GraphemeBreakClassTrieGenerator.cs 3.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109
  1. using System;
  2. using System.Collections.Generic;
  3. using System.IO;
  4. using System.Net.Http;
  5. using System.Text.RegularExpressions;
  6. using Avalonia.Media.TextFormatting.Unicode;
  7. namespace Avalonia.Base.UnitTests.Media.TextFormatting
  8. {
  9. public static class GraphemeBreakClassTrieGenerator
  10. {
  11. public static void Execute()
  12. {
  13. if (!Directory.Exists("Generated"))
  14. {
  15. Directory.CreateDirectory("Generated");
  16. }
  17. var trie = GenerateBreakTypeTrie();
  18. UnicodeDataGenerator.GenerateTrieClass("GraphemeBreak", trie);
  19. }
  20. private static UnicodeTrie GenerateBreakTypeTrie()
  21. {
  22. var trieBuilder = new UnicodeTrieBuilder();
  23. var graphemeBreakData = ReadBreakData(Path.Combine(UnicodeDataGenerator.Ucd, "auxiliary/GraphemeBreakProperty.txt"));
  24. var emojiBreakData = ReadBreakData(Path.Combine(UnicodeDataGenerator.Ucd, "emoji/emoji-data.txt"));
  25. foreach (var breakData in new [] { graphemeBreakData, emojiBreakData })
  26. {
  27. foreach (var (start, end, graphemeBreakType) in breakData)
  28. {
  29. if (!Enum.TryParse<GraphemeBreakClass>(graphemeBreakType.Replace("_", ""), out var value))
  30. {
  31. continue;
  32. }
  33. if (start == end)
  34. {
  35. trieBuilder.Set(start, (uint)value);
  36. }
  37. else
  38. {
  39. trieBuilder.SetRange(start, end, (uint)value);
  40. }
  41. }
  42. }
  43. return trieBuilder.Freeze();
  44. }
  45. public static List<(int, int, string)> ReadBreakData(string file)
  46. {
  47. var data = new List<(int, int, string)>();
  48. var rx = new Regex(@"([0-9A-F]+)(?:\.\.([0-9A-F]+))?\s*;\s*(\w+)\s*#.*", RegexOptions.Compiled);
  49. using (var client = new HttpClient())
  50. {
  51. using (var result = client.GetAsync(file).GetAwaiter().GetResult())
  52. {
  53. if (!result.IsSuccessStatusCode)
  54. {
  55. return data;
  56. }
  57. using (var stream = result.Content.ReadAsStreamAsync().GetAwaiter().GetResult())
  58. using (var reader = new StreamReader(stream))
  59. {
  60. while (!reader.EndOfStream)
  61. {
  62. var line = reader.ReadLine();
  63. if (string.IsNullOrEmpty(line))
  64. {
  65. continue;
  66. }
  67. var match = rx.Match(line);
  68. if (!match.Success)
  69. {
  70. continue;
  71. }
  72. var start = Convert.ToInt32(match.Groups[1].Value, 16);
  73. var end = start;
  74. if (!string.IsNullOrEmpty(match.Groups[2].Value))
  75. {
  76. end = Convert.ToInt32(match.Groups[2].Value, 16);
  77. }
  78. var breakType = match.Groups[3].Value;
  79. data.Add((start, end, breakType));
  80. }
  81. }
  82. }
  83. }
  84. return data;
  85. }
  86. }
  87. }