// Methods -------------------------------------------------------------
public List<SubtitleItem> ParseStream(Stream subStream, Encoding encoding)
{
// seek the beginning of the stream
subStream.Position = 0;
var reader = new StreamReader(subStream, encoding, true);
var firstLine = reader.ReadLine();
if (firstLine == FirstLine)
{
var line = reader.ReadLine();
var lineNumber = 2;
while (line != null && lineNumber <= MaxLineNumberForItems && !IsTimestampLine(line))
{
line = reader.ReadLine();
lineNumber++;
}
if (line != null && lineNumber <= MaxLineNumberForItems && IsTimestampLine(line))
{
// we parse all the lines
var items = new List<SubtitleItem>();
var timeCodeLine = line;
var textLine = reader.ReadLine();
var separatorLine = reader.ReadLine();
while (IsTimestampLine(timeCodeLine) && !string.IsNullOrEmpty(textLine))
{
// parse current line
var timeCodes = ParseTimecodeLine(timeCodeLine);
var lines = new List<string>(textLine.Split(_lineSeparator, StringSplitOptions.RemoveEmptyEntries));
var item = new SubtitleItem()
{
StartTime = timeCodes.Item1,
EndTime = timeCodes.Item2,
Lines = lines
};
items.Add(item);
// read next lines
timeCodeLine = reader.ReadLine();
textLine = reader.ReadLine();
separatorLine = reader.ReadLine();
}
return items;
}
else
{
var message = string.Format("Couldn't find the first timestamp line in the current sub file. " +
"Last line read: '{0}', line number #{1}", line, lineNumber);
throw new ArgumentException(message);
}
}
else
{
throw new ArgumentException("Wrong subtitles format for the SubViewerSubParser");
}
}
// Methods ------------------------------------------------------------------
public List<SubtitleItem> ParseStream(Stream ssaStream, Encoding encoding)
{
// test if stream if readable and seekable (just a check, should be good)
if (!ssaStream.CanRead || !ssaStream.CanSeek)
{
var message = string.Format("Stream must be seekable and readable in a subtitles parser. " +
"Operation interrupted; isSeekable: {0} - isReadable: {1}",
ssaStream.CanSeek, ssaStream.CanSeek);
throw new ArgumentException(message);
}
// seek the beginning of the stream
ssaStream.Position = 0;
var reader = new StreamReader(ssaStream, encoding, true);
var line = reader.ReadLine();
if (line == FirstLine)
{
var lineNumber = 1;
// read the line until the [Events] section
while (line != null && line != EventLine && lineNumber <= MaxLineNumberForEventSection)
{
line = reader.ReadLine();
lineNumber++;
}
if (line != null && lineNumber <= MaxLineNumberForEventSection)
{
// we are at the event section
var headerLine = reader.ReadLine();
if (!string.IsNullOrEmpty(headerLine))
{
var columnHeaders = headerLine.Split(Separator).Select(head => head.Trim()).ToList();
var startIndexColumn = columnHeaders.IndexOf(StartColumn);
var endIndexColumn = columnHeaders.IndexOf(EndColumn);
var textIndexColumn = columnHeaders.IndexOf(TextColumn);
if (startIndexColumn > 0 && endIndexColumn > 0 && textIndexColumn > 0)
{
var items = new List<SubtitleItem>();
line = reader.ReadLine();
while (line != null)
{
var columns = line.Split(Separator);
var startText = columns[startIndexColumn];
var endText = columns[endIndexColumn];
var textLine = columns[textIndexColumn];
var start = ParseSsaTimecode(startText);
var end = ParseSsaTimecode(endText);
// TODO: split text line?
var item = new SubtitleItem()
{
StartTime = start,
EndTime = end,
Lines = new List<string>() {textLine}
};
items.Add(item);
line = reader.ReadLine();
}
return items;
}
else
{
var message = string.Format("Couldn't find all the necessary columns " +
"headers ({0}, {1}, {2}) in header line {3}",
StartColumn, EndColumn, TextColumn, headerLine);
throw new ArgumentException(message);
}
}
else
{
var message = string.Format("The header line after the line '{0}' was null -> " +
"no need to continue parsing", line);
throw new ArgumentException(message);
}
}
else
{
var message = string.Format("We reached line '{0}' with line number #{1} without finding to " +
"Event section ({2})", line, lineNumber, EventLine);
throw new ArgumentException(message);
}
}
else
{
var message = string.Format("First subtitle line '{0}' was different from the one expected " +
"for a SSA file ({1}). We stop the parsing here.", line, FirstLine);
throw new ArgumentException(message);
}
}
/// <summary>
/// Parses one line of the .sub file
///
/// ex:
/// {0}{180}PIRATES OF THE CARIBBEAN|English subtitlez by tHe.b0dY
/// </summary>
/// <param name="line">The .sub file line</param>
/// <param name="frameRate">The frame rate with which the .sub file was created</param>
/// <returns>The corresponding SubtitleItem</returns>
private SubtitleItem ParseLine(string line, float frameRate)
{
const string regex = @"^[{\[](\d+)[}\]][{\[](\d+)[}\]](.*)";
var match = Regex.Match(line, regex);
if (match.Success && match.Groups.Count > 2)
{
var startFrame = match.Groups[1].Value;
var start = (int)(1000 * double.Parse(startFrame) / frameRate);
var endTime = match.Groups[2].Value;
var end = (int)(1000 * double.Parse(endTime) / frameRate);
var text = match.Groups[match.Groups.Count - 1].Value;
var lines = text.Split(_lineSeparators);
var nonEmptyLines = lines.Where(l => !string.IsNullOrEmpty(l)).ToList();
var item = new SubtitleItem
{
Lines = nonEmptyLines,
StartTime = start,
EndTime = end
};
return item;
}
else
{
var message = string.Format("The subtitle file line {0} is " +
"not in the .sub format. We stop the process.", line);
Console.WriteLine(message);
throw new InvalidDataException(message);
}
}
