageTag(lang_range_str, ec); | ||
| 223 | if (U_FAILURE(ec) || icu_locale.isBogus()) | ||
| 224 | continue; // ignore this locale | ||
| 225 | |||
| 226 | auto weight = 1.0f; | ||
| 227 | if (mweight_str && mweight_str->starts_with("q=")) | ||
| 228 | { | ||
| 229 | auto weight_str = mweight_str->substr(2, 4); | ||
| 230 | if (auto mweight = | ||
| 231 | util::parse_float(weight_str, std::chars_format::fixed); | ||
| 232 | mweight && 0.0f < *mweight && *mweight < 1.0f) | ||
| 233 | { | ||
| 234 | weight = *mweight; | ||
| 235 | } | ||
| 236 | } | ||
| 237 | |||
| 238 | if (weight > 0.0f) | ||
| 239 | { | ||
| 240 | dlpm[icu_locale] = { | ||
| 241 | .weight = weight, | ||
| 242 | .original_index = static_cast<std::size_t>(i), | ||
| 243 | }; | ||
| 244 | } | ||
| 245 | else | ||
| 246 | { | ||
| 247 | dlpm.erase(icu_locale); | ||
| 248 | } | ||
| 249 | } | ||
| 250 | |||
| 251 | auto desired_locales = | ||
| 252 | std::vector<icu_priority_locale>{dlpm.begin(), dlpm.end()}; | ||
| 253 | std::sort(desired_locales.begin(), desired_locales.end()); | ||
| 254 | auto it = icu_priority_locale_vec_iterator{std::move(desired_locales)}; | ||
| 255 | auto ec = UErrorCode::U_ZERO_ERROR; | ||
| 256 | auto res = matcher_.getBestMatchResult(it, ec); | ||
| 257 | if (U_FAILURE(ec)) | ||
| 258 | return default_; | ||
| 259 | auto resolved = res.makeResolvedLocale(ec); // TODO: maybe don't? | ||
| 260 | if (U_FAILURE(ec)) | ||
| 261 | return from_icu_locale(*res.getSupportedLocale()); | ||
| 262 | return from_icu_locale(resolved); | ||
| 263 | } | ||
| 264 | }; | 84 | }; |
| 265 | 85 | ||
| 266 | export auto to_bcp47_lang_tag(std::locale locale) -> std::optional<std::string> | 86 | auto to_bcp47_lang_tag(std::locale locale) -> std::optional<std::string>; |
| 267 | { | ||
| 268 | auto const& locale_info = std::use_facet<blocale::info>(locale); | ||
| 269 | auto ec = UErrorCode::U_ZERO_ERROR; | ||
| 270 | auto bcp47_lang_tag = | ||
| 271 | icu::Locale{locale_info.name().c_str()}.toLanguageTag<std::string>(ec); | ||
| 272 | if (U_FAILURE(ec)) | ||
| 273 | return std::nullopt; | ||
| 274 | return bcp47_lang_tag; | ||
| 275 | } | ||
| 276 | |||
| 277 | #ifdef LOCALEDIR | ||
| 278 | #define LOCALEDIR_AUX_XSTR(s) LOCALEDIR_AUX_STR(s) | ||
| 279 | #define LOCALEDIR_AUX_STR(s) #s | ||
| 280 | constexpr auto messages_path = std::string_view{LOCALEDIR_AUX_XSTR(LOCALEDIR)}; | ||
| 281 | #undef LOCALEDIR_AUX_STR | ||
| 282 | #undef LOCALEDIR_AUX_XSTR | ||
| 283 | #else // ifdef LOCALEDIR | ||
| 284 | constexpr auto messages_path = std::string_view{"locale/dev"}; | ||
| 285 | #endif // ifdef LOCALEDIR | ||
| 286 | 87 | ||
| 287 | export auto make_generator() -> std::shared_ptr<blocale::generator const> | 88 | export auto make_generator() -> std::shared_ptr<blocale::generator const>; |
| 288 | { | ||
| 289 | auto lgen = std::make_shared<blocale::generator>(); | ||
| 290 | lgen->add_messages_path(std::string{messages_path}); | ||
| 291 | lgen->add_messages_domain("routemon"); | ||
| 292 | return std::static_pointer_cast<blocale::generator const>(lgen); | ||
| 293 | } | ||
| 294 | 89 | ||
| 295 | } // namespace routemon::locale | 90 | } // namespace routemon::locale |