PDF XChange Forum

Code: Select all

'declare current page 
Dim doc As IPXV_Document = pdfCtl.Doc
Dim pl As PDFXEdit.IPXV_PagesLayoutManager = doc.ActiveView.PagesView.Layout
Dim pageNr = pl.CurrentPage
Dim curPage As PDFXEdit.IPXC_Page = doc.CoreDoc.Pages(pageNr)

'set the rect coordination
Dim rect As PXC_Rect
rect.left = dblLeft
rect.right = dblRight
rect.top = dblTop
rect.bottom = dblBottom
 
'declare for QCR (bottom part is no working we assume should assign the OCR to document we declare at upper part)
Dim OCRExt As IPXV_OCRExtension
Dim OCRRegions As IPXV_OCRRegions
Dim OCRRegion As IPXV_OCRRegion
Dim OCRTask As IPXV_OCRTask
Dim OCRResult As IPXV_OCRResult

OCRRegions = OCRExt.CreateOCRRegions()
OCRRegion = OCRRegions.InsertNew(PXV_OCRBlockType.PXV_OCRBlock_Text, rect)
OCRTask = OCRExt.CreateNewTask()
OCRResult = OCRTask.ProceedPage(curPage, OCRRegion, ,)
txtResult.Text = OCRResult.GetText(0, 0, )

Code: Select all

function TPXC_PageWraper.ExtractPageText(const fromRect: PXC_Rect): string;
const
  DELTA = 2;
var
  lineString : string;
  i, j, k : integer;
  tli : PXC_TextLineInfo;
  lineChars, nextChar : WideString;
  fcRect, charRect : PXC_Rect;
begin
  result := '';

  for i := 0 to -1 + fPageLineCount do
  begin
    fPageText.Get_LineInfo(i, tli);

    fPageText.GetChars(tli.nFirstCharIndex, tli.nCharsCount, lineChars);

    lineString := Trim(StringReplace(lineChars, ' ', '', [rfReplaceAll, rfIgnoreCase]));

    if lineString <> '' then
    begin
      fPageText.Get_CharRect(tli.nFirstCharIndex, fcRect);

      if (fcRect.top <= fromRect.top + DELTA) AND (fcRect.bottom >= fromRect.bottom - DELTA) then
      begin
        if result <> '' then result := result + ' ';

        for j := tli.nFirstCharIndex to tli.nFirstCharIndex + tli.nCharsCount do
        begin
          fPageText.Get_CharRect(j, charRect); //first char in line - could be lower than the rest

          if charRect.left >= fromRect.left - DELTA then
          begin
            k := j;
            while (charRect.right <= fromRect.right + DELTA) AND (k < tli.nFirstCharIndex + tli.nCharsCount) do
            begin
              fPageText.GetChars(k, 1, nextChar);

              result := result + nextChar;

              Inc(k);
              fPageText.Get_CharRect(k, charRect);
            end;
            Break; //for j
          end;
        end;
      end;
    end;
  end;

  result := Trim(result);
end;

Code: Select all

constructor TPXC_PageWraper.Create(const thePage: IPXC_Page);
begin
  fPage := thePage;

  fPage.GetText(nil, true, fPageText);

  fPageText.Get_LinesCount(fPageLineCount);
end;

Code: Select all

	Dim nId As Integer = pdfCtl.Inst.Str2ID("op.document.OCRPages", False)
        Dim Op As IOperation = pdfCtl.Inst.CreateOp(nId)
        Dim Input As ICabNode = Op.Params.Root
        Input("Input").v = pdfCtl.Doc

        Dim options As ICabNode = Input("Options")

        options("PagesRange.Type").v = "All"
        options("OutputType").v = 0
        options("OutputDPI").v = 300

        pdfCtl.Inst.AsyncDoAndWaitForFinish(Op) //hit error here