//This method changes the Header text when necessary and loading the other pdf in the pdfviewer
//TODO: this method should unload the documents from memory, will be fixed in the future
private void Change_Frame(object sender, RoutedEventArgs e)
{
if (orologio.Text.Equals("Ωρολόγιο Πρόγραμμα"))
{
orologio.Text = "Πρόγραμμα Εξεταστικής";
combost.Visibility = Visibility.Collapsed;
datepick.Visibility = Visibility.Visible;
cb1.SelectedIndex = 0;
cb2.SelectedIndex = 0;
pdfViewer.GotoPage(1);
// Load the Byte array
PdfLoadedDocument loadedDocument = new PdfLoadedDocument(contentBytesx2);
// Display the PDF document in PdfViewer
pdfViewer.LoadDocument(loadedDocument);
}
else
{
//closing
orologio.Text = "Ωρολόγιο Πρόγραμμα";
datepick.Visibility = Visibility.Collapsed;
combost.Visibility = Visibility.Visible;
exetastiki.Date = null;
pdfViewer.GotoPage(1);
// Load the Byte array
PdfLoadedDocument loadedDocument = new PdfLoadedDocument(contentBytesx1);
// Display the PDF document in PdfViewer
pdfViewer.LoadDocument(loadedDocument);
loadedDocument.Dispose();
}
}
private void CreateTxtFromPDF(string filename)
{
//ScanPageStatus scanPageStatus = new ScanPageStatus();
string tesseractPath = Path.Combine(AssemblyDirectory(), GetConstants.TesseractBinaries());
string tesseractData = Path.Combine(AssemblyDirectory(), GetConstants.TesseractData());
try {
using (OCRProcessor processor = new OCRProcessor(tesseractPath))
{
//Stream pdfStream2 = filename; // FileUpload1.PostedFile.InputStream;
// Read in PDF image file, and convert to searchable TXT pdf file
PdfLoadedDocument IDoc = new PdfLoadedDocument(filename);
processor.Settings.Language = Languages.English;
processor.Settings.Performance = Performance.Slow;
// var zz = processor.Settings.Performance;
//string tessdata = tesseractPath + @"\\Tessdata\\";
processor.PerformOCR(IDoc, tesseractData);
string outFileName = Path.GetFileName(filename) + "_OCR" + Path.GetExtension(filename);
string homePath = Path.GetDirectoryName(Path.GetDirectoryName(filename));
string savePath = Path.Combine(homePath, GetConstants.Directory("out"), outFileName);
// If file exists - delete it first.
if (File.Exists(savePath))
{
File.SetAttributes(savePath, FileAttributes.Normal);
File.Delete(savePath);
}
IDoc.Save(savePath);
IDoc.Close(true);
IDoc.Dispose();
scanPageStatus.scannedFileName = savePath;
scanPageStatus.rc = 0;
scanPageStatus.statusMessage = String.Format("File {0} scanned and saved to {1}", filename, scanPageStatus.scannedFileName);
}
}
catch (Exception ex)
{
scanPageStatus.scannedFileName = "";
scanPageStatus.statusMessage = String.Format("Error {0} when running OCR on source file {1}", ex, filename);
scanPageStatus.rc = -1;
}
// return scanPageStatus;
}
public static async void SaveChangesFile(string text)
{
PdfLoadedDocument loadedDocument = new PdfLoadedDocument();
await loadedDocument.OpenAsync(openFile).ConfigureAwait(true);
loadedDocument.Pages.RemoveAt(0);
Syncfusion.Pdf.PdfPage page = (Syncfusion.Pdf.PdfPage)loadedDocument.Pages.Add();
PdfGraphics graphics = page.Graphics;
