UnicodeEnumsGenerator.cs 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474
  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> CreateGeneralCategoryEnum()
  37. {
  38. var entries = new List<DataEntry> { new DataEntry("Other", "C", " Cc | Cf | Cn | Co | Cs") };
  39. ParseDataEntries("# General_Category (gc)", entries);
  40. using (var stream = File.Create("Generated\\GeneralCategory.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 GeneralCategory");
  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> CreateGraphemeBreakTypeEnum()
  58. {
  59. var entries = new List<DataEntry> { new DataEntry("Other", "XX", string.Empty) };
  60. ParseDataEntries("# Grapheme_Cluster_Break (GCB)", entries);
  61. using (var stream = File.Create("Generated\\GraphemeBreakClass.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 GraphemeBreakClass");
  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(" ExtendedPictographic");
  74. writer.WriteLine(" }");
  75. writer.WriteLine("}");
  76. }
  77. return entries;
  78. }
  79. private static List<string> GenerateBreakPairTable()
  80. {
  81. var rows = new List<string[]>();
  82. using (var stream =
  83. typeof(UnicodeEnumsGenerator).Assembly.GetManifestResourceStream(
  84. "Avalonia.Base.UnitTests.Media.TextFormatting.BreakPairTable.txt"))
  85. using (var reader = new StreamReader(stream))
  86. {
  87. while (!reader.EndOfStream)
  88. {
  89. var line = reader.ReadLine();
  90. var columns = line.Split('\t');
  91. rows.Add(columns);
  92. }
  93. }
  94. using (var stream = File.Create("Generated\\BreakPairTable.cs"))
  95. using (var writer = new StreamWriter(stream))
  96. {
  97. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  98. writer.WriteLine("{");
  99. writer.WriteLine(" internal static class BreakPairTable");
  100. writer.WriteLine(" {");
  101. writer.WriteLine(" private static readonly byte[][] s_breakPairTable = ");
  102. writer.WriteLine(" {");
  103. for (var i = 1; i < rows.Count; i++)
  104. {
  105. writer.Write(" new byte[] {");
  106. writer.Write($"{GetBreakPairType(rows[i][1])}");
  107. for (var index = 2; index < rows[i].Length; index++)
  108. {
  109. var column = rows[i][index];
  110. writer.Write($",{GetBreakPairType(column)}");
  111. }
  112. writer.Write("},");
  113. writer.Write(Environment.NewLine);
  114. }
  115. writer.WriteLine(" };");
  116. writer.WriteLine();
  117. writer.WriteLine(" public static PairBreakType Map(LineBreakClass first, LineBreakClass second)");
  118. writer.WriteLine(" {");
  119. writer.WriteLine(" return (PairBreakType)s_breakPairTable[(int)first][(int)second];");
  120. writer.WriteLine(" }");
  121. writer.WriteLine(" }");
  122. writer.WriteLine();
  123. writer.WriteLine(" internal enum PairBreakType : byte");
  124. writer.WriteLine(" {");
  125. writer.WriteLine(" DI = 0, // Direct break opportunity");
  126. writer.WriteLine(" IN = 1, // Indirect break opportunity");
  127. writer.WriteLine(" CI = 2, // Indirect break opportunity for combining marks");
  128. writer.WriteLine(" CP = 3, // Prohibited break for combining marks");
  129. writer.WriteLine(" PR = 4 // Prohibited break");
  130. writer.WriteLine(" }");
  131. writer.WriteLine("}");
  132. }
  133. return rows[0].Where(x => !string.IsNullOrEmpty(x)).ToList();
  134. }
  135. public const byte DI_BRK = 0; // Direct break opportunity
  136. public const byte IN_BRK = 1; // Indirect break opportunity
  137. public const byte CI_BRK = 2; // Indirect break opportunity for combining marks
  138. public const byte CP_BRK = 3; // Prohibited break for combining marks
  139. public const byte PR_BRK = 4; // Prohibited break
  140. private static byte GetBreakPairType(string type)
  141. {
  142. switch (type)
  143. {
  144. case "_":
  145. return DI_BRK;
