private string DetermineDocumentType(string projectId, string imageFile, string classifierPath)
{
Console.WriteLine("Determine Document Type is processing ...");
if (engine == null)
{
engine = LoadEngine(projectId);
}
if (processor == null)
{
processor = engine.CreateFlexiCaptureProcessor();
processor.AddClassificationTreeFile(classifierPath);
Console.WriteLine("Classifier template is added ...");
}
else
{
processor.ResetProcessing();
Console.WriteLine("Processor is reseted");
}
processor.AddImageFile(@imageFile);
var result = processor.ClassifyNextPage();
if (result != null && result.PageType == PageTypeEnum.PT_MeetsDocumentDefinition)
{
var names = result.GetClassNames();
Console.WriteLine("Determine Document Type is completed");
return(names.Item(0));
}
Console.WriteLine("Determine Document Type is not completed");
return("");
}
private bool tryProcessTask()
{
bool result = false;
if (requestsInWork.Count == 0 && requests.Count != 0)
{
TryToAddRequestsToWork();
}
//take actual request, cut, process task from it and put this request back to queue.
Request actualRequest;
bool actualRequestObtained = requestsInWork.TryDequeue(out actualRequest);
if (actualRequestObtained)
{
result = true;
List <byte[]> images;
bool isRequestPieceObtained = actualRequest.TryGetRequestPiece(taskSize, out images);
if (isRequestPieceObtained)
{
requestsInWork.Enqueue(actualRequest);
int procNumb;
IEngine engine;
IFlexiCaptureProcessor processor = null;
processor = fcProcessorsPool.GetProcessor(out engine, out procNumb);
Tuple <IFlexiCaptureProcessor, IEngine, int, List <byte[]>, Request> startParams =
Tuple.Create(processor, engine, procNumb, images, actualRequest);
Thread recognition = new Thread(Recognition);
recognition.Start(startParams);
}
}
return(result);
}
// USE CASE: Creating a Document Definition from a FlexiLayout (*.afl)
public static void Creating_a_Document_Definition_from_a_FlexiLayout(IEngine engine)
{
trace("Create a Document Definition from an *.afl file...");
string flexibleDescriptionFilePath = SamplesFolder + "\\SampleMisc\\Invoice_eng.afl";
IDocumentDefinition newDefinition = engine.CreateDocumentDefinitionFromAFL(flexibleDescriptionFilePath, "English");
// You can save the new Document Definition to a file or use it from memory
traceBegin("Use the Document Definition in FlexiCaptureProcessor...");
IFlexiCaptureProcessor processor = engine.CreateFlexiCaptureProcessor();
processor.AddDocumentDefinition(newDefinition);
// Add images for a single document
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Invoices_1.tif");
// Recognize the document and check the result
IDocument document = processor.RecognizeNextDocument();
assert(document != null);
assert(document.DocumentDefinition != null);
assert(document.Pages.Count == 1);
// Export the result
processor.ExportDocumentEx(document, SamplesFolder + "\\FCEExport", "Invoice", null);
traceEnd("OK");
}
// USE CASE: Creating a compound Document Definition
public static void Creating_a_compound_Document_Definition(IEngine engine)
{
trace("Create an empty Document Definition in memory...");
IDocumentDefinition newDefinition = engine.CreateDocumentDefinition();
assert(newDefinition != null);
trace("Set default language...");
ILanguage language = engine.PredefinedLanguages.FindLanguage("English");
assert(language != null);
newDefinition.DefaultLanguage = language;
trace("Create a new fixed section from an XFD file...");
newDefinition.DefaultTextType = TextTypeEnum.TT_Handprinted;
ISectionDefinition newSection1 = newDefinition.Sections.AddNew("Banking");
newSection1.LoadXFDDescription(SamplesFolder + "\\SampleMisc\\Banking_eng.xfd");
trace("Create a new flexible section from an AFL file...");
newDefinition.DefaultTextType = TextTypeEnum.TT_Normal;
ISectionDefinition newSection2 = newDefinition.Sections.AddNew("Invoice");
newSection2.LoadFlexibleDescription(SamplesFolder + "\\SampleMisc\\Invoice_eng.afl");
// Modify the template as required. In this sample we need to loosen some constraints
IPageAnalysisParams analysisParams = engine.CreatePageAnalysisParams();
