Feature: Global SkipRule or Generic BeforeRule preprocessor

[anorborg]

This is really an amazing tool, but would appreciate some sugar-syntax in my use-case to be able to vary in some null/unset values.

What I do today

new Faker<Test>()
     .RuleFor(x => x.Value1, f => {
          // Randomly set this value with a probability of .1 (10% of the time)
          return (f.Random.Bool(0.1f)) ? f.Random.Int(1, 5) : null;
      })
     .RuleFor(x => x.Value2, f => {
          // Randomly set this value with a probability of .1 (10% of the time)
          return (f.Random.Bool(0.1f)) ? f.Internet.Ip(): null;
      })

Suggested Feature Syntax

new Faker<Test>()
     .SkipRule((f, x) => f.Random.Bool(0.1f))
     .RuleFor(x => x.Value1, f => f.Random.Int(1, 5))
     .RuleFor(x => x.Value2, f => f.Internet.Ip());

I'm not sure if there is a better/current way to accomplish something similar, but generating a dataset with the occasional null value is pretty useful for my in testing scenarios. Again, great tool!

[bchavez]

Hi @anorborg,

Thank you for the complements! I'm really glad you like Bogus! It always makes me happy to hear from people using Bogus. 😎

To answer your question:
Bogus has a nifty .OrNull() helper extension to generate null values randomly; however, the feature is nestled away in an extension namespace called Bogus.Extensions. Below is an example of how to use the .OrNull() extension to generate randomly null values:

using Bogus.Extensions;

void Main()
{
   var fakeTests = new Faker<Test>()
      .RuleFor( x => x.Value1, f => f.Random.Int(1, 5).OrNull(f) )
      .RuleFor( x => x.Value2, f => f.Internet.Ip().OrNull(f) );
      
   fakeTests.Generate(10).Dump();
}

public class Test
{
   public int? Value1 {get; set;}
   public string Value2 {get;set;}
}

Currently, .OrNull() doesn't support a weighted probability of generated nulls; but that can be easily fixed by copying the current implementation and making your own extension method like this:

void Main()
{
   var fakeTests = new Faker<Test>()
      .RuleFor( x => x.Value1, f => f.Random.Int(1, 5).OrNull(f, 0.9f) )
      .RuleFor( x => x.Value2, f => f.Internet.Ip().OrNull(f, 0.1f) );
      
   fakeTests.Generate(10).Dump();
}

public static class MyBogusExtensions
{
   public static object OrNull(this object value, Faker f, float nullWeight)
   {
      return f.Random.Float() < nullWeight ? null : value;
   }
}

The code above is quite clear, elegant, terse, and very readable IMHO. Perhaps, it might be a good idea to make this change official and allow the user to optionally specify the weight of nullability. 😺

Hope that helps!

Thanks,
Brian

//cc related #124

⏳ 🔍 "But I still haven't found what I'm for..."

[anorborg]

Thanks! A small suggestion: use default instead of null which works for null but value-types as well:

public static object OrDefault(this object value, Faker f, double defaultValueWeight = 0.5f)
{
    if (defaultValueWeight > 1 || defaultValueWeight < 0) throw new ArgumentOutOfRangeException(nameof(defaultValueWeight), $".{nameof(OrDefault)}() {nameof(defaultValueWeight)} of '{defaultValueWeight}' must be between 1.0f and 0.0f. ");
        return f.Random.Float() > defaultValueWeight ? value : default;
   }
}

Although you can define zero in your range in most, if not all cases, since you've added the weight, its a nice option to only select that value with the same probability as null.

[bchavez]

Hi @anorborg,

That's a great idea! Thank you for the suggestion! I've released Bogus v24.1.0 with the latest suggestion and changes!

Hope that helps!

Thanks,
Brian

🚶 😎 "So don't delay... act now supplies are running out..."