  146. case "%":
  147. return IN_BRK;
  148. case "#":
  149. return CI_BRK;
  150. case "@":
  151. return CP_BRK;
  152. case "^":
  153. return PR_BRK;
  154. default:
  155. return byte.MaxValue;
  156. }
  157. }
  158. public static List<DataEntry> CreateLineBreakClassEnum()
  159. {
  160. var usedLineBreakClasses = GenerateBreakPairTable();
  161. var entries = new List<DataEntry> { new DataEntry("Unknown", "XX", string.Empty) };
  162. ParseDataEntries("# Line_Break (lb)", entries);
  163. var orderedLineBreakEntries = new Dictionary<string, DataEntry>();
  164. foreach (var tag in usedLineBreakClasses)
  165. {
  166. var entry = entries.Single(x => x.Tag == tag);
  167. orderedLineBreakEntries.Add(tag, entry);
  168. }
  169. foreach (var entry in entries)
  170. {
  171. if (orderedLineBreakEntries.ContainsKey(entry.Tag))
  172. {
  173. continue;
  174. }
  175. orderedLineBreakEntries.Add(entry.Tag, entry);
  176. }
  177. using (var stream = File.Create("Generated\\LineBreakClass.cs"))
  178. using (var writer = new StreamWriter(stream))
  179. {
  180. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  181. writer.WriteLine("{");
  182. writer.WriteLine(" public enum LineBreakClass");
  183. writer.WriteLine(" {");
  184. foreach (var entry in orderedLineBreakEntries.Values)
  185. {
  186. writer.WriteLine(" " + entry.Name.Replace("_", "") + ", //" + entry.Tag +
  187. (string.IsNullOrEmpty(entry.Comment) ? string.Empty : "#" + entry.Comment));
  188. }
  189. writer.WriteLine(" }");
  190. writer.WriteLine("}");
  191. }
  192. return orderedLineBreakEntries.Values.ToList();
  193. }
  194. public static List<DataEntry> CreateBidiClassEnum()
  195. {
  196. var entries = new List<DataEntry> { new DataEntry("Left_To_Right", "L", string.Empty) };
  197. ParseDataEntries("# Bidi_Class (bc)", entries);
  198. using (var stream = File.Create("Generated\\BidiClass.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 BidiClass");
  204. writer.WriteLine(" {");
  205. foreach (var entry in entries)
  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 entries;
  214. }
  215. public static List<DataEntry> CreateBiDiPairedBracketTypeEnum()
  216. {
  217. var entries = new List<DataEntry> { new DataEntry("None", "n", string.Empty) };
  218. ParseDataEntries("# Bidi_Paired_Bracket_Type (bpt)", entries);
  219. using (var stream = File.Create("Generated\\BidiPairedBracketType.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 BidiPairedBracketType");
  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 void CreatePropertyValueAliasHelper(UnicodeDataEntries unicodeDataEntries,
  237. BiDiDataEntries biDiDataEntries)
  238. {
  239. using (var stream = File.Create("Generated\\PropertyValueAliasHelper.cs"))
  240. using (var writer = new StreamWriter(stream))
  241. {
  242. writer.WriteLine("using System.Collections.Generic;");
  243. writer.WriteLine();
  244. writer.WriteLine("namespace Avalonia.Media.TextFormatting.Unicode");
  245. writer.WriteLine("{");
  246. writer.WriteLine(" internal static class PropertyValueAliasHelper");
  247. writer.WriteLine(" {");
  248. WritePropertyValueAliasGetTag(writer, unicodeDataEntries.Scripts, "Script", "Zzzz");
  249. WritePropertyValueAlias(writer, unicodeDataEntries.Scripts, "Script", "Unknown");
  250. WritePropertyValueAlias(writer, unicodeDataEntries.GeneralCategories, "GeneralCategory", "Other");
  251. WritePropertyValueAlias(writer, unicodeDataEntries.LineBreakClasses, "LineBreakClass", "Unknown");
  252. WritePropertyValueAlias(writer, biDiDataEntries.PairedBracketTypes, "BiDiPairedBracketType", "None");
  253. WritePropertyValueAlias(writer, biDiDataEntries.BiDiClasses, "BiDiClass", "LeftToRight");
  254. writer.WriteLine(" }");
  255. writer.WriteLine("}");
  256. }
  257. }
