View PDF files in C# using the Xpdf and muPDF library, Print PostScript.

• Download code from SVN (Code Google)   

Introduction 

Xpdf is an Open Source library released under GPL license; they have an ActiveX with commercial license, but some time ago, before I knew about this commercial control, I wrote this wrapper library to render PDF files in C#. 

Background

The basic idea is create a preview of PDF files in C#. After looking at many places in the internet, I found this wonderful library; the only problem is that the library uses XLib, and there is no XLib available for Windows. Fortunately, Xpdf can render the generated PDF into a Win32 DC.

Writting the wrapper

C++\CLI can mix managed and unmanaged code thanks to the IJW technology, so I was thinking that maybe I could to link the xpdf lib to the wrapper. The problem is that xpdf has some classes that are also declared in .Net, the solution was compile a C++ project with a class that includes only the necessary files to do the interop.
This library (AFPDFLib) contains a simple class that works like a proxy between C++ and C++\CLI, keeping the xpdf objects into the unmanaged Heap.
The C# wrapper is linked to AFPDFLib statically, and this only includes:
AFPDFDocInterop.h
OutlineItemInterop.h
SearchResultInterop.h
The classes:
AFPDFDoc -> Implement the methods that needs xpdf.
AFPDFDocInterop -> Write the methods to wrap into C#
PDFWrapper -> Wrapped methods

Marshal Strings:

IntPtr ptr = Marshal::StringToCoTaskMemAnsi(fileName);
char *singleByte= (char*)ptr.ToPointer();
try{
}finally{
     Marshal::FreeCoTaskMem(ptr);
}   

For releasing resources is necessary implement IDisposable:

!PDFWrapper()
{
   _pdfDoc->Dispose();
} 

Using the code  

The file xpdfWin-Interop.sln includes all the necessary files, you can also download the last version from http://www.foolabs.com/xpdf/ and recompile without the files that requires XLib.

The Build Project Order is as follows: freetype,xpdf,AFPDFLib, PDFLibNet. Once compiled PDFLibNet, it can be used in C# code:    

OpenFileDialog dlg = new OpenFileDialog();
dlg.Filter = "Portable Document Format (*.pdf)|*.pdf";
if (dlg.ShowDialog() == DialogResult.OK)
{
    _pdfDoc = new PDFLibNet.PDFWrapper();
    _pdfDoc.LoadPDF(dlg.FileName);
    _pdfDoc.CurrentPage = 1;

   PictureBox pic =new PictureBox();
   pic.Width=800;
   pic.Height=1024;
   _pdfDoc.FitToWidth(pic.Handle);
   pic.Height = _pdfDoc.PageHeight;
   _pdfDoc.RenderPage(pic.Handle);
   
   Bitmap _backbuffer = new Bitmap(_pdfDoc.PageWidth, _pdfDoc.PageHeight);
   using (Graphics g = Graphics.FromImage(_backbuffer))
   {
       _pdfDoc.RenderHDC(g.GetHdc);
       g.ReleaseHdc();
   }
   pic.Image = _backbuffer;
}   

It is necessary create a PictureBox because the class implements only a method that accepts an HWND, because in the first instance, I was trying to implement the scroll into the same control that the PDF is rendered. In the included sample, the scroll is controlled by a Panel container.

Xpdf can export the PDF to a PostScript file. For printing this is the best option if you have a PostScript Printer:       

PSOutputDev *psOut =new PSOutputDev((char *)fileName,m_PDFDoc->getXRef(),m_PDFDoc->getCatalog(),fromPage,toPage,psModePS);
if(psOut->isOk()){
    m_PDFDoc->displayPages(psOut,fromPage,toPage,PRINT_DPI,PRINT_DPI,0,gTrue,globalParams->getPSCrop(),gTrue);
}
delete psOut;  

The file must be sended in RAW format (http://support.microsoft.com/kb/322091)

JPG Export 

For async export: 

 _doc.ExportJpg(filename, 
1,        //From page
1,        //To page
150,      //Resolution in DPI
90,       //Jpg quality 

if you need a sync operation its posible especify a wait time:

 _doc.ExportJpg(filename, 
1,        //From page
1,        //To page
150,      //Resolution in DPI
90,       //Jpg quality
-1);      //Time to wait, -1 to infinite.

 If the file name does not contains a %d token (for the page number), then the procedure replaces .jpg with -page%d.jpg.  PDFWrapper exposes two events ExportJpgProgress and ExportJpgFinished. Both events are called from the exporting Thread, so it is necesary to make a security call using Invoke, check frmExportJpg for a sample.  

History:    

06\July\2009:

• Full deployed solution.
• Updated to xpdf 3.0.2 version.
• FreeType updated to 2.3.1
• When click in a bookmark and search, the page scroll to the correct position.
• PostScript implemented. 
• Now gets the Title, Author.  

08\July\2009

• Some memory leaks corrected 
• Prerender next page in new thread 
• Cache of pages  
• Mouse Scrolling 
• Mouse Navigation 
• Load links from page (LinkURI, LinkGoTo) 

11\July\2009 

• Using DIB Sections,  fixes the problem with the zoom.
• Added control PageViewer, now render only the viewable area
• Open password protected files.
• Export to txt
• Export to jpg

12\July\2009

• Added support for Unicode in Bookmarks, title, subject, keywords...
• Added support for named destinations

13\July\2009

• Fixed some bugs.
• Added support for unicode search

20\July\2009  

• Multithread jpg export
• Fixed others bugs

07\Nov\2009 

• Added MuPDF as second renderer. 

 26\NOV\2010

• Added support for swf convert using swftools (http://www.swftools.org/)
• Added support for html convert using html2text (http://pdftohtml.sourceforge.net/), I modified the HtmlOuputDev because originally it uses ghoscript to render the background. Now it uses SplashOutputDev. 

Know issues:     MuPDF has some problems with transparency, but is faster than xpdf.  A couple of memory leaks.

IMPORTANT:         

MuPDF uses recursion for analyze the tree document, so is necessary increment the Stack Size to at least 4mb to avoid problems with some complex files (editbin for C#, VB.Net exe's). Soon or later the recursion causes an stack overflow if the tree is so big, so while it is fixed that is the most important issue.

To Do:        

- Apply last xpdf patche - Show multiple pages in the viewer. - Improve user interface. - Implement LoadFromStream for MuPDF. There is missing some functionality that can be extracted from xpdf: - Enable selection, image extraction and instant snapshot. - Print in non PostScript printers.