SsaParser.cs 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477
  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(';'))
  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. trackEvents.Add(
  143. new SubtitleTrackEvent
  144. {
  145. StartPositionTicks = GetTimeCodeFromString(start),
  146. EndPositionTicks = GetTimeCodeFromString(end),
  147. Text = GetFormattedText(text)
  148. });
  149. }
  150. catch
  151. {
  152. }
  153. }
  154. }
  155. }
  156. // if (header.Length > 0)
  157. // subtitle.Header = header.ToString();
  158. // subtitle.Renumber(1);
  159. }
  160. trackInfo.TrackEvents = trackEvents.ToArray();
  161. return trackInfo;
  162. }
  163. private static long GetTimeCodeFromString(string time)
  164. {
  165. // h:mm:ss.cc
  166. string[] timeCode = time.Split(':', '.');
  167. return new TimeSpan(
  168. 0,
  169. int.Parse(timeCode[0], CultureInfo.InvariantCulture),
  170. int.Parse(timeCode[1], CultureInfo.InvariantCulture),
  171. int.Parse(timeCode[2], CultureInfo.InvariantCulture),
  172. int.Parse(timeCode[3], CultureInfo.InvariantCulture) * 10).Ticks;
  173. }
  174. private static string GetFormattedText(string text)
  175. {
  176. text = text.Replace("\\n", ParserValues.NewLine, StringComparison.OrdinalIgnoreCase);
  177. for (int i = 0; i < 10; i++) // just look ten times...
  178. {
  179. if (text.Contains(@"{\fn", StringComparison.Ordinal))
  180. {
  181. int start = text.IndexOf(@"{\fn", StringComparison.Ordinal);
  182. int end = text.IndexOf('}', start);
  183. if (end > 0 && !text.Substring(start).StartsWith("{\\fn}", StringComparison.Ordinal))
  184. {
  185. string fontName = text.Substring(start + 4, end - (start + 4));
  186. string extraTags = string.Empty;
  187. CheckAndAddSubTags(ref fontName, ref extraTags, out bool italic);
  188. text = text.Remove(start, end - start + 1);
  189. if (italic)
  190. {
  191. text = text.Insert(start, "<font face=\"" + fontName + "\"" + extraTags + "><i>");
  192. }
  193. else
  194. {
  195. text = text.Insert(start, "<font face=\"" + fontName + "\"" + extraTags + ">");
  196. }
  197. int indexOfEndTag = text.IndexOf("{\\fn}", start, StringComparison.Ordinal);
  198. if (indexOfEndTag > 0)
  199. {
  200. text = text.Remove(indexOfEndTag, "{\\fn}".Length).Insert(indexOfEndTag, "</font>");
  201. }
  202. else
  203. {
  204. text += "</font>";
  205. }
  206. }
  207. }
  208. if (text.Contains(@"{\fs", StringComparison.Ordinal))
  209. {
  210. int start = text.IndexOf(@"{\fs", StringComparison.Ordinal);
  211. int end = text.IndexOf('}', start);
  212. if (end > 0 && !text.Substring(start).StartsWith("{\\fs}", StringComparison.Ordinal))
  213. {
  214. string fontSize = text.Substring(start + 4, end - (start + 4));
  215. string extraTags = string.Empty;
  216. CheckAndAddSubTags(ref fontSize, ref extraTags, out bool italic);
  217. if (IsInteger(fontSize))
  218. {
  219. text = text.Remove(start, end - start + 1);
  220. if (italic)
  221. {
  222. text = text.Insert(start, "<font size=\"" + fontSize + "\"" + extraTags + "><i>");
  223. }
  224. else
  225. {
  226. text = text.Insert(start, "<font size=\"" + fontSize + "\"" + extraTags + ">");
  227. }
  228. int indexOfEndTag = text.IndexOf("{\\fs}", start, StringComparison.Ordinal);
  229. if (indexOfEndTag > 0)
  230. {
  231. text = text.Remove(indexOfEndTag, "{\\fs}".Length).Insert(indexOfEndTag, "</font>");
  232. }
  233. else
