Support for CP1252 Encoding

[codepwner]

I was getting an error on a particular file and when I made this change, it allowed it to proceed.

[AndrewRybka]

@codepwner, can you please attach that file? I would like to find out why the original code wasn't working correctly with it.

I that's not possible, can you please provide the strEncoding value which was failing? Current code works correctly for me with both "CP1252" and "cp1252". And even with "C̈P̈1252".

[codepwner]

Maybe it has to do with the specific versions of .NET I am using? Currently, my downstream application is targeting .NET Core 3.1.

Working on creating a sanitized version of the input file I used to be able to share.

[codepwner]

I was able to recreate the issue using this file.

Test Case.zip

The SPSS project has no other modifications than the one I checked in. I was not aiming to do any development in the project, but when the nuget version was throwing the error, I decided to pull the source and was able to make this hack of a fix to get past my issue.

[codepwner]

Here are the bytes of the string in the calling method too.

[codepwner]

I converted my project to .NET Standard 4.7 (for reasons outside this topic) and decided to run a quick additional test...

[AndrewRybka]

@codepwner, thank you for the file. But the library works correctly with that file on my side.

Text and byte representation of the encoding are the same as yours, but the encoding is correctly resolved by the following lines in the GetEncoding() method:

// Try to get encoding parsing codepage
int cp;
if (int.TryParse(Regex.Match(strEncoding, @"\d+").Value, out cp))
{
    info = encInfo.SingleOrDefault(ei => ei.CodePage == cp);
    if (info != null)
        return info.GetEncoding();
}

It seems that the source of the problem is in the Encoding.GetEncodings() method.
It returns 140 encodings in a project which targets .Net Framework 4.6.2.

But it seems than in projects which target .Net Core that method returns only a few default encodings (PowerShell/PowerShell#6580).

So it's better to make this fix more universal in order to support other encodings. Something like this:

// Try to get encoding parsing codepage
int cp;
if (int.TryParse(Regex.Match(strEncoding, @"\d+").Value, out cp))
{
    info = encInfo.SingleOrDefault(ei => ei.CodePage == cp);
    if (info != null)
        return info.GetEncoding();

    try
    {
        var enc = Encoding.GetEncoding(cp);
        return enc;
     }
    catch (ArgumentException)
    { }
}