TestUnicodeLocale.cpp 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266
  1. /*
  2. * Copyright (c) 2021, Tim Flynn <trflynn89@pm.me>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <LibTest/TestCase.h>
  7. #include <LibUnicode/Locale.h>
  8. TEST_CASE(is_unicode_language_subtag)
  9. {
  10. EXPECT(Unicode::is_unicode_language_subtag("aa"sv));
  11. EXPECT(Unicode::is_unicode_language_subtag("aaa"sv));
  12. EXPECT(Unicode::is_unicode_language_subtag("aaaaa"sv));
  13. EXPECT(Unicode::is_unicode_language_subtag("aaaaaa"sv));
  14. EXPECT(Unicode::is_unicode_language_subtag("aaaaaaa"sv));
  15. EXPECT(Unicode::is_unicode_language_subtag("aaaaaaaa"sv));
  16. EXPECT(!Unicode::is_unicode_language_subtag(""sv));
  17. EXPECT(!Unicode::is_unicode_language_subtag("a"sv));
  18. EXPECT(!Unicode::is_unicode_language_subtag("aaaa"sv));
  19. EXPECT(!Unicode::is_unicode_language_subtag("aaaaaaaaa"sv));
  20. EXPECT(!Unicode::is_unicode_language_subtag("123"sv));
  21. }
  22. TEST_CASE(is_unicode_script_subtag)
  23. {
  24. EXPECT(Unicode::is_unicode_script_subtag("aaaa"sv));
  25. EXPECT(!Unicode::is_unicode_script_subtag(""sv));
  26. EXPECT(!Unicode::is_unicode_script_subtag("a"sv));
  27. EXPECT(!Unicode::is_unicode_script_subtag("aa"sv));
  28. EXPECT(!Unicode::is_unicode_script_subtag("aaa"sv));
  29. EXPECT(!Unicode::is_unicode_script_subtag("aaaaa"sv));
  30. EXPECT(!Unicode::is_unicode_script_subtag("1234"sv));
  31. }
  32. TEST_CASE(is_unicode_region_subtag)
  33. {
  34. EXPECT(Unicode::is_unicode_region_subtag("aa"sv));
  35. EXPECT(Unicode::is_unicode_region_subtag("123"sv));
  36. EXPECT(!Unicode::is_unicode_region_subtag(""sv));
  37. EXPECT(!Unicode::is_unicode_region_subtag("a"sv));
  38. EXPECT(!Unicode::is_unicode_region_subtag("aaa"sv));
  39. EXPECT(!Unicode::is_unicode_region_subtag("12"sv));
  40. EXPECT(!Unicode::is_unicode_region_subtag("12a"sv));
  41. }
  42. TEST_CASE(is_unicode_variant_subtag)
  43. {
  44. EXPECT(Unicode::is_unicode_variant_subtag("aaaaa"sv));
  45. EXPECT(Unicode::is_unicode_variant_subtag("aaaaaa"sv));
  46. EXPECT(Unicode::is_unicode_variant_subtag("aaaaaaa"sv));
  47. EXPECT(Unicode::is_unicode_variant_subtag("aaaaaaaa"sv));
  48. EXPECT(Unicode::is_unicode_variant_subtag("1aaa"sv));
  49. EXPECT(Unicode::is_unicode_variant_subtag("12aa"sv));
  50. EXPECT(Unicode::is_unicode_variant_subtag("123a"sv));
  51. EXPECT(Unicode::is_unicode_variant_subtag("1234"sv));
  52. EXPECT(!Unicode::is_unicode_variant_subtag(""sv));
  53. EXPECT(!Unicode::is_unicode_variant_subtag("a"sv));
  54. EXPECT(!Unicode::is_unicode_variant_subtag("aa"sv));
  55. EXPECT(!Unicode::is_unicode_variant_subtag("aaa"sv));
  56. EXPECT(!Unicode::is_unicode_variant_subtag("aaaa"sv));
  57. EXPECT(!Unicode::is_unicode_variant_subtag("aaaaaaaaa"sv));
  58. EXPECT(!Unicode::is_unicode_variant_subtag("a234"sv));
