public override void LoadSubtitle(Subtitle subtitle, List <string> lines, string fileName) { _errorCount = 0; var sb = new StringBuilder(); lines.ForEach(line => sb.AppendLine(line)); var xml = new XmlDocument { XmlResolver = null }; xml.LoadXml(sb.ToString().Replace(" & ", " & ").Replace("Q&A", "Q&A").RemoveControlCharactersButWhiteSpace().Trim()); var nsmgr = new XmlNamespaceManager(xml.NameTable); nsmgr.AddNamespace("ttaf1", xml.DocumentElement.NamespaceURI); var div = xml.DocumentElement.SelectSingleNode("//ttaf1:body", nsmgr).SelectSingleNode("ttaf1:div", nsmgr); if (div == null) { div = xml.DocumentElement.SelectSingleNode("//ttaf1:body", nsmgr).FirstChild; } var styleDic = new Dictionary <string, string>(); foreach (XmlNode node in xml.DocumentElement.SelectNodes("//ttaf1:style", nsmgr)) { if (node.Attributes["tts:fontStyle"] != null && node.Attributes["xml:id"] != null) { styleDic.Add(node.Attributes["xml:id"].Value, node.Attributes["tts:fontStyle"].Value); } } bool couldBeFrames = true; bool couldBeMillisecondsWithMissingLastDigit = true; foreach (XmlNode node in div.ChildNodes) { try { var pText = new StringBuilder(); foreach (XmlNode innerNode in node.ChildNodes) { switch (innerNode.Name.Replace("tt:", string.Empty)) { case "br": pText.AppendLine(); break; case "span": bool italic = false; if (innerNode.Attributes["style"] != null && styleDic.ContainsKey(innerNode.Attributes["style"].Value)) { if (styleDic[innerNode.Attributes["style"].Value].Contains("italic")) { italic = true; pText.Append("<i>"); } } if (!italic && innerNode.Attributes != null) { var fs = innerNode.Attributes.GetNamedItem("tts:fontStyle"); if (fs != null && fs.Value == "italic") { italic = true; pText.Append("<i>"); } } if (innerNode.HasChildNodes) { foreach (XmlNode innerInnerNode in innerNode.ChildNodes) { if (innerInnerNode.Name == "br" || innerInnerNode.Name == "tt:br") { pText.AppendLine(); } else { pText.Append(innerInnerNode.InnerText); } } } else { pText.Append(innerNode.InnerText); } if (italic) { pText.Append("</i>"); } break; case "i": pText.Append("<i>" + innerNode.InnerText + "</i>"); break; case "b": pText.Append("<b>" + innerNode.InnerText + "</b>"); break; default: pText.Append(innerNode.InnerText); break; } } string start = null; string end = null; string dur = null; foreach (XmlAttribute attr in node.Attributes) { if (attr.Name.EndsWith("begin", StringComparison.Ordinal)) { start = attr.InnerText; } else if (attr.Name.EndsWith("end", StringComparison.Ordinal)) { end = attr.InnerText; } else if (attr.Name.EndsWith("duration", StringComparison.Ordinal)) { dur = attr.InnerText; } } string text = pText.ToString(); text = text.Replace(Environment.NewLine + "</i>", "</i>" + Environment.NewLine); text = text.Replace("<i></i>", string.Empty).Trim(); if (end != null) { if (end.Length != 11 || end.Substring(8, 1) != ":" || start == null || start.Length != 11 || start.Substring(8, 1) != ":") { couldBeFrames = false; } if (couldBeMillisecondsWithMissingLastDigit && (end.Length != 11 || start == null || start.Length != 11 || end.Substring(8, 1) != "." || start.Substring(8, 1) != ".")) { couldBeMillisecondsWithMissingLastDigit = false; } double dBegin, dEnd; if (!start.Contains(':') && Utilities.CountTagInText(start, '.') == 1 && !end.Contains(':') && Utilities.CountTagInText(end, '.') == 1 && double.TryParse(start, NumberStyles.Float, CultureInfo.InvariantCulture, out