analysisParams.CopyFrom(newDefinition.PageAnalysisParams);
analysisParams.MaxHorizontalShrinkPercent = 20;
analysisParams.MaxVerticalShrinkPercent = 20;
newDefinition.PageAnalysisParams = analysisParams;
trace("Check the Document Definition...");
assert(newDefinition.Check() == true);
// You can save the new Document Definition to a file or use it from memory
traceBegin("Use the Document Definition in FlexiCaptureProcessor...");
IFlexiCaptureProcessor processor = engine.CreateFlexiCaptureProcessor();
processor.AddDocumentDefinition(newDefinition);
// Add images for a single multipage document
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Banking_1.tif");
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Banking_2.tif");
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Banking_3.tif");
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Invoices_2.tif");
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Invoices_3.tif");
// Recognize the document
IDocument document = processor.RecognizeNextDocument();
assert(document != null);
assert(document.DocumentDefinition != null);
assert(document.Pages.Count == 5);
processor.ExportDocumentEx(document, SamplesFolder + "\\FCEExport", "Mixed", null);
traceEnd("OK");
}
internal static void DoConfigureProcessor(string docDifPath, IFlexiCaptureProcessor processor, IEngine engine)
{
IDocumentDefinition documentDefinition = createDocDifinitionFromXML(docDifPath, engine);
bool checkresult = documentDefinition.Check();
IStringsCollection errors = documentDefinition.Errors;
for (int i = 0; i < errors.Count; i++)
{
string str = errors[i];
}
bool isvalid = documentDefinition.IsValid;
processor.AddDocumentDefinition(documentDefinition);
processor.SetForceApplyDocumentDefinition(true);
}
// HOW-TO: Explicitly release a COM object from managed code
public static void How_to_explicitly_release_a_COM_object_from_managed_code(IEngine engine)
{
// When using COM objects from managed code, it is sometimes
// required to ensure that a COM object is released at a specific point
// and not left to the garbage collector (which might not
// release the object for a long time). This might be needed to release
// memory or other resources taken by the object if there is no other
// way to release them. This is effectively "DISPOSAL" of the COM object and
// the same rules should be followed as those used when deciding if
// Dispose method should be called on a disposable .NET Framework object.
// Releasing resources in timely manner is especially important in
// server applications. Garbage collector does not "know" the real
// amount of memory and resources used by COM objects and will
// not "know" that they should be garbage-collected. As the result, your server
// application might run low on non-managed memory while managed heap
// will be still almost empty and garbage-collection will not run.
// For example, the FlexiCapture processor object has some internal state,
// which consumes resources. If we just leave the object to the garbage
// collector, we cannot be sure when these resources are released.
// It might not be a problem in most cases, but if our application
// is to be used in high-performance scenarios we'd be better off
// explicitly releasing the object. The downside of this scenario is
// the complexity of the resulting code.
trace("Create a COM object...");
IFlexiCaptureProcessor processor = engine.CreateFlexiCaptureProcessor();
try {
// Some processing here
// ...
} finally {
trace("To explicitly release the COM object, call Marshal.ReleaseComObject.");
// We are sure that this instance of the processor will never be used again,
// so we are safe to release it here
Marshal.ReleaseComObject(processor);
}
// NB. Nulling the object WILL NOT do the trick as it did in Visual Basic 6.0.
// Calling GC.Collect might SEEM to work under some conditions but does not
// guarantee that a particular object will be freed (a stray reference).
// Moreover, GC.Collect might be a very lengthy operation (if your managed heap
// contains many objects) and calling it often might significantly degrade performance.
