public void Execute(Arguments arguments)
{
manager = AbbyyManager.Instance;
System.Drawing.Rectangle rectangle = !arguments.Relative.Value ? arguments.Area.Value : arguments.Area.Value.ToAbsoluteCoordinates();
System.Drawing.Bitmap partOfScreen = RobotWin32.GetPartOfScreen(rectangle);
IEngine engine = manager.Engine;
DocumentProcessingParams processingParams = engine.CreateDocumentProcessingParams();
RecognizerParams recognizingParams = processingParams.PageProcessingParams.RecognizerParams;
recognizingParams.SetPredefinedTextLanguage(arguments.Language.Value);
engine.LoadPredefinedProfile(AbbyyManager.TextAccuracyProfile);
FRDocument imageDocument = engine.CreateFRDocument();
using (System.IO.MemoryStream stream = new System.IO.MemoryStream())
{
partOfScreen.Save(stream, System.Drawing.Imaging.ImageFormat.Bmp);
stream.Position = 0;
IReadStream imageStream = new StreamNet2AbbyyAdapter(stream);
imageDocument.AddImageFileFromStream(imageStream);
}
imageDocument.Process(processingParams);
Scripter.Variables.SetVariableValue(arguments.Result.Value, new TextStructure(imageDocument.PlainText.Text));
}
public async Task ProcessDocument_ShouldBeOk()
{
// Arrange
var submitImageTask = await SubmitImageAsync(TestFile.Image);
var parameters = new DocumentProcessingParams
{
TaskId = submitImageTask.TaskId,
Language = "English",
ExportFormats = new[]
{
ExportFormat.PdfSearchable,
ExportFormat.Rtf,
},
};
// Act
var processDocumentTask = await ApiClient.ProcessDocumentAsync(
parameters,
true
);
// Assert
CheckResultTask(processDocumentTask, submitImageTask.TaskId, 2);
}
private DocumentProcessingParams GetProcessingParameters(string language, SettingsMode mode, List <string> dictionary, int languageWeight, int dictionaryWeight)
{
engine.LoadPredefinedProfile(DocumentConversionAccuracyProfile);
DocumentProcessingParams processingParams = engine.CreateDocumentProcessingParams();
SetPageProcessingParameters(processingParams, language, dictionary, languageWeight, dictionaryWeight);
SetSplittingParameters(processingParams, mode);
SetDocumentSynthesisParameters(processingParams, mode);
return(processingParams);
}
/// <inheritdoc />
public Task <TaskInfo> ProcessDocumentAsync(
DocumentProcessingParams parameters,
bool waitTaskFinished = false,
CancellationToken cancellationToken = default)
{
return(StartTaskAsync(
HttpMethod.Post,
Urls.Ocr.ProcessDocument,
parameters,
null,
null,
waitTaskFinished: waitTaskFinished,
cancellationToken: cancellationToken));
}
private void SetPageProcessingParameters(DocumentProcessingParams processingParams, string language, List <string> dictionary, int languageWeight, int dictionaryWeight)
{
PageProcessingParams pageProcessingParams = processingParams.PageProcessingParams ?? engine.CreatePageProcessingParams();
processingParams.PageProcessingParams = pageProcessingParams;
SetPreProcessingParameters(pageProcessingParams);
SetObjectExtractionParameters(pageProcessingParams);
SetPageAnalyzeParameters(pageProcessingParams);
SetPageSynthesisParameters(pageProcessingParams);
SetRecognitionParameters(language, pageProcessingParams, dictionary, languageWeight, dictionaryWeight);
}
private static async Task <List <string> > ProcessDocumentAsync(IOcrClient ocrClient)
{
var taskId = await UploadFilesAsync(ocrClient);
var processingParams = new DocumentProcessingParams
{
ExportFormats = new[] { ExportFormat.Docx, ExportFormat.Txt, },
Language = "English,French",
TaskId = taskId,
};
var taskInfo = await ocrClient.ProcessDocumentAsync(
processingParams,
waitTaskFinished : true);
return(taskInfo.ResultUrls);
}
private void SetDocumentSynthesisParameters(DocumentProcessingParams processingParams, SettingsMode mode)
{
SynthesisParamsForDocument documentSynthesisParams = processingParams.SynthesisParamsForDocument ?? engine.CreateSynthesisParamsForDocument();
switch (mode)
{
case SettingsMode.None:
documentSynthesisParams.DetectFontFormatting = true;
documentSynthesisParams.DetectDocumentStructure = true;
break;
case SettingsMode.LineByLine:
documentSynthesisParams.DetectFontFormatting = true;
documentSynthesisParams.DetectDocumentStructure = false;
break;
default:
break;
}
SetDocumentStructureDetectionParameters(documentSynthesisParams, mode);
SetFontFormatDetectionParameters(documentSynthesisParams, mode);
}
private void SetSplittingParameters(DocumentProcessingParams processingParams, SettingsMode mode)
{
PageSplittingParams pageSplittingParams = processingParams.SplittingParams ?? engine.CreatePageSplittingParams();
processingParams.SplittingParams = pageSplittingParams;
}
