| | | 1 | | // ----------------------------------------------------------------------- |
| | | 2 | | // <copyright file="WhitespaceNormalizationTransform.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.Text; |
| | | 10 | | |
| | | 11 | | namespace MyNet.Text.Normalization; |
| | | 12 | | |
| | | 13 | | /// <summary> |
| | | 14 | | /// Trims text and optionally collapses whitespace runs. |
| | | 15 | | /// </summary> |
| | | 16 | | public sealed class WhitespaceNormalizationTransform(TextNormalizationOptions options) : ITextNormalizationTransform |
| | | 17 | | { |
| | | 18 | | /// <inheritdoc/> |
| | | 19 | | public string Apply(string input, CultureInfo culture) |
| | | 20 | | { |
| | 6 | 21 | | ArgumentNullException.ThrowIfNull(input); |
| | 6 | 22 | | ArgumentNullException.ThrowIfNull(culture); |
| | | 23 | | |
| | 6 | 24 | | var value = options.Trim ? input.Trim() : input; |
| | 6 | 25 | | if (!options.CollapseWhitespace || value.Length == 0) |
| | 0 | 26 | | return value; |
| | | 27 | | |
| | 6 | 28 | | var sb = new StringBuilder(value.Length); |
| | 6 | 29 | | var seenWhitespace = false; |
| | | 30 | | |
| | 216 | 31 | | foreach (var ch in value) |
| | | 32 | | { |
| | 102 | 33 | | if (char.IsWhiteSpace(ch)) |
| | | 34 | | { |
| | 27 | 35 | | seenWhitespace = true; |
| | 27 | 36 | | continue; |
| | | 37 | | } |
| | | 38 | | |
| | 75 | 39 | | if (seenWhitespace && sb.Length > 0) |
| | 9 | 40 | | sb.Append(' '); |
| | | 41 | | |
| | 75 | 42 | | sb.Append(ch); |
| | 75 | 43 | | seenWhitespace = false; |
| | | 44 | | } |
| | | 45 | | |
| | 6 | 46 | | return sb.ToString(); |
| | | 47 | | } |
| | | 48 | | } |
| | | 49 | | |