SsaParser.cs 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395
  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. header.AppendLine(line);
  38. if (line.Trim().ToLowerInvariant() == "[events]")
  39. {
  40. eventsStarted = true;
  41. }
  42. else if (!string.IsNullOrEmpty(line) && line.Trim().StartsWith(";"))
  43. {
  44. // skip comment lines
  45. }
  46. else if (eventsStarted && line.Trim().Length > 0)
  47. {
  48. string s = line.Trim().ToLowerInvariant();
  49. if (s.StartsWith("format:"))
  50. {
  51. if (line.Length > 10)
  52. {
  53. format = line.ToLowerInvariant().Substring(8).Split(',');
  54. for (int i = 0; i < format.Length; i++)
  55. {
  56. if (format[i].Trim().ToLowerInvariant() == "layer")
  57. indexLayer = i;
  58. else if (format[i].Trim().ToLowerInvariant() == "start")
  59. indexStart = i;
  60. else if (format[i].Trim().ToLowerInvariant() == "end")
  61. indexEnd = i;
  62. else if (format[i].Trim().ToLowerInvariant() == "text")
  63. indexText = i;
  64. else if (format[i].Trim().ToLowerInvariant() == "effect")
  65. indexEffect = i;
  66. else if (format[i].Trim().ToLowerInvariant() == "style")
  67. indexStyle = i;
  68. }
  69. }
  70. }
  71. else if (!string.IsNullOrEmpty(s))
  72. {
  73. string text = string.Empty;
  74. string start = string.Empty;
  75. string end = string.Empty;
  76. string style = string.Empty;
  77. string layer = string.Empty;
  78. string effect = string.Empty;
  79. string name = string.Empty;
  80. string[] splittedLine;
  81. if (s.StartsWith("dialogue:"))
  82. splittedLine = line.Substring(10).Split(',');
  83. else
  84. splittedLine = line.Split(',');
  85. for (int i = 0; i < splittedLine.Length; i++)
  86. {
  87. if (i == indexStart)
  88. start = splittedLine[i].Trim();
  89. else if (i == indexEnd)
  90. end = splittedLine[i].Trim();
  91. else if (i == indexLayer)
  92. layer = splittedLine[i];
  93. else if (i == indexEffect)
  94. effect = splittedLine[i];
  95. else if (i == indexText)
  96. text = splittedLine[i];
  97. else if (i == indexStyle)
  98. style = splittedLine[i];
  99. else if (i == indexName)
  100. name = splittedLine[i];
  101. else if (i > indexText)
  102. text += "," + splittedLine[i];
  103. }
  104. try
  105. {
  106. var p = new SubtitleTrackEvent();
  107. p.StartPositionTicks = GetTimeCodeFromString(start);
  108. p.EndPositionTicks = GetTimeCodeFromString(end);
  109. p.Text = GetFormattedText(text);
  110. trackEvents.Add(p);
  111. }
  112. catch
  113. {
  114. }
  115. }
  116. }
  117. }
  118. //if (header.Length > 0)
  119. //subtitle.Header = header.ToString();
  120. //subtitle.Renumber(1);
  121. }
  122. trackInfo.TrackEvents = trackEvents.ToArray();
  123. return trackInfo;
  124. }
  125. private static long GetTimeCodeFromString(string time)
  126. {
  127. // h:mm:ss.cc
  128. string[] timeCode = time.Split(':', '.');
  129. return new TimeSpan(0, int.Parse(timeCode[0]),
  130. int.Parse(timeCode[1]),
  131. int.Parse(timeCode[2]),
  132. int.Parse(timeCode[3]) * 10).Ticks;
  133. }
  134. public static string GetFormattedText(string text)
  135. {
  136. text = text.Replace("\\n", ParserValues.NewLine, StringComparison.OrdinalIgnoreCase);
  137. bool italic = false;
  138. for (int i = 0; i < 10; i++) // just look ten times...
