private PdfOCRResponse HandleOCRPDF(PDFApi pdfApiInstance, PDFOCRActionConfiguration actionConfiguration, FileToProcess fileToProcess, string fileID, int workerNumber)
{
// First get the number of page of the PDF
PdfGetInfoResponse getInfoResponse = PassportPDFRequestsUtilities.SendGetInfoRequest(pdfApiInstance, new PdfGetInfoParameters(fileID), workerNumber, fileToProcess.FileAbsolutePath, FileOperationStartEventHandler);// todo: use appropriate event handler
if (getInfoResponse.Error != null)
{
return(null);
}
PdfOCRParameters ocrParameters = PassportPDFParametersUtilities.GetOCRParameters(actionConfiguration, fileID);
int pageCount = getInfoResponse.PageCount;
int chunkLength = Math.Min(getInfoResponse.PageCount, FrameworkGlobals.PAGE_CHUNK_LENGTH_FOR_OCR_ACTION);
int chunkCount = getInfoResponse.PageCount > FrameworkGlobals.PAGE_CHUNK_LENGTH_FOR_OCR_ACTION ? (int)Math.Ceiling((double)getInfoResponse.PageCount / FrameworkGlobals.PAGE_CHUNK_LENGTH_FOR_OCR_ACTION) : 1;
PdfOCRResponse ocrResponse = null;
for (int chunkNumber = 1; chunkNumber <= chunkCount; chunkNumber++)
{
ocrParameters.PageRange = PassportPDFParametersUtilities.GetChunkProcessingPageRange(pageCount, chunkLength, chunkNumber, chunkCount);
ocrResponse = PassportPDFRequestsUtilities.SendOCRRequest(pdfApiInstance, ocrParameters, workerNumber, fileToProcess.FileAbsolutePath, ocrParameters.PageRange, pageCount, FileChunkProcessingProgressEventHandler);
if (_cancellationPending || ocrResponse == null)
{
return(ocrResponse);
}
}
return(ocrResponse);
}
public static OperationsWorkflow CreatePDFOCRWorkflow(PDFOCRActionConfiguration ocrActionConfiguration)
{
List <Operation> actionsToBePerformed = new List <Operation>
{
new Operation(Operation.OperationType.LoadPDF, PdfVersion.PdfVersionRetainExisting),
new Operation(Operation.OperationType.OCRPDF, ocrActionConfiguration)
};
return(new OperationsWorkflow(actionsToBePerformed, OperationsWorkflow.SaveOperationType.SavePDF));
}
public static void ParseCommandLineArgs(string[] args, ApplicationConfiguration applicationConfiguration, PDFReduceActionConfiguration reduceActionConfiguration = null, PDFOCRActionConfiguration ocrActionConfiguration = null, ImageSaveAsPDFMRCActionConfiguration imageSaveAsPdfMrcActionConfiguration = null, DocumentLoadAsPDFActionConfiguration loadAsPdfActionConfiguration = null)
{
for (int index = 0; index < args.Length; index++)
{
if (!ParseApplicationConfigurationArgument(args, index, applicationConfiguration))
{
if (reduceActionConfiguration != null)
{
ParseReduceActionConfigurationArgument(args, index, reduceActionConfiguration);
}
if (ocrActionConfiguration != null)
{
ParseOCRActionConfigurationArgument(args, index, ocrActionConfiguration);
}
if (imageSaveAsPdfMrcActionConfiguration != null)
{
ParseSaveImageAsPdfMrcConfigurationArgument(args, index, imageSaveAsPdfMrcActionConfiguration);
}
if (loadAsPdfActionConfiguration != null)
{
ParseDocumentLoadAsPdfConfigurationArgument(args, index, loadAsPdfActionConfiguration);
}
}
}
}
private WorkflowProcessingResult ProcessWorkflow(PDFApi pdfApiInstance, ImageApi imageApiInstance, OperationsWorkflow workflow, FileToProcess fileToProcess, int workerNumber)
{
List <string> warningMessages = new List <string>();
bool contentRemoved = false;
bool versionChanged = false;
bool linearized = false;
string fileID = null;
foreach (Operation operation in workflow.OperationsToBePerformed)
{
Error actionError = null;
ReduceErrorInfo reduceErrorInfo = null;
long remainingTokens = 0;
if (_cancellationPending)
{
return(null);
