SsaParser.cs 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469
  1. using System;
  2. using System.Collections.Generic;
  3. using System.Globalization;
  4. using System.IO;
  5. using System.Text;
  6. using System.Threading;
  7. using MediaBrowser.Model.MediaInfo;
  8. namespace MediaBrowser.MediaEncoding.Subtitles
  9. {
  10. /// <summary>
  11. /// <see href="https://github.com/SubtitleEdit/subtitleedit/blob/a299dc4407a31796364cc6ad83f0d3786194ba22/src/Logic/SubtitleFormats/SubStationAlpha.cs">Credit</see>.
  12. /// </summary>
  13. public class SsaParser : ISubtitleParser
  14. {
  15. /// <inheritdoc />
  16. public SubtitleTrackInfo Parse(Stream stream, CancellationToken cancellationToken)
  17. {
  18. var trackInfo = new SubtitleTrackInfo();
  19. var trackEvents = new List<SubtitleTrackEvent>();
  20. using (var reader = new StreamReader(stream))
  21. {
  22. bool eventsStarted = false;
  23. string[] format = "Marked, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text".Split(',');
  24. int indexLayer = 0;
  25. int indexStart = 1;
  26. int indexEnd = 2;
  27. int indexStyle = 3;
  28. int indexName = 4;
  29. int indexEffect = 8;
  30. int indexText = 9;
  31. int lineNumber = 0;
  32. var header = new StringBuilder();
  33. string line;
  34. while ((line = reader.ReadLine()) != null)
  35. {
  36. cancellationToken.ThrowIfCancellationRequested();
  37. lineNumber++;
  38. if (!eventsStarted)
  39. {
  40. header.AppendLine(line);
  41. }
  42. if (string.Equals(line.Trim(), "[events]", StringComparison.OrdinalIgnoreCase))
  43. {
  44. eventsStarted = true;
  45. }
  46. else if (!string.IsNullOrEmpty(line) && line.Trim().StartsWith(";", StringComparison.Ordinal))
  47. {
  48. // skip comment lines
  49. }
  50. else if (eventsStarted && line.Trim().Length > 0)
  51. {
  52. string s = line.Trim().ToLowerInvariant();
  53. if (s.StartsWith("format:", StringComparison.Ordinal))
  54. {
  55. if (line.Length > 10)
  56. {
  57. format = line.ToLowerInvariant().Substring(8).Split(',');
  58. for (int i = 0; i < format.Length; i++)
  59. {
  60. if (string.Equals(format[i].Trim(), "layer", StringComparison.OrdinalIgnoreCase))
  61. {
  62. indexLayer = i;
  63. }
  64. else if (string.Equals(format[i].Trim(), "start", StringComparison.OrdinalIgnoreCase))
  65. {
  66. indexStart = i;
  67. }
  68. else if (string.Equals(format[i].Trim(), "end", StringComparison.OrdinalIgnoreCase))
  69. {
  70. indexEnd = i;
  71. }
  72. else if (string.Equals(format[i].Trim(), "text", StringComparison.OrdinalIgnoreCase))
  73. {
  74. indexText = i;
  75. }
  76. else if (string.Equals(format[i].Trim(), "effect", StringComparison.OrdinalIgnoreCase))
  77. {
  78. indexEffect = i;
  79. }
  80. else if (string.Equals(format[i].Trim(), "style", StringComparison.OrdinalIgnoreCase))
  81. {
  82. indexStyle = i;
  83. }
  84. }
  85. }
  86. }
  87. else if (!string.IsNullOrEmpty(s))
  88. {
  89. string text = string.Empty;
  90. string start = string.Empty;
  91. string end = string.Empty;
  92. string style = string.Empty;
  93. string layer = string.Empty;
  94. string effect = string.Empty;
  95. string name = string.Empty;
  96. string[] splittedLine;
  97. if (s.StartsWith("dialogue:", StringComparison.Ordinal))
  98. {
  99. splittedLine = line.Substring(10).Split(',');
  100. }
  101. else
  102. {
  103. splittedLine = line.Split(',');
  104. }
  105. for (int i = 0; i < splittedLine.Length; i++)
  106. {
  107. if (i == indexStart)
  108. {
