private bool DownloadAndSaveDocument(PDFApi pdfApi, ImageApi imageApi, string fileId, int workerNumber, OperationsWorkflow.SaveOperationType saveOperationType, object saveOperationConfiguration, ProgressDelegate downloadOperationStartEventHandler, string inputFileAbsolutePath, out string downloadedDocumentFileName)
{
downloadedDocumentFileName = Path.GetTempFileName();
using (FileStream outputFileStream = new FileStream(downloadedDocumentFileName, FileMode.Open))
{
try
{
if (saveOperationType == OperationsWorkflow.SaveOperationType.SavePDF)
{
PassportPDFRequestsUtilities.DownloadPDF(pdfApi, new PdfSaveDocumentParameters(fileId), workerNumber, inputFileAbsolutePath, outputFileStream, downloadOperationStartEventHandler);
}
else if (saveOperationType == OperationsWorkflow.SaveOperationType.SaveImageAsPDFMRC)
{
PassportPDFRequestsUtilities.DownloadImageAsPDFMRC(imageApi, PassportPDFParametersUtilities.GetImageSaveAsPDFMRCParameters((ImageSaveAsPDFMRCActionConfiguration)saveOperationConfiguration, fileId),
workerNumber, inputFileAbsolutePath, outputFileStream, downloadOperationStartEventHandler);
}
return(true);
}
catch (Exception ex)
{
OnError(LogMessagesUtils.ReplaceMessageSequencesAndReferences(FrameworkGlobals.MessagesLocalizer.GetString("message_output_file_download_failure", FrameworkGlobals.ApplicationLanguage), inputFileAbsolutePath, additionalMessage: ex.Message));
return(false);
}
}
}
public static void DownloadPDF(PDFApi apiInstance, PdfSaveDocumentParameters saveDocumentParameters, int workerNumber, string inputFilePath, Stream destinationStream, OperationsManager.ProgressDelegate downloadOperationStartEventHandler)
{
Exception e = null;
int pausems = 5000;
for (int i = 0; i < FrameworkGlobals.MAX_RETRYING_REQUESTS; i++)
{
downloadOperationStartEventHandler.Invoke(workerNumber, inputFilePath, i);
try
{
apiInstance.SaveDocumentToFile(saveDocumentParameters, destinationStream);
return;
}
catch (Exception ex)
{
if (i < FrameworkGlobals.MAX_RETRYING_REQUESTS - 1)
{
Thread.Sleep(pausems); //marking a pause in case of cnx temporarily out and to avoid overhead.
pausems += 2000;
}
else
{//last iteration
e = ex;
}
}
}
throw e;
}
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);
}
private PdfLoadDocumentResponse HandleLoadPDF(PDFApi pdfApiInstance, PdfVersion outputVersion, FileToProcess fileToProcess, int workerNumber)
{
FileStream inputFileStream = null;
try
{
PassportPDFParametersUtilities.GetLoadDocumentMultipartParameters(fileToProcess.FileAbsolutePath, outputVersion, out inputFileStream, out PdfConformance conformance, out string fileName);
using (FileStream tmpFile = File.Create(Path.GetTempFileName(), 4096, FileOptions.DeleteOnClose))
{
using (GZipStream dataStream = new GZipStream(tmpFile, CompressionLevel.Optimal, true))
{
inputFileStream.CopyTo(dataStream);
inputFileStream.Dispose();
inputFileStream = null;
}
tmpFile.Seek(0, SeekOrigin.Begin);
return(PassportPDFRequestsUtilities.SendLoadPDFMultipartRequest(pdfApiInstance, workerNumber, fileToProcess.FileAbsolutePath, fileName, conformance, fileToProcess.Password, tmpFile, ContentEncoding.Gzip, UploadOperationStartEventHandler));
}
}
catch
{
if (inputFileStream != null)
{
inputFileStream.Dispose();
}
throw;
}
}
public void Init()
{
Configuration.Default.Username = "******";
Configuration.Default.Password = "******";
Configuration.Default.BasePath = "http://api.docspring.local:31337/api/v1";
instance = new PDFApi();
}
public static PdfGetInfoResponse SendGetInfoRequest(PDFApi apiInstance, PdfGetInfoParameters getInfoParameters, int workerNumber, string inputFilePath, OperationsManager.ProgressDelegate getInfoOperationStartEventHandler)
{
Exception e = null;
int pausems = 5000;
for (int i = 0; i < FrameworkGlobals.MAX_RETRYING_REQUESTS; i++)
{
getInfoOperationStartEventHandler.Invoke(workerNumber, inputFilePath, i);
try
{
return(apiInstance.GetInfo(getInfoParameters));
}
catch (Exception ex)
{
if (i < FrameworkGlobals.MAX_RETRYING_REQUESTS - 1)
{
Thread.Sleep(pausems); //marking a pause in case of cnx temporarily out and to avoid overhead.
