Question

Let’s suppose I have such “helper” methods used by some object.

    private int MatchRegex(string regex, string input)
    {
        var match = Regex.Match(input, regex);
        return match.Success ? Convert.ToInt32(match.Groups[1].Value) : 0;
    }

    private string Exec(string arguments, string path = "", bool oneLine = false)
    {
        var p = new Process();
        p.StartInfo.UseShellExecute = false;
        p.StartInfo.RedirectStandardOutput = true;
        p.StartInfo.CreateNoWindow = true;

        if (path != "")
            p.StartInfo.WorkingDirectory = path;

        p.StartInfo.FileName = "binary.exe";
        p.StartInfo.Arguments = arguments;
        p.Start();

        string output = oneLine ? p.StandardOutput.ReadLine() : p.StandardOutput.ReadToEnd();
        p.WaitForExit();

        return output;
    }

Which would you choose to move out them: another class, partial class or extension methods? And why?

Answer 1

If they accessed private state they would have to be methods in a partial class fragment. Extension methods are useful it it can support a range of objects, or the type can’t be used as a partial class (an interface being the most likely example, or outside of your assembly).

Looking at those methods, they don’t really seem to relate to any given object, so I would do neither, and just expose them as static methods on a utility class. For example:

public static class ProcessUtils {
    public static string Exec(...) {...}
}

The regex one is a different case; obtaining group 1 as an int seems such a specific scenario (unless there is something domain-specific in your project that makes this common-place), and the code is so trivial, that I would just let the calling code use the existing static Regex.Match. In particular, I would want the caller think about whether a static pre-compiled regex might be appropriate, which your utility method doesn’t allow.