UnicodeEnumsGenerator.cs 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501
  1. using System;
  2. using System.Collections.Generic;
  3. using System.IO;
  4. using System.Linq;
  5. using System.Net.Http;
  6. namespace Avalonia.Base.UnitTests.Media.TextFormatting
  7. {
  8. internal static class UnicodeEnumsGenerator
  9. {
  10. public static List<DataEntry> CreateScriptEnum()
  11. {
  12. var entries = new List<DataEntry>
  13. {
  14. new DataEntry("Unknown", "Zzzz", string.Empty),
  15. new DataEntry("Common", "Zyyy", string.Empty),
  16. new DataEntry("Inherited", "Zinh", string.Empty)
  17. };
  18. ParseDataEntries("# Script (sc)", entries);
  19. using (var stream = File.Create("Generated\\Script.cs"))
  20. using (var writer = new StreamWriter(stream))
  21. {
  22. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  23. writer.WriteLine("{");
  24. writer.WriteLine(" public enum Script");
  25. writer.WriteLine(" {");
  26. foreach (var entry in entries)
  27. {
  28. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  29. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  30. }
  31. writer.WriteLine(" }");
  32. writer.WriteLine("}");
  33. }
  34. return entries;
  35. }
  36. public static List<DataEntry> CreateEastAsianWidthClassEnum()
  37. {
  38. var entries = new List<DataEntry>();
  39. ParseDataEntries("# East_Asian_Width (ea)", entries);
  40. using (var stream = File.Create("Generated\\EastAsianWidthClass.cs"))
  41. using (var writer = new StreamWriter(stream))
  42. {
  43. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  44. writer.WriteLine("{");
  45. writer.WriteLine(" public enum EastAsianWidthClass");
  46. writer.WriteLine(" {");
  47. foreach (var entry in entries)
  48. {
  49. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  50. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  51. }
  52. writer.WriteLine(" }");
  53. writer.WriteLine("}");
  54. }
  55. return entries;
  56. }
  57. public static List<DataEntry> CreateGeneralCategoryEnum()
  58. {
  59. var entries = new List<DataEntry> { new DataEntry("Other", "C", " Cc | Cf | Cn | Co | Cs") };
  60. ParseDataEntries("# General_Category (gc)", entries);
  61. using (var stream = File.Create("Generated\\GeneralCategory.cs"))
  62. using (var writer = new StreamWriter(stream))
  63. {
  64. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  65. writer.WriteLine("{");
  66. writer.WriteLine(" public enum GeneralCategory");
  67. writer.WriteLine(" {");
  68. foreach (var entry in entries)
  69. {
  70. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  71. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  72. }
  73. writer.WriteLine(" }");
  74. writer.WriteLine("}");
  75. }
  76. return entries;
  77. }
  78. public static List<DataEntry> CreateGraphemeBreakTypeEnum()
  79. {
  80. var entries = new List<DataEntry> { new DataEntry("Other", "XX", string.Empty) };
  81. ParseDataEntries("# Grapheme_Cluster_Break (GCB)", entries);
  82. using (var stream = File.Create("Generated\\GraphemeBreakClass.cs"))
  83. using (var writer = new StreamWriter(stream))
  84. {
  85. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  86. writer.WriteLine("{");
  87. writer.WriteLine(" public enum GraphemeBreakClass");
  88. writer.WriteLine(" {");
  89. foreach (var entry in entries)
  90. {
  91. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  92. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  93. }
  94. writer.WriteLine(" ExtendedPictographic");
  95. writer.WriteLine(" }");
  96. writer.WriteLine("}");
  97. }
  98. return entries;
  99. }
  100. private static List<string> GenerateBreakPairTable()