}
public MainForm(string workDirectory, string fileLoad)
{
InitializeComponent();
// инициализация движка ABBYY
try
{
_engineLoader = new FCEngine.InprocLoader
{
CustomerProjectId = FceConfig.GetCustomerProjectId(), LicensePassword = "", LicensePath = ""
};
_engine = _engineLoader.GetEngine();
//engine = engineLoader.Load(FceConfig.GetCustomerProjectId(), "");
_processor = _engine.CreateFlexiCaptureProcessor();
// Initialize DocumentView component
documentView.Engine = _engine;
_pathFolderWork = workDirectory;
if (_pathFolderWork.EndsWith("\\"))
{
_pathFolderWork = _pathFolderWork.Substring(0, _pathFolderWork.Length - 1);
}
if (String.IsNullOrEmpty(Path.GetDirectoryName(_pathFolderWork)))
{
throw new Exception(_pathFolderWork + " не является корректной директорией");
}
_pathFolderWork += "\\visualeditorfiles\\";
if (!String.IsNullOrEmpty(fileLoad))
{
_nameCurrentFile = Path.GetFileNameWithoutExtension(fileLoad);
}
}
catch (Exception ex)
{
MessageBox.Show("Не удалось загрузить компоненту ABBYY по причине: " + ex.Message);
Application.Exit();
}
//updateButtonsState();
LoadWorkDocument();
}
static IDocument PrepareNewRecognizedDocument(IEngine engine, out IFlexiCaptureProcessor processor)
{
// Create and configure an instance of FlexiCapture processor
processor = engine.CreateFlexiCaptureProcessor();
processor.AddDocumentDefinitionFile(SamplesFolder + "\\SampleProject\\Templates\\Invoice_eng.fcdot");
// Add images for a single multipage document
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Invoices_2.tif");
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Invoices_3.tif");
// Recognize the document
IDocument document = processor.RecognizeNextDocument();
assert(document != null);
assert(document.DocumentDefinition != null);
assert(document.Pages.Count == 2);
return(document);
}
// USE CASE: Using image processing tools in custom preprocessing
public static void Using_image_processing_tools_in_custom_preprocessing(IEngine engine)
{
trace("Create and configure an instance of FlexiCapture processor...");
IFlexiCaptureProcessor processor = engine.CreateFlexiCaptureProcessor();
processor.AddDocumentDefinitionFile(SamplesFolder + "\\SampleProject\\Templates\\Invoice_eng.fcdot");
trace("Set up an image source with custom preprocessing...");
// Create and configure sample image source. ALL PREPROCESSING IS DONE IN THE IMAGE SOURCE
// (see SampleImageSource class for details)
CustomPreprocessingImageSource imageSource = new CustomPreprocessingImageSource(engine);
imageSource.AddImageFile(SamplesFolder + "\\SampleImages\\Invoices_1.tif");
imageSource.AddImageFile(SamplesFolder + "\\SampleImages\\Invoices_2.tif");
imageSource.AddImageFile(SamplesFolder + "\\SampleImages\\Invoices_3.tif");
processor.SetCustomImageSource(imageSource);
traceBegin("Process the images...");
int count = 0;
while (true)
{
IDocument document = processor.RecognizeNextDocument();
if (document == null)
{
IProcessingError error = processor.GetLastProcessingError();
assert(error == null); // No errors are expected in this sample
break;
}
else
{
// We expect that this will never happen in this sample
assert(document.DocumentDefinition != null);
}
processor.ExportDocumentEx(document, SamplesFolder + "\\FCEExport", "NextDocument_" + count, null);
count++;
}
traceEnd("OK");
trace("Check the results...");
assert(count == 2);
}
static void CheckTrainedDocumentDefinition(IEngine engine, IDocumentDefinition newDocumentDefinition, string rootFolder)
{
IFlexiCaptureProcessor processor = engine.CreateFlexiCaptureProcessor();
processor.AddDocumentDefinition(newDocumentDefinition);
processor.AddImageFile(rootFolder + "\\02.jpg");
processor.AddImageFile(rootFolder + "\\03.jpg");
IDocument document = processor.RecognizeNextDocument();
assert(document.DocumentDefinition != null);
assert(document.Pages.Count == 1);
