/// <summary>
/// Loads full-context label or mono align label from master label file.
/// </summary>
/// <param name="masterLabelFile">The file name of the master label file.</param>
public void Load(string masterLabelFile)
{
using (StreamReader reader = new StreamReader(masterLabelFile))
{
// This is the header of master label file.
string line = reader.ReadLine();
if (line != MasterLabelFileHeader)
{
throw new InvalidDataException(Helper.NeutralFormat("Master label file header expected, but input \"{0}\"", line));
}
while (!reader.EndOfStream)
{
// Read the line for sentence id.
line = reader.ReadLine();
Match match = RegexOfSentId.Match(line);
if (match.Success)
{
bool endOfSentenceExist;
string sentId = match.Groups[1].Value;
if (_idKeyedSentences.ContainsKey(sentId))
{
// Not the first time to load the sentence.
endOfSentenceExist = _idKeyedSentences[sentId].Load(reader);
}
else
{
// Load the sentence in first time.
Sentence sentence = new Sentence { Id = sentId, TrainingSet = this };
endOfSentenceExist = sentence.Load(reader);
_idKeyedSentences[sentId] = sentence;
}
if (!endOfSentenceExist)
{
throw new InvalidDataException("Sentence end is expected");
}
}
else
{
throw new InvalidDataException(Helper.NeutralFormat("Sentence id expected, but input \"{0}\"", line));
}
}
}
}
/// <summary>
/// Converts label files to master label file.
/// </summary>
/// <param name="alignmentDir">The directory of alignment files.</param>
/// <param name="mlfFileName">The name of target master label file.</param>
public static void ConvertLabelFilesToMlf(string alignmentDir, string mlfFileName)
{
TrainingSentenceSet set = new TrainingSentenceSet();
foreach (string labelFile in Directory.GetFiles(alignmentDir, Helper.NeutralFormat("*.{0}", FileExtensions.LabelFile)))
{
Sentence sentence = new Sentence();
using (StreamReader sr = new StreamReader(labelFile))
{
if (sentence.Load(sr))
{
throw new InvalidDataException("Sentence end is not expected");
}
}
string id = Path.GetFileNameWithoutExtension(labelFile);
set.Sentences.Add(id, sentence);
}
set.Save(mlfFileName, LabelTypeOptions.FullContext, LabelAlignOptions.StateAlign, true);
}