Syncfusion.Pdf.Graphics.PdfFont font = new PdfStandardFont(PdfFontFamily.Helvetica, 20);
graphics.DrawString(text, font, PdfBrushes.Black, new PointF(0, 0));
await loadedDocument.Save().ConfigureAwait(true);
loadedDocument.Dispose();
}
public static async Task <string> Read(StorageFile openFile)
{
SaveFilePdf.openFile = openFile;
string extractedText = "";
if (openFile != null)
{
PdfLoadedDocument loadedDocument = new PdfLoadedDocument();
await loadedDocument.OpenAsync(openFile).ConfigureAwait(true);
PdfPageBase page = loadedDocument.Pages[0];
extractedText = page.ExtractText();
loadedDocument.Close(true);
loadedDocument.Dispose();
var mru = Windows.Storage.AccessCache.StorageApplicationPermissions.MostRecentlyUsedList;
string mruToken = mru.Add(openFile, "Pdf file");
}
return(extractedText);
}
public string SaveRightMovePdf(RightMoveModel rightMoveModel)
{
try
{
PdfLoadedDocument loadedDocument = new PdfLoadedDocument(new FileStream(templatePath, FileMode.Open));
if (!Directory.Exists(archiveFolder))
{
Directory.CreateDirectory(archiveFolder);
}
if (loadedDocument.PageCount > 0)
{
PdfLoadedPage pdfLoadedPage = loadedDocument.Pages[1] as PdfLoadedPage;
PdfTemplate pdfTemplate = new PdfTemplate(900, 600);
PdfFont pdfFont = new PdfStandardFont(PdfFontFamily.Helvetica, 15);
PdfBrush brush = new PdfSolidBrush(SfDrawing.Color.Black);
byte[] imageBytes = new WebClient().DownloadData(rightMoveModel.PropertyMainPicture);
Stream imageStream = new MemoryStream(imageBytes);
pdfTemplate.Graphics.DrawString($"Property Address: {rightMoveModel.PropertyAddress}", pdfFont, brush, 100, 30);
pdfTemplate.Graphics.DrawString($"Property Type: {rightMoveModel.PropertyType}", pdfFont, brush, 100, 50);
pdfTemplate.Graphics.DrawString($"PropertyPrice: {rightMoveModel.PropertyPrice} ", pdfFont, brush, 100, 70);
pdfTemplate.Graphics.DrawImage(PdfImage.FromStream(imageStream), new SfDrawing.PointF(100, 100), new SfDrawing.SizeF(400, 400));
pdfLoadedPage.Graphics.DrawPdfTemplate(pdfTemplate, SfDrawing.PointF.Empty);
string rawName = rightMoveModel
.PropertyUrl
.Replace("/", "")
.Replace("-", "")
.Replace(".", "")
.Replace(":", "")
.Replace("//", "");
string fileName = Regex.Match(rawName, @"(\d+(?:\.\d{1,2})?)").Value;
PdfDocument propertyHeatMapPdfDocument = htmlConverter.Convert(rightMoveModel.PropertyHeatHtmlString, string.Empty);
PdfDocument homeCoUKHtmlPdfDocument = htmlConverter.Convert(rightMoveModel.HomeCoUKHtmlString, string.Empty);
string tempPropertyHeatMap = Path.Combine(tempFolder, $"propertyHeatMap{fileName}.pdf");
using (FileStream propertyHeatMapStream = new FileStream(tempPropertyHeatMap, FileMode.Create))
{
propertyHeatMapPdfDocument.Save(propertyHeatMapStream);
propertyHeatMapPdfDocument.Close(true);
propertyHeatMapPdfDocument.Dispose();
propertyHeatMapStream.Close();
propertyHeatMapStream.Dispose();
}
string tempHomeCoUK = Path.Combine(tempFolder, $"homeCoUK{fileName}.pdf");
using (FileStream homeCoUKHtmlPdfStream = new FileStream(tempHomeCoUK, FileMode.Create))
{
homeCoUKHtmlPdfDocument.Save(homeCoUKHtmlPdfStream);
homeCoUKHtmlPdfDocument.Close(true);
homeCoUKHtmlPdfDocument.Dispose();