assert(document.Sections[0].Children[0].Name == "ISBN");
assert(document.Sections[0].Children[0].Value.AsString == "0-517-59939-2");
document = processor.RecognizeNextDocument();
assert(document.DocumentDefinition != null);
assert(document.Pages.Count == 1);
assert(document.Sections[0].Children[0].Name == "ISBN");
assert(document.Sections[0].Children[0].Value.AsString == "0-8050-6176-2");
}
// USE CASE: Creating a Document Definition from an XML Form Definition
public static void Creating_a_Document_Definition_from_an_XML_Form_Definition(IEngine engine)
{
trace("Create a Document Definition from an *.xfd file...");
string formDescriptionFilePath = SamplesFolder + "\\SampleMisc\\Banking_eng.xfd";
IDocumentDefinition newDefinition = engine.CreateDocumentDefinitionFromXFD(formDescriptionFilePath, "English");
// Modify the template as required. In this sample we need to loosen some constraints
IPageAnalysisParams analysisParams = engine.CreatePageAnalysisParams();
analysisParams.CopyFrom(newDefinition.PageAnalysisParams);
analysisParams.MaxHorizontalShrinkPercent = 20;
analysisParams.MaxVerticalShrinkPercent = 20;
newDefinition.PageAnalysisParams = analysisParams;
// You can save the new Document Definition to a file or use it from memory
traceBegin("Use the Document Definition in FlexiCaptureProcessor...");
IFlexiCaptureProcessor processor = engine.CreateFlexiCaptureProcessor();
processor.AddDocumentDefinition(newDefinition);
// Add images for a single multipage document
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Banking_1.tif");
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Banking_2.tif");
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Banking_3.tif");
// Recognize the document and check the result
IDocument document = processor.RecognizeNextDocument();
assert(document != null);
assert(document.DocumentDefinition != null);
assert(document.Pages.Count == 3);
// Export the result
processor.ExportDocumentEx(document, SamplesFolder + "\\FCEExport", "Banking", null);
traceEnd("OK");
}
public static Tuple <Dictionary <string, object>, List <object> > GetData(string imgPath)
{
if (engine == null)
{
engine = loadEngine();
}
if (processor == null)
{
processor = engine.CreateFlexiCaptureProcessor();
processor.AddDocumentDefinitionFile("C:\\ProgramData\\ABBYY\\SDK\\11\\FlexiCapture Engine\\Samples\\SampleMisc\\Invoice.fcdot");
}
else
{
processor.ResetProcessing();
}
processor.AddImageFile(imgPath);
IDocument document = processor.RecognizeNextDocument();
if (document != null && document.DocumentDefinition != null)
{
return(ReadDocData(document));
}
return(null);
}
static IDocument PrepareNewRecognizedDocument( IEngine engine, out IFlexiCaptureProcessor processor )
{
// Create and configure an instance of FlexiCapture processor
processor = engine.CreateFlexiCaptureProcessor();
processor.AddDocumentDefinitionFile( SamplesFolder + "\\SampleProject\\Templates\\Invoice_eng.fcdot" );
// Add images for a single multipage document
processor.AddImageFile( SamplesFolder + "\\SampleImages\\Invoices_2.tif" );
processor.AddImageFile( SamplesFolder + "\\SampleImages\\Invoices_3.tif" );
// Recognize the document
IDocument document = processor.RecognizeNextDocument();
assert( document != null );
assert( document.DocumentDefinition != null );
assert( document.Pages.Count == 2 );
return document;
}
// USE CASE: Configuring fields for better recognition results
public static void Configuring_fields_for_better_recognition_results(IEngine engine)
{
trace("Create a Document Definition from a FlexiLayout...");
IDocumentDefinition newDefinition = engine.CreateDocumentDefinitionFromAFL(SamplesFolder + "\\SampleMisc\\Invoice_eng.afl", "English");
assert(newDefinition != null);
trace("Configure data types...");
setFieldValueType(newDefinition, "InvoiceDate", FieldValueTypeEnum.FVT_DateTime);
