private void button4_Click(object sender, EventArgs e)
{
DateTime startTime = DateTime.Now;
PreprocessingUtility.SumWords(folderTextBox.Text, PreprocessingConsts.CategoryFilePattern, folderTextBox.Text + "\\summary.all");
MessageBox.Show("All done in:" + (DateTime.Now.Subtract(startTime)).ToString());
}
private void button14_Click(object sender, EventArgs e)
{
DocumentClass.LoadFromFiles(folderTextBox.Text, PreprocessingConsts.CategoryFilePattern);
CtfIdfDictionary dictionary = new CtfIdfDictionary(folderTextBox.Text, folderTextBox.Text + "\\" + PreprocessingConsts.SummaryFileName, 1000);
String summaryFilePath = Application.StartupPath + "\\Preprocessing\\" + PreprocessingConsts.SummaryFileName;
LearningDocInfo learningDocInfo = new LearningDocInfo(folderTextBox.Text, summaryFilePath);
PreprocessingUtility.CreateLearningDocumentList(folderTextBox.Text, dictionary, DocumentRepresentationType.TfIdf, learningDocInfo);
MessageBox.Show("Done");
}
private void button10_Click(object sender, EventArgs e)
{
folderBrowserDialog1.SelectedPath = Application.StartupPath;
if (folderBrowserDialog1.ShowDialog() == DialogResult.OK)
{
Dictionary <int, String> stopWords = PreprocessingUtility.LoadStopWords("Preprocessing\\stopwords.txt");
DirectoryInfo rootDirInfo = new DirectoryInfo(folderBrowserDialog1.SelectedPath);
DateTime startTime = DateTime.Now;
foreach (DirectoryInfo sourceDirInfo in rootDirInfo.GetDirectories())
{
PreprocessingUtility.StemDir(sourceDirInfo.FullName, stopWords, null);
PreprocessingUtility.SumWords(sourceDirInfo.FullName + "\\stem\\", PreprocessingConsts.StemmedFilePattern, rootDirInfo + "\\" + sourceDirInfo.Name + PreprocessingConsts.CategoryFileExtension);
}
//sumowanie kategorii
PreprocessingUtility.SumWords(folderBrowserDialog1.SelectedPath, PreprocessingConsts.CategoryFilePattern, rootDirInfo + "\\" + PreprocessingConsts.SummaryFileName);
MessageBox.Show("All done in:" + (DateTime.Now.Subtract(startTime)).ToString());
}
}
private void button2_Click(object sender, EventArgs e)
{
Dictionary <int, String> stopWords = PreprocessingUtility.LoadStopWords("Preprocessing\\stopwords.txt");
DateTime startTime = DateTime.Now;
/*
* DirectoryInfo sourceDir = new DirectoryInfo(folderTextBox.Text);
* DirectoryInfo destDir = Directory.CreateDirectory(sourceDir.FullName + "\\stem");
* String destFile;
* DateTime startTime = DateTime.Now;
* foreach (FileInfo sourceFile in sourceDir.GetFiles())
* {
* Console.WriteLine("Processing file: " + sourceFile);
* destFile = sourceFile.Name + ".stm";
* FileConverter.StemFile(sourceFile.FullName,destDir.FullName + "\\" + destFile, stopWords);
* }
*/
PreprocessingUtility.StemDir(folderTextBox.Text, stopWords, null);
MessageBox.Show("All done in:" + (DateTime.Now.Subtract(startTime)).ToString());
}
/// <summary>
/// Stemuje wszystkie pliki w danym katalogu. Nie uwzglêdnia podkatalogów.
/// Pliki po stemmingu s¹ zapisywane do podkatalogu "stem", który zostaje utworzony.
/// Wynikowy plk ma nazwê pliku wejœciowego i rozszerzenie ".stm".
/// </summary>
/// <param name="sourceDir">Katalog zawieraj¹ce pliki wejœciowe.</param>
/// <param name="stopWords">Lista "stop words"</param>
public static void StemDir(String sourceDir, Dictionary <int, String> stopWords, BackgroundWorker bWorker)
{
DirectoryInfo sourceDirInfo = new DirectoryInfo(sourceDir);
DirectoryInfo destDirInfo = Directory.CreateDirectory(sourceDirInfo.FullName + "\\stem");
String destFile;
foreach (FileInfo sourceFile in sourceDirInfo.GetFiles())
{
Console.WriteLine("Processing file: " + sourceFile);
destFile = sourceFile.Name + PreprocessingConsts.StemmedFileExtension;//".stm";
PreprocessingUtility.StemFile(sourceFile.FullName, destDirInfo.FullName + "\\" + destFile, stopWords);
//worer
if (bWorker != null)
{
if (bWorker.CancellationPending)
{
return;
}
bWorker.ReportProgress(1);
}
}
}
