Cannot stamp certain PDFs - pdf

Question

PDF stamping works for nearly every document I have tried. However, a client scanned some pages and his computer generated a PDF document that is resistant to stamping. The embedded image files are in JBIG2 format, but I am not sure if that is important. I have debugged the PDF with Apache's pdfbox, and I can see the text is embedded. It just doesn't show up.
Here is the PDF that won't stamp: http://demo.clearvillageinc.com/plans.pdf
And my code:
static void Main(string[] args) {
string stamp = "<div style=\"color:#F00;\">Reviewed for Code Compliance</div>";
string fileName = #"C:\temp\source.pdf";
string outputFileName = #"C:\temp\source-output.pdf";
// Open a destination stream.
using (var destStream = new System.IO.MemoryStream()) {
using (var sourceReader = new PdfReader(fileName)) {
// Convert the HTML into a stamp.
using (var stampData = FromHtml(stamp)) {
using (var stampReader = new PdfReader(stampData)) {
using (var stamper = new PdfStamper(sourceReader, destStream)) {
stamper.Writer.CloseStream = false;
// Add the stamp stream to the source document.
var stampPage = stamper.GetImportedPage(stampReader, 1);
// Process all of the pages in the source document.
for (int i = 1; i <= sourceReader.NumberOfPages; i++) {
var canvas = stamper.GetOverContent(i);
canvas.AddTemplate(stampPage, 0, -50);
}
}
}
}
}
// Finished. Save the file.
using (var fs = new System.IO.FileStream(outputFileName, FileMode.Create)) {
destStream.Position = 0;
destStream.CopyTo(fs);
}
}
}
public static System.IO.Stream FromHtml(string html) {
var ms = new System.IO.MemoryStream();
// Convert html to pdf.
using (var document = new iTextSharp.text.Document()) {
var writer = iTextSharp.text.pdf.PdfWriter.GetInstance(document, ms);
writer.CloseStream = false;
document.Open();
using (var sr = new System.IO.StringReader(html)) {
XMLWorkerHelper.GetInstance().ParseXHtml(writer, document, sr);
}
}
ms.Position = 0; // Reset for reading.
return ms;
}

Answer 1

One part of a page definition is the "MediaBox" which controls the page's size. This property takes two locations that specify the coordinates of two opposite corners of a rectangle. Although not required, most PDFs specify the lower left corner first followed by the upper right corner. Also, most PDF use 0x0 for the lower left and then whatever the page's width and height for the top corner. So an 8.5x11 inch PDF would be 0,0 and 612,792 (8.5 * 72 = 612 and 11 * 72 = 792) and this would be written as 0,0,612,792.
Your scanned PDF, however, has for whatever reason decided to treat 0,7072 as the lower left corner and 614,7864 as the top right corner. That still gives us (almost) an 8.5x11 page size but if you try to draw something at 0,0 it will be 7,072 pixels below the actual page. You can see this in Acrobat Pro by zooming out very far (1% for me), picking Tools, Edit Object and then doing a Select All. You should see something way far down selected, too.
To get around this, you need to respect the page's boundaries.
for (int i = 1; i <= sourceReader.NumberOfPages; i++) {
//Get the page to be stamped
var pageToBeStamped = sourceReader.GetPageSize(i);
var canvas = stamper.GetOverContent(i);
//Offset our new page by 50 pixels off of the destination page's bottom
canvas.AddTemplate(stampPage, pageToBeStamped.Left, pageToBeStamped.Bottom - 50);
}
The code above gets the rectangle for the imported page and uses bottom offset by 50 pixels (from your original code). Also, although not a problem in your case, we use the imported page's actual left edge instead of just zero.
This code can still break, however. The math in the first paragraph uses 72 which is the default for PDFs but this can be changed. Most people don't change it but most people also don't change 0,0. Currently your -50 assumes the 72 which gives the visual perception of moving the stamp about seven-tenths of an inch from the top edge. If you run into this scenario you'll want to look into retrieving the user unit.
Also, as I said in the first paragraph, most applications use lower left upper right but this isn't a hard rule. Someone could specify upper right and bottom left or even top left and bottom right. This is a hard one to take into account but it is something that you should at least be aware of.