Handle line feed character in EOD marker for base85 decoder

[michael-michalski]

Hello :)
We are receiving base64 encoded pdfs from our partner, which has an additional non-spec line feed character. We use pdfsharp to open them with a memory stream, and perform some edits.

The adobe documentation outlines the EOD marker as ~>: https://opensource.adobe.com/dc-acrobat-sdk-docs/pdfstandards/PDF32000_2008.pdf

But if I open the pdf in several popular pdf viewers/browsers it correctly interprets the pdfs, contrary pdfsharp raises an invalid character exception.

Should this path be less strict, or is there something I am misunderstanding?

[ThomasHoevel]

Looks like a bug in PDFsharp. We will fix it. Thanks for letting us know.
Any whitespace should be ignored, so your PR is not the ultimate fix.

It would be good to have a PDF for testing. You can attach the PDF here, send it by email or upload to our FTP server.
https://docs.pdfsharp.net/General/Issue-Reporting/FTP-uploads.html

[michael-michalski]

Unfortunately the PDF's are filled with customer data, so I would have to scramble an example together.
I will try to write an example PDF.

[StLange]

The bug you found was not the only one. I rewrote the whole code. See here.

[michael-michalski]

If you need it for tracking. The pdfs encountered where Fedex freight labels.

[michael-michalski]

When can we look forward to a Nuget prerelease? :)

[michael-michalski]

Hey Team,
Is there a planned release of a new preview for 6.2, with the new base85 decoder?