  139. {
  140. if (text.Contains(@"{\fn"))
  141. {
  142. int start = text.IndexOf(@"{\fn");
  143. int end = text.IndexOf('}', start);
  144. if (end > 0 && !text.Substring(start).StartsWith("{\\fn}"))
  145. {
  146. string fontName = text.Substring(start + 4, end - (start + 4));
  147. string extraTags = string.Empty;
  148. CheckAndAddSubTags(ref fontName, ref extraTags, out italic);
  149. text = text.Remove(start, end - start + 1);
  150. if (italic)
  151. text = text.Insert(start, "<font face=\"" + fontName + "\"" + extraTags + "><i>");
  152. else
  153. text = text.Insert(start, "<font face=\"" + fontName + "\"" + extraTags + ">");
  154. int indexOfEndTag = text.IndexOf("{\\fn}", start);
  155. if (indexOfEndTag > 0)
  156. text = text.Remove(indexOfEndTag, "{\\fn}".Length).Insert(indexOfEndTag, "</font>");
  157. else
  158. text += "</font>";
  159. }
  160. }
  161. if (text.Contains(@"{\fs"))
  162. {
  163. int start = text.IndexOf(@"{\fs");
  164. int end = text.IndexOf('}', start);
  165. if (end > 0 && !text.Substring(start).StartsWith("{\\fs}"))
  166. {
  167. string fontSize = text.Substring(start + 4, end - (start + 4));
  168. string extraTags = string.Empty;
  169. CheckAndAddSubTags(ref fontSize, ref extraTags, out italic);
  170. if (IsInteger(fontSize))
  171. {
  172. text = text.Remove(start, end - start + 1);
  173. if (italic)
  174. text = text.Insert(start, "<font size=\"" + fontSize + "\"" + extraTags + "><i>");
  175. else
  176. text = text.Insert(start, "<font size=\"" + fontSize + "\"" + extraTags + ">");
  177. int indexOfEndTag = text.IndexOf("{\\fs}", start);
  178. if (indexOfEndTag > 0)
  179. text = text.Remove(indexOfEndTag, "{\\fs}".Length).Insert(indexOfEndTag, "</font>");
  180. else
  181. text += "</font>";
  182. }
  183. }
  184. }
  185. if (text.Contains(@"{\c"))
  186. {
  187. int start = text.IndexOf(@"{\c");
  188. int end = text.IndexOf('}', start);
  189. if (end > 0 && !text.Substring(start).StartsWith("{\\c}"))
  190. {
  191. string color = text.Substring(start + 4, end - (start + 4));
  192. string extraTags = string.Empty;
  193. CheckAndAddSubTags(ref color, ref extraTags, out italic);
  194. color = color.Replace("&", string.Empty).TrimStart('H');
  195. color = color.PadLeft(6, '0');
  196. // switch to rrggbb from bbggrr
  197. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  198. color = color.ToLowerInvariant();
  199. text = text.Remove(start, end - start + 1);
  200. if (italic)
  201. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + "><i>");
  202. else
  203. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + ">");
  204. int indexOfEndTag = text.IndexOf("{\\c}", start);
  205. if (indexOfEndTag > 0)
  206. text = text.Remove(indexOfEndTag, "{\\c}".Length).Insert(indexOfEndTag, "</font>");
  207. else
  208. text += "</font>";
  209. }
  210. }
  211. if (text.Contains(@"{\1c")) // "1" specifices primary color
  212. {
  213. int start = text.IndexOf(@"{\1c");
  214. int end = text.IndexOf('}', start);
  215. if (end > 0 && !text.Substring(start).StartsWith("{\\1c}"))
  216. {
  217. string color = text.Substring(start + 5, end - (start + 5));
  218. string extraTags = string.Empty;
  219. CheckAndAddSubTags(ref color, ref extraTags, out italic);
  220. color = color.Replace("&", string.Empty).TrimStart('H');
  221. color = color.PadLeft(6, '0');
  222. // switch to rrggbb from bbggrr
  223. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  224. color = color.ToLowerInvariant();
  225. text = text.Remove(start, end - start + 1);
  226. if (italic)
  227. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + "><i>");
  228. else
  229. text = text.Insert(start, "<font color=\"" + color + "\"" + extraTags + ">");
  230. text += "</font>";
