Merging N pdf files, created from html using IText

问题:

I need to merge N PDF files into one. I create a blank file first

byte[] pdfBytes = null;

var ms = new MemoryStream();
var doc = new iTextSharp.text.Document();
var cWriter = new PdfCopy(doc, ms);

Later I cycle through html strings array

foreach (NBElement htmlString in someElement.Children())
                    {
                        byte[] msTempDoc = getPdfDocFrom(htmlString.GetString(), cssString.GetString());
                        addPagesToPdf(cWriter, msTempDoc);
                    }

In getPdfDocFrom I create pdf file using XMLWorkerHelper and return it as byte array

private byte[] getPdfDocFrom(string htmlString, string cssString)
    {
        var tempMs = new MemoryStream();
        byte[] tempMsBytes;
        var tempDoc = new iTextSharp.text.Document();
        var tempWriter = PdfWriter.GetInstance(tempDoc, tempMs);
        tempDoc.Open();

        using (var msCss = new MemoryStream(System.Text.Encoding.UTF8.GetBytes(cssString)))
        {
            using (var msHtml = new MemoryStream(System.Text.Encoding.UTF8.GetBytes(htmlString)))
            {
                //Parse the HTML
                iTextSharp.tool.xml.XMLWorkerHelper.GetInstance().ParseXHtml(tempWriter, tempDoc, msHtml, msCss);
                tempMsBytes = tempMs.ToArray();
            }
        }

        tempDoc.Close();
        return tempMsBytes;
    }

Later on I try to add pages from this PDF file to the blank one.

private static void addPagesToPdf(PdfCopy mainDocWriter,  byte[] sourceDocBytes)
    {

        using (var msOut = new MemoryStream())
        {
            PdfReader reader = new PdfReader(new MemoryStream(sourceDocBytes));
            int n = reader.NumberOfPages;
            PdfImportedPage page;
            for (int i = 1; i <= n; i++)
            {
                page = mainDocWriter.GetImportedPage(reader, i);
                mainDocWriter.AddPage(page);
            }
        }}

It breaks when it tries to create a PdfReader from the byte array I pass to the function. "Rebuild failed: trailer not found.; Original message: PDF startxref not found."

I used another library to work with PDF before. I passed 2 PdfDocuments as an objects and just added pages from one to another in cycle. It didn't support Css though, so I had to switch to ITextSharp.

I don't quite get the difference between PdfWriter and PdfCopy.

回答1:

There a logical error in your code. When you create a document from scratch as is done in the getPdfDocFrom() method, the document isn't complete until you've triggered the Close() method. In this Close() method, a trailer is created as well as a cross-reference (xref) table. The error tells you that those are missing.

Indeed, you do call the Close() method:

tempDoc.Close();

But by the time you Close() the document, it's too late: you have already created the tempMsBytes array. You need to create that array after you close the document.

Edit: I don't know anything about C#, but if MemoryStream clears its buffer after closing it, you could use mainDocWriter.CloseStream = false; so that the MemoryStream isn't closed when you close the document.

In Java, it would be a bad idea to set the "close stream" parameter to false. When I read the answers to the question Create PDF in memory instead of physical file I see that C# probably doesn't always require this extra line.

Remark: merging files by adding PdfImportedPage instances to a PdfWriter is an example of bad taste. If you are using iTextSharp 5 or earlier, you should use PdfCopy or PdfSmartCopy to do that. If you use PdfWriter, you throw away a lot of information (e.g. link annotations).