}
switch (operation.Type)
{
case Operation.OperationType.LoadPDF:
PdfVersion outputVersion = (PdfVersion)operation.Parameters;
PdfLoadDocumentResponse loadDocumentResponse = HandleLoadPDF(pdfApiInstance, outputVersion, fileToProcess, workerNumber);
if (loadDocumentResponse == null)
{
OnError(LogMessagesUtils.ReplaceMessageSequencesAndReferences(FrameworkGlobals.MessagesLocalizer.GetString("message_invalid_response_received", FrameworkGlobals.ApplicationLanguage), actionName: "Load"));
return(null);
}
remainingTokens = loadDocumentResponse.RemainingTokens;
actionError = loadDocumentResponse.Error;
fileID = loadDocumentResponse.FileId;
break;
case Operation.OperationType.LoadImage:
ImageLoadResponse imageLoadResponse = HandleLoadImage(imageApiInstance, fileToProcess, workerNumber);
if (imageLoadResponse == null)
{
OnError(LogMessagesUtils.ReplaceMessageSequencesAndReferences(FrameworkGlobals.MessagesLocalizer.GetString("message_invalid_response_received", FrameworkGlobals.ApplicationLanguage), actionName: "Load"));
return(null);
}
remainingTokens = imageLoadResponse.RemainingTokens;
actionError = imageLoadResponse.Error;
fileID = imageLoadResponse.FileId;
break;
case Operation.OperationType.ReducePDF:
PDFReduceActionConfiguration reduceActionConfiguration = (PDFReduceActionConfiguration)operation.Parameters;
PdfReduceResponse reduceResponse = HandleReducePDF(pdfApiInstance, reduceActionConfiguration, fileToProcess, fileID, workerNumber, warningMessages);
if (reduceResponse == null)
{
OnError(LogMessagesUtils.ReplaceMessageSequencesAndReferences(FrameworkGlobals.MessagesLocalizer.GetString("message_invalid_response_received", FrameworkGlobals.ApplicationLanguage), actionName: "Reduce"));
return(null);
}
remainingTokens = reduceResponse.RemainingTokens;
contentRemoved = reduceResponse.ContentRemoved;
versionChanged = reduceResponse.VersionChanged;
actionError = reduceResponse.Error;
reduceErrorInfo = reduceResponse.ErrorInfo;
linearized = reduceActionConfiguration.FastWebView;
break;
case Operation.OperationType.OCRPDF:
PDFOCRActionConfiguration ocrActionConfiguration = (PDFOCRActionConfiguration)operation.Parameters;
PdfOCRResponse ocrResponse = HandleOCRPDF(pdfApiInstance, ocrActionConfiguration, fileToProcess, fileID, workerNumber);
if (ocrResponse == null)
{
OnError(LogMessagesUtils.ReplaceMessageSequencesAndReferences(FrameworkGlobals.MessagesLocalizer.GetString("message_invalid_response_received", FrameworkGlobals.ApplicationLanguage), actionName: "OCR"));
return(null);
}
remainingTokens = ocrResponse.RemainingTokens;
actionError = ocrResponse.Error;
break;
}
if (actionError != null)
{
string errorMessage = reduceErrorInfo != null && reduceErrorInfo.ErrorCode != ReduceErrorCode.OK ? ErrorManager.GetMessageFromReduceActionError(reduceErrorInfo, fileToProcess.FileAbsolutePath) : ErrorManager.GetMessageFromPassportPDFError(actionError, operation.Type, fileToProcess.FileAbsolutePath);
OnError(errorMessage);
return(null);
}
else
{
RemainingTokensUpdateEventHandler.Invoke(remainingTokens);
}
}
return(new WorkflowProcessingResult(contentRemoved, versionChanged, linearized, fileID, warningMessages));
}
private static void ParseOCRActionConfigurationArgument(string[] args, int index, PDFOCRActionConfiguration ocrActionConfiguration)
{
switch (args[index].ToUpper())
{
case "/L": //language is the next parameter
if (args.Length > index + 1)
{
ocrActionConfiguration.OCRLanguage = args[index + 1].Trim();
}
break;
case "/SKIPPAGESWITHTEXT":
ocrActionConfiguration.SkipPagesWithText = true;
break;
}
}
