Enter Pipe class between any of the existing ones

[KaloyanYosifov]

Hey I have a feature request which I would like to discuss before I create a PR.

---

So this idea comes from an issue I stumbled today while I was working on a project. We have a DTO which has a variable type that gets filled by an optional route parameter.

class SomeDTO extends Data {
    public function __construct(
        #[FromRouteParameter('type')]
        public ?SomeEnum $type
    ) {
    
    }
}

This is working very well so far. But the issue is that our team has agreed to a spec where we send type as null in the url (Don't ask me why 💀 ). Example: https://example.org/posts/category/null

Now in this case I would want the type to be null in the DTO, so I don't get any cast or validation errors. This is not the case so I created myself a pipe which I pass to the pipeline that checks if type is a string null and if it is I just set it to null type.

Here is how this works:

class SomeDTO extends Data {
    public function __construct(
        #[FromRouteParameter('type')]
        public ?SomeEnum $type
    ) {
    
    }

    public static function pipeline(): DataPipeline
    {
        return DataPipeline::create()
            ->into(static::class)
            ->through(AuthorizedDataPipe::class)
            ->through(MapPropertiesDataPipe::class)
            ->through(FillRouteParameterPropertiesDataPipe::class)
            ->through(FixTypeRouteParameterPropertyDataPipe::class)
            ->through(ValidatePropertiesDataPipe::class)
            ->through(DefaultValuesDataPipe::class)
            ->through(CastPropertiesDataPipe::class);
    }
}

Notice how I add FixTypeRouteParameterPropertyDataPipe::class just after FillRouteParameterPropertiesDataPipe::class.

---

Now with that story done 😅 . My whole problem with the above solution is that I have to copy the pipeline definition of the parent to add the pipe after FillRoute. I would like to propose a method on DataPipeline where we can choose where we can add a new Pipe which allows to use the default pipes without copying. This also allows to use the new default pipes introduced into the library

As an example:

    public static function pipeline(): DataPipeline
    {
        return parent::pipeline()
            ->throughAfter(FixTypeRouteParameterPropertyDataPipe::class, FillRouteParameterPropertiesDataPipe::class);
    }

or maybe have DataPipeline support setting pipes array with a setter and have a getter for the user to manipulate.

Let me know if this proposal makes sense

[rubenvanassche]

In such a case I would create a new abstract data object which has the pipeline like you've described, then every data object could extend from that data object and would run the extra pipe.