  258. public static void ParseDataEntries(string property, List<DataEntry> entries)
  259. {
  260. using (var client = new HttpClient())
  261. {
  262. var url = Path.Combine(UnicodeDataGenerator.Ucd, "PropertyValueAliases.txt");
  263. using (var result = client.GetAsync(url).GetAwaiter().GetResult())
  264. {
  265. if (!result.IsSuccessStatusCode)
  266. {
  267. return;
  268. }
  269. using (var stream = result.Content.ReadAsStreamAsync().GetAwaiter().GetResult())
  270. using (var reader = new StreamReader(stream))
  271. {
  272. while (!reader.EndOfStream)
  273. {
  274. var line = reader.ReadLine();
  275. if (string.IsNullOrEmpty(line))
  276. {
  277. continue;
  278. }
  279. if (line != property)
  280. {
  281. continue;
  282. }
  283. reader.ReadLine();
  284. break;
  285. }
  286. while (!reader.EndOfStream)
  287. {
  288. var line = reader.ReadLine();
  289. if (string.IsNullOrEmpty(line) || line.StartsWith("#"))
  290. {
  291. break;
  292. }
  293. var elements = line.Split(';');
  294. var tag = elements[1].Trim();
  295. elements = elements[2].Split('#');
  296. var name = elements[0].Trim();
  297. if (entries.Any(x => x.Name == name))
  298. {
  299. continue;
  300. }
  301. var comment = string.Empty;
  302. if (elements.Length > 1)
  303. {
  304. comment = elements[1];
  305. }
  306. var entry = new DataEntry(name, tag, comment);
  307. entries.Add(entry);
  308. }
  309. }
  310. }
  311. }
  312. }
  313. private static void WritePropertyValueAliasGetTag(TextWriter writer, IEnumerable<DataEntry> entries,
  314. string typeName, string defaultValue)
  315. {
  316. writer.WriteLine(
  317. $" private static readonly Dictionary<{typeName}, string> s_{typeName.ToLower()}ToTag = ");
  318. writer.WriteLine($" new Dictionary<{typeName}, string>{{");
  319. foreach (var entry in entries)
  320. {
  321. writer.WriteLine($" {{ {typeName}.{entry.Name.Replace("_", "")}, \"{entry.Tag}\"}},");
  322. }
  323. writer.WriteLine(" };");
  324. writer.WriteLine();
  325. writer.WriteLine($" public static string GetTag({typeName} {typeName.ToLower()})");
  326. writer.WriteLine(" {");
  327. writer.WriteLine($" if(!s_{typeName.ToLower()}ToTag.ContainsKey({typeName.ToLower()}))");
  328. writer.WriteLine(" {");
  329. writer.WriteLine($" return \"{defaultValue}\";");
  330. writer.WriteLine(" }");
  331. writer.WriteLine($" return s_{typeName.ToLower()}ToTag[{typeName.ToLower()}];");
  332. writer.WriteLine(" }");
  333. writer.WriteLine();
  334. }
  335. private static void WritePropertyValueAlias(TextWriter writer, IEnumerable<DataEntry> entries, string typeName,
  336. string defaultValue)
  337. {
  338. writer.WriteLine($" private static readonly Dictionary<string, {typeName}> s_tagTo{typeName} = ");
  339. writer.WriteLine($" new Dictionary<string,{typeName}>{{");
  340. foreach (var entry in entries)
  341. {
  342. writer.WriteLine($" {{ \"{entry.Tag}\", {typeName}.{entry.Name.Replace("_", "")}}},");
  343. }
  344. writer.WriteLine(" };");
  345. writer.WriteLine();
  346. writer.WriteLine($" public static {typeName} Get{typeName}(string tag)");
  347. writer.WriteLine(" {");
  348. writer.WriteLine($" if(!s_tagTo{typeName}.ContainsKey(tag))");
  349. writer.WriteLine(" {");
  350. writer.WriteLine($" return {typeName}.{defaultValue};");
  351. writer.WriteLine(" }");
  352. writer.WriteLine($" return s_tagTo{typeName}[tag];");
  353. writer.WriteLine(" }");
  354. writer.WriteLine();
  355. }
  356. }
  357. public readonly struct DataEntry
  358. {
  359. public DataEntry(string name, string tag, string comment)
  360. {
  361. Name = name;
  362. Tag = tag;
  363. Comment = comment;
  364. }
  365. public string Name { get; }
  366. public string Tag { get; }
  367. public string Comment { get; }
  368. }
  369. }