Streams are cut off at Length which extracts incomplete files

[qbq-ber-robin-klimonow]

Hi,

I stumbled upon an issue while using PdfPig for extracting attachments from a PDF file. I have attached a sample PDF to reproduce the error:

0_ZUGFeRD.pdf

After debugging through the code, I found that the /Length attribute of the stream is set to a wrong value of 11417 bytes, however there is no endstream at the expected position. Instead, there are more bytes, followed by endstream.

PdfPig now seems to just cut all additional bytes off, which seems reasonable at first. However, other Pdf Libraries we tested can handle this pdf file just fine, and so can Adobe Acrobat Reader itself.

Therefore, this seems to be an issue with PdfPig in my oppinion.

The fix for me was to remove the if check in PdfTokenScanner.cs, Line 437

                if (length.HasValue && memoryStream.Length >= length)
                {
                    // Use the declared length to copy just the data we want.
                    byte[] data = new byte[read];
                    memoryStream.Read(data, 0, (int)read);

                    stream = new StreamToken(streamDictionaryToken, data);
                }
                else
                {

And just use the else block which tries to find endobj or endstream and cut off there - which works with the PDF I attached.

Another reason could be the special characters in the attachment, which might have lead to this Length attribute value.

Would this be a valid fix in your oppinion?

[sbruyere]

Dears,

I have the same issue and came to same conclusion. Any hope this can be fixed ?
Thanks !