pausems += 2000;
}
else
{//last iteration
e = ex;
}
}
}
throw e;
}
public static PdfOCRResponse SendOCRRequest(PDFApi apiInstance, PdfOCRParameters ocrParameters, int workerNumber, string inputFilePath, string pageRange, int pageCount, OperationsManager.ChunkProgressDelegate chunkProgressEventHandler)
{
Exception e = null;
int pausems = 5000;
for (int i = 0; i < FrameworkGlobals.MAX_RETRYING_REQUESTS; i++)
{
chunkProgressEventHandler.Invoke(workerNumber, inputFilePath, pageRange, pageCount, i);
try
{
PdfOCRResponse response = apiInstance.OCR(ocrParameters);
return(response);
}
catch (Exception ex)
{
if (i < FrameworkGlobals.MAX_RETRYING_REQUESTS - 1)
{
Thread.Sleep(pausems); //marking a pause in case of cnx temporarily out and to avoid overhead.
pausems += 2000;
}
else
{//last iteration
e = ex;
}
}
}
throw e;
}
public static string[] GetPdfApiSupportedFileExtensions()
{
PDFApi apiInstance = new PDFApi(FrameworkGlobals.PassportPdfApiUri);
Exception e = null;
int pauseMs = 5000;
for (int i = 0; i < FrameworkGlobals.MAX_RETRYING_REQUESTS; i++)
{
try
{
return(apiInstance.GetPDFImportSupportedFileExtensions().Value.ToArray());
}
catch (Exception ex)
{
if (i < FrameworkGlobals.MAX_RETRYING_REQUESTS - 1)
{
Thread.Sleep(pauseMs); //marking a pause in case of cnx temporarily out and to avoid overhead.
pauseMs += 2000;
}
else
{//last iteration
e = ex;
}
}
}
throw e;
}
private void Process(PDFApi pdfApi, ImageApi imageApi, int workerNumber, FileProductionRules fileProductionRules, OperationsWorkflow workflow, string destinationFolder, bool fileSizeReductionIsIntended)
{
while (PickFile(out FileToProcess fileToProcess))
{
if (_cancellationPending)
{
break;
}
try
{
long inputFileSize = FileUtils.GetFileSize(fileToProcess.FileAbsolutePath);
bool inputIsPDF = Path.GetExtension(fileToProcess.FileAbsolutePath).ToUpper() == ".PDF";
if (!CheckInputFileSizeValidity(inputFileSize, fileToProcess.FileAbsolutePath))
{
continue;
}
WorkflowProcessingResult workFlowProcessingResult = ProcessWorkflow(pdfApi, imageApi, workflow, fileToProcess, workerNumber);
if (workFlowProcessingResult != null)
{
string outputFileAbsolutePath = destinationFolder + fileToProcess.FileRelativePath;
bool fileSuccesfullyProcessed = HandleOutputFileProduction(fileToProcess, workFlowProcessingResult.FileID, workerNumber, workflow.SaveOperation, workflow.SaveOperationConfiguration, pdfApi, imageApi, fileProductionRules, workFlowProcessingResult, fileSizeReductionIsIntended, inputIsPDF, inputFileSize, outputFileAbsolutePath);
TryCloseDocumentAsync(pdfApi, workFlowProcessingResult.FileID);
if (fileSuccesfullyProcessed)
{
OnFileSuccesfullyProcessed(new FileOperationsResult(fileToProcess.FileAbsolutePath, inputFileSize, FileUtils.GetFileSize(outputFileAbsolutePath), !inputIsPDF), workFlowProcessingResult.WarningMessages);
}
else
{
continue;
}
}
}
catch (Exception exception)
{
OnError(ErrorManager.GetMessageFromException(exception, fileToProcess.FileAbsolutePath));
}
if (_workPaused && !_cancellationPending)
{
