UnicodeEnumsGenerator.cs 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503
  1. using System;
  2. using System.Collections.Generic;
  3. using System.IO;
  4. using System.Linq;
  5. using System.Net.Http;
  6. using Xunit;
  7. namespace Avalonia.Base.UnitTests.Media.TextFormatting
  8. {
  9. internal static class UnicodeEnumsGenerator
  10. {
  11. public static List<DataEntry> CreateScriptEnum()
  12. {
  13. var entries = new List<DataEntry>
  14. {
  15. new DataEntry("Unknown", "Zzzz", string.Empty),
  16. new DataEntry("Common", "Zyyy", string.Empty),
  17. new DataEntry("Inherited", "Zinh", string.Empty)
  18. };
  19. ParseDataEntries("# Script (sc)", entries);
  20. using (var stream = File.Create("Generated\\Script.cs"))
  21. using (var writer = new StreamWriter(stream))
  22. {
  23. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  24. writer.WriteLine("{");
  25. writer.WriteLine(" public enum Script");
  26. writer.WriteLine(" {");
  27. foreach (var entry in entries)
  28. {
  29. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  30. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  31. }
  32. writer.WriteLine(" }");
  33. writer.WriteLine("}");
  34. }
  35. return entries;
  36. }
  37. public static List<DataEntry> CreateEastAsianWidthClassEnum()
  38. {
  39. var entries = new List<DataEntry>();
  40. ParseDataEntries("# East_Asian_Width (ea)", entries);
  41. using (var stream = File.Create("Generated\\EastAsianWidthClass.cs"))
  42. using (var writer = new StreamWriter(stream))
  43. {
  44. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  45. writer.WriteLine("{");
  46. writer.WriteLine(" public enum EastAsianWidthClass");
  47. writer.WriteLine(" {");
  48. foreach (var entry in entries)
  49. {
  50. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  51. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  52. }
  53. writer.WriteLine(" }");
  54. writer.WriteLine("}");
  55. }
  56. return entries;
  57. }
  58. public static List<DataEntry> CreateGeneralCategoryEnum()
  59. {
  60. var entries = new List<DataEntry> { new DataEntry("Other", "C", " Cc | Cf | Cn | Co | Cs") };
  61. ParseDataEntries("# General_Category (gc)", entries);
  62. using (var stream = File.Create("Generated\\GeneralCategory.cs"))
  63. using (var writer = new StreamWriter(stream))
  64. {
  65. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  66. writer.WriteLine("{");
  67. writer.WriteLine(" public enum GeneralCategory");
  68. writer.WriteLine(" {");
  69. foreach (var entry in entries)
  70. {
  71. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  72. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  73. }
  74. writer.WriteLine(" }");
  75. writer.WriteLine("}");
  76. }
  77. return entries;
  78. }
  79. public static List<DataEntry> CreateGraphemeBreakTypeEnum()
  80. {
  81. var entries = new List<DataEntry> { new DataEntry("Other", "XX", string.Empty) };
  82. ParseDataEntries("# Grapheme_Cluster_Break (GCB)", entries);
  83. using (var stream = File.Create("Generated\\GraphemeBreakClass.cs"))
  84. using (var writer = new StreamWriter(stream))
  85. {
  86. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  87. writer.WriteLine("{");
  88. writer.WriteLine(" public enum GraphemeBreakClass");
  89. writer.WriteLine(" {");
  90. foreach (var entry in entries)
  91. {
  92. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  93. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  94. }
  95. writer.WriteLine(" ExtendedPictographic");
  96. writer.WriteLine(" }");
  97. writer.WriteLine("}");
  98. }
  99. return entries;
  100. }
  101. private static List<string> GenerateBreakPairTable()
  102. {
  103. var rows = new List<string[]>();
  104. using (var stream =
  105. typeof(UnicodeEnumsGenerator).Assembly.GetManifestResourceStream(
  106. "Avalonia.Base.UnitTests.Media.TextFormatting.BreakPairTable.txt")!)
