SsaParser.cs 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472
  1. using System;
  2. using System.Collections.Generic;
  3. using System.IO;
  4. using System.Text;
  5. using System.Threading;
  6. using MediaBrowser.Model.MediaInfo;
  7. namespace MediaBrowser.MediaEncoding.Subtitles
  8. {
  9. /// <summary>
  10. /// Credit to https://github.com/SubtitleEdit/subtitleedit/blob/a299dc4407a31796364cc6ad83f0d3786194ba22/src/Logic/SubtitleFormats/SubStationAlpha.cs
  11. /// </summary>
  12. public class SsaParser : ISubtitleParser
  13. {
  14. public SubtitleTrackInfo Parse(Stream stream, CancellationToken cancellationToken)
  15. {
  16. var trackInfo = new SubtitleTrackInfo();
  17. var trackEvents = new List<SubtitleTrackEvent>();
  18. using (var reader = new StreamReader(stream))
  19. {
  20. bool eventsStarted = false;
  21. string[] format = "Marked, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text".Split(',');
  22. int indexLayer = 0;
  23. int indexStart = 1;
  24. int indexEnd = 2;
  25. int indexStyle = 3;
  26. int indexName = 4;
  27. int indexEffect = 8;
  28. int indexText = 9;
  29. int lineNumber = 0;
  30. var header = new StringBuilder();
  31. string line;
  32. while ((line = reader.ReadLine()) != null)
  33. {
  34. cancellationToken.ThrowIfCancellationRequested();
  35. lineNumber++;
  36. if (!eventsStarted)
  37. {
  38. header.AppendLine(line);
  39. }
  40. if (line.Trim().ToLowerInvariant() == "[events]")
  41. {
  42. eventsStarted = true;
  43. }
  44. else if (!string.IsNullOrEmpty(line) && line.Trim().StartsWith(";"))
  45. {
  46. // skip comment lines
  47. }
  48. else if (eventsStarted && line.Trim().Length > 0)
  49. {
  50. string s = line.Trim().ToLowerInvariant();
  51. if (s.StartsWith("format:"))
  52. {
  53. if (line.Length > 10)
  54. {
  55. format = line.ToLowerInvariant().Substring(8).Split(',');
  56. for (int i = 0; i < format.Length; i++)
  57. {
  58. if (format[i].Trim().ToLowerInvariant() == "layer")
  59. {
  60. indexLayer = i;
  61. }
  62. else if (format[i].Trim().ToLowerInvariant() == "start")
  63. {
  64. indexStart = i;
  65. }
  66. else if (format[i].Trim().ToLowerInvariant() == "end")
  67. {
  68. indexEnd = i;
  69. }
  70. else if (format[i].Trim().ToLowerInvariant() == "text")
  71. {
  72. indexText = i;
  73. }
  74. else if (format[i].Trim().ToLowerInvariant() == "effect")
  75. {
  76. indexEffect = i;
  77. }
  78. else if (format[i].Trim().ToLowerInvariant() == "style")
  79. {
  80. indexStyle = i;
  81. }
  82. }
  83. }
  84. }
  85. else if (!string.IsNullOrEmpty(s))
  86. {
  87. string text = string.Empty;
  88. string start = string.Empty;
  89. string end = string.Empty;
  90. string style = string.Empty;
  91. string layer = string.Empty;
  92. string effect = string.Empty;
  93. string name = string.Empty;
  94. string[] splittedLine;
  95. if (s.StartsWith("dialogue:"))
  96. {
  97. splittedLine = line.Substring(10).Split(',');
  98. }
  99. else
  100. {
  101. splittedLine = line.Split(',');
  102. }
  103. for (int i = 0; i < splittedLine.Length; i++)
  104. {
  105. if (i == indexStart)
  106. {
  107. start = splittedLine[i].Trim();
  108. }
  109. else if (i == indexEnd)
  110. {
  111. end = splittedLine[i].Trim();
  112. }
  113. else if (i == indexLayer)
  114. {
  115. layer = splittedLine[i];
  116. }
  117. else if (i == indexEffect)
  118. {
  119. effect = splittedLine[i];
  120. }
  121. else if (i == indexText)
  122. {
  123. text = splittedLine[i];
  124. }
  125. else if (i == indexStyle)
  126. {
  127. style = splittedLine[i];
  128. }
  129. else if (i == indexName)
  130. {
  131. name = splittedLine[i];
  132. }
  133. else if (i > indexText)
  134. {
  135. text += "," + splittedLine[i];
  136. }
  137. }
  138. try
  139. {
  140. var p = new SubtitleTrackEvent();
  141. p.StartPositionTicks = GetTimeCodeFromString(start);
  142. p.EndPositionTicks = GetTimeCodeFromString(end);
  143. p.Text = GetFormattedText(text);
  144. trackEvents.Add(p);
  145. }
  146. catch
  147. {
  148. }
  149. }
  150. }
  151. }
  152. // if (header.Length > 0)
  153. // subtitle.Header = header.ToString();
  154. // subtitle.Renumber(1);
  155. }
  156. trackInfo.TrackEvents = trackEvents.ToArray();
  157. return trackInfo;
  158. }
  159. private static long GetTimeCodeFromString(string time)
  160. {
  161. // h:mm:ss.cc
  162. string[] timeCode = time.Split(':', '.');
  163. return new TimeSpan(0, int.Parse(timeCode[0]),
  164. int.Parse(timeCode[1]),
  165. int.Parse(timeCode[2]),
  166. int.Parse(timeCode[3]) * 10).Ticks;
  167. }
  168. public static string GetFormattedText(string text)
  169. {
  170. text = text.Replace("\\n", ParserValues.NewLine, StringComparison.OrdinalIgnoreCase);
  171. bool italic = false;
  172. for (int i = 0; i < 10; i++) // just look ten times...