homeCoUKHtmlPdfStream.Close();
homeCoUKHtmlPdfStream.Dispose();
}
using (FileStream propertyHeatMapReadStream = new FileStream(tempPropertyHeatMap, FileMode.Open))
{
PdfLoadedDocument tempPropertyHeatMapDocument = new PdfLoadedDocument(propertyHeatMapReadStream);
loadedDocument.ImportPage(tempPropertyHeatMapDocument, 0);
propertyHeatMapReadStream.Close();
propertyHeatMapReadStream.Dispose();
}
using (FileStream homeCoUKReadStream = new FileStream(tempHomeCoUK, FileMode.Open))
{
PdfLoadedDocument tempHomeCoUKDocument = new PdfLoadedDocument(homeCoUKReadStream);
loadedDocument.ImportPage(tempHomeCoUKDocument, 0);
homeCoUKReadStream.Close();
homeCoUKReadStream.Dispose();
}
string savePath = Path.Combine(archiveFolder, $"{fileName}.pdf");
using (FileStream saveStream = new FileStream(savePath, FileMode.Create))
{
loadedDocument.Save(saveStream);
loadedDocument.Close(true);
loadedDocument.Dispose();
saveStream.Close();
saveStream.Dispose();
}
return($"file successfully saved at: {savePath}");
}
else
{
Console.WriteLine("Invalid PDF file");
return($"Invalid PDF file");
}
}
catch (Exception ex)
{
Console.WriteLine($"Unable to save the file. {ex.Message}");
return($"Unable to save the file");
}
}
private async void Extract_Click(object sender, RoutedEventArgs e)
{
if (_cts != null)
{
_cts.Cancel();
return;
}
try
{
_cts = new CancellationTokenSource();
ChangeEnability(false);
var source = SourceTextBox.Text;
var destination = DestinationTextBox.Text;
if (string.IsNullOrWhiteSpace(source) || !Directory.Exists(source))
{
MessageBox.Show("Please specify a valid source folder!");
return;
}
else if (string.IsNullOrWhiteSpace(destination))
{
MessageBox.Show("Please specify destination folder!");
return;
}
var skipDuplicateFiles = SkipRadioButton.IsChecked == true;
if (!Directory.Exists(source))
{
MessageBox.Show("This folder doesn't exist!");
return;
}
var pdfs = Directory.EnumerateFiles(source, "*.pdf", SearchOption.AllDirectories).ToList();
progressBar.Value = 0;
int processed = 0;
statusTextBlock.Text = "Starting...";
await Task.Run(async() =>
{
try
{
foreach (var pdf in pdfs)
{
_cts.Token.ThrowIfCancellationRequested();
var name = pdf.Split('\\', '/').Last();
var destinationFile = Path.Combine(destination, name + ".txt");
if (skipDuplicateFiles && File.Exists(destinationFile))
{
continue;
}
var builder = new StringBuilder();
PdfLoadedDocument loadedDocument = new PdfLoadedDocument(pdf);
foreach (PdfLoadedPage page in loadedDocument.Pages)
{
builder.AppendLine(page.ExtractText());
}
loadedDocument.Dispose();
var document = builder.ToString();
File.WriteAllText(Path.Combine(destination, name + ".txt"), document);
Interlocked.Increment(ref processed);
await Dispatcher.InvokeAsync(() =>
{
progressBar.Value = (double)processed / pdfs.Count;
statusTextBlock.Text = $"{processed} / {pdfs.Count} ({progressBar.Value * 100:N2}%)";
});
}
}
catch (OperationCanceledException)
{
}
});
}
catch (OperationCanceledException)
{
}
catch (Exception ex)
{
MessageBox.Show(ex.ToString());
}
finally
{
ChangeEnability(true);
statusTextBlock.Text = "";
progressBar.Value = 0;
_cts = null;
}
}
