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

using it.stefanochizzolini.clown.documents;
using it.stefanochizzolini.clown.documents.contents;
using it.stefanochizzolini.clown.documents.contents.objects;
using it.stefanochizzolini.clown.documents.interchange.metadata;
using it.stefanochizzolini.clown.files;
using it.stefanochizzolini.clown.objects;

using System;
using System.Collections.Generic;

namespace it.stefanochizzolini.clown.samples{
  /**
    <summary>This sample demonstrates how to inspect the structure of a PDF
    document.</summary>
    <remarks>
      <para>This implementation is just a limited exercise: see the API documentation
      to perform all the possible access functionalities.</para>
    </remarks>
  */
  public class ParsingSample
    : 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!
      File file = new File(filePath);

      // 2. Parsing the document...
      // Get the PDF document!
      Document document = file.Document;
      // 2.1. Showing basic metadata...
      Console.WriteLine("\nDocument information:");
      Information info = document.Information;
      if(info == null)
      {Console.WriteLine("No information available (Info dictionary doesn't exist).");}
      else
      {
        Console.WriteLine("Author: " + info.Author);
        Console.WriteLine("Title: " + info.Title);
        Console.WriteLine("Subject: " + info.Subject);
        Console.WriteLine("CreationDate: " + info.CreationDate);
      }

      Console.WriteLine("\nIterating through the indirect-object collection (please wait)...");

      // 2.2. Counting the indirect objects, grouping them by type...
      Dictionary<string,int> objCounters = new Dictionary<string,int>();
      objCounters["xref free entry"] = 0;
      foreach(PdfIndirectObject obj in file.IndirectObjects)
      {
        if(obj.IsInUse()) // In-use entry.
        {
          string typeName = obj.DataObject.GetType().Name;
          if(objCounters.ContainsKey(typeName))
          {objCounters[typeName]++;}
          else
          {objCounters[typeName] = 1;}
        }
        else // Free entry.
        {objCounters["xref free entry"]++;}
      }
      Console.WriteLine("\nIndirect objects partial counts (grouped by PDF object type):");
      foreach(KeyValuePair<string,int> keyValuePair in objCounters)
      {Console.WriteLine(" " + keyValuePair.Key + ": " + keyValuePair.Value);}
      Console.WriteLine("Indirect objects total count: " + file.IndirectObjects.Count);

      // 2.3. Showing some page information...
      Pages pages = document.Pages;
      int pageCount = pages.Count;
      Console.WriteLine("\nPage count: " + pageCount);
      int pageIndex = (int)Math.Round((float)pageCount / 2);
      Console.WriteLine("Mid page:");
      PrintPageInfo(pages[pageIndex],pageIndex);

      pageIndex++;
      if(pageIndex < pageCount)
      {
        Console.WriteLine("Next page:");
        PrintPageInfo(pages[pageIndex],pageIndex);
      }

      file.Dispose();
    }

    private void PrintPageInfo(
      Page page,
      int index
      )
    {
      // 1. Showing basic page information...
      Console.WriteLine(" Index (calculated): " + page.Index + " (should be " + index + ")");
      Console.WriteLine(" ID: " + ((PdfReference)page.BaseObject).ID);
      PdfDictionary pageDictionary = page.BaseDataObject;
      Console.WriteLine(" Dictionary entries:");
      foreach(PdfName key in pageDictionary.Keys)
      {Console.WriteLine("  " + key.Value);}

      // 2. Showing page contents information...
      Contents contents = page.Contents;
      Console.WriteLine(" Content objects count: " + contents.Count);
      Console.WriteLine(" Content head (operations):");
      {
        int i = 0,
          count = contents.Count;
        while(i < 10
          && i < count)
        {i = PrintContentObject(contents[i],i,0);}
      }

      // 3. Showing page resources information...
      {
        Resources resources = page.Resources;
        Console.WriteLine(" Resources:");
        try{Console.WriteLine("  Font count: " + resources.Fonts.Count);}catch{}
        try{Console.WriteLine("  XObjects count: " + resources.XObjects.Count);}catch{}
        try{Console.WriteLine("  ColorSpaces count: " + resources.ColorSpaces.Count);}catch{}
      }
    }

    private int PrintContentObject(
      ContentObject content,
      int index,
      int level
      )
    {
      string indentation = new String(' ',level);

      /*
        NOTE: Contents are expressed through both simple operations and composite objects.
      */
      if(content is Operation)
      {Console.WriteLine("   " + indentation + (++index) + ": " + content.ToString());}
      else if(content is ContainerObject)
      {
        Console.WriteLine(
          "   " + indentation + content.GetType().Name
            + "\n   " + indentation + "{"
          );
        foreach(ContentObject obj in ((ContainerObject)content).Objects)
        {if((index = PrintContentObject(obj,index,level+1)) > 9) break;}
        Console.WriteLine("   " + indentation + "}");
      }
      else if(content is GraphicsObject)
      {
        Console.WriteLine(
          "   " + indentation + content.GetType().Name
            + "\n   " + indentation + "{"
          );
        foreach(Operation obj in ((GraphicsObject)content).Objects)
        {if((index = PrintContentObject(obj,index,level+1)) > 9) break;}
        Console.WriteLine("   " + indentation + "}");
      }
      return index;
    }
  }
}