private async Task <PdfPageSizeInfo> GetPdfPageSize(string pdfFilePath)
{
var info = await GetPdfInfo(pdfFilePath);
var infos = info.Split(Environment.NewLine.ToCharArray(), StringSplitOptions.RemoveEmptyEntries);
var sizeString = infos.Where(x => x.StartsWith("Page size:")).FirstOrDefault();
var retVal = new PdfPageSizeInfo();
retVal.OriginalSizeString = sizeString;
var splitSize = sizeString.Split(new char[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
retVal.SizeX = float.Parse(splitSize[2].Replace('.', ','), new System.Globalization.CultureInfo("de-DE"));
retVal.SizeY = float.Parse(splitSize[4].Replace('.', ','), new System.Globalization.CultureInfo("de-DE"));
retVal.Unit = splitSize[5];
return(retVal);
}
/// <summary>
/// Gets text from a PDF based that is within a given area.
/// </summary>
/// <param name="pdfFilePath">PDF location on disk.</param>
/// <param name="cropAreaInfo">Percentual area which is to be extracted.</param>
/// <param name="pdfPageSizeInfo">Size information of the PDF file. Used to calculate absolute area from percental area.</param>
/// <returns></returns>
public async Task <string> GetTextFromPdf(string pdfFilePath, PercentalAreaInfo cropAreaInfo, PdfPageSizeInfo pdfPageSizeInfo)
{
var pdfInfo = pdfPageSizeInfo;
var x = (int)Math.Round(cropAreaInfo.TopLeftX * pdfInfo.SizeX, 0);
var y = (int)Math.Round(cropAreaInfo.TopLeftY * pdfInfo.SizeY, 0);
var W = (int)Math.Round(cropAreaInfo.Width * pdfInfo.SizeX, 0);
var H = (int)Math.Round(cropAreaInfo.Height * pdfInfo.SizeY, 0);
var pdfToTextOptions = " -f " + cropAreaInfo.PageNumber + " -l " + cropAreaInfo.PageNumber + " -x " + x + " -y " + y + " -W " + W + " -H " + H + " -layout -nopgbrk ";
var retVal = await GetTextFromPdf(pdfFilePath, false, pdfToTextOptions);
// Remove last line break, as it is added by poppler and does not represent the selected area
if (retVal.Length > 1)
{
retVal = retVal.Remove(retVal.Length - 2);
}
if (retVal is null)
{
retVal = string.Empty;
}
return(retVal);
}
