public static List <Sentence> GetSentencesFromSMI(string fileName, Encoding enc)
{
List <Sentence> sents = new List <Sentence>();
using (TextReader tr = new StreamReader(fileName, enc)) //, true))
{
double frameStart = 0;
SentenceVideo sentPrev = null;
bool isBody = false;
string line = "";
while ((line = tr.ReadLine()) != null)
{
if (!isBody)
{
if (line.Contains("<BODY>"))
{
isBody = true;
}
continue;
}
else
if (line.Contains("</BODY>"))
{
break;
}
if (line.Contains("<SYNC"))
{
int i; // <SYNC Start=67860><P Class=ENCC>
if (int.TryParse(line.Split(new string[] { "Start=", ">" }, StringSplitOptions.RemoveEmptyEntries)[1].Trim(), out i))
{
frameStart = i / 1000;
if (sentPrev != null)
{
sentPrev.InitLength(frameStart - sentPrev.Start);
}
}
else
{
System.Diagnostics.Debug.Fail(string.Format("Error on parsing line '{0}'", line));
}
}
else
{
line = line.Replace("<br>", Environment.NewLine);
sentPrev = new SentenceVideo(line, sents, frameStart, frameStart + DefaultSubtitleLength);
sents.Add(sentPrev);
}
}
// в SMI нет конца, поэтому после цикла последнее предложение остается с длиной по умолчанию
//if (sentPrev != null)
// sentPrev.InitLength(sentPrev.End - sentPrev.Start);
}
return(sents);
}
public static List <Sentence> GetSentencesFromSSA(string fileName, Encoding enc)
{
List <Sentence> sents = new List <Sentence>();
using (TextReader tr = new StreamReader(fileName, enc)) //, true))
{
bool isBody = false;
bool isFormat = false;
string line;
SentenceVideo sent = null;
while ((line = tr.ReadLine()) != null)
{
if (!isBody || !isFormat)
{
if (line.Contains("[Events]"))
{
isBody = true;
}
if (isBody && line.Contains("Format:"))
{
isFormat = true;
}
continue;
}
// do split Dialogue: Marked=0,1:55:09.19,1:55:12.94,StyleB,NTP,0000,0000,0000,!Effect,...и я не чувствую ничего,\Nкроме благодарности...
string[] parts = line.Split(new string[] { "=0,", ",", "Effect," }, StringSplitOptions.RemoveEmptyEntries);
string frameStartRaw = parts[1].Replace(".", ","); // 1:55:09.19 => 1:55:09,19
string frameEndRaw = parts[2].Replace(".", ",");
double frameStart = ConvertTimeSpanFromString(frameStartRaw);
double frameEnd = ConvertTimeSpanFromString(frameEndRaw);
string text = parts[parts.Length - 1].Replace("\\N", Environment.NewLine);
sent = new SentenceVideo(text, sents, frameStart, frameEnd);
sent.InitLength(sent.End - sent.Start);
sents.Add(sent);
}
}
return(sents);
}
public static List <Sentence> GetSentencesFromTXT(string fileName)
{
List <Sentence> sents = new List <Sentence>();
using (TextReader tr = new StreamReader(fileName, Encoding.Default)) //, true))
{
string line;
SentenceVideo sent = null;
while ((line = tr.ReadLine()) != null)
{
if (sents.Count == 0 && line.StartsWith("["))
{
continue; // for [INFORMATION][AUTHOR] .... [COLF]&HFFFFFF,[STYLE]no,[SIZE]18,[FONT]Arial
}
string[] parts = line.Split(new char[] { '{', '}' }, StringSplitOptions.RemoveEmptyEntries);
double frameStart; // = Convert.ToDouble(parts[0]);
if (!Double.TryParse(parts[0], out frameStart))
{
throw new ApplicationException(string.Format("Error in file '{0}'", fileName));
}
frameStart = Math.Round(frameStart / 25, 3);
if (sent != null)
{
sent.InitLength(frameStart - sent.Start);
}
double frameEnd = Math.Round(Convert.ToDouble(parts[1]) / 25, 3);
//double frameEnd = Convert.ToDouble(parts[1]) / 25;
sent = new SentenceVideo(parts[2].Replace("|", Environment.NewLine), sents, frameStart, frameEnd);
sents.Add(sent);
}
if (sent != null)
{
sent.InitLength(sent.End - sent.Start);
}
}
return(sents);
}
public static List <Sentence> GetSentencesFromSRT(string fileName, Encoding enc)
{
string[] timeDelimiters = new string[] { "-->", "," };
List <Sentence> sents = new List <Sentence>();
using (TextReader tr = new StreamReader(fileName, enc, true)) //, true))
{
bool isNumber = true;
double frameStart = 0;
double frameEnd = 0;
string line, lineStore = "";
SentenceVideo sentPrev = null;
//while ((line = tr.ReadLine()) != null)
while (true)
{
line = tr.ReadLine();
if (line == null)
{
break;
}
if (sents.Count == 0 && line.StartsWith("["))
{
continue; // for [INFORMATION][AUTHOR] .... [COLF]&HFFFFFF,[STYLE]no,[SIZE]18,[FONT]Arial
}
if (isNumber && line != null && (line.Contains(timeDelimiters[0]) || line.Contains(timeDelimiters[1])))
{
string timeDelimiter = timeDelimiters[0]; // frequently Delimiter == "-->"
if (!line.Contains(timeDelimiters[0]))
{
timeDelimiter = timeDelimiters[1];
}
string[] times = line.Split(new string[] { timeDelimiter }, StringSplitOptions.RemoveEmptyEntries);
frameStart = ConvertTimeSpanFromString(times[0]);
frameEnd = ConvertTimeSpanFromString(times[1]);
isNumber = false;
continue;
}
if (isNumber && line != null)
{
continue;
}
//if (line == null && lineStore == null)
// break;
if (string.IsNullOrEmpty(line))
{
if (sentPrev != null)
{
sentPrev.InitLength(frameStart - sentPrev.Start);
// Console.WriteLine("length - " + sent.Length.ToString());
}
sentPrev = new SentenceVideo(lineStore, sents, frameStart, frameEnd);
sents.Add(sentPrev);
// Console.WriteLine(sents.Count.ToString() + ") " + lineStore + " frame: " + frameStart.ToString());
lineStore = null;
isNumber = true;
}
else
{
if (sents.Count == 0 && line.StartsWith("["))
{
continue; // for [INFORMATION][AUTHOR] .... [COLF]&HFFFFFF,[STYLE]no,[SIZE]18,[FONT]Arial
}
lineStore += Environment.NewLine + " " + line;
//lineStore += " " + line;
}
}
if (sentPrev != null)
{
sentPrev.InitLength(sentPrev.End - sentPrev.Start);
}
}
return(sents);
}