  59. }
  60. TEST_CASE(parse_unicode_locale_id)
  61. {
  62. auto fail = [](StringView locale) {
  63. auto locale_id = Unicode::parse_unicode_locale_id(locale);
  64. EXPECT(!locale_id.has_value());
  65. };
  66. auto pass = [](StringView locale, Optional<StringView> expected_language, Optional<StringView> expected_script, Optional<StringView> expected_region, Vector<StringView> expected_variants) {
  67. auto locale_id = Unicode::parse_unicode_locale_id(locale);
  68. VERIFY(locale_id.has_value());
  69. EXPECT_EQ(locale_id->language_id.language, expected_language);
  70. EXPECT_EQ(locale_id->language_id.script, expected_script);
  71. EXPECT_EQ(locale_id->language_id.region, expected_region);
  72. EXPECT_EQ(locale_id->language_id.variants, expected_variants);
  73. };
  74. fail("a"sv);
  75. fail("1234"sv);
  76. fail("aaa-"sv);
  77. fail("aaa-cc-"sv);
  78. fail("aaa-bbbb-cc-"sv);
  79. fail("aaa-bbbb-cc-123"sv);
  80. pass("aaa"sv, "aaa"sv, {}, {}, {});
  81. pass("aaa-bbbb"sv, "aaa"sv, "bbbb"sv, {}, {});
  82. pass("aaa-cc"sv, "aaa"sv, {}, "cc"sv, {});
  83. pass("aaa-bbbb-cc"sv, "aaa"sv, "bbbb"sv, "cc"sv, {});
  84. pass("aaa-bbbb-cc-1234"sv, "aaa"sv, "bbbb"sv, "cc"sv, { "1234"sv });
  85. pass("aaa-bbbb-cc-1234-5678"sv, "aaa"sv, "bbbb"sv, "cc"sv, { "1234"sv, "5678"sv });
  86. }
  87. TEST_CASE(parse_unicode_locale_id_with_unicode_locale_extension)
  88. {
  89. auto fail = [](StringView locale) {
  90. auto locale_id = Unicode::parse_unicode_locale_id(locale);
  91. EXPECT(!locale_id.has_value());
  92. };
  93. auto pass = [](StringView locale, Unicode::LocaleExtension const& expected_extension) {
  94. auto locale_id = Unicode::parse_unicode_locale_id(locale);
  95. VERIFY(locale_id.has_value());
  96. EXPECT_EQ(locale_id->extensions.size(), 1u);
  97. auto const& actual_extension = locale_id->extensions[0].get<Unicode::LocaleExtension>();
  98. VERIFY(actual_extension.attributes == expected_extension.attributes);
  99. EXPECT_EQ(actual_extension.keywords.size(), expected_extension.keywords.size());
  100. for (size_t i = 0; i < actual_extension.keywords.size(); ++i) {
  101. auto const& actual_keyword = actual_extension.keywords[i];
  102. auto const& expected_keyword = expected_extension.keywords[i];
  103. EXPECT_EQ(actual_keyword.key, expected_keyword.key);
  104. EXPECT_EQ(actual_keyword.types, expected_keyword.types);
  105. }
  106. };
  107. fail("en-u"sv);
  108. fail("en-u-"sv);
  109. fail("en-u-x"sv);
  110. fail("en-u-xx-"sv);
  111. fail("en-u--xx"sv);
  112. fail("en-u-xx-xxxxx-"sv);
  113. fail("en-u-xx--xxxxx"sv);
  114. fail("en-u-xx-xxxxxxxxx"sv);
  115. fail("en-u-xxxxx-"sv);
  116. fail("en-u-xxxxxxxxx"sv);
  117. pass("en-u-xx"sv, { {}, { { "xx"sv, {} } } });
  118. pass("en-u-xx-yyyy"sv, { {}, { { "xx"sv, { "yyyy"sv } } } });
  119. pass("en-u-xx-yyyy-zzzz"sv, { {}, { { "xx"sv, { "yyyy"sv, "zzzz"sv } } } });
