private void getTbSentences_Click(object sender, EventArgs e)
{
ClearList();
string mrgFile = null;
try
{
mrgFile = mrgFileCombo.Text;
mrgFile = mrgFile.Substring(0, mrgFile.IndexOf(' '));
mrgFile = _treeBankEngine.GetFullMrgPath(mrgFile);
}
catch (Exception)
{
MessageBox.Show("Invalid MRG file");
return;
}
try
{
foreach (int sentNum in _treeBankEngine.GetSentenceNumbers(mrgFile))
{
AppendOutput(_treeBankEngine.GetParseTree(mrgFile, sentNum));
}
}
catch (Exception ex)
{
MessageBox.Show(ex.ToString());
}
}
/// <summary>
/// Gets the next training node
/// </summary>
/// <returns>Training node</returns>
public override ClassifiableEntity GetNextInstance()
{
// try to move to next node in current sentence
while (!_nodeEnum.MoveNext())
{
// try to move to next sentence in current MRG file
while (!_sentEnum.MoveNext())
{
// try to move to next MRG file...if there are none, we're done
if (!MoveToNextValidMrgFile())
{
return(null);
}
// start at first sentence of next MRG file
_sentEnum = TreeBankEngine.GetSentenceNumbers(CurrentMrgFile).GetEnumerator();
}
// filter all nodes in the tree, keeping the good ones
TreeBankNode root = GetTrainingInstanceParseTree(CurrentMrgFile, _sentEnum.Current);
List <TreeBankNode> filteredNodes = new List <TreeBankNode>();
foreach (TreeBankNode n in root.AllNodes)
{
if (Filter(n))
{
filteredNodes.Add(n);
}
}
_nodeEnum = filteredNodes.GetEnumerator();
}
return(_nodeEnum.Current);
}
private void mrgFile_SelectedIndexChanged(object sender, EventArgs e)
{
if (mrgFile.SelectedItem != null)
{
sentence.Items.Clear();
string mrgFilePath = _treeBankEngine.GetFullMrgPath(mrgFile.SelectedItem.ToString() + ".mrg");
foreach (int sentNum in _treeBankEngine.GetSentenceNumbers(mrgFilePath))
{
sentence.Items.Add(sentNum);
}
}
}
private void loadTbBtn_Click(object sender, EventArgs e)
{
string root = Directory.GetDirectoryRoot(".");
_treeBankEngine = new TreeBankEngine(root + @"NLP\Resources\PennTreeBank_3\PARSED\MRG\WSJ",
root + @"NLP\Resources\Indexes\treebank_index");
// populate MRG file combo box
foreach (string mrgPath in _treeBankEngine.IndexedMrgFiles)
{
mrgFileCombo.Items.Add(Path.GetFileName(mrgPath) + " (" + _treeBankEngine.GetSentenceNumbers(mrgPath).Count + " sentences)");
}
treeBankGroupBox.Enabled = true;
loadTbBtn.Enabled = false;
}