// 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);
}
protected override void Execute(CodeActivityContext context)
{
Console.WriteLine("Loading FlexiCapture Engine for Classification ...");
bool isClassified = false;
List <LogMessage> logList = new List <LogMessage>();
List <string> filenameList = new List <string>();
List <string> annexPages = new List <string>();
List <string> missingInvoices = new List <string>();
int totalClassified = 0;
int totalReject = 0;
string name = "";
bool isInvoice = false;
bool isPurchaseOrder = false;
try
{
//get Document Project Id
string documentProjectId = DocumentProjectId.Get(context);
Console.WriteLine("Document Project ID: " + documentProjectId);
logList.Add(new LogMessage("Document Project ID: " + documentProjectId, LogType.Information));
//get classifierTemplate path
string classifierTemplatePath = ClassifierTemplate.Get(context);
Console.WriteLine("Classifier path: " + classifierTemplatePath);
logList.Add(new LogMessage("Classifier path: " + classifierTemplatePath, LogType.Information));
//get image path
string sourceFolder = FolderFile.Get(context);
Console.WriteLine("image path: " + sourceFolder);
logList.Add(new LogMessage("image path: " + sourceFolder, LogType.Information));
//get unknown folder
string rejectFolder = RejectFolder.Get(context);
Console.WriteLine("unknown folder: " + rejectFolder);
logList.Add(new LogMessage("image path: " + sourceFolder, LogType.Information));
if (!Directory.Exists(rejectFolder))
{
Directory.CreateDirectory(rejectFolder);
}
//get classified Folder
string classifiedFolder = ClassifiedFolder.Get(context);
Console.WriteLine("Classified folder: " + classifiedFolder);
logList.Add(new LogMessage("Classified folder: " + classifiedFolder, LogType.Information));
if (!Directory.Exists(classifiedFolder))
{
Directory.CreateDirectory(classifiedFolder);
}
Console.WriteLine("Adding images to process...");
string[] files = Directory.GetFiles(sourceFolder, "*.pdf", SearchOption.AllDirectories);
List <string> pdfFileList = new List <string>();
if (files.Length == 0)
{
Console.WriteLine("pdf Files *.pdf is not found");
logList.Add(new LogMessage("pdf Files *.pdf is not found", LogType.Error));
throw new Exception("pdf Files *.pdf is not found");
}
foreach (string pdf in files)
{
Console.WriteLine("pdf: " + pdf);
logList.Add(new LogMessage("pdf: " + pdf, LogType.Information));
pdfFileList.Add(pdf);
string[] splitstr = pdf.Split('_');
string compare = splitstr[0];
if (pdf.Contains(compare))
{
filenameList.Add(compare);
}
}
Console.WriteLine("Added images to process...");
List <string> invoiceList = new List <string>();
List <string> poList = new List <string>();
foreach (string obj in filenameList.Distinct())
{
int total = getTotalPage(files, obj);
isInvoice = false;
isPurchaseOrder = false;
invoiceList.Clear();
poList.Clear();
for (int i = 1; i <= total; i++)
{
string pdfFile = obj + "_" + i + ".pdf";
name = this.DetermineDocumentType(documentProjectId, pdfFile, classifierTemplatePath);
Console.WriteLine("executing pdf: " + pdfFile + " as " + name);
logList.Add(new LogMessage("executing pdf: " + pdfFile, LogType.Information));
if (Object.Equals(name.ToUpper(), "INVOICE"))
{
invoiceList.Add(pdfFile);
isInvoice = true;
continue;
}
if (Object.Equals(name.ToUpper(), "PURCHASEORDER"))
{
poList.Add(pdfFile);
isPurchaseOrder = true;
continue;
}
if (name.Length == 0)
{
Console.WriteLine("annex page is found as " + pdfFile);
logList.Add(new LogMessage("annex page is found as " + pdfFile, LogType.Error));
annexPages.Add(obj);
Console.WriteLine("move to reject folder " + rejectFolder);
string filenamewithoutextention = Path.GetFileNameWithoutExtension(pdfFile);
logList.Add(new LogMessage("move to reject folder", LogType.Information));
MoveFileToDestinationFolder(sourceFolder, rejectFolder, pdfFile, pdfFile, filenamewithoutextention);