// If pause has been requested, wait for resume signal
WorkerPauseEventHandler.Invoke(workerNumber);
_waitHandle.WaitOne();
}
}
OnWorkerWorkCompletion(workerNumber);
}
private void InitializeApiInstances(out PDFApi pdfApiInstance, out ImageApi imageApiInstance, string apiKey)
{
pdfApiInstance = new PDFApi(apiKey)
{
BasePath = FrameworkGlobals.PassportPdfApiUri
};
imageApiInstance = new ImageApi(apiKey)
{
BasePath = FrameworkGlobals.PassportPdfApiUri
};
PassportPDF.Client.GlobalConfiguration.Timeout = FrameworkGlobals.PassportPDFConfiguration.SuggestedClientTimeout;
}
private PdfReduceResponse HandleReducePDF(PDFApi pdfApiInstance, PDFReduceActionConfiguration actionConfiguration, FileToProcess fileToProcess, string fileID, int workerNumber, List <string> warnings)
{
PdfReduceParameters reduceParameters = PassportPDFParametersUtilities.GetReduceParameters(actionConfiguration, fileID);
PdfReduceResponse reduceResponse = PassportPDFRequestsUtilities.SendReduceRequest(pdfApiInstance, reduceParameters, workerNumber, fileToProcess.FileAbsolutePath, FileOperationStartEventHandler);
if (reduceResponse.WarningsInfo != null)
{
foreach (ReduceWarningInfo warning in reduceResponse.WarningsInfo)
{
warnings.Add(LogMessagesUtils.GetWarningStatustext(warning, fileToProcess.FileAbsolutePath));
}
}
return(reduceResponse);
}
public static PdfLoadDocumentResponse SendLoadDocumentRequest(PDFApi apiInstance, int workerNumber, string inputFilePath, string fileName, PdfConformance conformance, string password, Stream fileStream, ContentEncoding contentEncoding, OperationsManager.ProgressDelegate uploadOperationStartEventHandler)
{
Exception e = null;
int pausems = 5000;
if (fileStream.Length > int.MaxValue)
{
throw new OutOfMemoryException();
}
for (int i = 0; i < FrameworkGlobals.MAX_RETRYING_REQUESTS; i++)
{
uploadOperationStartEventHandler.Invoke(workerNumber, inputFilePath, i);
try
{
fileStream.Seek(0, SeekOrigin.Begin);
byte[] data = new byte[fileStream.Length];
fileStream.Read(data, 0, (int)fileStream.Length);
PdfLoadDocumentFromByteArrayParameters pdfLoadDocumentFromByteArrayParameters = new PdfLoadDocumentFromByteArrayParameters(data)
{
FileName = fileName,
Password = password,
Conformance = conformance,
ContentEncoding = contentEncoding
};
PdfLoadDocumentResponse response = apiInstance.LoadDocumentAsPDF(pdfLoadDocumentFromByteArrayParameters);
return(response);
}
catch (Exception ex)
{
if (i < FrameworkGlobals.MAX_RETRYING_REQUESTS - 1)
{
Thread.Sleep(pausems); //marking a pause in case of cnx temporarily out and to avoid overhead.
pausems += 2000;
}
else
{//last iteration
e = ex;
}
}
}
throw e;
}
private static async void TryCloseDocumentAsync(PDFApi pdfApiInstance, string fileID)
{
if (string.IsNullOrWhiteSpace(fileID))
{
throw new ArgumentNullException("FileID");
}
PdfCloseDocumentParameters closeDocumentParameters = new PdfCloseDocumentParameters(fileID);
try
{
await pdfApiInstance.ClosePDFAsync(closeDocumentParameters); //we do not want to stop the process by waiting such response.