  109. start = splittedLine[i].Trim();
  110. }
  111. else if (i == indexEnd)
  112. {
  113. end = splittedLine[i].Trim();
  114. }
  115. else if (i == indexLayer)
  116. {
  117. layer = splittedLine[i];
  118. }
  119. else if (i == indexEffect)
  120. {
  121. effect = splittedLine[i];
  122. }
  123. else if (i == indexText)
  124. {
  125. text = splittedLine[i];
  126. }
  127. else if (i == indexStyle)
  128. {
  129. style = splittedLine[i];
  130. }
  131. else if (i == indexName)
  132. {
  133. name = splittedLine[i];
  134. }
  135. else if (i > indexText)
  136. {
  137. text += "," + splittedLine[i];
  138. }
  139. }
  140. try
  141. {
  142. var p = new SubtitleTrackEvent();
  143. p.StartPositionTicks = GetTimeCodeFromString(start);
  144. p.EndPositionTicks = GetTimeCodeFromString(end);
  145. p.Text = GetFormattedText(text);
  146. trackEvents.Add(p);
  147. }
  148. catch
  149. {
  150. }
  151. }
  152. }
  153. }
  154. // if (header.Length > 0)
  155. // subtitle.Header = header.ToString();
  156. // subtitle.Renumber(1);
  157. }
  158. trackInfo.TrackEvents = trackEvents.ToArray();
  159. return trackInfo;
  160. }
  161. private static long GetTimeCodeFromString(string time)
  162. {
  163. // h:mm:ss.cc
  164. string[] timeCode = time.Split(':', '.');
  165. return new TimeSpan(
  166. 0,
  167. int.Parse(timeCode[0], CultureInfo.InvariantCulture),
  168. int.Parse(timeCode[1], CultureInfo.InvariantCulture),
  169. int.Parse(timeCode[2], CultureInfo.InvariantCulture),
  170. int.Parse(timeCode[3], CultureInfo.InvariantCulture) * 10).Ticks;
  171. }
  172. private static string GetFormattedText(string text)
  173. {
  174. text = text.Replace("\\n", ParserValues.NewLine, StringComparison.OrdinalIgnoreCase);
  175. for (int i = 0; i < 10; i++) // just look ten times...
  176. {
  177. if (text.Contains(@"{\fn", StringComparison.Ordinal))
  178. {
  179. int start = text.IndexOf(@"{\fn", StringComparison.Ordinal);
  180. int end = text.IndexOf('}', start);
  181. if (end > 0 && !text.Substring(start).StartsWith("{\\fn}", StringComparison.Ordinal))
  182. {
  183. string fontName = text.Substring(start + 4, end - (start + 4));
  184. string extraTags = string.Empty;
  185. CheckAndAddSubTags(ref fontName, ref extraTags, out bool italic);
  186. text = text.Remove(start, end - start + 1);
  187. if (italic)
  188. {
  189. text = text.Insert(start, "<font face=\"" + fontName + "\"" + extraTags + "><i>");
  190. }
  191. else
  192. {
  193. text = text.Insert(start, "<font face=\"" + fontName + "\"" + extraTags + ">");
  194. }
  195. int indexOfEndTag = text.IndexOf("{\\fn}", start, StringComparison.Ordinal);
  196. if (indexOfEndTag > 0)
  197. {
  198. text = text.Remove(indexOfEndTag, "{\\fn}".Length).Insert(indexOfEndTag, "</font>");
  199. }
  200. else
  201. {
  202. text += "</font>";
  203. }
  204. }
  205. }
  206. if (text.Contains(@"{\fs", StringComparison.Ordinal))
  207. {
  208. int start = text.IndexOf(@"{\fs", StringComparison.Ordinal);
  209. int end = text.IndexOf('}', start);
  210. if (end > 0 && !text.Substring(start).StartsWith("{\\fs}", StringComparison.Ordinal))
  211. {
  212. string fontSize = text.Substring(start + 4, end - (start + 4));
  213. string extraTags = string.Empty;
  214. CheckAndAddSubTags(ref fontSize, ref extraTags, out bool italic);
  215. if (IsInteger(fontSize))
  216. {
  217. text = text.Remove(start, end - start + 1);
  218. if (italic)
  219. {
  220. text = text.Insert(start, "<font size=\"" + fontSize + "\"" + extraTags + "><i>");