totalReject++;
}
}
if (isInvoice && isPurchaseOrder)
{
foreach (string pdfFile in invoiceList)
{
Console.WriteLine("move to classified folder " + classifiedFolder);
string filenamewithoutextention = Path.GetFileNameWithoutExtension(pdfFile);
logList.Add(new LogMessage("move to classified folder " + classifiedFolder, LogType.Information));
MoveFileToDestinationFolder(sourceFolder, classifiedFolder, pdfFile, pdfFile, filenamewithoutextention);
totalClassified++;
}
foreach (string pdfFile in poList)
{
Console.WriteLine("move to reject folder " + rejectFolder);
string filenamewithoutextention = Path.GetFileNameWithoutExtension(pdfFile);
logList.Add(new LogMessage("move to reject folder", LogType.Information));
MoveFileToDestinationFolder(sourceFolder, rejectFolder, pdfFile, pdfFile, filenamewithoutextention);
totalReject++;
}
}
else
{
missingInvoices.Add(obj);
foreach (string pdfFile in invoiceList)
{
Console.WriteLine("move to reject folder " + rejectFolder);
string filenamewithoutextention = Path.GetFileNameWithoutExtension(pdfFile);
logList.Add(new LogMessage("move to reject folder", LogType.Information));
MoveFileToDestinationFolder(sourceFolder, rejectFolder, pdfFile, pdfFile, filenamewithoutextention);
missingInvoices.Add(pdfFile);
totalReject++;
}
foreach (string pdfFile in poList)
{
Console.WriteLine("move to reject folder " + rejectFolder);
string filenamewithoutextention = Path.GetFileNameWithoutExtension(pdfFile);
logList.Add(new LogMessage("move to reject folder", LogType.Information));
MoveFileToDestinationFolder(sourceFolder, rejectFolder, pdfFile, pdfFile, filenamewithoutextention);
totalReject++;
}
}
}
//foreach (string pdfFile in pdfFileList)
//{
// Console.WriteLine("file: " + pdfFile);
// name = this.DetermineDocumentType(documentProjectId, pdfFile, classifierTemplatePath);
// string filenamewithoutextention = Path.GetFileNameWithoutExtension(pdfFile);
// Console.WriteLine("executing pdf: " + pdfFile);
// logList.Add(new LogMessage("executing pdf: " + pdfFile, LogType.Information));
// if (name.Length == 0)
// {
// Console.WriteLine("move to reject folder " + rejectFolder);
// logList.Add(new LogMessage("move to reject folder", LogType.Information));
// MoveFileToDestinationFolder(sourceFolder, rejectFolder, pdfFile, pdfFile, filenamewithoutextention);
// NoRejected++;
// }
// else
// {
// Console.WriteLine("move to classified folder "+classifiedFolder);
// logList.Add(new LogMessage("move to classified folder "+ classifiedFolder, LogType.Information));
// MoveFileToDestinationFolder(sourceFolder, classifiedFolder, pdfFile, pdfFile, filenamewithoutextention);
// NoClassified++;
// }
//}
isClassified = true;
string message = String.Format("Total Number of classified Invoice is {0} and Rejected is {1}", totalClassified, totalReject);
logList.Add(new LogMessage(message, LogType.Information));
var msg = processor.GetLastProcessingError();
if (msg != null)
{
var msgError = msg.MessageText();
logList.Add(new LogMessage(msgError, LogType.Error));
Console.WriteLine(msgError);
}
}
#pragma warning disable CS0618 // Type or member is obsolete
catch (ExecutionEngineException e)
#pragma warning restore CS0618 // Type or member is obsolete
{
isClassified = false;
Console.WriteLine("error: " + e.Message);
logList.Add(new LogMessage("throw exception because of " + e.Message, LogType.Error));
}
finally
{
Console.WriteLine("Released FlexiCapture Engine ...");
UnloadEngine(ref engine);
logList.Add(new LogMessage("Released FlexiCapture Engine ...", LogType.Information));
IsClassified.Set(context, isClassified);
LogMessages.Set(context, logList);
AnnexPageList.Set(context, annexPages);
MissingInvoiceList.Set(context, missingInvoices);
}
}
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);
}
}