dBegin) && double.TryParse(end, NumberStyles.Float, CultureInfo.InvariantCulture, out dEnd)) { subtitle.Paragraphs.Add(new Paragraph(text, dBegin * TimeCode.BaseUnit, dEnd * TimeCode.BaseUnit)); } else { if (start.Length == 8 && start[2] == ':' && start[5] == ':' && end.Length == 8 && end[2] == ':' && end[5] == ':') { var p = new Paragraph(); var parts = start.Split(SplitCharColon); p.StartTime = new TimeCode(int.Parse(parts[0]), int.Parse(parts[1]), int.Parse(parts[2]), 0); parts = end.Split(SplitCharColon); p.EndTime = new TimeCode(int.Parse(parts[0]), int.Parse(parts[1]), int.Parse(parts[2]), 0); p.Text = text; subtitle.Paragraphs.Add(p); } else { subtitle.Paragraphs.Add(new Paragraph(TimedText10.GetTimeCode(start, false), TimedText10.GetTimeCode(end, false), text)); } } } else if (dur != null) { if (dur.Length != 11 || dur.Substring(8, 1) != ":" || start == null || start.Length != 11 || start.Substring(8, 1) != ":") { couldBeFrames = false; } if (couldBeMillisecondsWithMissingLastDigit && (dur.Length != 11 || start == null || start.Length != 11 || dur.Substring(8, 1) != "." || start.Substring(8, 1) != ".")) { couldBeMillisecondsWithMissingLastDigit = false; } TimeCode duration = TimedText10.GetTimeCode(dur, false); TimeCode startTime = TimedText10.GetTimeCode(start, false); var endTime = new TimeCode(startTime.TotalMilliseconds + duration.TotalMilliseconds); subtitle.Paragraphs.Add(new Paragraph(startTime, endTime, text)); } } catch (Exception ex) { System.Diagnostics.Debug.WriteLine(ex.Message); _errorCount++; } } subtitle.RemoveEmptyLines(); if (couldBeFrames) { bool all30OrBelow = true; foreach (Paragraph p in subtitle.Paragraphs) { if (p.StartTime.Milliseconds > 30 || p.EndTime.Milliseconds > 30) { all30OrBelow = false; break; } } if (all30OrBelow) { foreach (Paragraph p in subtitle.Paragraphs) { p.StartTime.Milliseconds = SubtitleFormat.FramesToMillisecondsMax999(p.StartTime.Milliseconds); p.EndTime.Milliseconds = SubtitleFormat.FramesToMillisecondsMax999(p.EndTime.Milliseconds); } } } else if (couldBeMillisecondsWithMissingLastDigit && Configuration.Settings.SubtitleSettings.TimedText10TimeCodeFormatSource != "hh:mm:ss.ms-two-digits") { foreach (Paragraph p in subtitle.Paragraphs) { p.StartTime.Milliseconds *= 10; p.EndTime.Milliseconds *= 10; } } subtitle.Renumber(); }
public override void LoadSubtitle(Subtitle subtitle, List <string> lines, string fileName) { _errorCount = 0; var sb = new StringBuilder(); lines.ForEach(line => sb.AppendLine(line)); var xml = new XmlDocument { XmlResolver = null }; xml.LoadXml(sb.ToString().Trim()); var nsmgr = new XmlNamespaceManager(xml.NameTable); nsmgr.AddNamespace("ttaf1", xml.DocumentElement.NamespaceURI); XmlNode div = xml.DocumentElement.SelectSingleNode("//ttaf1:body", nsmgr).SelectSingleNode("ttaf1:div", nsmgr); if (div == null) { div = xml.DocumentElement.SelectSingleNode("//ttaf1:body", nsmgr).FirstChild; } var styleDic = new Dictionary <string, string>(); foreach (XmlNode node in xml.DocumentElement.SelectNodes("//ttaf1:style", nsmgr)) { if (node.Attributes["tts:fontStyle"] != null && node.Attributes["xml:id"] != null) { styleDic.Add(node.Attributes["xml:id"].Value, node.Attributes["tts:fontStyle"].Value); } } foreach (XmlNode node in div.ChildNodes) { try { var pText = new StringBuilder(); foreach (XmlNode innerNode in node.ChildNodes) { switch (innerNode.Name) { case "br": pText.AppendLine(); break; case "span": bool italic = false; if (innerNode.Attributes["style"] != null && styleDic.ContainsKey(innerNode.Attributes["style"].Value)) { if (styleDic[innerNode.Attributes["style"].Value].Contains("italic")) { italic = true; pText.Append("<i>"); } } if (!italic && innerNode.Attributes != null) { var fs = innerNode.Attributes.GetNamedItem("tts:fontStyle"); if (fs != null && fs.Value == "italic") { italic = true; pText.Append("<i>"); } } if (innerNode.HasChildNodes) { foreach (XmlNode innerInnerNode in innerNode.ChildNodes) { if (innerInnerNode.Name == "br") { pText.AppendLine(); } else { pText.Append(innerInnerNode.InnerText); } } } else { pText.Append(innerNode.InnerText); } if (italic) { pText.Append("</i>"); } break; case "i": pText.Append("<i>" + innerNode.InnerText + "</i>"); break; case "b": pText.Append("<b>" + innerNode.InnerText + "</b>"); break; default: pText.Append(innerNode.InnerText); break; } } string start = null; // = node.Attributes["begin"].InnerText; string end = null; // = node.Attributes["begin"].InnerText; string dur = null; // = node.Attributes["begin"].InnerText; foreach (XmlAttribute attr in node.Attributes) { if (attr.Name.EndsWith("begin", StringComparison.Ordinal)) { start = attr.InnerText; } else if (attr.Name.EndsWith("end", StringComparison.Ordinal)) { end = attr.InnerText; } else if (attr.Name.EndsWith("duration", StringComparison.Ordinal)) { dur = attr.InnerText; } } //string start = node.Attributes["begin"].InnerText; string text = pText.ToString(); text = text.Replace(Environment.NewLine + "</i>", "</i>" + Environment.NewLine); text = text.Replace("<i></i>", string.Empty).Trim(); if (end != null) { //string end = node.Attributes["end"].InnerText; double dBegin, dEnd; if (!start.Contains(':') && Utilities.CountTagInText(start, '.') == 1 && !end.Contains(':') && Utilities.CountTagInText(end, '.') == 1 && double.TryParse(start, out dBegin) && double.TryParse(end, out dEnd)) { subtitle.Paragraphs.Add(new Paragraph(text, dBegin * TimeCode.BaseUnit, dEnd * TimeCode.BaseUnit)); } else { if (start.Length == 8 && start[2] == ':' && start[5] == ':' && end.Length == 8 && end[2] == ':' && end[5] == ':') { var p = new Paragraph(); var parts = start.Split(new[] { ':' }); p.StartTime = new TimeCode(int.Parse(parts[0]), int.Parse(parts[1]), int.Parse(parts[2]), 0); parts = end.Split(new[] { ':' }); p.EndTime = new TimeCode(int.Parse(parts[0]), int.Parse(parts[1]), int.Parse(parts[2]), 0); p.Text = text; subtitle.Paragraphs.Add(p); } else { subtitle.Paragraphs.Add(new Paragraph(TimedText10.GetTimeCode(start, false), TimedText10.GetTimeCode(end, false), text)); } } } else if (dur != null) { TimeCode duration = TimedText10.GetTimeCode(dur, false); TimeCode startTime = TimedText10.GetTimeCode(start, false); var endTime = new TimeCode(startTime.TotalMilliseconds + duration.TotalMilliseconds); subtitle.Paragraphs.Add(new Paragraph(startTime, endTime, text)); } } catch (Exception ex) { System.Diagnostics.Debug.WriteLine(ex.Message); _errorCount++; } } subtitle.RemoveEmptyLines(); subtitle.Renumber(); }