  107. using (var reader = new StreamReader(stream))
  108. {
  109. while (!reader.EndOfStream)
  110. {
  111. var line = reader.ReadLine();
  112. Assert.NotNull(line);
  113. var columns = line.Split('\t');
  114. rows.Add(columns);
  115. }
  116. }
  117. using (var stream = File.Create("Generated\\BreakPairTable.cs"))
  118. using (var writer = new StreamWriter(stream))
  119. {
  120. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  121. writer.WriteLine("{");
  122. writer.WriteLine(" internal static class BreakPairTable");
  123. writer.WriteLine(" {");
  124. writer.WriteLine(" private static readonly byte[][] s_breakPairTable = ");
  125. writer.WriteLine(" {");
  126. for (var i = 1; i < rows.Count; i++)
  127. {
  128. writer.Write(" new byte[] {");
  129. writer.Write($"{GetBreakPairType(rows[i][1])}");
  130. for (var index = 2; index < rows[i].Length; index++)
  131. {
  132. var column = rows[i][index];
  133. writer.Write($",{GetBreakPairType(column)}");
  134. }
  135. writer.Write("},");
  136. writer.Write(Environment.NewLine);
  137. }
  138. writer.WriteLine(" };");
  139. writer.WriteLine();
  140. writer.WriteLine(" public static PairBreakType Map(LineBreakClass first, LineBreakClass second)");
  141. writer.WriteLine(" {");
  142. writer.WriteLine(" return (PairBreakType)s_breakPairTable[(int)first][(int)second];");
  143. writer.WriteLine(" }");
  144. writer.WriteLine(" }");
  145. writer.WriteLine();
  146. writer.WriteLine(" internal enum PairBreakType : byte");
  147. writer.WriteLine(" {");
  148. writer.WriteLine(" DI = 0, // Direct break opportunity");
  149. writer.WriteLine(" IN = 1, // Indirect break opportunity");
  150. writer.WriteLine(" CI = 2, // Indirect break opportunity for combining marks");
  151. writer.WriteLine(" CP = 3, // Prohibited break for combining marks");
  152. writer.WriteLine(" PR = 4 // Prohibited break");
  153. writer.WriteLine(" }");
  154. writer.WriteLine("}");
  155. }
  156. return rows[0].Where(x => !string.IsNullOrEmpty(x)).ToList();
  157. }
  158. public const byte DI_BRK = 0; // Direct break opportunity
  159. public const byte IN_BRK = 1; // Indirect break opportunity
  160. public const byte CI_BRK = 2; // Indirect break opportunity for combining marks
  161. public const byte CP_BRK = 3; // Prohibited break for combining marks
  162. public const byte PR_BRK = 4; // Prohibited break
  163. private static byte GetBreakPairType(string type)
  164. {
  165. switch (type)
  166. {
  167. case "_":
  168. return DI_BRK;
  169. case "%":
  170. return IN_BRK;
  171. case "#":
  172. return CI_BRK;
  173. case "@":
  174. return CP_BRK;
  175. case "^":
  176. return PR_BRK;
  177. default:
  178. return byte.MaxValue;
  179. }
  180. }
  181. public static List<DataEntry> CreateLineBreakClassEnum()
  182. {
  183. var usedLineBreakClasses = GenerateBreakPairTable();
  184. var entries = new List<DataEntry> { new DataEntry("Unknown", "XX", string.Empty) };
  185. ParseDataEntries("# Line_Break (lb)", entries);
  186. var orderedLineBreakEntries = new Dictionary<string, DataEntry>();
  187. foreach (var tag in usedLineBreakClasses)
  188. {
  189. var entry = entries.Single(x => x.Tag == tag);
  190. orderedLineBreakEntries.Add(tag, entry);
  191. }
  192. foreach (var entry in entries)
  193. {
  194. if (orderedLineBreakEntries.ContainsKey(entry.Tag))
  195. {
  196. continue;
  197. }
  198. orderedLineBreakEntries.Add(entry.Tag, entry);
  199. }
  200. using (var stream = File.Create("Generated\\LineBreakClass.cs"))
  201. using (var writer = new StreamWriter(stream))
