Keyword "validate" suggestion

[srburton]

I have suggestion process of new keywords for Sin data manipulation in C#

keywords:

validate : Keyword for validations of a variable .

with : keyword to signal which class will be initialized.

any: Accepts any given type as [int, string, bool ... ] Only when together with the keyword validate

Example Class of a validator

public class MyValidate
{

  public void NotEmpty(validate any param,string errorMsg)
  {
       if(param == typeof(string))
       {
            if(param != String.Empty)
            {
                param += " very good";
            }
       }           

       throw new Exception(errorMsg); 

   }

  public void MaxCaracter(this string param,string errorMsg,int max)
  {
     //....
  }

}

Example validator

string name = "c# is ";
validate name with (MyValidate) 
{
  NotEmpty("Not empty value!")
  MaxCaracter("Maximum caracter",8)
};

or

string name = "c# is ";
with name validate (MyValidate) 
{
  NotEmpty("Not empty value!")
  MaxCaracter("Maximum caracter",8)
};

Output
Exception: Maximum caracter

In this example step a string to validate with less than 8 characters but in the validation is added more then when the second method is executed that counts the number of characters it already has a number greater than 8 then causes an exception

[Unknown6656]

What would be the advantage of your proposal compared to a combination of generics, pattern matching and maybe dynamic (yeah, yeah, I know it's slow and bad .....)?

You could rewrite MyValidate.NotEmpty as:

public void NotEmpty<T>(T param, string errorMsg)
{
    if (param is string s && s != "")
    {
        s += " very good";

        // do something with s here (typesafe!)
    }

    throw new Exception(errorMsg);
}

or

public void NotEmpty(dynamic param, string errorMsg)
{
    if (param is string s && s != "")
        param += " very good";

    // do something with param here (not typesafe)

    throw new Exception(errorMsg);
}

[srburton]

KEYWORD sequential

Model People

public class People
{
    public string Name {get; set;}
    public string LastName {get; set;}
    public string State {get; set;}
    public string City {get; set;}
}

Validator methods

public class Validator
{
     public static void NotEmpty<T>(T param, string error){  ...  }
     public static void MaxCaracter<T>(T param, string error,int max){  ...  }
     public static void CheckSocialSegure<T>(T socialNumber, T state, T city, string error) { ... }
}

Object

var people = new People();
var socialNumber = 232324;

sequential people in p 
{
   Validator.NotEmpty(p.Name,"Not Empty!");
   Validator.MaxCaracter(p,Name,"Max caracter is 8",8);
   Validator.NotEmpty(p.LastName,"Not Empty!");
   Validator.CheckSocialSegure(socialNumber, p.State, p.City, "Not is social segure!" );
}

List

var peoples = new List<People>()
{
    new People()
};
var socialNumber = 232324;

sequential peoples in p
{
   Validator.NotEmpty(p.Name,"Not Empty!");
   Validator.MaxCaracter(p,Name,"Max caracter is 8",8);
   Validator.NotEmpty(p.LastName,"Not Empty!");
   Validator.CheckSocialSegure(socialNumber, p.State, p.City, "Not is social segure!" );
}

Primitive[]

var myString = new string[] { "Hello", "C#"};

sequential myString in m
{
   Validator.NotEmpty(m,"Not Empty!");
}

dynamic

       var myDynamic = new
            {
                Name = "",
                LastName = "",
                Address = new
                {
                    City = ""
                }
            };
            

sequential myDynamic in m
{
   Validator.NotEmpty(m.Name,"Not Empty!");
   
   sequential m.Address in a
    {
             Validator.NotEmpty(a.City,"Not Empty city!");
             //Example "NameAndCity"
             //Validator.NameAndCity(m.Name,a.City,"City is valid!");
    }

}

In this example I want to demonstrate that a method if called within the sequential it does not run on how much it does not reach its turn, this scenario would be for validations and testing in methods a native form of summarizing code and grouping a particular operation.

[AustinBryan]

I don't get the point of it at all.