setFieldValueType(newDefinition, "Quantity", FieldValueTypeEnum.FVT_Number);
setFieldValueType(newDefinition, "UnitPrice", FieldValueTypeEnum.FVT_Currency);
setFieldValueType(newDefinition, "Total", FieldValueTypeEnum.FVT_Currency);
setFieldValueType(newDefinition, "TotalAmount", FieldValueTypeEnum.FVT_Currency);
trace("Configure recognition languages for text fields ...");
IFieldDefinition fieldDef = findFieldDef(newDefinition, "InvoiceNumber");
assert(fieldDef != null);
ITextRecognitionParams textParams = fieldDef.RecognitionParams.AsTextParams();
ILanguage newLanguage = textParams.CreateEmbeddedLanguageByDataType(FieldValueTypeEnum.FVT_DateTime);
textParams.Language = newLanguage;
newLanguage = textParams.CreateEmbeddedLanguage(textParams.Language.Type, textParams.Language);
assert(newLanguage != textParams.Language);
assert(newLanguage.LanguageCategory == LanguageCategoryEnum.LC_DataType);
assert(newLanguage.DatatypeCategory == DatatypeCategoryEnum.TC_DateTime);
textParams.Language = newLanguage;
newLanguage = textParams.CreateEmbeddedLanguage(LanguageTypeEnum.LT_Group, null);
newLanguage.AsGroupLanguage().Add(engine.PredefinedLanguages.FindLanguage("English"));
newLanguage.AsGroupLanguage().Add(engine.PredefinedLanguages.FindLanguage("Russian"));
textParams.Language = newLanguage;
assert(textParams.Language.Type == LanguageTypeEnum.LT_Group);
assert(textParams.Language.AsGroupLanguage().Count == 2);
assert(textParams.Language.AsGroupLanguage().Item(0).InternalName == "English");
assert(textParams.Language.AsGroupLanguage().Item(1).InternalName == "Russian");
newLanguage = textParams.CreateEmbeddedLanguage(LanguageTypeEnum.LT_Simple, null);
newLanguage.AsSimpleLanguage().set_LetterSet(LanguageLetterSetEnum.LLS_Alphabet, "ABCDEFGHIJKLMNOPQRSTUVWXYZ");
newLanguage.AsSimpleLanguage().RegularExpression = "[A-Z]{1-}";
textParams.Language = newLanguage;
assert(textParams.Language.AsSimpleLanguage().RegularExpression.Length > 0);
newLanguage = textParams.CreateEmbeddedLanguage(LanguageTypeEnum.LT_Simple, engine.PredefinedLanguages.FindLanguage("English"));
assert(newLanguage.AsSimpleLanguage().UsePredefinedDictionary == true);
assert(newLanguage.AsSimpleLanguage().UseUserDefinedDictionary == false);
assert(newLanguage.AsSimpleLanguage().UserDefinedDictionary == null);
newLanguage.AsSimpleLanguage().UseUserDefinedDictionary = true;
FCEngine.IDictionary dictionary = newLanguage.AsSimpleLanguage().UserDefinedDictionary;
assert(dictionary != null);
assert(dictionary.WordsCount == 0);
dictionary.AddWord("ONE", 1);
dictionary.AddWord("TWO", 1);
dictionary.AddWord("THREE", 1);
assert(dictionary.WordsCount == 3);
IEnumDictionaryWords enumWords = dictionary.EnumWords();
for (int i = 0; i < 10; i++)
{
int confidence = 0;
string word = enumWords.Next(out confidence);
if (confidence == 0)
{
break;
}
trace(word);
}
textParams.Language = newLanguage;
trace("Check the Document Definition...");
assert(newDefinition.Check() == true);
traceBegin("Use the Document Definition in FlexiCaptureProcessor...");
IFlexiCaptureProcessor processor = engine.CreateFlexiCaptureProcessor();
processor.AddDocumentDefinition(newDefinition);
// Add images for a single document
processor.AddImageFile(SamplesFolder + "\\SampleImages\\Invoices_1.tif");
// Recognize the document
IDocument document = processor.RecognizeNextDocument();
assert(document != null);
assert(document.DocumentDefinition != null);
assert(document.Pages.Count == 1);
processor.ExportDocumentEx(document, SamplesFolder + "\\FCEExport", "Invoice", null);
traceEnd("OK");
}
static void processError( IProcessingError error, IFlexiCaptureProcessor processor, ErrorHandlingStrategy strategy )
{
assert( error != null );
// You can store the problematic image for later analysis
if( error.ImageFile() != null ) {
string imageReference = error.ImageFile().GetFileReference();
}
// And use different strategies to handle the error
