StringExtensions.cs 1.6 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253
  1. #nullable enable
  2. #pragma warning disable CS1591
  3. using System;
  4. using System.Globalization;
  5. using System.Linq;
  6. using System.Text;
  7. using System.Text.RegularExpressions;
  8. namespace MediaBrowser.Controller.Extensions
  9. {
  10. /// <summary>
  11. /// Class BaseExtensions.
  12. /// </summary>
  13. public static class StringExtensions
  14. {
  15. public static string RemoveDiacritics(this string text)
  16. {
  17. var chars = Normalize(text, NormalizationForm.FormD)
  18. .Where(ch => CharUnicodeInfo.GetUnicodeCategory(ch) != UnicodeCategory.NonSpacingMark);
  19. return Normalize(string.Concat(chars), NormalizationForm.FormC);
  20. }
  21. private static string Normalize(string text, NormalizationForm form, bool stripStringOnFailure = true)
  22. {
  23. if (stripStringOnFailure)
  24. {
  25. try
  26. {
  27. return text.Normalize(form);
  28. }
  29. catch (ArgumentException)
  30. {
  31. // will throw if input contains invalid unicode chars
  32. // https://mnaoumov.wordpress.com/2014/06/14/stripping-invalid-characters-from-utf-16-strings/
  33. text = Regex.Replace(text, "([\ud800-\udbff](?![\udc00-\udfff]))|((?<![\ud800-\udbff])[\udc00-\udfff])", "");
  34. return Normalize(text, form, false);
  35. }
  36. }
  37. try
  38. {
  39. return text.Normalize(form);
  40. }
  41. catch (ArgumentException)
  42. {
  43. // if it still fails, return the original text
  44. return text;
  45. }
  46. }
  47. }
  48. }