In your example:

public class MyValidate
{
    public void NotEmpty(validate any param, string errorMsg)
    {
       if (param == typeof(string))
       {
            if (param != String.Empty)
                param += " very good";
       }           

       throw new Exception(errorMsg); 
    }
}

There's so many flaws here. For one, checking if param is a string and then you're adding " very good" if it is, and then you throw an exception, regardless if it is a string or not?

Also, for param to equal typeof(string) then it must be of type Type or a child class, which means param += " very good" probably wouldn't be valid. It depends, what type is param? Is it a string, Type or something that mimicks both interfaces?

The biggest reason why I dislike the feature, is why not just do this:

public class MyValidate
{
    public void NotEmpty(string param, string errorMsg)
    {
        if (param != String.Empty)
            param += " very good";
        throw new Exception(errorMsg); 
    }
}

Or use object or dynamic instead of making up so many keywords? I see literally nothing being gained here.

[srburton]

@AustinBryan
My independent approach to code written above is that C# has something native to automation and data validation easily and simplistically I would like to contribute ideas so that we can find a good implementation for this functionality.

[CyrusNajmabadi]

Could you explain what you find insufficient in the existing mechanisms the language already has for "data validation"?

[srburton]

@CyrusNajmabadi

Current Code

public bool ValidEmailAddress(string emailAddress, out string errorMessage)
{
   // Confirm that the e-mail address string is not empty.
   if(emailAddress.Length == 0)
   {
      errorMessage = "e-mail address is required.";
         return false;
   }

   // Confirm that there is an "@" and a "." in the e-mail address, and in the correct order.
   if(emailAddress.IndexOf("@") > -1)
   {
      if(emailAddress.IndexOf(".", emailAddress.IndexOf("@") ) > emailAddress.IndexOf("@") )
      {
         errorMessage = "";
         return true;
      }
   }

   errorMessage = "e-mail address must be valid e-mail address format.\n" +
      "For example '[email protected]' ";
      return false;
}

My approach - Implementation example

In this case, validate returns a Boolean if all the test is true so ´ was satisfied with validation if a test is false then everything else is invalid!

public bool ValidEmailAddress(string emailAddress)
{
    
    return validate emailAddress in e
    {
          e.Lenght == 0 error "E-mail address is required."; 
          e.IndexOf("@") == -1 error "Email require @";
          e.Regex("pattern_valid_email") error "Email invalid!";
    }

}

[CyrusNajmabadi]

1. your example is very strange since you're checking different things in each form. However, you could accomplish what you want in the second form with just:

          if (e.Lenght == 0) throw new ArgumentException("E-mail address is required.");
          if (e.IndexOf("@") == -1) throw new ArgumentException("Email require @");
          if (e.Regex("pattern_valid_email")) throw new ArgumentException("Email invalid!");
          return true;

This has the virtues of:

1. being idiomatic.
2. working in C# since the very first version of the language
3. being shorter than the form you specified.

[SilentSin]

@CyrusNajmabadi I think that example is supposed to return false if one of the tests fails (rather than throwing exceptions) and provide the error message as an out parameter. The problem being that the method doesn't appear to have an out parameter so I can't figure out of error is supposed to be that parameter (in which case it's just the current syntax without the if, parentheses, or equals sign) or if error is supposed to be a new keyword that somehow magics up an out parameter or something.

[CyrusNajmabadi]

in that case, you can just do:

          if (e.Lenght == 0) return (false, "E-mail address is required.");
          if (e.IndexOf("@") == -1) return (false, "Email require @");
          if (e.Regex("pattern_valid_email")) return (false, "Email invalid!");
          return (true, null);

[theunrepentantgeek]

It can also be done with a simple semantic type:

public ValidationResult ValidateEmailAddress(string emailAddress)
{
    return Validation.ErrorWhen( emailAddress.Length == 0, E-mail address is required.")
        && Validation.ErrorWhen( emailAddress.IndexOf("@") == -1, "Email require @")
        && Validation.ErrorWhen( emailAddress.Regex("pattern_valid_email"), "Email invalid!");
}

The && does the usual short circuiting, so the expression exits early as soon as there is a problem.

This isn't theoretical code - it's written with a semantic type for validation that I wrote some time ago for my blog.

So, no need for special syntax in C# - just some library design. My ValidationResult is just one approach.