  101. {
  102. var rows = new List<string[]>();
  103. using (var stream =
  104. typeof(UnicodeEnumsGenerator).Assembly.GetManifestResourceStream(
  105. "Avalonia.Base.UnitTests.Media.TextFormatting.BreakPairTable.txt"))
  106. using (var reader = new StreamReader(stream))
  107. {
  108. while (!reader.EndOfStream)
  109. {
  110. var line = reader.ReadLine();
  111. var columns = line.Split('\t');
  112. rows.Add(columns);
  113. }
  114. }
  115. using (var stream = File.Create("Generated\\BreakPairTable.cs"))
  116. using (var writer = new StreamWriter(stream))
  117. {
  118. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  119. writer.WriteLine("{");
  120. writer.WriteLine(" internal static class BreakPairTable");
  121. writer.WriteLine(" {");
  122. writer.WriteLine(" private static readonly byte[][] s_breakPairTable = ");
  123. writer.WriteLine(" {");
  124. for (var i = 1; i < rows.Count; i++)
  125. {
  126. writer.Write(" new byte[] {");
  127. writer.Write($"{GetBreakPairType(rows[i][1])}");
  128. for (var index = 2; index < rows[i].Length; index++)
  129. {
  130. var column = rows[i][index];
  131. writer.Write($",{GetBreakPairType(column)}");
  132. }
  133. writer.Write("},");
  134. writer.Write(Environment.NewLine);
  135. }
  136. writer.WriteLine(" };");
  137. writer.WriteLine();
  138. writer.WriteLine(" public static PairBreakType Map(LineBreakClass first, LineBreakClass second)");
  139. writer.WriteLine(" {");
  140. writer.WriteLine(" return (PairBreakType)s_breakPairTable[(int)first][(int)second];");
  141. writer.WriteLine(" }");
  142. writer.WriteLine(" }");
  143. writer.WriteLine();
  144. writer.WriteLine(" internal enum PairBreakType : byte");
  145. writer.WriteLine(" {");
  146. writer.WriteLine(" DI = 0, // Direct break opportunity");
  147. writer.WriteLine(" IN = 1, // Indirect break opportunity");
  148. writer.WriteLine(" CI = 2, // Indirect break opportunity for combining marks");
  149. writer.WriteLine(" CP = 3, // Prohibited break for combining marks");
  150. writer.WriteLine(" PR = 4 // Prohibited break");
  151. writer.WriteLine(" }");
  152. writer.WriteLine("}");
  153. }
  154. return rows[0].Where(x => !string.IsNullOrEmpty(x)).ToList();
  155. }
  156. public const byte DI_BRK = 0; // Direct break opportunity
  157. public const byte IN_BRK = 1; // Indirect break opportunity
  158. public const byte CI_BRK = 2; // Indirect break opportunity for combining marks
  159. public const byte CP_BRK = 3; // Prohibited break for combining marks
  160. public const byte PR_BRK = 4; // Prohibited break
  161. private static byte GetBreakPairType(string type)
  162. {
  163. switch (type)
  164. {
  165. case "_":
  166. return DI_BRK;
  167. case "%":
  168. return IN_BRK;
  169. case "#":
  170. return CI_BRK;
  171. case "@":
  172. return CP_BRK;
  173. case "^":
  174. return PR_BRK;
  175. default:
  176. return byte.MaxValue;
  177. }
  178. }
  179. public static List<DataEntry> CreateLineBreakClassEnum()
  180. {
  181. var usedLineBreakClasses = GenerateBreakPairTable();
  182. var entries = new List<DataEntry> { new DataEntry("Unknown", "XX", string.Empty) };
  183. ParseDataEntries("# Line_Break (lb)", entries);
  184. var orderedLineBreakEntries = new Dictionary<string, DataEntry>();
  185. foreach (var tag in usedLineBreakClasses)
  186. {
  187. var entry = entries.Single(x => x.Tag == tag);
  188. orderedLineBreakEntries.Add(tag, entry);
  189. }
  190. foreach (var entry in entries)
  191. {
  192. if (orderedLineBreakEntries.ContainsKey(entry.Tag))
  193. {
  194. continue;
  195. }
  196. orderedLineBreakEntries.Add(entry.Tag, entry);