switch( strategy ) {
// 1) Log the error and continue
case ErrorHandlingStrategy.LogAndContinue:
trace( "Processing error: " + error.MessageText() );
return;
// 2) Retry
case ErrorHandlingStrategy.LogAndRetry:
trace( "Processing error: " + error.MessageText() );
// Resume processing from the position where error occurred
processor.ResumeProcessing( true );
return;
// 3) Break processing
case ErrorHandlingStrategy.ThrowException:
// This will reset the processing state (image queue, error condition, etc.) but keep
// the processing configuration (loaded templates and processing params). You can omit this
// call if you do not plan to reuse the processor instance or reset the processor elsewhere
processor.ResetProcessing();
throw new Exception( error.MessageText() );
}
}
protected override void Execute(CodeActivityContext context)
{
string message = "";
List <LogMessage> logList = new List <LogMessage>();
List <string> nonInvoices = new List <string>();
List <string> notConfidenceList = new List <string>();
message = "Loading FlexiCapture Engine for Recognition ...";
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
string documentProjectId = DocumentProjectId.Get(context);
message = "Get Document Project ID: " + documentProjectId;
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
int confidenceLevelLimitation = ConfidenceLevelLimitation.Get(context);
message = string.Format("Confidence Level Limitation: {0}", confidenceLevelLimitation);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
string sourceFolder = SourceFolder.Get(context);
message = string.Format("Get PDF folder: {0}", sourceFolder);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
string ocrTemplate = OCRTemplateFolder.Get(context);
message = string.Format("OCR Template Folder: {0}", ocrTemplate);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
string exportFolder = ExportFolder.Get(context);
message = string.Format("Get Export Folder: {0}", exportFolder);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
if (!Directory.Exists(exportFolder))
{
Directory.CreateDirectory(exportFolder);
message = string.Format("Folder {0} is created completely", exportFolder);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
}
string recognizedFolder = RecognizedFolder.Get(context);
Console.WriteLine("Get Recognize Folder: " + recognizedFolder);
logList.Add(new LogMessage("Recognize Folder is: " + recognizedFolder, LogType.Information));
if (!Directory.Exists(recognizedFolder))
{
Directory.CreateDirectory(recognizedFolder);
message = string.Format("Folder {0} is created completely", recognizedFolder);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
}
string notConfidenceFolder = NotConfidenceFolder.Get(context);
Console.WriteLine("Get Not Confidence Folder: " + notConfidenceFolder);
logList.Add(new LogMessage("Not Confidence Folder is: " + notConfidenceFolder, LogType.Information));
if (!Directory.Exists(notConfidenceFolder))
{
Directory.CreateDirectory(notConfidenceFolder);
message = string.Format("Folder {0} is created completely", notConfidenceFolder);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
}
string rejectFolder = RejectFolder.Get(context);
Console.WriteLine("Get Reject Folder: " + rejectFolder);
logList.Add(new LogMessage("Reject Folder is: " + rejectFolder, LogType.Information));
if (!Directory.Exists(rejectFolder))
{
Directory.CreateDirectory(rejectFolder);
message = string.Format("Folder {0} is created completely", rejectFolder);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
}
engine = LoadEngine(documentProjectId);
Boolean isRecognized = false;
bool isValidInvoice = false;
int count = 0;
int noError = 0;
int noSuccess = 0;
int noNotConfidence = 0;
try
{
message = "Creating and configuring the FlexiCapture Processor...";
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
