public OcrResult Ocr(string pdfPath, ResultTypeEnum resultTypeEnum, string language = "tur")
{
var result = new OcrResult();
var doc = PDFDoc.Open(pdfPath);
if (!string.IsNullOrEmpty(doc.GetText()))
{
return(result);
}
doc.Ocr(Clock.Utils.OcrMode.Tesseract, language, WriteTextMode.Word, null);
if (resultTypeEnum.HasFlag(ResultTypeEnum.OcrCompressed) || resultTypeEnum.HasFlag(ResultTypeEnum.All))
{
var reader = new PdfReader(doc.PDFBytes);
var stamper = new PdfStamper(reader, new FileStream(Path.Combine(Path.GetTempPath(), "Clock.hocr", Path.GetFileNameWithoutExtension(pdfPath) + "ocrCompressed" + Path.GetExtension(pdfPath)), FileMode.Create), PdfWriter.VERSION_1_5);
var pageNum = reader.NumberOfPages;
for (var i = 1; i <= pageNum; i++)
{
reader.SetPageContent(i, reader.GetPageContent(i));
}
stamper.SetFullCompression();
stamper.Close();
}
if (resultTypeEnum.HasFlag(ResultTypeEnum.Text) || resultTypeEnum.HasFlag(ResultTypeEnum.All))
{
result.Text = doc.GetText();
}
if (resultTypeEnum.HasFlag(ResultTypeEnum.OcrBest) || resultTypeEnum.HasFlag(ResultTypeEnum.All))
{
result.OcredBest = doc.PDFBytes;
}
if (resultTypeEnum.HasFlag(ResultTypeEnum.OcrCompressed) || resultTypeEnum.HasFlag(ResultTypeEnum.All))
{
result.OcredCompressed = System.IO.File.ReadAllBytes(Path.Combine(Path.GetTempPath(), "Clock.hocr", Path.GetFileNameWithoutExtension(pdfPath) + "ocrCompressed" + Path.GetExtension(pdfPath)));
}
_fileHelper.DeleteDirectory(Path.Combine(Path.GetTempPath(), "Clock.hocr"));
return(result);
}
protected override void Execute(NativeActivityContext context)
{
WorkflowDataContext dataContext = context.DataContext;
PropertyDescriptorCollection propertyDescriptorCollection = dataContext.GetProperties();
string sessionId = string.Empty;
int fileSystemItemId = 0;
foreach (PropertyDescriptor propertyDesc in propertyDescriptorCollection)
{
if (propertyDesc.Name == "ArgSessionId")
{
sessionId = propertyDesc.GetValue(dataContext) as string;
break;
}
}
foreach (PropertyDescriptor propertyDesc in propertyDescriptorCollection)
{
if (propertyDesc.Name == "ArgFileSystemItemId")
{
fileSystemItemId = (int)propertyDesc.GetValue(dataContext);
break;
}
}
if (string.IsNullOrWhiteSpace(sessionId))
{
LogHelper.AddGeneralLog(GeneralLogTypeEnum.ActivityError, $"SessionId is null.");
return;
}
if (fileSystemItemId == 0)
{
LogHelper.AddSessionLog(SessionLogTypeEnum.ActivityError, sessionId, $"FileSystemItemId is null.");
return;
}
FileSystemItemDto fileSystemItem = null;
try
{
using (var sqlDbContext = new SqlDbContext())
{
fileSystemItem = sqlDbContext.FileSystemItems.FirstOrDefault(x => x.Id == fileSystemItemId);
}
}
catch (Exception ex)
{
LogHelper.AddSessionLog(SessionLogTypeEnum.ActivityError, sessionId, $"Exception has been thrown when getting fileSystemItem. FileSystemItemId: {fileSystemItemId}", ex);
return;
}
var targetFile = TargetFilePathForTextFile.Get(context);
int counter = 0;
while (true)
{
counter++;
if (File.Exists(targetFile))
{
var directory = Path.GetDirectoryName(targetFile);
var fileName = Path.GetFileNameWithoutExtension(targetFile);
var extension = Path.GetExtension(targetFile);
targetFile = Path.Combine(directory, fileName + counter + extension);
}
else
{
break;
}
}
try
{
TempData.Instance.TempPath = Path.GetTempPath();
using (PDFDoc doc = PDFDoc.Open(fileSystemItem.FullPath))
{
if (doc.GetText() == string.Empty)
{
doc.Ocr(OcrMode.Tesseract, "tur", WriteTextMode.Word);
doc.Save(TargetFilePathForPDFFile.Get(context));
var ocrText = doc.GetText();
File.WriteAllText(TargetFilePathForTextFile.Get(context), ocrText);
}
}
}
catch (Exception ex)
{
LogHelper.AddFileSystemItemLog(fileSystemItemId, fileSystemItem.SourceId, sessionId, FileSystemItemLogTypeEnum.StoppedWithError, $"Exception has been thrown in OCR operation. FileSystemItemId: {fileSystemItemId}", ex);
return;
}
LogHelper.AddFileSystemItemLog(fileSystemItem.Id, fileSystemItem.SourceId, sessionId, FileSystemItemLogTypeEnum.TesseractOcrOk);
Result.Set(context, true);
}