  120. pass("en-u-xx-yyyy-zzzz-aa"sv, { {}, { { "xx"sv, { "yyyy"sv, "zzzz"sv } }, { "aa"sv, {} } } });
  121. pass("en-u-xxx"sv, { { "xxx"sv }, {} });
  122. pass("en-u-fff-gggg"sv, { { "fff"sv, "gggg"sv }, {} });
  123. pass("en-u-fff-xx"sv, { { "fff"sv }, { { "xx"sv, {} } } });
  124. pass("en-u-fff-xx-yyyy"sv, { { "fff"sv }, { { "xx"sv, { "yyyy"sv } } } });
  125. pass("en-u-fff-gggg-xx-yyyy"sv, { { "fff"sv, "gggg"sv }, { { "xx"sv, { "yyyy"sv } } } });
  126. }
  127. TEST_CASE(parse_unicode_locale_id_with_transformed_extension)
  128. {
  129. auto fail = [](StringView locale) {
  130. auto locale_id = Unicode::parse_unicode_locale_id(locale);
  131. EXPECT(!locale_id.has_value());
  132. };
  133. auto pass = [](StringView locale, Unicode::TransformedExtension const& expected_extension) {
  134. auto locale_id = Unicode::parse_unicode_locale_id(locale);
  135. VERIFY(locale_id.has_value());
  136. EXPECT_EQ(locale_id->extensions.size(), 1u);
  137. auto const& actual_extension = locale_id->extensions[0].get<Unicode::TransformedExtension>();
  138. VERIFY(actual_extension.language.has_value() == expected_extension.language.has_value());
  139. if (actual_extension.language.has_value()) {
  140. EXPECT_EQ(actual_extension.language->language, expected_extension.language->language);
  141. EXPECT_EQ(actual_extension.language->script, expected_extension.language->script);
  142. EXPECT_EQ(actual_extension.language->region, expected_extension.language->region);
  143. EXPECT_EQ(actual_extension.language->variants, expected_extension.language->variants);
  144. }
  145. EXPECT_EQ(actual_extension.fields.size(), expected_extension.fields.size());
  146. for (size_t i = 0; i < actual_extension.fields.size(); ++i) {
  147. auto const& actual_field = actual_extension.fields[i];
  148. auto const& expected_field = expected_extension.fields[i];
  149. EXPECT_EQ(actual_field.key, expected_field.key);
  150. EXPECT_EQ(actual_field.values, expected_field.values);
  151. }
  152. };
  153. fail("en-t"sv);
  154. fail("en-t-"sv);
  155. fail("en-t-a"sv);
  156. fail("en-t-en-"sv);
  157. fail("en-t-root"sv);
  158. fail("en-t-aaaaaaaaa"sv);
  159. fail("en-t-en-aaa"sv);
  160. fail("en-t-en-latn-latn"sv);
  161. fail("en-t-en-a"sv);
  162. fail("en-t-en-00"sv);
  163. fail("en-t-en-latn-0"sv);
  164. fail("en-t-en-latn-00"sv);
  165. fail("en-t-en-latn-xyz"sv);
  166. fail("en-t-en-aaaaaaaaa"sv);
  167. fail("en-t-en-latn-gb-aaaa"sv);
  168. fail("en-t-en-latn-gb-aaaaaaaaa"sv);
  169. fail("en-t-k0"sv);
  170. fail("en-t-k0-aa"sv);
  171. fail("en-t-k0-aaaaaaaaa"sv);
  172. pass("en-t-en"sv, { Unicode::LanguageID { false, "en"sv }, {} });
  173. pass("en-t-en-latn"sv, { Unicode::LanguageID { false, "en"sv, "latn"sv }, {} });
  174. pass("en-t-en-us"sv, { Unicode::LanguageID { false, "en"sv, {}, "us"sv }, {} });
