TestUnicodeNormalization.cpp 3.9 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495
  1. /*
  2. * Copyright (c) 2022, mat
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <LibTest/TestCase.h>
  7. #include <LibUnicode/Normalize.h>
  8. using namespace Unicode;
  9. TEST_CASE(normalize_nfd)
  10. {
  11. EXPECT_EQ(normalize(""sv, NormalizationForm::NFD), ""sv);
  12. EXPECT_EQ(normalize("Hello"sv, NormalizationForm::NFD), "Hello"sv);
  13. EXPECT_EQ(normalize("Amélie"sv, NormalizationForm::NFD), "Ame\u0301lie"sv);
  14. EXPECT_EQ(normalize("Office"sv, NormalizationForm::NFD), "Office"sv);
  15. EXPECT_EQ(normalize("\u1E9B\u0323"sv, NormalizationForm::NFD), "\u017F\u0323\u0307"sv);
  16. EXPECT_EQ(normalize("\u0112\u0300"sv, NormalizationForm::NFD), "\u0045\u0304\u0300"sv);
  17. EXPECT_EQ(normalize("\u03D3"sv, NormalizationForm::NFD), "\u03D2\u0301"sv);
  18. EXPECT_EQ(normalize("\u03D4"sv, NormalizationForm::NFD), "\u03D2\u0308"sv);
  19. EXPECT_EQ(normalize("닭"sv, NormalizationForm::NFD), "\u1103\u1161\u11B0"sv);
  20. EXPECT_EQ(normalize("\u1100\uAC00\u11A8"sv, NormalizationForm::NFD), "\u1100\u1100\u1161\u11A8"sv);
  21. // Composition exclusions.
  22. EXPECT_EQ(normalize("\u0958"sv, NormalizationForm::NFD), "\u0915\u093C"sv);
  23. EXPECT_EQ(normalize("\u2126"sv, NormalizationForm::NFD), "\u03A9"sv);
  24. }
  25. TEST_CASE(normalize_nfc)
  26. {
  27. EXPECT_EQ(normalize(""sv, NormalizationForm::NFC), ""sv);
  28. EXPECT_EQ(normalize("Hello"sv, NormalizationForm::NFC), "Hello"sv);
  29. EXPECT_EQ(normalize("Office"sv, NormalizationForm::NFC), "Office"sv);
  30. EXPECT_EQ(normalize("\u1E9B\u0323"sv, NormalizationForm::NFC), "\u1E9B\u0323"sv);
  31. EXPECT_EQ(normalize("\u0044\u0307"sv, NormalizationForm::NFC), "\u1E0A"sv);
  32. EXPECT_EQ(normalize("\u0044\u0307\u0323"sv, NormalizationForm::NFC), "\u1E0C\u0307"sv);
  33. EXPECT_EQ(normalize("\u0044\u0323\u0307"sv, NormalizationForm::NFC), "\u1E0C\u0307"sv);
  34. EXPECT_EQ(normalize("\u0112\u0300"sv, NormalizationForm::NFC), "\u1E14"sv);
  35. EXPECT_EQ(normalize("\u1E14\u0304"sv, NormalizationForm::NFC), "\u1E14\u0304"sv);
  36. EXPECT_EQ(normalize("\u05B8\u05B9\u05B1\u0591\u05C3\u05B0\u05AC\u059F"sv, NormalizationForm::NFC), "\u05B1\u05B8\u05B9\u0591\u05C3\u05B0\u05AC\u059F"sv);
  37. EXPECT_EQ(normalize("\u0592\u05B7\u05BC\u05A5\u05B0\u05C0\u05C4\u05AD"sv, NormalizationForm::NFC), "\u05B0\u05B7\u05BC\u05A5\u0592\u05C0\u05AD\u05C4"sv);
  38. EXPECT_EQ(normalize("\u03D3"sv, NormalizationForm::NFC), "\u03D3"sv);
  39. EXPECT_EQ(normalize("\u03D4"sv, NormalizationForm::NFC), "\u03D4"sv);
  40. EXPECT_EQ(normalize("\u0958"sv, NormalizationForm::NFC), "\u0915\u093C"sv);
  41. EXPECT_EQ(normalize("\u2126"sv, NormalizationForm::NFC), "\u03A9"sv);
  42. EXPECT_EQ(normalize("\u1103\u1161\u11B0"sv, NormalizationForm::NFC), "닭"sv);
  43. EXPECT_EQ(normalize("\u1100\uAC00\u11A8"sv, NormalizationForm::NFC), "\u1100\uAC01"sv);
  44. EXPECT_EQ(normalize("\u1103\u1161\u11B0\u11B0"sv, NormalizationForm::NFC), "닭\u11B0");
  45. }
  46. TEST_CASE(normalize_nfkd)
  47. {
  48. EXPECT_EQ(normalize(""sv, NormalizationForm::NFKD), ""sv);
  49. EXPECT_EQ(normalize("Office"sv, NormalizationForm::NFKD), "Office"sv);
  50. EXPECT_EQ(normalize("¼"sv, NormalizationForm::NFKD), "1\u20444"sv);
  51. EXPECT_EQ(normalize("\u03D3"sv, NormalizationForm::NFKD), "\u03A5\u0301"sv);
  52. EXPECT_EQ(normalize("\u03D4"sv, NormalizationForm::NFKD), "\u03A5\u0308"sv);
  53. EXPECT_EQ(normalize("\u0958"sv, NormalizationForm::NFKD), "\u0915\u093C"sv);
  54. EXPECT_EQ(normalize("\u2126"sv, NormalizationForm::NFKD), "\u03A9"sv);
  55. EXPECT_EQ(normalize("\uFDFA"sv, NormalizationForm::NFKD), "\u0635\u0644\u0649\u0020\u0627\u0644\u0644\u0647\u0020\u0639\u0644\u064A\u0647\u0020\u0648\u0633\u0644\u0645"sv);
  56. }
  57. TEST_CASE(normalize_nfkc)
  58. {
  59. EXPECT_EQ(normalize(""sv, NormalizationForm::NFKC), ""sv);
  60. EXPECT_EQ(normalize("\u03D3"sv, NormalizationForm::NFKC), "\u038E"sv);
  61. EXPECT_EQ(normalize("\u03D4"sv, NormalizationForm::NFKC), "\u03AB"sv);
  62. EXPECT_EQ(normalize("\u0958"sv, NormalizationForm::NFKC), "\u0915\u093C"sv);
  63. EXPECT_EQ(normalize("\u2126"sv, NormalizationForm::NFKC), "\u03A9"sv);
  64. }