Atalasoft DPI

We use a software library from Atalasoft in our product to allow users to add annotations to PDFs.

One of our Indian developers posted on Slack to ask a question about a bug he was assigned to fix. It was quite hard to understand what he wanted but it sounded like the quality of the users PDFs were lowered to a point that they were blurry and unusable.

Hi Everyone, Here is my doubt was more of a generic one. In Document Attachment Module, I’m trying to attach a PDF. The attached PDF gets depreciated in the doc viewer.. After analysis came to a conclusion that, the Atalasoft Viewer we are using Document Attachment viewer should pass only with 96dpi(dots per inch).

However in the Atalasoft Documentation itself was given that inorder to increase the quality of the document inside the Document Viewer of Atalasoft we need to pass on the default or hardcoded resolution as attached.

With respect to this have attempting a bug in which need to fix this depreciation not in a hardcoded format.

Is there any way to calculate a PDF file’s DPI through its file size. (Note: Since PDF file was vector based and doesn’t posses any information related to dpi).Can anyone please guide me on this ? Apart from hardcoding and passing on a resolution value.

After struggling with it, another developer started working on it, but then went on annual leave so yet another developer took over. None of them had put much thought into what they were doing because when I asked them to explain the code, they couldn’t seem to. I then googled the code and found it on the Atalasoft website. https://www.atalasoft.com/kb2/KB/50067/HOWTO-Safely-Change-Set-Resolution-of-PdfDecoder

using (var annotateViewer = new AnnotateViewer())
{
    annotateViewer.DataImporters.Add(new Atalasoft.Annotate.Importers.PdfAnnotationDataImporter { SkipUnknownAnnotationTypes = false });                
    using (var pdfDec = new PdfDecoder())
    {
        pdfDec.RenderSettings = new RenderSettings { AnnotationSettings = AnnotationRenderSettings.RenderAll };
        Atalasoft.Imaging.Codec.RegisteredDecoders.Decoders.Add(pdfDec);
        SetPdfDecoderResolution();
    }     
    
    annotateViewer.Open(filePath);                
    var printer = new Printer(annotateViewer);
    printer.Print(printerSettings, documentName, printContext);
} 
 
 
static readonly object pdfLock = new object();

private static void SetPdfDecoderResolution()
{
    int standardResolution = 300;
    lock (pdfLock)
    {
        foreach (Atalasoft.Imaging.Codec.ImageDecoder rawDecoder in Atalasoft.Imaging.Codec.RegisteredDecoders.Decoders)
        {
            if (rawDecoder is PdfDecoder)
            {
                //By default PdfDecoder sets to lower resolution of 96 dpi
                //Reason for PDF depreciation
                ((PdfDecoder)rawDecoder).Resolution = standardResolution;
                return;
            }
        }
        Atalasoft.Imaging.Codec.RegisteredDecoders.Decoders.Add(new PdfDecoder() { Resolution = standardResolution });
    }
}

The code instantly stood out for being convoluted because we are creating a PdfDecoder called pdfDec, then instead of just setting properties on it, we add it to the RegisteredDecoders, then call our SetPdfDecoderResolution which loops through the decoders to find the one we added. If it can’t find it (which surely is impossible) it will add one.

I was talking to a Lead Developer about a completely different bug fix, and he says

“People just don’t think about what they write”

Lead Developer

So I decided to bring up this Atalasoft problem…

He said 

When I saw the lock I wanted to ask, “Which Stack Overflow post did you find this on?”

Lead Developer

So I told him they got it from the Atalasoft website!

So they had blindly pasted this Atalasoft code in without thinking. They could just set the Resolution property in the existing code since we already create the object; so already hold a reference to it. If this code can mean we can add multiple decoders (which you aren’t supposed to do), then we could create a method similar to SetPdfDecoderResolution where it checks if there is a decoder or add one if none exists. Then we ensure all the correct properties are set. 

They need to think

Lead Developer

I think the problem the Lead Developer had with the lock is that you use lock when you want to guarantee that only one thread is accessing a resource/section of code at any time; but this code wasn’t used in a multi-threaded context. So by blindly pasting in code without thinking, they were adding redundant lines and creating confusion. 

The actual fix was just

private const int highDotsPerInch = 300;
pdfDec.Resolution = highDotsPerInch;

But to reach this outcome, it took 3 developers to look at it, then 2 to review it.