  197. }
  198. using (var stream = File.Create("Generated\\LineBreakClass.cs"))
  199. using (var writer = new StreamWriter(stream))
  200. {
  201. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  202. writer.WriteLine("{");
  203. writer.WriteLine(" public enum LineBreakClass");
  204. writer.WriteLine(" {");
  205. foreach (var entry in orderedLineBreakEntries.Values)
  206. {
  207. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  208. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  209. }
  210. writer.WriteLine(" }");
  211. writer.WriteLine("}");
  212. }
  213. return orderedLineBreakEntries.Values.ToList();
  214. }
  215. public static List<DataEntry> CreateBidiClassEnum()
  216. {
  217. var entries = new List<DataEntry> { new DataEntry("Left_To_Right", "L", string.Empty) };
  218. ParseDataEntries("# Bidi_Class (bc)", entries);
  219. using (var stream = File.Create("Generated\\BidiClass.cs"))
  220. using (var writer = new StreamWriter(stream))
  221. {
  222. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  223. writer.WriteLine("{");
  224. writer.WriteLine(" public enum BidiClass");
  225. writer.WriteLine(" {");
  226. foreach (var entry in entries)
  227. {
  228. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  229. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  230. }
  231. writer.WriteLine(" }");
  232. writer.WriteLine("}");
  233. }
  234. return entries;
  235. }
  236. public static List<DataEntry> CreateBiDiPairedBracketTypeEnum()
  237. {
  238. var entries = new List<DataEntry> { new DataEntry("None", "n", string.Empty) };
  239. ParseDataEntries("# Bidi_Paired_Bracket_Type (bpt)", entries);
  240. using (var stream = File.Create("Generated\\BidiPairedBracketType.cs"))
  241. using (var writer = new StreamWriter(stream))
  242. {
  243. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  244. writer.WriteLine("{");
  245. writer.WriteLine(" public enum BidiPairedBracketType");
  246. writer.WriteLine(" {");
  247. foreach (var entry in entries)
  248. {
  249. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  250. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  251. }
  252. writer.WriteLine(" }");
  253. writer.WriteLine("}");
  254. }
  255. return entries;
  256. }
  257. public static void CreatePropertyValueAliasHelper(UnicodeDataEntries unicodeDataEntries,
  258. BiDiDataEntries biDiDataEntries)
  259. {
  260. using (var stream = File.Create("Generated\\PropertyValueAliasHelper.cs"))
  261. using (var writer = new StreamWriter(stream))
  262. {
  263. writer.WriteLine("using System.Collections.Generic;");
  264. writer.WriteLine();
  265. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  266. writer.WriteLine("{");
  267. writer.WriteLine(" internal static class PropertyValueAliasHelper");
  268. writer.WriteLine(" {");
  269. WritePropertyValueAliasGetTag(writer, unicodeDataEntries.Scripts, "Script", "Zzzz");
  270. WritePropertyValueAlias(writer, unicodeDataEntries.Scripts, "Script", "Unknown");
  271. WritePropertyValueAlias(writer, unicodeDataEntries.GeneralCategories, "GeneralCategory", "Other");
  272. WritePropertyValueAlias(writer, unicodeDataEntries.LineBreakClasses, "LineBreakClass", "Unknown");
  273. WritePropertyValueAlias(writer, biDiDataEntries.PairedBracketTypes, "BidiPairedBracketType", "None");
  274. WritePropertyValueAlias(writer, biDiDataEntries.BiDiClasses, "BidiClass", "LeftToRight");
  275. writer.WriteLine(" }");
  276. writer.WriteLine("}");
  277. }
  278. }
  279. public static void ParseDataEntries(string property, List<DataEntry> entries)
  280. {