  234. {
  235. text += "</font>";
  236. }
  237. }
  238. }
  239. }
  240. if (text.Contains(@"{\c", StringComparison.Ordinal))
  241. {
  242. int start = text.IndexOf(@"{\c", StringComparison.Ordinal);
  243. int end = text.IndexOf('}', start);
  244. if (end > 0 && !text.Substring(start).StartsWith("{\\c}", StringComparison.Ordinal))
  245. {
  246. string color = text.Substring(start + 4, end - (start + 4));
  247. string extraTags = string.Empty;
  248. CheckAndAddSubTags(ref color, ref extraTags, out bool italic);
  249. color = color.Replace("&", string.Empty, StringComparison.Ordinal).TrimStart('H');
  250. color = color.PadLeft(6, '0');
  251. // switch to rrggbb from bbggrr
  252. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  253. color = color.ToLowerInvariant();
  254. text = text.Remove(start, end - start + 1);
  255. if (italic)
  256. {
  257. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + "><i>");
  258. }
  259. else
  260. {
  261. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + ">");
  262. }
  263. int indexOfEndTag = text.IndexOf("{\\c}", start, StringComparison.Ordinal);
  264. if (indexOfEndTag > 0)
  265. {
  266. text = text.Remove(indexOfEndTag, "{\\c}".Length).Insert(indexOfEndTag, "</font>");
  267. }
  268. else
  269. {
  270. text += "</font>";
  271. }
  272. }
  273. }
  274. if (text.Contains(@"{\1c", StringComparison.Ordinal)) // "1" specifices primary color
  275. {
  276. int start = text.IndexOf(@"{\1c", StringComparison.Ordinal);
  277. int end = text.IndexOf('}', start);
  278. if (end > 0 && !text.Substring(start).StartsWith("{\\1c}", StringComparison.Ordinal))
  279. {
  280. string color = text.Substring(start + 5, end - (start + 5));
  281. string extraTags = string.Empty;
  282. CheckAndAddSubTags(ref color, ref extraTags, out bool italic);
  283. color = color.Replace("&", string.Empty, StringComparison.Ordinal).TrimStart('H');
  284. color = color.PadLeft(6, '0');
  285. // switch to rrggbb from bbggrr
  286. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  287. color = color.ToLowerInvariant();
  288. text = text.Remove(start, end - start + 1);
  289. if (italic)
  290. {
  291. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + "><i>");
  292. }
  293. else
  294. {
  295. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + ">");
  296. }
  297. int indexOfEndTag = text.IndexOf("{\\1c}", start, StringComparison.Ordinal);
  298. if (indexOfEndTag > 0)
  299. {
  300. text = text.Remove(indexOfEndTag, "{\\1c}".Length).Insert(indexOfEndTag, "</font>");
  301. }
  302. else
  303. {
  304. text += "</font>";
  305. }
  306. }
  307. }
  308. }
  309. text = text.Replace(@"{\i1}", "<i>", StringComparison.Ordinal);
  310. text = text.Replace(@"{\i0}", "</i>", StringComparison.Ordinal);
  311. text = text.Replace(@"{\i}", "</i>", StringComparison.Ordinal);
  312. if (CountTagInText(text, "<i>") > CountTagInText(text, "</i>"))
  313. {
  314. text += "</i>";
  315. }
  316. text = text.Replace(@"{\u1}", "<u>", StringComparison.Ordinal);
  317. text = text.Replace(@"{\u0}", "</u>", StringComparison.Ordinal);
  318. text = text.Replace(@"{\u}", "</u>", StringComparison.Ordinal);
  319. if (CountTagInText(text, "<u>") > CountTagInText(text, "</u>"))
  320. {
  321. text += "</u>";
  322. }
  323. text = text.Replace(@"{\b1}", "<b>", StringComparison.Ordinal);
  324. text = text.Replace(@"{\b0}", "</b>", StringComparison.Ordinal);