  231. }
  232. }
  233. }
  234. text = text.Replace(@"{\i1}", "<i>");
  235. text = text.Replace(@"{\i0}", "</i>");
  236. text = text.Replace(@"{\i}", "</i>");
  237. if (CountTagInText(text, "<i>") > CountTagInText(text, "</i>"))
  238. text += "</i>";
  239. text = text.Replace(@"{\u1}", "<u>");
  240. text = text.Replace(@"{\u0}", "</u>");
  241. text = text.Replace(@"{\u}", "</u>");
  242. if (CountTagInText(text, "<u>") > CountTagInText(text, "</u>"))
  243. text += "</u>";
  244. text = text.Replace(@"{\b1}", "<b>");
  245. text = text.Replace(@"{\b0}", "</b>");
  246. text = text.Replace(@"{\b}", "</b>");
  247. if (CountTagInText(text, "<b>") > CountTagInText(text, "</b>"))
  248. text += "</b>";
  249. return text;
  250. }
  251. private static bool IsInteger(string s)
  252. {
  253. if (int.TryParse(s, out var i))
  254. return true;
  255. return false;
  256. }
  257. private static int CountTagInText(string text, string tag)
  258. {
  259. int count = 0;
  260. int index = text.IndexOf(tag);
  261. while (index >= 0)
  262. {
  263. count++;
  264. if (index == text.Length)
  265. return count;
  266. index = text.IndexOf(tag, index + 1);
  267. }
  268. return count;
  269. }
  270. private static void CheckAndAddSubTags(ref string tagName, ref string extraTags, out bool italic)
  271. {
  272. italic = false;
  273. int indexOfSPlit = tagName.IndexOf(@"\");
  274. if (indexOfSPlit > 0)
  275. {
  276. string rest = tagName.Substring(indexOfSPlit).TrimStart('\\');
  277. tagName = tagName.Remove(indexOfSPlit);
  278. for (int i = 0; i < 10; i++)
  279. {
  280. if (rest.StartsWith("fs") && rest.Length > 2)
  281. {
  282. indexOfSPlit = rest.IndexOf(@"\");
  283. string fontSize = rest;
  284. if (indexOfSPlit > 0)
  285. {
  286. fontSize = rest.Substring(0, indexOfSPlit);
  287. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  288. }
  289. else
  290. {
  291. rest = string.Empty;
  292. }
  293. extraTags += " size=\"" + fontSize.Substring(2) + "\"";
  294. }
  295. else if (rest.StartsWith("fn") && rest.Length > 2)
  296. {
  297. indexOfSPlit = rest.IndexOf(@"\");
  298. string fontName = rest;
  299. if (indexOfSPlit > 0)
  300. {
  301. fontName = rest.Substring(0, indexOfSPlit);
  302. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  303. }
  304. else
  305. {
  306. rest = string.Empty;
  307. }
  308. extraTags += " face=\"" + fontName.Substring(2) + "\"";
  309. }
  310. else if (rest.StartsWith("c") && rest.Length > 2)
  311. {
  312. indexOfSPlit = rest.IndexOf(@"\");
  313. string fontColor = rest;
  314. if (indexOfSPlit > 0)
  315. {
  316. fontColor = rest.Substring(0, indexOfSPlit);
  317. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  318. }
  319. else
  320. {
  321. rest = string.Empty;
  322. }
  323. string color = fontColor.Substring(2);
  324. color = color.Replace("&", string.Empty).TrimStart('H');
  325. color = color.PadLeft(6, '0');
  326. // switch to rrggbb from bbggrr
  327. color = "#" + color.Remove(color.Length - 6) + color.Substring(color.Length - 2, 2) + color.Substring(color.Length - 4, 2) + color.Substring(color.Length - 6, 2);
  328. color = color.ToLowerInvariant();
  329. extraTags += " color=\"" + color + "\"";
  330. }
  331. else if (rest.StartsWith("i1") && rest.Length > 1)
  332. {
  333. indexOfSPlit = rest.IndexOf(@"\");
  334. italic = true;
  335. if (indexOfSPlit > 0)
  336. {
  337. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  338. }
  339. else
  340. {
  341. rest = string.Empty;
  342. }
  343. }
  344. else if (rest.Length > 0 && rest.Contains("\\"))
  345. {
  346. indexOfSPlit = rest.IndexOf(@"\");
  347. rest = rest.Substring(indexOfSPlit).TrimStart('\\');
  348. }
  349. }
  350. }
  351. }
  352. }
  353. }