processor = engine.CreateFlexiCaptureProcessor();
message = "Adding Document Definition to process...";
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
string[] TotalOCRFiles = Directory.GetFiles(ocrTemplate, "*.fcdot", SearchOption.AllDirectories);
if (TotalOCRFiles.Length == 0)
{
message = string.Format("OCR Template *.fcdot is not found in the folder {0}", ocrTemplate);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Error));
throw new Exception(message);
}
foreach (string ocr in TotalOCRFiles)
{
processor.AddDocumentDefinitionFile(ocr);
message = string.Format("OCR Temhplate {0} is added", ocr);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
}
message = string.Format("Adding images to process...");
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
string[] TotalFiles = Directory.GetFiles(sourceFolder, "*.pdf", SearchOption.AllDirectories);
if (TotalFiles.Length == 0)
{
message = string.Format("PDF Files *.pdf is not found in the folder {0}.", sourceFolder);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Error));
throw new Exception(message);
}
foreach (string pdfFile in TotalFiles)
{
processor.AddImageFile(pdfFile);
message = string.Format("PDF Files {0} is added", pdfFile);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
}
message = "Recognizing the images and exporting the results...";
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
while (true)
{
// Recognize next document
confidenceLevel = 0;
confidenceDetailLevel = 0;
confidenceHeaderLevel = 0;
totalConfidenceLevelHD = 0;
IDocument document = processor.RecognizeNextDocument();
if (document == null)
{
IProcessingError error = processor.GetLastProcessingError();
if (error != null)
{
// Processing error
message = string.Format("processing error because of {0}.", error.MessageText());
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Error));
continue;
}
else
{
// No more images
message = string.Format("all PDF Files has been executed or no PDF file in the folder {0}", sourceFolder);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
break;
}
}
else if (document.DocumentDefinition == null)
{
// Couldn't find matching template for the image. In this sample this is an error.
// In other scenarios this might be normal
message = string.Format("PDF file is not matched with existing OCR Templates.");
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Error));
//string tempPage = document.Pages[0].OriginalImagePath;
//if(tempPage != null)
//{
// string movefile = Path.GetFileName(tempPage);
// string tempfilename = Path.GetFileNameWithoutExtension(tempPage);
// logList.Add(new LogMessage("Move to Reject Folder", LogType.Information));
// if (File.Exists(tempPage))
// {
// MoveFileToDestinationFolder(sourceFolder, rejectFolder, tempPage, tempfilename);
// }
// else
// {
// MoveFileToDestinationFolder(notConfidenceFolder, rejectFolder, notConfidenceFolder+"\\"+movefile, tempfilename);
// }
//}
continue;
}
string originalPath = document.Pages[0].OriginalImagePath;
string file = Path.GetFileName(originalPath);
string filenamewithoutextention = Path.GetFileNameWithoutExtension(originalPath);
message = string.Format("Recognizing pdf {0} is started.", originalPath);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
//set confident level and status
message = string.Format("Extracting data from pdf {0} is started", originalPath);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
for (int i = 0; i < document.Sections.Count; i++)
{ // extracing
var section = document.Sections[i];
if (object.ReferenceEquals(section, null))
{
continue;
}
for (int d = 0; d < section.Children.Count; d++)
{
var child = section.Children[d];
if (object.ReferenceEquals(child, null))
{
continue;
}
var field = ((IField)child);
message = string.Format("Extracting column {0} = {1}", field.Name, field.Value.AsText);
Console.WriteLine(message);