  173. {
  174. if (text.Contains(@"{\fn"))
  175. {
  176. int start = text.IndexOf(@"{\fn");
  177. int end = text.IndexOf('}', start);
  178. if (end > 0 && !text.Substring(start).StartsWith("{\\fn}"))
  179. {
  180. string fontName = text.Substring(start + 4, end - (start + 4));
  181. string extraTags = string.Empty;
  182. CheckAndAddSubTags(ref fontName, ref extraTags, out italic);
  183. text = text.Remove(start, end - start + 1);
  184. if (italic)
  185. {
  186. text = text.Insert(start, "<font face=\"" + fontName + "\"" + extraTags + "><i>");
  187. }
  188. else
  189. {
  190. text = text.Insert(start, "<font face=\"" + fontName + "\"" + extraTags + ">");
  191. }
  192. int indexOfEndTag = text.IndexOf("{\\fn}", start);
  193. if (indexOfEndTag > 0)
  194. {
  195. text = text.Remove(indexOfEndTag, "{\\fn}".Length).Insert(indexOfEndTag, "</font>");
  196. }
  197. else
  198. {
  199. text += "</font>";
  200. }
  201. }
  202. }
  203. if (text.Contains(@"{\fs"))
  204. {
  205. int start = text.IndexOf(@"{\fs");
  206. int end = text.IndexOf('}', start);
  207. if (end > 0 && !text.Substring(start).StartsWith("{\\fs}"))
  208. {
  209. string fontSize = text.Substring(start + 4, end - (start + 4));
  210. string extraTags = string.Empty;
  211. CheckAndAddSubTags(ref fontSize, ref extraTags, out italic);
  212. if (IsInteger(fontSize))
  213. {
  214. text = text.Remove(start, end - start + 1);
  215. if (italic)
  216. {
  217. text = text.Insert(start, "<font size=\"" + fontSize + "\"" + extraTags + "><i>");
  218. }
  219. else
  220. {
  221. text = text.Insert(start, "<font size=\"" + fontSize + "\"" + extraTags + ">");
  222. }
  223. int indexOfEndTag = text.IndexOf("{\\fs}", start);
  224. if (indexOfEndTag > 0)
  225. {
  226. text = text.Remove(indexOfEndTag, "{\\fs}".Length).Insert(indexOfEndTag, "</font>");
  227. }
  228. else
  229. {
  230. text += "</font>";
  231. }
  232. }
  233. }
  234. }
  235. if (text.Contains(@"{\c"))
  236. {
  237. int start = text.IndexOf(@"{\c");
  238. int end = text.IndexOf('}', start);
  239. if (end > 0 && !text.Substring(start).StartsWith("{\\c}"))
  240. {
  241. string color = text.Substring(start + 4, end - (start + 4));
  242. string extraTags = string.Empty;
  243. CheckAndAddSubTags(ref color, ref extraTags, out italic);
  244. color = color.Replace("&", string.Empty).TrimStart('H');
  245. color = color.PadLeft(6, '0');
  246. // switch to rrggbb from bbggrr
  247. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  248. color = color.ToLowerInvariant();
  249. text = text.Remove(start, end - start + 1);
  250. if (italic)
  251. {
  252. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + "><i>");
  253. }
  254. else
  255. {
  256. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + ">");
  257. }
  258. int indexOfEndTag = text.IndexOf("{\\c}", start);
  259. if (indexOfEndTag > 0)
  260. {
  261. text = text.Remove(indexOfEndTag, "{\\c}".Length).Insert(indexOfEndTag, "</font>");
  262. }
  263. else
  264. {
  265. text += "</font>";
  266. }
  267. }
  268. }
  269. if (text.Contains(@"{\1c")) // "1" specifices primary color
  270. {
  271. int start = text.IndexOf(@"{\1c");
  272. int end = text.IndexOf('}', start);
  273. if (end > 0 && !text.Substring(start).StartsWith("{\\1c}"))
  274. {
  275. string color = text.Substring(start + 5, end - (start + 5));
  276. string extraTags = string.Empty;
  277. CheckAndAddSubTags(ref color, ref extraTags, out italic);
  278. color = color.Replace("&", string.Empty).TrimStart('H');
  279. color = color.PadLeft(6, '0');
  280. // switch to rrggbb from bbggrr
  281. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  282. color = color.ToLowerInvariant();
  283. text = text.Remove(start, end - start + 1);
  284. if (italic)
  285. {
  286. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + "><i>");
