Question

I would like to open an existing pdf, add some text and then output as content disposition using itext sharp. I have the following code. Where it falls down it is that i want to output as memory stream but need to filestream to open the original file.

Here’s what i have. Obviously defining PdfWriter twice won’t work.

   public static void Create(string path)
    {
        var Response = HttpContext.Current.Response;
        Response.Clear();
        Response.ContentType = "application/pdf";
        System.IO.MemoryStream m = new System.IO.MemoryStream();
        Document document = new Document();
        PdfWriter wri = PdfWriter.GetInstance(document, new FileStream(path, FileMode.Create));
        PdfWriter.GetInstance(document, m);
        document.Open();
        document.Add(new Paragraph(DateTime.Now.ToString()));
        document.NewPage();
        document.Add(new Paragraph("Hello World"));
        document.Close();
        Response.OutputStream.Write(m.GetBuffer(), 0, m.GetBuffer().Length);
        Response.OutputStream.Flush();
        Response.OutputStream.Close();
        Response.End();
    } 

Answer 1

You’ve got a couple of problems that I’ll try to walk you through.

First, the Document object is only for working with new PDFs, not modifying existing ones. Basically the Document object is a bunch of wrapper classes that abstract away the underlying parts of the PDF spec and allow you to work with higher level things like paragraphs and reflowable content. These abstractions turn what you think of “paragraphs” into raw commands that write the paragraph one line at a time with no relationship between lines. When working with an existing document there’s no safe way to say how to reflow text so these abstractions aren’t used.

Instead you want to use the PdfStamper object. When working with this object you have two choices for how to work with potentially overlapping content, either your new text gets written on top of existing content or your text gets written below it. The two methods GetOverContent() or GetUnderContent() of an instantiated PdfStamper object will return a PdfContentByte object that you can then write text with.

There’s two main ways to write text, either manually or through a ColumnText object. If you’ve done HTML you can think of the ColumnText object as using a big fixed-position single row, single column <TABLE>. The advantage of the ColumnText is that you can use the higher level abstractions such as Paragraph.

Below is a full working C# 2010 WinForms app targeting iTextSharp 5.1.2.0 that show off the above. See the code comments for any questions. It should be pretty easy to convert this to ASP.Net.

using System;
using System.IO;
using System.Windows.Forms;
using iTextSharp.text;
using iTextSharp.text.pdf;

namespace WindowsFormsApplication1 {
    public partial class Form1 : Form {
        public Form1() {
            InitializeComponent();
        }

        private void Form1_Load(object sender, EventArgs e) {
            string existingFile = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Desktop), "file1.pdf");
            string newFile = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Desktop), "file2.pdf");
            using (FileStream fs = new FileStream(existingFile, FileMode.Create, FileAccess.Write, FileShare.None)) {
                using (Document doc = new Document(PageSize.LETTER)) {
                    using (PdfWriter writer = PdfWriter.GetInstance(doc, fs)) {
                        doc.Open();

                        doc.Add(new Paragraph("This is a test"));

                        doc.Close();
                    }
                }
            }

            //Bind a PdfReader to our first document
            PdfReader reader = new PdfReader(existingFile);
            //Create a new stream for our output file (this could be a MemoryStream, too)
            using (FileStream fs = new FileStream(newFile, FileMode.Create, FileAccess.Write, FileShare.None)) {
                //Use a PdfStamper to bind our source file with our output file
                using (PdfStamper stamper = new PdfStamper(reader, fs)) {

                    //In case of conflict we want our new text to be written "on top" of any existing content
                    //Get the "Over" state for page 1
                    PdfContentByte cb = stamper.GetOverContent(1);

                    //Begin text command
                    cb.BeginText();
                    //Set the font information
                    cb.SetFontAndSize(BaseFont.CreateFont(BaseFont.HELVETICA, BaseFont.CP1250, false), 16f);
                    //Position the cursor for drawing
                    cb.MoveText(50, 50);
                    //Write some text
                    cb.ShowText("This was added manually");
                    //End text command
                    cb.EndText();

                    //Create a new ColumnText object to write to
                    ColumnText ct = new ColumnText(cb);
                    //Create a single column who's lower left corner is at 100x100 and upper right is at 500x200
                    ct.SetSimpleColumn(100,100,500,200);
                    //Add a higher level object
                    ct.AddElement(new Paragraph("This was added using ColumnText"));
                    //Flush the text buffer
                    ct.Go();

                }
            }

            this.Close();
        }
    }
}

As to your second problem about the FileStream vs MemoryStream, if you look at the method signature for almost every (actually all as far as I know) method within iTextSharp you’ll see that they all take a Stream object and not just a FileStream object. Any time you see this, even outside of iTextSharp, this means that you can pass in any subclass of Stream which includes the MemoryStream object, everything else stays the same.

The code below is a slightly modified version of the one above. I’ve removed most of the comments to make it shorter. The main change is that we’re using a MemoryStream instead of a FileStream. Also, when we’re done with the PDF when need to close the PdfStamper object before accessing the raw binary data. (The using statment will do this for us automatically later but it also closes the stream so we need to manually do it here.)

One other thing, never, ever use the GetBuffer() method of the MemoryStream. It sounds like what you want (and I have mistakenly used it, too) but instead you want to use ToArray(). GetBuffer() includes uninitialized bytes which usually produces corrupt PDFs. Also, instead of writing to the HTTP Response stream I’m saving the bytes to array first. From a debugging perspective this allows me to finish all of my iTextSharp and System.IO code and make sure that it is correct, then do whatever I want with the raw byte array. In my case I don’t have a web server handy so I’m writing them to disk but you could just as easily call Response.BinaryWrite(bytes)

string existingFile = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Desktop), "file1.pdf");
string newFile = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.Desktop), "file2.pdf");
PdfReader reader = new PdfReader(existingFile);
byte[] bytes;
using(MemoryStream ms = new MemoryStream()){
    using (PdfStamper stamper = new PdfStamper(reader, ms)) {
        PdfContentByte cb = stamper.GetOverContent(1);
        ColumnText ct = new ColumnText(cb);
        ct.SetSimpleColumn(100,100,500,200);
        ct.AddElement(new Paragraph("This was added using ColumnText"));
        ct.Go();

        //Flush the PdfStamper's buffer
        stamper.Close();
        //Get the raw bytes of the PDF
        bytes = ms.ToArray();
    }
}

//Do whatever you want with the bytes
//Below I'm writing them to disk but you could also write them to the output buffer, too
using (FileStream fs = new FileStream(newFile, FileMode.Create, FileAccess.Write, FileShare.None)) {
    fs.Write(bytes, 0, bytes.Length);
}