  325. text = text.Replace(@"{\b}", "</b>", StringComparison.Ordinal);
  326. if (CountTagInText(text, "<b>") > CountTagInText(text, "</b>"))
  327. {
  328. text += "</b>";
  329. }
  330. return text;
  331. }
  332. private static bool IsInteger(string s)
  333. => int.TryParse(s, out _);
  334. private static int CountTagInText(string text, string tag)
  335. {
  336. int count = 0;
  337. int index = text.IndexOf(tag, StringComparison.Ordinal);
  338. while (index >= 0)
  339. {
  340. count++;
  341. if (index == text.Length)
  342. {
  343. return count;
  344. }
  345. index = text.IndexOf(tag, index + 1, StringComparison.Ordinal);
  346. }
  347. return count;
  348. }
  349. private static void CheckAndAddSubTags(ref string tagName, ref string extraTags, out bool italic)
  350. {
  351. italic = false;
  352. int indexOfSPlit = tagName.IndexOf('\\', StringComparison.Ordinal);
  353. if (indexOfSPlit > 0)
  354. {
  355. string rest = tagName.Substring(indexOfSPlit).TrimStart('\\');
  356. tagName = tagName.Remove(indexOfSPlit);
  357. for (int i = 0; i < 10; i++)
  358. {
  359. if (rest.StartsWith("fs", StringComparison.Ordinal) && rest.Length > 2)
  360. {
  361. indexOfSPlit = rest.IndexOf('\\', StringComparison.Ordinal);
  362. string fontSize = rest;
  363. if (indexOfSPlit > 0)
  364. {
  365. fontSize = rest.Substring(0, indexOfSPlit);
  366. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  367. }
  368. else
  369. {
  370. rest = string.Empty;
  371. }
  372. extraTags += " size=\"" + fontSize.Substring(2) + "\"";
  373. }
  374. else if (rest.StartsWith("fn", StringComparison.Ordinal) && rest.Length > 2)
  375. {
  376. indexOfSPlit = rest.IndexOf('\\', StringComparison.Ordinal);
  377. string fontName = rest;
  378. if (indexOfSPlit > 0)
  379. {
  380. fontName = rest.Substring(0, indexOfSPlit);
  381. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  382. }
  383. else
  384. {
  385. rest = string.Empty;
  386. }
  387. extraTags += " face=\"" + fontName.Substring(2) + "\"";
  388. }
  389. else if (rest.StartsWith("c", StringComparison.Ordinal) && rest.Length > 2)
  390. {
  391. indexOfSPlit = rest.IndexOf('\\', StringComparison.Ordinal);
  392. string fontColor = rest;
  393. if (indexOfSPlit > 0)
  394. {
  395. fontColor = rest.Substring(0, indexOfSPlit);
  396. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  397. }
  398. else
  399. {
  400. rest = string.Empty;
  401. }
  402. string color = fontColor.Substring(2);
  403. color = color.Replace("&", string.Empty, StringComparison.Ordinal).TrimStart('H');
  404. color = color.PadLeft(6, '0');
  405. // switch to rrggbb from bbggrr
  406. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  407. color = color.ToLowerInvariant();
  408. extraTags += " color=\"" + color + "\"";
  409. }
  410. else if (rest.StartsWith("i1", StringComparison.Ordinal) && rest.Length > 1)
  411. {
  412. indexOfSPlit = rest.IndexOf('\\', StringComparison.Ordinal);
  413. italic = true;
  414. if (indexOfSPlit > 0)
  415. {
  416. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  417. }
  418. else
  419. {
  420. rest = string.Empty;
  421. }
  422. }
  423. else if (rest.Length > 0 && rest.Contains('\\', StringComparison.Ordinal))
  424. {
  425. indexOfSPlit = rest.IndexOf('\\', StringComparison.Ordinal);
  426. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  427. }
  428. }
  429. }
  430. }
  431. }
  432. }