/**
* /// Internal routine used when loading Lattices from .LAT files
*
* /// @param lattice
* /// @param tokens
*/
public static void Load(Lattice lattice, StringTokenizer tokens)
{
string id = tokens.nextToken();
string label = tokens.nextToken();
lattice.AddNode(id, label, 0, 0);
}
public static Lattice ReadSlf(String fileName)
{
var lattice = new Lattice();
var @in = new LineNumberReader(new StreamReader(fileName));
String line;
var readingNodes = false;
var readingEdges = false;
var startIdx = 0;
var endIdx = 1;
var lmscale = 9.5;
while ((line = @in.ReadLine()) != null)
{
if (line.Contains("Node definitions"))
{
readingEdges = false;
readingNodes = true;
continue;
}
if (line.Contains("Link definitions"))
{
readingEdges = true;
readingNodes = false;
continue;
}
if (line.StartsWith("#"))
{
//skip commented line
continue;
}
if (readingNodes)
{
//reading node info, format:
//I=id t=start_time_sec W=word_transcription
var parts = line.Split("\\s+");
if (parts.Length != 3 || !parts[0].StartsWith("I=") || !parts[1].StartsWith("t=") || !parts[2].StartsWith("W="))
{
@in.Close();
throw new IOException("Unknown node definition: " + line);
}
var idx = Convert.ToInt32(parts[0].Substring(2), CultureInfo.InvariantCulture.NumberFormat);
//convert to milliseconds inplace
var beginTime = (long)(Convert.ToDouble(parts[1].Substring(2), CultureInfo.InvariantCulture.NumberFormat) * 1000);
var wordStr = parts[2].Substring(2);
var isFiller = false;
if (idx == startIdx || wordStr.Equals("!ENTER"))
{
wordStr = "<s>";
isFiller = true;
}
if (idx == endIdx || wordStr.Equals("!EXIT"))
{
wordStr = "</s>";
isFiller = true;
}
if (wordStr.Equals("!NULL"))
{
wordStr = "<sil>";
isFiller = true;
}
if (wordStr.StartsWith("["))
{
isFiller = true;
}
var word = new Word(wordStr, new Pronunciation[0], isFiller);
var node = lattice.AddNode(idx.ToString(CultureInfo.InvariantCulture), word, beginTime, -1);
if (wordStr.Equals("<s>"))
{
lattice.InitialNode = node;
}
if (wordStr.Equals("</s>"))
{
lattice.TerminalNode = node;
}
}
else if (readingEdges)
{
//reading edge info, format:
//J=id S=from_node E=to_node a=acoustic_score l=language_score
var parts = line.Split("\\s+");
if (parts.Length != 5 || !parts[1].StartsWith("S=") || !parts[2].StartsWith("E=") ||
!parts[3].StartsWith("a=") || !parts[4].StartsWith("l="))
{
@in.Close();
throw new IOException("Unknown edge definition: " + line);
}
var fromId = parts[1].Substring(2);
var toId = parts[2].Substring(2);
var ascore = Convert.ToDouble(parts[3].Substring(2), CultureInfo.InvariantCulture.NumberFormat);
var lscore = Convert.ToDouble(parts[4].Substring(2), CultureInfo.InvariantCulture.NumberFormat) * lmscale;
lattice.AddEdge(lattice.Nodes.Get(fromId), lattice.Nodes.Get(toId), ascore, lscore);
}
else
{
//reading header here if needed
if (line.StartsWith("start="))
{
startIdx = Convert.ToInt32(line.Replace("start=", ""), CultureInfo.InvariantCulture.NumberFormat);
}
if (line.StartsWith("end="))
{
endIdx = Convert.ToInt32(line.Replace("end=", ""), CultureInfo.InvariantCulture.NumberFormat);
}
if (line.StartsWith("lmscale="))
{
lmscale = Convert.ToDouble(line.Replace("lmscale=", ""), CultureInfo.InvariantCulture.NumberFormat);
}
}
}
foreach (var node in lattice.Nodes.Values)
{
//calculate end time of nodes depending successors begin time
foreach (var edge in node.LeavingEdges)
{
if (node.EndTime < 0 || node.EndTime > edge.ToNode.BeginTime)
{
node.EndTime = edge.ToNode.BeginTime;
}
}
}
@in.Close();
return(lattice);
}