  221. }
  222. else
  223. {
  224. text = text.Insert(start, "<font size=\"" + fontSize + "\"" + extraTags + ">");
  225. }
  226. int indexOfEndTag = text.IndexOf("{\\fs}", start, StringComparison.Ordinal);
  227. if (indexOfEndTag > 0)
  228. {
  229. text = text.Remove(indexOfEndTag, "{\\fs}".Length).Insert(indexOfEndTag, "</font>");
  230. }
  231. else
  232. {
  233. text += "</font>";
  234. }
  235. }
  236. }
  237. }
  238. if (text.Contains(@"{\c", StringComparison.Ordinal))
  239. {
  240. int start = text.IndexOf(@"{\c", StringComparison.Ordinal);
  241. int end = text.IndexOf('}', start);
  242. if (end > 0 && !text.Substring(start).StartsWith("{\\c}", StringComparison.Ordinal))
  243. {
  244. string color = text.Substring(start + 4, end - (start + 4));
  245. string extraTags = string.Empty;
  246. CheckAndAddSubTags(ref color, ref extraTags, out bool italic);
  247. color = color.Replace("&", string.Empty, StringComparison.Ordinal).TrimStart('H');
  248. color = color.PadLeft(6, '0');
  249. // switch to rrggbb from bbggrr
  250. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  251. color = color.ToLowerInvariant();
  252. text = text.Remove(start, end - start + 1);
  253. if (italic)
  254. {
  255. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + "><i>");
  256. }
  257. else
  258. {
  259. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + ">");
  260. }
  261. int indexOfEndTag = text.IndexOf("{\\c}", start, StringComparison.Ordinal);
  262. if (indexOfEndTag > 0)
  263. {
  264. text = text.Remove(indexOfEndTag, "{\\c}".Length).Insert(indexOfEndTag, "</font>");
  265. }
  266. else
  267. {
  268. text += "</font>";
  269. }
  270. }
  271. }
  272. if (text.Contains(@"{\1c", StringComparison.Ordinal)) // "1" specifices primary color
  273. {
  274. int start = text.IndexOf(@"{\1c", StringComparison.Ordinal);
  275. int end = text.IndexOf('}', start);
  276. if (end > 0 && !text.Substring(start).StartsWith("{\\1c}", StringComparison.Ordinal))
  277. {
  278. string color = text.Substring(start + 5, end - (start + 5));
  279. string extraTags = string.Empty;
  280. CheckAndAddSubTags(ref color, ref extraTags, out bool italic);
  281. color = color.Replace("&", string.Empty, StringComparison.Ordinal).TrimStart('H');
  282. color = color.PadLeft(6, '0');
  283. // switch to rrggbb from bbggrr
  284. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  285. color = color.ToLowerInvariant();
  286. text = text.Remove(start, end - start + 1);
  287. if (italic)
  288. {
  289. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + "><i>");
  290. }
  291. else
  292. {
  293. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + ">");
  294. }
  295. text += "</font>";
  296. }
  297. }
  298. }
  299. text = text.Replace(@"{\i1}", "<i>", StringComparison.Ordinal);
  300. text = text.Replace(@"{\i0}", "</i>", StringComparison.Ordinal);
  301. text = text.Replace(@"{\i}", "</i>", StringComparison.Ordinal);
  302. if (CountTagInText(text, "<i>") > CountTagInText(text, "</i>"))
  303. {
  304. text += "</i>";
  305. }
  306. text = text.Replace(@"{\u1}", "<u>", StringComparison.Ordinal);
  307. text = text.Replace(@"{\u0}", "</u>", StringComparison.Ordinal);
  308. text = text.Replace(@"{\u}", "</u>", StringComparison.Ordinal);
  309. if (CountTagInText(text, "<u>") > CountTagInText(text, "</u>"))
  310. {
  311. text += "</u>";
  312. }
  313. text = text.Replace(@"{\b1}", "<b>", StringComparison.Ordinal);