public override void LoadSubtitle(Subtitle subtitle, List <string> lines, string fileName) { _errorCount = 0; var sb = new StringBuilder(); lines.ForEach(line => sb.AppendLine(line)); var xml = new XmlDocument { XmlResolver = null }; xml.LoadXml(sb.ToString().RemoveControlCharactersButWhiteSpace().Trim().Replace("http://www.w3.org/2006/04/ttaf1#styling\"xml:lang", "http://www.w3.org/2006/04/ttaf1#styling\" xml:lang")); var nsmgr = new XmlNamespaceManager(xml.NameTable); nsmgr.AddNamespace("ttaf1", xml.DocumentElement.NamespaceURI); foreach (XmlNode node in xml.DocumentElement.SelectNodes("//ttaf1:p", nsmgr)) { try { var pText = new StringBuilder(); foreach (XmlNode innerNode in node.ChildNodes) { switch (innerNode.Name) { case "br": pText.AppendLine(); break; case "span": bool italic = false; if (innerNode.Attributes != null) { var fs = innerNode.Attributes.GetNamedItem("tts:fontStyle"); if (fs != null && fs.Value == "italic") { italic = true; pText.Append("<i>"); } } if (innerNode.HasChildNodes) { foreach (XmlNode innerInnerNode in innerNode.ChildNodes) { if (innerInnerNode.Name == "br") { pText.AppendLine(); } else { pText.Append(innerInnerNode.InnerText); } } } else { pText.Append(innerNode.InnerText); } if (italic) { pText.Append("</i>"); } break; default: pText.Append(innerNode.InnerText); break; } } string start = node.Attributes["begin"].InnerText; string text = pText.ToString(); text = text.Replace(Environment.NewLine + "</i>", "</i>" + Environment.NewLine); text = text.Replace("<i></i>", string.Empty); if (node.Attributes["end"] != null) { string end = node.Attributes["end"].InnerText; subtitle.Paragraphs.Add(new Paragraph(TimedText10.GetTimeCode(start, false), TimedText10.GetTimeCode(end, false), text)); } else if (node.Attributes["dur"] != null) { TimeCode duration = TimedText10.GetTimeCode(node.Attributes["dur"].InnerText, false); TimeCode startTime = TimedText10.GetTimeCode(start, false); var endTime = new TimeCode(startTime.TotalMilliseconds + duration.TotalMilliseconds); subtitle.Paragraphs.Add(new Paragraph(startTime, endTime, text)); } } catch (Exception ex) { System.Diagnostics.Debug.WriteLine(ex.Message); _errorCount++; } } bool allBelow100 = true; foreach (Paragraph p in subtitle.Paragraphs) { p.Text = Utilities.RemoveUnneededSpaces(p.Text, null).Replace(Environment.NewLine + Environment.NewLine, Environment.NewLine); if (p.StartTime.Milliseconds >= 100 || p.EndTime.Milliseconds >= 100) { allBelow100 = false; } } if (allBelow100) { foreach (Paragraph p in subtitle.Paragraphs) { p.StartTime.Milliseconds *= 10; p.EndTime.Milliseconds *= 10; } } subtitle.Renumber(); }
public override void LoadSubtitle(Subtitle subtitle, List <string> lines, string fileName) { _errorCount = 0; var sb = new StringBuilder(); lines.ForEach(line => sb.AppendLine(line)); var xml = new XmlDocument { XmlResolver = null }; xml.LoadXml(sb.ToString().RemoveControlCharactersButWhiteSpace().Trim()); var nsmgr = new XmlNamespaceManager(xml.NameTable); nsmgr.AddNamespace("tt", xml.DocumentElement.NamespaceURI); XmlNode body = xml.DocumentElement.SelectSingleNode("//tt:body", nsmgr); if (body == null) { body = xml.DocumentElement.SelectSingleNode("//tt:body", nsmgr); } bool couldBeFrames = true; bool couldBeMillisecondsWithMissingLastDigit = true; foreach (XmlNode node in body.ChildNodes) { try { var pText = new StringBuilder(); foreach (XmlNode innerNode in node.ChildNodes) { if (innerNode.Name == "image" || innerNode.Name.EndsWith(":image", StringComparison.Ordinal)) { var src = innerNode.Attributes["src"]; if (src != null) { pText.Append(src.InnerText); } break; } } string start = null; string end = null; string dur = null; foreach (XmlAttribute attr in node.Attributes) { if (attr.Name.EndsWith("begin", StringComparison.Ordinal)) { start = attr.InnerText; } else if (attr.Name.EndsWith("end", StringComparison.Ordinal)) { end = attr.InnerText; } else if (attr.Name.EndsWith("duration", StringComparison.Ordinal)) { dur = attr.InnerText; } } if (start == null) { continue; } string text = pText.ToString(); text = text.Replace(Environment.NewLine + "</i>", "</i>" + Environment.NewLine); text = text.Replace("<i></i>", string.Empty).Trim(); if (end != null) { if (end.Length != 11 || end.Substring(8, 1) != ":" || start == null || start.Length != 11 || start.Substring(8, 1) != ":") { couldBeFrames = false; } if (couldBeMillisecondsWithMissingLastDigit && (end.Length != 11 || start == null || start.Length != 11 || end.Substring(8, 1) != "." || start.Substring(8, 1) != ".")) { couldBeMillisecondsWithMissingLastDigit = false; } double dBegin, dEnd; if (!start.Contains(':') && Utilities.CountTagInText(start, '.') == 1 && !end.Contains(':') && Utilities.CountTagInText(end, '.') == 1 && double.TryParse(start, NumberStyles.Float, CultureInfo.InvariantCulture, out dBegin) && double.TryParse(end, NumberStyles.Float, CultureInfo.InvariantCulture, out dEnd)) { subtitle.Paragraphs.Add(new Paragraph(text, dBegin * TimeCode.BaseUnit, dEnd * TimeCode.BaseUnit)); } else { if (!couldBeFrames) { if (end.Length == 11 && end[8] == '.') // 00:05:36.92 { end += "0"; } else if (end.Length == 10 && end[8] == '.') // 00:05:36.9 { end += "00"; } if (start.Length == 11 && start[8] == '.') // 00:05:36.92 { start += "0"; } else if (start.Length == 10 && start[8] == '.') // 00:05:36.9 { start += "00"; } couldBeMillisecondsWithMissingLastDigit = false; } if (start.Length == 8 && start[2] == ':' && start[5] == ':' && end.Length == 8 && end[2] == ':' && end[5] == ':') { var p = new Paragraph(); var parts = start.Split(SplitCharColon); p.StartTime = new TimeCode(int.Parse(parts[0]), int.Parse(parts[1]), int.Parse(parts[2]), 0); parts = end.Split(SplitCharColon); p.EndTime = new TimeCode(int.Parse(parts[0]), int.Parse(parts[1]), int.Parse(parts[2]), 0); p.Text = text; subtitle.Paragraphs.Add(p); } else { subtitle.Paragraphs.Add(new Paragraph(TimedText10.GetTimeCode(start, false), TimedText10.GetTimeCode(end, false), text)); } } } else if (dur != null) { if (dur.Length != 11 || dur.Substring(8, 1) != ":" || start == null || start.Length != 11 || start.Substring(8, 1) != ":") { couldBeFrames = false; } if (couldBeMillisecondsWithMissingLastDigit && (dur.Length != 11 || start == null || start.Length != 11 || dur.Substring(8, 1) != "." || start.Substring(8, 1) != ".")) { couldBeMillisecondsWithMissingLastDigit = false; } TimeCode duration = TimedText10.GetTimeCode(dur, false); TimeCode startTime = TimedText10.GetTimeCode(start, false); var endTime = new TimeCode(startTime.TotalMilliseconds + duration.TotalMilliseconds); subtitle.Paragraphs.Add(new Paragraph(startTime, endTime, text)); } } catch (Exception ex) { System.Diagnostics.Debug.WriteLine(ex.Message); _errorCount++; } } subtitle.RemoveEmptyLines(); if (couldBeFrames) { bool all30OrBelow = true; foreach (Paragraph p in subtitle.Paragraphs) { if (p.StartTime.Milliseconds > 30 || p.EndTime.Milliseconds > 30) { all30OrBelow = false; } } if (all30OrBelow) { foreach (Paragraph p in subtitle.Paragraphs) { p.StartTime.Milliseconds = FramesToMillisecondsMax999(p.StartTime.Milliseconds); p.EndTime.Milliseconds = FramesToMillisecondsMax999(p.EndTime.Milliseconds); } } } else if (couldBeMillisecondsWithMissingLastDigit) { foreach (Paragraph p in subtitle.Paragraphs) { p.StartTime.Milliseconds *= 10; p.EndTime.Milliseconds *= 10; } } subtitle.Renumber(); }