Locale.h 4.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155
  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. struct ListPatterns {
  65. StringView start;
  66. StringView middle;
  67. StringView end;
  68. StringView pair;
  69. };
  70. // Note: These methods only verify that the provided strings match the EBNF grammar of the
  71. // Unicode identifier subtag (i.e. no validation is done that the tags actually exist).
  72. constexpr bool is_unicode_language_subtag(StringView subtag)
  73. {
  74. // unicode_language_subtag = alpha{2,3} | alpha{5,8}
  75. if ((subtag.length() < 2) || (subtag.length() == 4) || (subtag.length() > 8))
  76. return false;
  77. return all_of(subtag, is_ascii_alpha);
  78. }
  79. constexpr bool is_unicode_script_subtag(StringView subtag)
  80. {
  81. // unicode_script_subtag = alpha{4}
  82. if (subtag.length() != 4)
  83. return false;
  84. return all_of(subtag, is_ascii_alpha);
  85. }
  86. constexpr bool is_unicode_region_subtag(StringView subtag)
  87. {
  88. // unicode_region_subtag = (alpha{2} | digit{3})
  89. if (subtag.length() == 2)
  90. return all_of(subtag, is_ascii_alpha);
  91. if (subtag.length() == 3)
  92. return all_of(subtag, is_ascii_digit);
  93. return false;
  94. }
  95. constexpr bool is_unicode_variant_subtag(StringView subtag)
  96. {
  97. // unicode_variant_subtag = (alphanum{5,8} | digit alphanum{3})
  98. if ((subtag.length() >= 5) && (subtag.length() <= 8))
  99. return all_of(subtag, is_ascii_alphanumeric);
  100. if (subtag.length() == 4)
  101. return is_ascii_digit(subtag[0]) && all_of(subtag.substring_view(1), is_ascii_alphanumeric);
  102. return false;
  103. }
  104. bool is_type_identifier(StringView);
  105. Optional<LanguageID> parse_unicode_language_id(StringView);
  106. Optional<LocaleID> parse_unicode_locale_id(StringView);
  107. void canonicalize_unicode_extension_values(StringView key, String& value, bool remove_true);
  108. Optional<String> canonicalize_unicode_locale_id(LocaleID&);
  109. String const& default_locale();
  110. bool is_locale_available(StringView locale);
  111. Optional<Locale> locale_from_string(StringView locale);
  112. Optional<StringView> get_locale_language_mapping(StringView locale, StringView language);
  113. Optional<StringView> get_locale_territory_mapping(StringView locale, StringView territory);
  114. Optional<StringView> get_locale_script_mapping(StringView locale, StringView script);
  115. Optional<StringView> get_locale_currency_mapping(StringView locale, StringView currency);
  116. Vector<StringView> get_locale_key_mapping(StringView locale, StringView keyword);
  117. Optional<ListPatterns> get_locale_list_patterns(StringView locale, StringView type, StringView style);
  118. Optional<StringView> resolve_language_alias(StringView language);
  119. Optional<StringView> resolve_territory_alias(StringView territory);
  120. Optional<StringView> resolve_script_tag_alias(StringView script_tag);
  121. Optional<StringView> resolve_variant_alias(StringView variant);
  122. Optional<StringView> resolve_subdivision_alias(StringView subdivision);
  123. Optional<LanguageID> add_likely_subtags(LanguageID const& language_id);
  124. Optional<LanguageID> remove_likely_subtags(LanguageID const& language_id);
  125. String resolve_most_likely_territory(LanguageID const& language_id, StringView territory_alias);
  126. }