public Normalizer(RawDocument document, string format)
{
// Construimos un StyleReader para obtener las regexp.
StyleReader style = new StyleReader (format);
this.format = format;
rules = style.GetRules ();
// Si el estilo tiene mas de una columna se rompe y se convierte a una
// sola columna.
if (style.GetNumColumns () > 1)
document.BreakColumns ();
EncodeText (document.GetText ());
}
public void GetText()
{
string rawtext;
int count = 0;
foreach (PDFPoppler doc in test_docs) {
RawDocument rdoc = new RawDocument (doc);
rawtext = rdoc.GetText ();
Assert.AreEqual (raw_docs[count], rawtext, "GT" + count);
count += 1;
}
}
private void OnOpenActivated(object sender, System.EventArgs e)
{
OpenPDFDialog dialog = new OpenPDFDialog ();
if (dialog.Run () == (int) ResponseType.Ok) {
Uri uri = new Uri (dialog.Document);
PDFPoppler reader = new PDFPoppler (uri);
//Extracting images from document
reader.GetNonText ();
//Extracting text from document
rdocument = reader.CreateRawDocument ();
textview.Buffer.Text = rdocument.GetText ();
Markup.Sensitive = true;
Normalize.Sensitive = true;
store.Clear ();
// Logger.ClearList ();
}
dialog.Destroy ();
}