Can pdf_crypto load all kinds of pdf files?

[kaciula]

There are some PDF files that cannot be rendered at all on iOS but can be rendered on Android. I want to know if pdf_crypto can load these particular PDF files. I think it's a compression-related issue. If I read the pdf file with ghostscript and write it back, it successfully renders.

If pdf_crypto can read these PDF files, my client would like to purchase the licence. But first we need to know if it works. I can privately share some of these PDFs and you can confirm that it works. How should we proceed?

[DavBfr]

pdf_crypto doesn't render pdf, it only allow to modify them and add some cryptography.

[kaciula]

@DavBfr I misspoke. I don't need pdf_crypto to render the PDF. I just need to parse it so I can modify it so it can become renderable using other pdf rendering plugins out there. I just want to know if it can parse these PDFs.

[kaciula]

@DavBfr Can I send you some pdfs that you can test if loading with pdf_crypto and then saving them with the pdf plugin works?

For example, the pdfs that don't render on iOS become renderable if I simply load and rewrite the pdf with Ghostscript.

gs -dNOPAUSE -dBATCH -sDEVICE=pdfwrite -sOutputFile=output.pdf blank.pdf

https://teikningar.rnb.is/rnes_kort/Rnb----9721.pdf
https://teikningar.rnb.is/rnes_kort/Rnb-----9724.pdf
https://teikningar.rnb.is/rnes_kort/Rnb-----9730.pdf
https://teikningar.rnb.is/rnes_kort/Rnb-----9732.pdf

[DavBfr]

I tested your files:

    final original = PdfDocumentParser(
        await File('Rnb-----9730.pdf').readAsBytes());

    final tools = PdfTools(
      compress: false,
      verbose: true,
    );
    tools.addPages(original);

    for (final o in tools.xref.objects) {
      final p = o.params;
      if (p is PdfDictStream) {
        if (p.values['/Subtype'] == PdfName('/Image') &&
            p.values['/Filter'] == PdfName('/DCTDecode')) {
          print(p.values);
          File('image_${o.objser}.jpg').writeAsBytesSync(p.decode().data);
          print('Saved image_${o.objser}.jpg');
        }
      }
    }

    final result = tools.save();
    final file = File('template.pdf');
    await file.writeAsBytes(result);

The DCTDecode filter represents an embedded Jpeg image. This particular image is 10012x7136 pixels, but the file seems corrupted. the iOS/MacOS jpeg decoder can't read it. Some software can open it, but shows an image of 10012x65280 pixels.

If you can find a library on iOS that can read this jpeg data, it should be possible to rewrite the PDF with some correct data.

[DavBfr]

Chrome / Safari can't display the attached image, but Firefox can.

[kaciula]

@DavBfr Thanks for looking into it. Is the image link you included still valid? Firefox does not show anything for me for the image url.
https://private-user-images.githubusercontent.com/1387855/419457337-94e8b1ff-e692-4f7a-9188-c0eab14d3082.jpg?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.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.WdzVuXuNs_MXT7Wx83rLEUwDPeXcQAcDMldEymZYzCs

I'm also wondering why GhostScript managed to decode the pdf and jpeg correctly. It looks like there is a way to decode the jpeg so that it then renders correctly everywhere.

gs -dNOPAUSE -dBATCH -sDEVICE=pdfwrite -sOutputFile=output.pdf blank.pdf