  287. }
  288. else
  289. {
  290. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + ">");
  291. }
  292. text += "</font>";
  293. }
  294. }
  295. }
  296. text = text.Replace(@"{\i1}", "<i>");
  297. text = text.Replace(@"{\i0}", "</i>");
  298. text = text.Replace(@"{\i}", "</i>");
  299. if (CountTagInText(text, "<i>") > CountTagInText(text, "</i>"))
  300. {
  301. text += "</i>";
  302. }
  303. text = text.Replace(@"{\u1}", "<u>");
  304. text = text.Replace(@"{\u0}", "</u>");
  305. text = text.Replace(@"{\u}", "</u>");
  306. if (CountTagInText(text, "<u>") > CountTagInText(text, "</u>"))
  307. {
  308. text += "</u>";
  309. }
  310. text = text.Replace(@"{\b1}", "<b>");
  311. text = text.Replace(@"{\b0}", "</b>");
  312. text = text.Replace(@"{\b}", "</b>");
  313. if (CountTagInText(text, "<b>") > CountTagInText(text, "</b>"))
  314. {
  315. text += "</b>";
  316. }
  317. return text;
  318. }
  319. private static bool IsInteger(string s)
  320. {
  321. if (int.TryParse(s, out var i))
  322. {
  323. return true;
  324. }
  325. return false;
  326. }
  327. private static int CountTagInText(string text, string tag)
  328. {
  329. int count = 0;
  330. int index = text.IndexOf(tag);
  331. while (index >= 0)
  332. {
  333. count++;
  334. if (index == text.Length)
  335. {
  336. return count;
  337. }
  338. index = text.IndexOf(tag, index + 1);
  339. }
  340. return count;
  341. }
  342. private static void CheckAndAddSubTags(ref string tagName, ref string extraTags, out bool italic)
  343. {
  344. italic = false;
  345. int indexOfSPlit = tagName.IndexOf(@"\");
  346. if (indexOfSPlit > 0)
  347. {
  348. string rest = tagName.Substring(indexOfSPlit).TrimStart('\\');
  349. tagName = tagName.Remove(indexOfSPlit);
  350. for (int i = 0; i < 10; i++)
  351. {
  352. if (rest.StartsWith("fs") && rest.Length > 2)
  353. {
  354. indexOfSPlit = rest.IndexOf(@"\");
  355. string fontSize = rest;
  356. if (indexOfSPlit > 0)
  357. {
  358. fontSize = rest.Substring(0, indexOfSPlit);
  359. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  360. }
  361. else
  362. {
  363. rest = string.Empty;
  364. }
  365. extraTags += " size=\"" + fontSize.Substring(2) + "\"";
  366. }
  367. else if (rest.StartsWith("fn") && rest.Length > 2)
  368. {
  369. indexOfSPlit = rest.IndexOf(@"\");
  370. string fontName = rest;
  371. if (indexOfSPlit > 0)
  372. {
  373. fontName = rest.Substring(0, indexOfSPlit);
  374. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  375. }
  376. else
  377. {
  378. rest = string.Empty;
  379. }
  380. extraTags += " face=\"" + fontName.Substring(2) + "\"";
  381. }
  382. else if (rest.StartsWith("c") && rest.Length > 2)
  383. {
  384. indexOfSPlit = rest.IndexOf(@"\");
  385. string fontColor = rest;
  386. if (indexOfSPlit > 0)
  387. {
  388. fontColor = rest.Substring(0, indexOfSPlit);
  389. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  390. }
  391. else
  392. {
  393. rest = string.Empty;
  394. }
  395. string color = fontColor.Substring(2);
  396. color = color.Replace("&", string.Empty).TrimStart('H');
  397. color = color.PadLeft(6, '0');
  398. // switch to rrggbb from bbggrr
  399. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  400. color = color.ToLowerInvariant();
  401. extraTags += " color=\"" + color + "\"";
  402. }
  403. else if (rest.StartsWith("i1") && rest.Length > 1)
  404. {
  405. indexOfSPlit = rest.IndexOf(@"\");
  406. italic = true;
  407. if (indexOfSPlit > 0)
  408. {
  409. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  410. }
  411. else
  412. {
  413. rest = string.Empty;
  414. }
  415. }
  416. else if (rest.Length > 0 && rest.Contains("\\"))
  417. {
  418. indexOfSPlit = rest.IndexOf(@"\");
  419. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  420. }
  421. }
  422. }
  423. }
  424. }
  425. }