  175. pass("en-t-en-latn-us"sv, { Unicode::LanguageID { false, "en"sv, "latn"sv, "us"sv }, {} });
  176. pass("en-t-en-posix"sv, { Unicode::LanguageID { false, "en"sv, {}, {}, { "posix"sv } }, {} });
  177. pass("en-t-en-latn-posix"sv, { Unicode::LanguageID { false, "en"sv, "latn"sv, {}, { "posix"sv } }, {} });
  178. pass("en-t-en-us-posix"sv, { Unicode::LanguageID { false, "en"sv, {}, "us"sv, { "posix"sv } }, {} });
  179. pass("en-t-en-latn-us-posix"sv, { Unicode::LanguageID { false, "en"sv, "latn"sv, "us"sv, { "posix"sv } }, {} });
  180. pass("en-t-k0-aaa"sv, { {}, { { "k0"sv, { "aaa"sv } } } });
  181. pass("en-t-k0-aaa-bbbb"sv, { {}, { { "k0"sv, { "aaa"sv, "bbbb" } } } });
  182. pass("en-t-k0-aaa-k1-bbbb"sv, { {}, { { "k0"sv, { "aaa"sv } }, { "k1"sv, { "bbbb"sv } } } });
  183. pass("en-t-en-k0-aaa"sv, { Unicode::LanguageID { false, "en"sv }, { { "k0"sv, { "aaa"sv } } } });
  184. }
  185. TEST_CASE(parse_unicode_locale_id_with_other_extension)
  186. {
  187. auto fail = [](StringView locale) {
  188. auto locale_id = Unicode::parse_unicode_locale_id(locale);
  189. EXPECT(!locale_id.has_value());
  190. };
  191. auto pass = [](StringView locale, Unicode::OtherExtension const& expected_extension) {
  192. auto locale_id = Unicode::parse_unicode_locale_id(locale);
  193. VERIFY(locale_id.has_value());
  194. EXPECT_EQ(locale_id->extensions.size(), 1u);
  195. auto const& actual_extension = locale_id->extensions[0].get<Unicode::OtherExtension>();
  196. EXPECT_EQ(actual_extension.key, expected_extension.key);
  197. EXPECT_EQ(actual_extension.values, expected_extension.values);
  198. };
  199. fail("en-z"sv);
  200. fail("en-0"sv);
  201. fail("en-z-"sv);
  202. fail("en-0-"sv);
  203. fail("en-z-a"sv);
  204. fail("en-0-a"sv);
  205. fail("en-z-aaaaaaaaa"sv);
  206. fail("en-0-aaaaaaaaa"sv);
  207. fail("en-z-aaa-"sv);
  208. fail("en-0-aaa-"sv);
  209. fail("en-z-aaa-a"sv);
  210. fail("en-0-aaa-a"sv);
  211. pass("en-z-aa", { 'z', { "aa"sv } });
  212. pass("en-z-aa-bbb", { 'z', { "aa"sv, "bbb"sv } });
  213. pass("en-z-aa-bbb-cccccccc", { 'z', { "aa"sv, "bbb"sv, "cccccccc"sv } });
  214. }
  215. TEST_CASE(canonicalize_unicode_locale_id)
  216. {
  217. auto test = [](StringView locale, StringView expected_canonical_locale) {
  218. auto locale_id = Unicode::parse_unicode_locale_id(locale);
  219. VERIFY(locale_id.has_value());
  220. auto canonical_locale = Unicode::canonicalize_unicode_locale_id(*locale_id);
  221. EXPECT_EQ(canonical_locale, expected_canonical_locale);
  222. };
  223. test("aaa"sv, "aaa"sv);
  224. test("AaA"sv, "aaa"sv);
  225. test("aaa-bbbb"sv, "aaa-Bbbb"sv);
  226. test("aaa-cc"sv, "aaa-CC"sv);
  227. test("aaa-bBBB-cC"sv, "aaa-Bbbb-CC"sv);
  228. test("aaa-bbbb-cc-1234"sv, "aaa-Bbbb-CC-1234"sv);
  229. test("aaa-bbbb-cc-ABCDE"sv, "aaa-Bbbb-CC-abcde"sv);
  230. }