  314. text = text.Replace(@"{\b0}", "</b>", StringComparison.Ordinal);
  315. text = text.Replace(@"{\b}", "</b>", StringComparison.Ordinal);
  316. if (CountTagInText(text, "<b>") > CountTagInText(text, "</b>"))
  317. {
  318. text += "</b>";
  319. }
  320. return text;
  321. }
  322. private static bool IsInteger(string s)
  323. => int.TryParse(s, out _);
  324. private static int CountTagInText(string text, string tag)
  325. {
  326. int count = 0;
  327. int index = text.IndexOf(tag, StringComparison.Ordinal);
  328. while (index >= 0)
  329. {
  330. count++;
  331. if (index == text.Length)
  332. {
  333. return count;
  334. }
  335. index = text.IndexOf(tag, index + 1, StringComparison.Ordinal);
  336. }
  337. return count;
  338. }
  339. private static void CheckAndAddSubTags(ref string tagName, ref string extraTags, out bool italic)
  340. {
  341. italic = false;
  342. int indexOfSPlit = tagName.IndexOf('\\', StringComparison.Ordinal);
  343. if (indexOfSPlit > 0)
  344. {
  345. string rest = tagName.Substring(indexOfSPlit).TrimStart('\\');
  346. tagName = tagName.Remove(indexOfSPlit);
  347. for (int i = 0; i < 10; i++)
  348. {
  349. if (rest.StartsWith("fs", StringComparison.Ordinal) && rest.Length > 2)
  350. {
  351. indexOfSPlit = rest.IndexOf('\\', StringComparison.Ordinal);
  352. string fontSize = rest;
  353. if (indexOfSPlit > 0)
  354. {
  355. fontSize = rest.Substring(0, indexOfSPlit);
  356. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  357. }
  358. else
  359. {
  360. rest = string.Empty;
  361. }
  362. extraTags += " size=\"" + fontSize.Substring(2) + "\"";
  363. }
  364. else if (rest.StartsWith("fn", StringComparison.Ordinal) && rest.Length > 2)
  365. {
  366. indexOfSPlit = rest.IndexOf('\\', StringComparison.Ordinal);
  367. string fontName = rest;
  368. if (indexOfSPlit > 0)
  369. {
  370. fontName = rest.Substring(0, indexOfSPlit);
  371. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  372. }
  373. else
  374. {
  375. rest = string.Empty;
  376. }
  377. extraTags += " face=\"" + fontName.Substring(2) + "\"";
  378. }
  379. else if (rest.StartsWith("c", StringComparison.Ordinal) && rest.Length > 2)
  380. {
  381. indexOfSPlit = rest.IndexOf('\\', StringComparison.Ordinal);
  382. string fontColor = rest;
  383. if (indexOfSPlit > 0)
  384. {
  385. fontColor = rest.Substring(0, indexOfSPlit);
  386. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  387. }
  388. else
  389. {
  390. rest = string.Empty;
  391. }
  392. string color = fontColor.Substring(2);
  393. color = color.Replace("&", string.Empty, StringComparison.Ordinal).TrimStart('H');
  394. color = color.PadLeft(6, '0');
  395. // switch to rrggbb from bbggrr
  396. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  397. color = color.ToLowerInvariant();
  398. extraTags += " color=\"" + color + "\"";
  399. }
  400. else if (rest.StartsWith("i1", StringComparison.Ordinal) && rest.Length > 1)
  401. {
  402. indexOfSPlit = rest.IndexOf('\\', StringComparison.Ordinal);
  403. italic = true;
  404. if (indexOfSPlit > 0)
  405. {
  406. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  407. }
  408. else
  409. {
  410. rest = string.Empty;
  411. }
  412. }
  413. else if (rest.Length > 0 && rest.Contains('\\', StringComparison.Ordinal))
  414. {
  415. indexOfSPlit = rest.IndexOf('\\', StringComparison.Ordinal);
  416. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  417. }
  418. }
  419. }
  420. }
  421. }
  422. }