  202. {
  203. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  204. writer.WriteLine("{");
  205. writer.WriteLine(" public enum LineBreakClass");
  206. writer.WriteLine(" {");
  207. foreach (var entry in orderedLineBreakEntries.Values)
  208. {
  209. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  210. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  211. }
  212. writer.WriteLine(" }");
  213. writer.WriteLine("}");
  214. }
  215. return orderedLineBreakEntries.Values.ToList();
  216. }
  217. public static List<DataEntry> CreateBidiClassEnum()
  218. {
  219. var entries = new List<DataEntry> { new DataEntry("Left_To_Right", "L", string.Empty) };
  220. ParseDataEntries("# Bidi_Class (bc)", entries);
  221. using (var stream = File.Create("Generated\\BidiClass.cs"))
  222. using (var writer = new StreamWriter(stream))
  223. {
  224. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  225. writer.WriteLine("{");
  226. writer.WriteLine(" public enum BidiClass");
  227. writer.WriteLine(" {");
  228. foreach (var entry in entries)
  229. {
  230. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  231. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  232. }
  233. writer.WriteLine(" }");
  234. writer.WriteLine("}");
  235. }
  236. return entries;
  237. }
  238. public static List<DataEntry> CreateBiDiPairedBracketTypeEnum()
  239. {
  240. var entries = new List<DataEntry> { new DataEntry("None", "n", string.Empty) };
  241. ParseDataEntries("# Bidi_Paired_Bracket_Type (bpt)", entries);
  242. using (var stream = File.Create("Generated\\BidiPairedBracketType.cs"))
  243. using (var writer = new StreamWriter(stream))
  244. {
  245. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  246. writer.WriteLine("{");
  247. writer.WriteLine(" public enum BidiPairedBracketType");
  248. writer.WriteLine(" {");
  249. foreach (var entry in entries)
  250. {
  251. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  252. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  253. }
  254. writer.WriteLine(" }");
  255. writer.WriteLine("}");
  256. }
  257. return entries;
  258. }
  259. public static void CreatePropertyValueAliasHelper(UnicodeDataEntries unicodeDataEntries,
  260. BiDiDataEntries biDiDataEntries)
  261. {
  262. using (var stream = File.Create("Generated\\PropertyValueAliasHelper.cs"))
  263. using (var writer = new StreamWriter(stream))
  264. {
  265. writer.WriteLine("using System.Collections.Generic;");
  266. writer.WriteLine();
  267. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  268. writer.WriteLine("{");
  269. writer.WriteLine(" internal static class PropertyValueAliasHelper");
  270. writer.WriteLine(" {");
  271. WritePropertyValueAliasGetTag(writer, unicodeDataEntries.Scripts, "Script", "Zzzz");
  272. WritePropertyValueAlias(writer, unicodeDataEntries.Scripts, "Script", "Unknown");
  273. WritePropertyValueAlias(writer, unicodeDataEntries.GeneralCategories, "GeneralCategory", "Other");
  274. WritePropertyValueAlias(writer, unicodeDataEntries.LineBreakClasses, "LineBreakClass", "Unknown");
  275. WritePropertyValueAlias(writer, biDiDataEntries.PairedBracketTypes, "BidiPairedBracketType", "None");
  276. WritePropertyValueAlias(writer, biDiDataEntries.BiDiClasses, "BidiClass", "LeftToRight");
  277. writer.WriteLine(" }");
  278. writer.WriteLine("}");
  279. }
  280. }
  281. public static void ParseDataEntries(string property, List<DataEntry> entries)
  282. {
  283. using (var client = new HttpClient())
  284. {
