Locale.h 6.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209
  1. /*
  2. * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #pragma once
  7. #include <AK/CharacterTypes.h>
  8. #include <AK/Optional.h>
  9. #include <AK/String.h>
  10. #include <AK/StringView.h>
  11. #include <AK/Variant.h>
  12. #include <AK/Vector.h>
  13. #include <LibUnicode/Forward.h>
  14. namespace Unicode {
  15. struct LanguageID {
  16. String to_string() const;
  17. bool operator==(LanguageID const&) const = default;
  18. bool is_root { false };
  19. Optional<String> language {};
  20. Optional<String> script {};
  21. Optional<String> region {};
  22. Vector<String> variants {};
  23. };
  24. struct Keyword {
  25. String key {};
  26. String value {};
  27. };
  28. struct LocaleExtension {
  29. Vector<String> attributes {};
  30. Vector<Keyword> keywords {};
  31. };
  32. struct TransformedField {
  33. String key {};
  34. String value {};
  35. };
  36. struct TransformedExtension {
  37. Optional<LanguageID> language {};
  38. Vector<TransformedField> fields {};
  39. };
  40. struct OtherExtension {
  41. char key {};
  42. String value {};
  43. };
  44. using Extension = Variant<LocaleExtension, TransformedExtension, OtherExtension>;
  45. struct LocaleID {
  46. String to_string() const;
  47. template<typename ExtensionType>
  48. Vector<Extension> remove_extension_type()
  49. {
  50. Vector<Extension> removed_extensions {};
  51. auto tmp_extensions = move(extensions);
  52. for (auto& extension : tmp_extensions) {
  53. if (extension.has<ExtensionType>())
  54. removed_extensions.append(move(extension));
  55. else
  56. extensions.append(move(extension));
  57. }
  58. return removed_extensions;
  59. }
  60. LanguageID language_id {};
  61. Vector<Extension> extensions {};
  62. Vector<String> private_use_extensions {};
  63. };
  64. enum class Style : u8 {
  65. Long,
  66. Short,
  67. Narrow,
  68. Numeric,
  69. };
  70. struct NumberGroupings {
  71. u8 primary_grouping_size { 0 };
  72. u8 secondary_grouping_size { 0 };
  73. };
  74. enum class StandardNumberFormatType : u8 {
  75. Decimal,
  76. Currency,
  77. Accounting,
  78. Percent,
  79. Scientific,
  80. };
  81. enum class CompactNumberFormatType : u8 {
  82. DecimalLong,
  83. DecimalShort,
  84. CurrencyUnit,
  85. CurrencyShort,
  86. };
  87. struct NumberFormat {
  88. enum class Plurality : u8 {
  89. Other,
  90. Zero,
  91. Single,
  92. One,
  93. Two,
  94. Few,
  95. Many,
  96. };
  97. u8 magnitude { 0 };
  98. u8 exponent { 0 };
  99. Plurality plurality { Plurality::Other };
  100. StringView zero_format {};
  101. StringView positive_format {};
  102. StringView negative_format {};
  103. Vector<StringView> identifiers {};
  104. };
  105. struct ListPatterns {
  106. StringView start;
  107. StringView middle;
  108. StringView end;
  109. StringView pair;
  110. };
  111. // Note: These methods only verify that the provided strings match the EBNF grammar of the
  112. // Unicode identifier subtag (i.e. no validation is done that the tags actually exist).
  113. constexpr bool is_unicode_language_subtag(StringView subtag)
  114. {
  115. // unicode_language_subtag = alpha{2,3} | alpha{5,8}
  116. if ((subtag.length() < 2) || (subtag.length() == 4) || (subtag.length() > 8))
  117. return false;
  118. return all_of(subtag, is_ascii_alpha);
  119. }
  120. constexpr bool is_unicode_script_subtag(StringView subtag)
  121. {
  122. // unicode_script_subtag = alpha{4}
  123. if (subtag.length() != 4)
  124. return false;
  125. return all_of(subtag, is_ascii_alpha);
  126. }
  127. constexpr bool is_unicode_region_subtag(StringView subtag)
  128. {
  129. // unicode_region_subtag = (alpha{2} | digit{3})
  130. if (subtag.length() == 2)
  131. return all_of(subtag, is_ascii_alpha);
  132. if (subtag.length() == 3)
  133. return all_of(subtag, is_ascii_digit);
  134. return false;
  135. }
  136. constexpr bool is_unicode_variant_subtag(StringView subtag)
  137. {
  138. // unicode_variant_subtag = (alphanum{5,8} | digit alphanum{3})
  139. if ((subtag.length() >= 5) && (subtag.length() <= 8))
  140. return all_of(subtag, is_ascii_alphanumeric);
  141. if (subtag.length() == 4)
  142. return is_ascii_digit(subtag[0]) && all_of(subtag.substring_view(1), is_ascii_alphanumeric);
  143. return false;
  144. }
  145. bool is_type_identifier(StringView);
  146. Optional<LanguageID> parse_unicode_language_id(StringView);
  147. Optional<LocaleID> parse_unicode_locale_id(StringView);
  148. void canonicalize_unicode_extension_values(StringView key, String& value, bool remove_true);
  149. Optional<String> canonicalize_unicode_locale_id(LocaleID&);
  150. String const& default_locale();
  151. bool is_locale_available(StringView locale);
  152. Optional<Locale> locale_from_string(StringView locale);
  153. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  154. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  155. Optional<StringView> get_locale_script_mapping(StringView locale, StringView script);
  156. Optional<StringView> get_locale_currency_mapping(StringView locale, StringView currency, Style style);
  157. Vector<StringView> get_locale_key_mapping(StringView locale, StringView keyword);
  158. Optional<StringView> get_number_system_symbol(StringView locale, StringView system, StringView symbol);
  159. Optional<NumberGroupings> get_number_system_groupings(StringView locale, StringView system);
  160. Optional<NumberFormat> get_standard_number_system_format(StringView locale, StringView system, StandardNumberFormatType type);
  161. Vector<NumberFormat> get_compact_number_system_formats(StringView locale, StringView system, CompactNumberFormatType type);
  162. Optional<ListPatterns> get_locale_list_patterns(StringView locale, StringView type, StringView style);
  163. Optional<StringView> resolve_language_alias(StringView language);
  164. Optional<StringView> resolve_territory_alias(StringView territory);
  165. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  166. Optional<StringView> resolve_variant_alias(StringView variant);
  167. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  168. Optional<LanguageID> add_likely_subtags(LanguageID const& language_id);
  169. Optional<LanguageID> remove_likely_subtags(LanguageID const& language_id);
  170. String resolve_most_likely_territory(LanguageID const& language_id, StringView territory_alias);
  171. Optional<NumberFormat> select_pattern_with_plurality(Vector<NumberFormat> const& formats, double number);
  172. Optional<String> augment_currency_format_pattern(StringView currency_display, StringView base_pattern);
  173. }