}
catch
{
return;
}
}
public static PdfLoadDocumentResponse SendLoadPDFMultipartRequest(PDFApi apiInstance, int workerNumber, string inputFilePath, string fileName, PdfConformance conformance, string password, Stream fileStream, ContentEncoding contentEncoding, OperationsManager.ProgressDelegate uploadOperationStartEventHandler)
{
Exception e = null;
int pausems = 5000;
for (int i = 0; i < FrameworkGlobals.MAX_RETRYING_REQUESTS; i++)
{
uploadOperationStartEventHandler.Invoke(workerNumber, inputFilePath, i);
try
{
fileStream.Seek(0, SeekOrigin.Begin);
PdfLoadDocumentResponse response = apiInstance.LoadDocumentAsPDFMultipart(fileStream,
new PdfLoadDocumentParameters()
{
ContentEncoding = contentEncoding,
Conformance = conformance,
Password = password,
FileName = fileName
});
return(response);
}
catch (Exception ex)
{
if (i < FrameworkGlobals.MAX_RETRYING_REQUESTS - 1)
{
Thread.Sleep(pausems); //marking a pause in case of cnx temporarily out and to avoid overhead.
pausems += 2000;
}
else
{//last iteration
e = ex;
}
}
}
throw e;
}
private bool HandleOutputFileProduction(FileToProcess fileToProcess, string fileId, int workerNumber, OperationsWorkflow.SaveOperationType saveOperationType, object saveOperationConfiguration, PDFApi pdfApi, ImageApi imageApi, FileProductionRules fileProductionRules, WorkflowProcessingResult workflowProcessingResult, bool fileSizeReductionIsIntended, bool inputIsPDF, long inputFileSize, string outputFileAbsolutePath)
{
if (!DownloadAndSaveDocument(pdfApi, imageApi, fileId, workerNumber, saveOperationType, saveOperationConfiguration, DownloadOperationStartEventHandler, fileToProcess.FileAbsolutePath, out string downloadedDocumentFileName))
{
return(false);
}
bool keepProducedFile = MustProducedFileBeKept(workflowProcessingResult, fileSizeReductionIsIntended, inputIsPDF, inputFileSize, FileUtils.GetFileSize(downloadedDocumentFileName));
bool outputIsInput = FileUtils.AreSamePath(fileToProcess.FileAbsolutePath, outputFileAbsolutePath);
if (keepProducedFile)
{
if (fileProductionRules.DeleteOriginalFileOnSuccess && !outputIsInput)
{
try
{
FileUtils.DeleteFile(fileToProcess.FileAbsolutePath);
}
catch (Exception exception)
{
OnWarning(LogMessagesUtils.ReplaceMessageSequencesAndReferences(FrameworkGlobals.MessagesLocalizer.GetString("message_original_file_deletion_failure", FrameworkGlobals.ApplicationLanguage), fileName: fileToProcess.FileAbsolutePath, additionalMessage: exception.Message));
}
}
FileUtils.MoveFile(downloadedDocumentFileName, outputFileAbsolutePath);
File.SetCreationTime(outputFileAbsolutePath, File.GetCreationTime(fileToProcess.FileAbsolutePath));
}
else
{
if (!outputIsInput)
{
FileUtils.CopyFile(fileToProcess.FileAbsolutePath, outputFileAbsolutePath);
}
if (fileSizeReductionIsIntended)
{
// Inform file size reduction failure
workflowProcessingResult.WarningMessages.Add(LogMessagesUtils.GetWarningStatustext(new ReduceWarningInfo()
{
WarningCode = ReduceWarningCode.FileSizeReductionFailure
}, fileToProcess.FileAbsolutePath));
}
FileUtils.DeleteFile(downloadedDocumentFileName);
}
if (fileProductionRules.KeepWriteAndAccessTime)
{
FileUtils.SetOriginalLastAccessTime(fileToProcess.FileAbsolutePath, outputFileAbsolutePath);
}
return(true);
}
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));
}