  281. using (var client = new HttpClient())
  282. {
  283. var url = Path.Combine(UnicodeDataGenerator.Ucd, "PropertyValueAliases.txt");
  284. using (var result = client.GetAsync(url).GetAwaiter().GetResult())
  285. {
  286. if (!result.IsSuccessStatusCode)
  287. {
  288. return;
  289. }
  290. using (var stream = result.Content.ReadAsStreamAsync().GetAwaiter().GetResult())
  291. using (var reader = new StreamReader(stream))
  292. {
  293. while (!reader.EndOfStream)
  294. {
  295. var line = reader.ReadLine();
  296. if (string.IsNullOrEmpty(line))
  297. {
  298. continue;
  299. }
  300. if (line != property)
  301. {
  302. continue;
  303. }
  304. reader.ReadLine();
  305. break;
  306. }
  307. while (!reader.EndOfStream)
  308. {
  309. var line = reader.ReadLine();
  310. if (string.IsNullOrEmpty(line) || line.StartsWith("#"))
  311. {
  312. break;
  313. }
  314. var elements = line.Split(';');
  315. var tag = elements[1].Trim();
  316. elements = elements[2].Split('#');
  317. var name = elements[0].Trim();
  318. if (entries.Any(x => x.Name == name))
  319. {
  320. continue;
  321. }
  322. var comment = string.Empty;
  323. if (elements.Length > 1)
  324. {
  325. comment = elements[1];
  326. }
  327. var entry = new DataEntry(name, tag, comment);
  328. entries.Add(entry);
  329. }
  330. }
  331. }
  332. }
  333. }
  334. private static void WritePropertyValueAliasGetTag(TextWriter writer, IEnumerable<DataEntry> entries,
  335. string typeName, string defaultValue)
  336. {
  337. writer.WriteLine(
  338. $" private static readonly Dictionary<{typeName}, string> s_{typeName.ToLower()}ToTag = ");
  339. writer.WriteLine($" new Dictionary<{typeName}, string>{{");
  340. foreach (var entry in entries)
  341. {
  342. writer.WriteLine($" {{ {typeName}.{entry.Name.Replace("_", "")}, \"{entry.Tag}\"}},");
  343. }
  344. writer.WriteLine(" };");
  345. writer.WriteLine();
  346. writer.WriteLine($" public static string GetTag({typeName} {typeName.ToLower()})");
  347. writer.WriteLine(" {");
  348. writer.WriteLine($" if (!s_{typeName.ToLower()}ToTag.TryGetValue({typeName.ToLower()}, out var value))");
  349. writer.WriteLine(" {");
  350. writer.WriteLine($" return \"{defaultValue}\";");
  351. writer.WriteLine(" }");
  352. writer.WriteLine($" return value;");
  353. writer.WriteLine(" }");
  354. writer.WriteLine();
  355. }
  356. private static void WritePropertyValueAlias(TextWriter writer, IEnumerable<DataEntry> entries, string typeName,
  357. string defaultValue)
  358. {
  359. writer.WriteLine($" private static readonly Dictionary<string, {typeName}> s_tagTo{typeName} = ");
  360. writer.WriteLine($" new Dictionary<string,{typeName}>{{");
  361. foreach (var entry in entries)
  362. {
  363. writer.WriteLine($" {{ \"{entry.Tag}\", {typeName}.{entry.Name.Replace("_", "")}}},");
  364. }
  365. writer.WriteLine(" };");
  366. writer.WriteLine();
  367. writer.WriteLine($" public static {typeName} Get{typeName}(string tag)");
  368. writer.WriteLine(" {");
  369. writer.WriteLine($" if (!s_tagTo{typeName}.TryGetValue(tag, out var value))");
  370. writer.WriteLine(" {");
  371. writer.WriteLine($" return {typeName}.{defaultValue};");
  372. writer.WriteLine(" }");
  373. writer.WriteLine($" return value;");
  374. writer.WriteLine(" }");
  375. writer.WriteLine();
  376. }
  377. }
  378. public readonly struct DataEntry
  379. {
  380. public DataEntry(string name, string tag, string comment)
  381. {
  382. Name = name;
  383. Tag = tag;
  384. Comment = comment;
  385. }
  386. public string Name { get; }
  387. public string Tag { get; }
  388. public string Comment { get; }
  389. }
  390. }