| | | 1 | | // ----------------------------------------------------------------------- |
| | | 2 | | // <copyright file="DiacriticsRemovalTransform.cs" company="Stéphane ANDRE"> |
| | | 3 | | // Copyright (c) Stéphane ANDRE. All rights reserved. |
| | | 4 | | // </copyright> |
| | | 5 | | // ----------------------------------------------------------------------- |
| | | 6 | | |
| | | 7 | | using System; |
| | | 8 | | using System.Globalization; |
| | | 9 | | using System.Linq; |
| | | 10 | | using System.Text; |
| | | 11 | | |
| | | 12 | | namespace MyNet.Text.Normalization; |
| | | 13 | | |
| | | 14 | | /// <summary> |
| | | 15 | | /// Removes diacritical marks from text. |
| | | 16 | | /// </summary> |
| | | 17 | | public sealed class DiacriticsRemovalTransform : ITextNormalizationTransform |
| | | 18 | | { |
| | | 19 | | /// <inheritdoc/> |
| | | 20 | | public string Apply(string input, CultureInfo culture) |
| | | 21 | | { |
| | 24 | 22 | | ArgumentNullException.ThrowIfNull(input); |
| | 24 | 23 | | ArgumentNullException.ThrowIfNull(culture); |
| | | 24 | | |
| | 24 | 25 | | if (string.IsNullOrEmpty(input)) |
| | 0 | 26 | | return input; |
| | | 27 | | |
| | 24 | 28 | | var normalized = input.Normalize(NormalizationForm.FormD); |
| | 24 | 29 | | var buffer = new StringBuilder(normalized.Length); |
| | | 30 | | |
| | 696 | 31 | | foreach (var c in from c in normalized let category = CharUnicodeInfo.GetUnicodeCategory(c) where category != Un |
| | | 32 | | { |
| | 324 | 33 | | buffer.Append(c); |
| | | 34 | | } |
| | | 35 | | |
| | 24 | 36 | | return buffer.ToString().Normalize(NormalizationForm.FormC); |
| | | 37 | | } |
| | | 38 | | } |
| | | 39 | | |