  285. var url = Path.Combine(UnicodeDataGenerator.Ucd, "PropertyValueAliases.txt");
  286. using (var result = client.GetAsync(url).GetAwaiter().GetResult())
  287. {
  288. if (!result.IsSuccessStatusCode)
  289. {
  290. return;
  291. }
  292. using (var stream = result.Content.ReadAsStreamAsync().GetAwaiter().GetResult())
  293. using (var reader = new StreamReader(stream))
  294. {
  295. while (!reader.EndOfStream)
  296. {
  297. var line = reader.ReadLine();
  298. if (string.IsNullOrEmpty(line))
  299. {
  300. continue;
  301. }
  302. if (line != property)
  303. {
  304. continue;
  305. }
  306. reader.ReadLine();
  307. break;
  308. }
  309. while (!reader.EndOfStream)
  310. {
  311. var line = reader.ReadLine();
  312. if (string.IsNullOrEmpty(line) || line.StartsWith("#"))
  313. {
  314. break;
  315. }
  316. var elements = line.Split(';');
  317. var tag = elements[1].Trim();
  318. elements = elements[2].Split('#');
  319. var name = elements[0].Trim();
  320. if (entries.Any(x => x.Name == name))
  321. {
  322. continue;
  323. }
  324. var comment = string.Empty;
  325. if (elements.Length > 1)
  326. {
  327. comment = elements[1];
  328. }
  329. var entry = new DataEntry(name, tag, comment);
  330. entries.Add(entry);
  331. }
  332. }
  333. }
  334. }
  335. }
  336. private static void WritePropertyValueAliasGetTag(TextWriter writer, IEnumerable<DataEntry> entries,
  337. string typeName, string defaultValue)
  338. {
  339. writer.WriteLine(
  340. $" private static readonly Dictionary<{typeName}, string> s_{typeName.ToLower()}ToTag = ");
  341. writer.WriteLine($" new Dictionary<{typeName}, string>{{");
  342. foreach (var entry in entries)
  343. {
  344. writer.WriteLine($" {{ {typeName}.{entry.Name.Replace("_", "")}, \"{entry.Tag}\"}},");
  345. }
  346. writer.WriteLine(" };");
  347. writer.WriteLine();
  348. writer.WriteLine($" public static string GetTag({typeName} {typeName.ToLower()})");
  349. writer.WriteLine(" {");
  350. writer.WriteLine($" if (!s_{typeName.ToLower()}ToTag.TryGetValue({typeName.ToLower()}, out var value))");
  351. writer.WriteLine(" {");
  352. writer.WriteLine($" return \"{defaultValue}\";");
  353. writer.WriteLine(" }");
  354. writer.WriteLine($" return value;");
  355. writer.WriteLine(" }");
  356. writer.WriteLine();
  357. }
  358. private static void WritePropertyValueAlias(TextWriter writer, IEnumerable<DataEntry> entries, string typeName,
  359. string defaultValue)
  360. {
  361. writer.WriteLine($" private static readonly Dictionary<string, {typeName}> s_tagTo{typeName} = ");
  362. writer.WriteLine($" new Dictionary<string,{typeName}>{{");
  363. foreach (var entry in entries)
  364. {
  365. writer.WriteLine($" {{ \"{entry.Tag}\", {typeName}.{entry.Name.Replace("_", "")}}},");
  366. }
  367. writer.WriteLine(" };");
  368. writer.WriteLine();
  369. writer.WriteLine($" public static {typeName} Get{typeName}(string tag)");
  370. writer.WriteLine(" {");
  371. writer.WriteLine($" if (!s_tagTo{typeName}.TryGetValue(tag, out var value))");
  372. writer.WriteLine(" {");
  373. writer.WriteLine($" return {typeName}.{defaultValue};");
  374. writer.WriteLine(" }");
  375. writer.WriteLine($" return value;");
  376. writer.WriteLine(" }");
  377. writer.WriteLine();
  378. }
  379. }
  380. public readonly struct DataEntry
  381. {
  382. public DataEntry(string name, string tag, string comment)
  383. {
  384. Name = name;
  385. Tag = tag;
  386. Comment = comment;
  387. }
  388. public string Name { get; }
  389. public string Tag { get; }
  390. public string Comment { get; }
  391. }
  392. }