if (field.Name.ToUpper().Trim() == "INV_CONFIDENCE_LEVEL")
{
var value = TextFieldHelper.GetConfidenLevel(engine, document);
totalConfidenceLevelHD = value;
var data = engine.CreateText(value.ToString(), null);
field.Value.AsInteger = value;
confidenceLevel = value;
message = string.Format("Confidence level of {0} is {1}", file, value);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
}
if (field.Name.ToUpper().Trim() == "INV_STATUS")
{
var value = "Recognized";
var data = engine.CreateText(value, null);
field.Value.AsText.Delete(0, field.Value.AsText.Length);
field.Value.AsText.Insert(data, 0);
}
if (field.Name.ToUpper().Trim() == "FILE_NAME")
{
var data = engine.CreateText(file, null);
field.Value.AsText.Delete(0, field.Value.AsText.Length);
field.Value.AsText.Insert(data, 0);
}
}
} //end extracting
isValidInvoice = true;
if (isValidInvoice)
{
message = string.Format("Total Confidence is {0} ", totalConfidenceLevelHD);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
//check confidence level validation
if (totalConfidenceLevelHD >= confidenceLevelLimitation)
{
try
{
message = string.Format("Exporting process for pdf {0} is started ...", file);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
//IFileExportParams exportParams = engine.CreateFileExportParams();
//Console.WriteLine("XLS");
//exportParams.FileFormat = FileExportFormatEnum.FEF_XLS;
processor.ExportDocument(document, exportFolder);
Console.WriteLine("Exporting process is completed ...");
MoveFileToDestinationFolder(sourceFolder, exportFolder, originalPath, filenamewithoutextention);
Console.WriteLine(string.Format("Moving {0} to Export folder {1} is completed", file, exportFolder));
logList.Add(new LogMessage("Exporting process is ended ...", LogType.Information));
noSuccess++;
}
catch (Exception e)
{
noError++;
Console.WriteLine(string.Format("exporting is failed because of {0}.", e.Message));
logList.Add(new LogMessage(message, LogType.Error));
MoveFileToDestinationFolder(sourceFolder, rejectFolder, originalPath, filenamewithoutextention);
message = string.Format("Moving pdf {0} to Reject folder {1} is completed", file, rejectFolder);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
continue;
}
}
else
{
message = string.Format("Confidence of PDF {0} is {1} less than target confidence {2}", file, totalConfidenceLevelHD, confidenceLevelLimitation);
logList.Add(new LogMessage(message, LogType.Error));
notConfidenceList.Add(file);
logList.Add(new LogMessage(string.Format("Total number of not confidence is {0}", notConfidenceList.Count), LogType.Information));
MoveFileToDestinationFolder(sourceFolder, notConfidenceFolder, originalPath, filenamewithoutextention);
message = string.Format("Moving pdf {0} to Not Confidence folder {1} is completed", file, notConfidenceFolder);
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
noNotConfidence++;
}
}
count++;
}
var msg = processor.GetLastProcessingError();
if (msg != null)
{
var msgError = string.Format("the processing error because of {0}.", msg.MessageText());
Console.WriteLine(msgError);
logList.Add(new LogMessage(msgError, LogType.Error));
noError++;
}
message = string.Format("No. of Not Confidence {0} and No. of Error {1} and No. of Exported to DB {2}", noNotConfidence, noError, noSuccess);
logList.Add(new LogMessage(message, LogType.Information));
isRecognized = true;
}
finally
{
UnloadEngine(ref engine);
message = string.Format("Released FlexiCapture Engine for Recognition...");
Console.WriteLine(message);
logList.Add(new LogMessage(message, LogType.Information));
IsRecognized.Set(context, isRecognized);
NonInvoiceList.Set(context, nonInvoices);
Messages.Set(context, logList);
TotalExported.Set(context, noSuccess);
TotalError.Set(context, noError);
TotalNotConfidence.Set(context, noNotConfidence);
NotConfidenceList.Set(context, notConfidenceList);
}
}
