ImageExtractionSample.cs :  » PDF » PDF-Clown » it » stefanochizzolini » clown » samples » C# / CSharp Open Source

using it.stefanochizzolini.clown.bytes;
using it.stefanochizzolini.clown.documents;
using filesit.stefanochizzolini.clown.files;
using it.stefanochizzolini.clown.objects;

using System;
using System.IO;

namespace it.stefanochizzolini.clown.samples{
  /**
    <summary>This sample demonstrates how to extract XObject images from a PDF file.</summary>
    <remarks>
      <para>Inline images are ignored.</para>
      <para>XObject images other than JPEG aren't currently supported for handling.</para>
    </remarks>
  */
  public class ImageExtractionSample
    : ISample
  {
    #region dynamic
    #region interface
    #region public
    #region ISample
    public void Run(
      SampleLoader loader
      )
    {
      // (boilerplate user choice -- ignore it)
      string filePath = loader.GetPdfFileChoice("Please select a PDF file");

      // 1. Open the PDF file!
      files::File file = new files::File(filePath);

      // 2.1. Iterating through the indirect object collection...
      int index = 0;
      foreach(PdfIndirectObject indirectObject in file.IndirectObjects)
      {
        // Get the data object associated to the indirect object!
        PdfDataObject dataObject = indirectObject.DataObject;
        // Is this data object a stream?
        if(dataObject is PdfStream)
        {
          PdfDictionary header = ((PdfStream)dataObject).Header;
          // Is this stream an image?
          if(header.ContainsKey(PdfName.Type)
            && header[PdfName.Type].Equals(PdfName.XObject)
            && header[PdfName.Subtype].Equals(PdfName.Image))
          {
            // 2.1.1. Is this image a pass-through JPEG image?
            if(header[PdfName.Filter].Equals(PdfName.DCTDecode)) // JPEG image.
            {
              // Get the image data (keeping it encoded)!
              IBuffer body = ((PdfStream)dataObject).GetBody(false);
              // Export the image!
              ExportImage(
                body,
                loader.OutputPath + Path.DirectorySeparatorChar + "ImageExtractionSample_" + (index++) + ".jpg"
                );
            }
            else // Unsupported image.
            {
              Console.WriteLine("Image XObject " + indirectObject.Reference + " couldn't be extracted (filter: " + header[PdfName.Filter] + ")");
            }
          }
        }
      }
    }
    #endregion
    #endregion

    #region private
    private void ExportImage(
      IBuffer data,
      string outputPath
      )
    {
      FileStream outputStream;
      try
      {outputStream = new FileStream(outputPath, FileMode.CreateNew);}
      catch(Exception e)
      {throw new Exception(outputPath + " file couldn't be created.",e);}

      try
      {
        BinaryWriter writer = new BinaryWriter(outputStream);
        writer.Write(data.ToByteArray());
        writer.Close();
        outputStream.Close();
      }
      catch(Exception e)
      {throw new Exception(outputPath + " file writing has failed.",e);}

      Console.WriteLine("Output: " + outputPath);
    }
    #endregion
    #endregion
    #endregion
  }
}