protected override string DoScan(string filename, LoggingSection log)
{
DateTime beginning = DateTime.Now;
try
{
log.Debug($"Extracting text from file");
string text = File.ReadAllText(filename);
log.Debug("Text extracted");
return(text);
}
catch (Exception)
{
throw;
//return new ScanResult(null, filename, this, beginning, DateTime.Now, false, e);
}
}
public CommandHandler()
{
logging = new LoggingSection(this);
commands = new List <ICommand>();
hasControl = false;
logging.Info("Initializing commands");
foreach (Type type in AppDomain.CurrentDomain.GetAssemblies()
.SelectMany(s => s.GetTypes())
.Where(p => !p.IsInterface && typeof(ICommand).IsAssignableFrom(p))
.Where(t => t.GetConstructors().All(c => c.GetParameters().Length == 0)))
{
ICommand cmd = (ICommand)Activator.CreateInstance(type);
logging.Debug($"Initialized command {cmd.Name}");
commands.Add(cmd);
}
logging.Debug($"Initialized {commands.Count} commands");
}
public void TakeControl()
{
if (!hasControl)
{
logging.Debug($"Took control over Thread \"{Thread.CurrentThread.Name}\" and handling console input");
hasControl = true;
while (Program.IsRunning)
{
HandleInput(Console.ReadLine());
}
hasControl = false;
}
}
protected override async Task <string> DoScanAsync(string filename, LoggingSection log)
{
DateTime beginning = DateTime.Now;
try
{
string text;
log.Debug($"Extracting text from file");
using (StreamReader streamReader = File.OpenText(filename))
{
text = await streamReader.ReadToEndAsync();
log.Debug("Text extracted");
}
return(text);
}
catch (Exception)
{
throw;
//return new ScanResult(null, filename, this, beginning, DateTime.Now, false, e);
}
}
public void Log_ReturnedLogMessage_ShouldHaveLogLevelDebug()
{
//Arrange
LoggingSection loggingSection = new LoggingSection(this);
LogMessage logMessage;
//Act
logMessage = loggingSection.Debug("Test");
//Assert
Assert.NotNull(logMessage);
Assert.Equal(LogLevel.DEBUG, logMessage.LogLevel);
}
public Scanners()
{
logging = new LoggingSection(this);
scanners = new List <Scanner>();
logging.Info("Initializing scanners");
foreach (Type type in AppDomain.CurrentDomain.GetAssemblies()
.SelectMany(s => s.GetTypes())
.Where(p => !p.IsInterface && p != typeof(Scanner) && typeof(Scanner).IsAssignableFrom(p))
.Where(t => t.GetConstructors().All(c => c.GetParameters().Length == 0)))
{
Scanner scanner = (Scanner)Activator.CreateInstance(type);
logging.Debug($"Created instance of {type.Name}");
scanners.Add(scanner);
}
}
public static void Disable(IDisposable service)
{
try
{
service.Dispose();
servicesToBeDisposed.Remove(service);
logging.Debug($"{service.GetType().Name + (service.GetType().GetTypeInfo().GenericTypeArguments.Length > 0 ? $"<{string.Join(", ", service.GetType().GenericTypeArguments.Select(type => type.Name))}>" : "")} disposed");
}
catch (Exception e)
{
logging.Error($"" +
$"Unhandled exception while disposing Service {service.GetType().Name}: {e.Message}" +
$"\nSource: {(e.Source != null ? e.Source : "Unknown")}" +
$"\nStackTrace: {e.StackTrace}");
throw;
}
}
protected override string DoScan(string filename, LoggingSection log)
{
PdfDocument pdfDocument = PdfReader.Open(filename);
StringBuilder stringBuilder = new StringBuilder();
for (int pageIndex = 0; pageIndex < pdfDocument.PageCount; pageIndex++)
{
log.Verbose($"Scanning page {pageIndex + 1} of {pdfDocument.PageCount}");
PdfPage pdfPage = pdfDocument.Pages[pageIndex];
//Extract text from text elements
stringBuilder.Append($"{ExtractTextFromPdfPage(pdfPage)}{Environment.NewLine}");
//Extract text from image elements with Tesseract OCR - awesome! :)
PdfDictionary resources = pdfPage.Elements.GetDictionary("/Resources");
if (resources != null)
{
PdfDictionary xObjects = resources.Elements.GetDictionary("/XObject");
if (xObjects != null)
{
ICollection <PdfItem> items = xObjects.Elements.Values;
foreach (PdfItem item in items)
{
PdfReference reference = item as PdfReference;
if (reference != null)
{
PdfDictionary xObject = reference.Value as PdfDictionary;
if (xObject != null && xObject.Elements.GetString("/Subtype") == "/Image")
{
Bitmap bitmap = PdfImageToBitmap(xObject);
if (bitmap == null)
{
log.Error("Could not extract bitmap from PDF image element. Seems like the PDF image filter type is not supported. Skipping element!");
continue;
}
log.Debug("Rotating image");
bitmap.RotateFlip(RotateFlipType.Rotate90FlipNone);
log.Debug("Upscaling image 2x");
BitmapUtils.Scale(ref bitmap, 2);
log.Debug("Grayscaling image");
BitmapUtils.GrayscaleWithLockBits(bitmap);
log.Debug("Denoising image");
BitmapUtils.DenoiseWithLockBits(bitmap);
log.Debug("Applying OCR on image");
Pix pix = PixConverter.ToPix(bitmap);
TesseractEngine tesseractEngine = Services.OCRProvider.AwaitResource();
Page tesseractPage = tesseractEngine.Process(pix);
try
{
string text = tesseractPage.GetText();
log.Debug($"Text is {text.Length} characters long");
if (!string.IsNullOrWhiteSpace(text) && text != "\n")
{
stringBuilder.Append(text.Replace("\n", " "));
}
}
catch (InvalidOperationException e)
{
log.Error($"OCR failed on Page {pageIndex} of file {filename}:\n{e.StackTrace}");
}
Services.OCRProvider.Feed(tesseractEngine);
pix.Dispose();
}
}
}
}
}
stringBuilder.Append("\n");
}
log.Debug("Trimming text");
string documentText = stringBuilder.ToString();
documentText = documentText.Trim();
while (documentText.Contains(" "))
{
documentText = documentText.Replace(" ", " ");
}
while (documentText.Contains("\n\n"))
{
documentText = documentText.Replace("\n\n", "\n");
}
return(stringBuilder.ToString());
}