Locale.h 6.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193
  1. /*
  2. * Copyright (c) 2021-2022, 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. };
  69. struct DisplayPattern {
  70. StringView locale_pattern;
  71. StringView locale_separator;
  72. };
  73. struct ListPatterns {
  74. StringView start;
  75. StringView middle;
  76. StringView end;
  77. StringView pair;
  78. };
  79. // Note: These methods only verify that the provided strings match the EBNF grammar of the
  80. // Unicode identifier subtag (i.e. no validation is done that the tags actually exist).
  81. constexpr bool is_unicode_language_subtag(StringView subtag)
  82. {
  83. // unicode_language_subtag = alpha{2,3} | alpha{5,8}
  84. if ((subtag.length() < 2) || (subtag.length() == 4) || (subtag.length() > 8))
  85. return false;
  86. return all_of(subtag, is_ascii_alpha);
  87. }
  88. constexpr bool is_unicode_script_subtag(StringView subtag)
  89. {
  90. // unicode_script_subtag = alpha{4}
  91. if (subtag.length() != 4)
  92. return false;
  93. return all_of(subtag, is_ascii_alpha);
  94. }
  95. constexpr bool is_unicode_region_subtag(StringView subtag)
  96. {
  97. // unicode_region_subtag = (alpha{2} | digit{3})
  98. if (subtag.length() == 2)
  99. return all_of(subtag, is_ascii_alpha);
  100. if (subtag.length() == 3)
  101. return all_of(subtag, is_ascii_digit);
  102. return false;
  103. }
  104. constexpr bool is_unicode_variant_subtag(StringView subtag)
  105. {
  106. // unicode_variant_subtag = (alphanum{5,8} | digit alphanum{3})
  107. if ((subtag.length() >= 5) && (subtag.length() <= 8))
  108. return all_of(subtag, is_ascii_alphanumeric);
  109. if (subtag.length() == 4)
  110. return is_ascii_digit(subtag[0]) && all_of(subtag.substring_view(1), is_ascii_alphanumeric);
  111. return false;
  112. }
  113. bool is_type_identifier(StringView);
  114. Optional<LanguageID> parse_unicode_language_id(StringView);
  115. Optional<LocaleID> parse_unicode_locale_id(StringView);
  116. void canonicalize_unicode_extension_values(StringView key, String& value, bool remove_true);
  117. Optional<String> canonicalize_unicode_locale_id(LocaleID&);
  118. String const& default_locale();
  119. bool is_locale_available(StringView locale);
  120. Style style_from_string(StringView style);
  121. StringView style_to_string(Style style);
  122. Optional<Locale> locale_from_string(StringView locale);
  123. Optional<Language> language_from_string(StringView language);
  124. Optional<Territory> territory_from_string(StringView territory);
  125. Optional<ScriptTag> script_tag_from_string(StringView script_tag);
  126. Optional<Currency> currency_from_string(StringView currency);
  127. Optional<CalendarName> calendar_name_from_string(StringView calendar);
  128. Optional<DateField> date_field_from_string(StringView calendar);
  129. Optional<Key> key_from_string(StringView key);
  130. Optional<ListPatternType> list_pattern_type_from_string(StringView list_pattern_type);
  131. Optional<DisplayPattern> get_locale_display_patterns(StringView locale);
  132. Optional<String> format_locale_for_display(StringView locale, LocaleID locale_id);
  133. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  134. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  135. Optional<StringView> get_locale_script_mapping(StringView locale, StringView script);
  136. Optional<StringView> get_locale_long_currency_mapping(StringView locale, StringView currency);
  137. Optional<StringView> get_locale_short_currency_mapping(StringView locale, StringView currency);
  138. Optional<StringView> get_locale_narrow_currency_mapping(StringView locale, StringView currency);
  139. Optional<StringView> get_locale_numeric_currency_mapping(StringView locale, StringView currency);
  140. Optional<StringView> get_locale_calendar_mapping(StringView locale, StringView calendar);
  141. Optional<StringView> get_locale_long_date_field_mapping(StringView locale, StringView date_field);
  142. Optional<StringView> get_locale_short_date_field_mapping(StringView locale, StringView date_field);
  143. Optional<StringView> get_locale_narrow_date_field_mapping(StringView locale, StringView date_field);
  144. Optional<StringView> get_locale_key_mapping(StringView locale, StringView keyword);
  145. Vector<StringView> get_locale_key_mapping_list(StringView locale, StringView keyword);
  146. Optional<ListPatterns> get_locale_list_patterns(StringView locale, StringView type, Style style);
  147. Optional<StringView> resolve_language_alias(StringView language);
  148. Optional<StringView> resolve_territory_alias(StringView territory);
  149. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  150. Optional<StringView> resolve_variant_alias(StringView variant);
  151. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  152. void resolve_complex_language_aliases(LanguageID& language_id);
  153. Optional<LanguageID> add_likely_subtags(LanguageID const& language_id);
  154. Optional<LanguageID> remove_likely_subtags(LanguageID const& language_id);
  155. Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id);
  156. String resolve_most_likely_territory_alias(LanguageID const& language_id, StringView territory_alias);
  157. }