Summary

This package provides a basic forward pipe operator. This is a commonly requested language feature in C#, but appears to not currently be on the schedule.

Here are currently open tickets requesting the feature:

If you find you like this package, please take the time to upvote these tickets.

Install

NuGet page

What is a forward pipe operator?

A forward pipe operator is a way to visualize the call of functions from top to bottom instead of inside to outside. A typical call structure in an imperative language might look like:

var model = Query(input);
Validate(model);
var outputModel = Transform(model);
Submit(outputModel);

Or even worse:

Submit(Transform(Validate(Query(input);

In a functional language (example in F#) there is a way to "pipe" the results of one function to another:

input
  |> Query
  |> Validate
  |> Transform
  |> Submit

The above syntax results in, "Take input, pass it to Query, then pass the result to Validate, then pass the result to Transform, then pass the result to Submit."

What does a forward pipe operator look like in C#?

Barring it being added to C# natively, the best way to simulate a forward pipe operator is with extension methods.

Implementation

The code for this is dead simple:

public static U Pipe<T, U>(this T input, Func<T, U> @operator)
{
    return @operator(input);
}

In reality, there are a lot of permutations of the above function to handle all async and dyadic/triadic use cases.

Using the pipe operator

To use the Pipe operator, you would just call .Pipe on whatever object you're using and pass in the function you're hoping to run.

input
  .Pipe(Query)
  .Pipe(Validate)
  .Pipe(Transform)
  .Pipe(Submit);

Asynchronous processing

You can also use async methods. Because of how C# conceptualizes tasks, you must use all .PipeAsync after your first async method regardless of whether your subsequent method is async. PipeAsync refers to the return type of the method, which will always be asynchronous after the first asynchronous request. Under the covers, PipeAsync awaits the result of Task and passes T to the non-async method.

await input
    .Pipe(Query)
    .Pipe(Validate)
    .PipeAsync(TransformAsync)
    // Notice this is PipeAsync even though Submit is synchronous.
    .PipeAsync(Submit);

Cancellation Tokens

Cancellation Tokens are available as of version 1.1.0. To use them, pass in the token after passing in the function to operate on. For this to compile, the function must accept a cancellation token (Func<T, CancellationToken, TResult>).

await input
    .Pipe(Query)
    .Pipe(Validate)
    // Note the cancellation token.
    .PipeAsync(TransformAsync, cancellationToken)
    .PipeAsync(Submit);

Passing multiple arguments.

This package has opted to retain a consistent "Take what's on the left and pass it to the right" syntax. This falls in-line with the original operator design for F#. However, as of version 1.3.0, There is a way to use dyadic and triadic functions. All you have to do is operate on a tuple with 2 or 3 parameters. Internally, .Pipe destructures the tuple and passes it to your function. Here's an example

// Function with 2 parameters:
bool Validate(int id, string name)
  => id > 0 && name != "invalid";

// Pass arguments to Validate through a tuple.
// This is valid syntax as of v1.3.0.
var isValid = (1, "Charlie").Pipe(Validate);

// Function with 3 parameters:
bool Validate(int id, string name, short age)
  => id > 0 && name != "invalid" && age < 175;

// Pass arguments to Validate through a tuple.
// This is valid syntax as of v1.3.0.
var isValid = (1, "Charlie", 57).Pipe(Validate);

Why do we need a forward pipe operator?

There are three basic arguments for using .Pipe over plain function calls - readability, breaking dependencies, garbage collection.

Readability: When you use a forward pipe operator, you reduce the noise created by temporary variables. Your eye is drawn to the sequence of operations instead of bouncing back and forth between variable and function call.
Breaking dependencies: A major problem with large functions is that any line can depend on any line before it. If a function is 30 lines, line 29 can depend on line 15, 2, 1, 4, or any other. When a developer thinks in sequence with a forward pipe operator, the next line can only depend on the line before it.
Garbage collection: Memory allocation/deallocation is one of the hardest things for a developer to get right consistently. Only being able to depend on the previous line means that an object has to be allocated/deallocated within a small scope. Having a small scope means that a developer can more easily see where an object needs to be deallocated and when memory leaks occur, it is easy to spot where a variable is not being disposed of properly.

Example user flow

public class UserFlow
{
  public async Task<ActionResult> Pipe_UserFlow(Input input)
  {
    try 
    {
      await input
        .Pipe(Query)
        .Pipe(Validate)
        .Pipe(Transform)
        .PipeAsync(SubmitAsync);
        
      return Ok();
    }
    catch(ValidationException validationException)
    {
      return BadRequest(validationException.Message);
    }
  }
  
  private Model Query(Input input)
  {
    // Your query to a database to add extra information.
    // I've called the new model "Model", but it's a terrible name.
    
    return new Model(input);
  }
  
  private Model Validate(Model model)
  {
    // Your validation which throws an exception when it's not right.
    
    return model;
  }
  
  private Output Transform(Model model)
  {
    // Often you need to provide a different model to the universe than 
    // the model you use to complete the current operation. You could
    // Do this as part of the submit if you have lots of submit operations,
    // But I've elected to do it here.
    
    return new Output(model);
  }
  
  private async Task SubmitAsync(Output output)
  {
    // This could be any operation. In this case, we'll just wait for two seconds
    // to simulate saving something to a database.
    
    await Task.Delay(2000);
  }
}

Other forward pipe operator projects

After I released this package, I found out there is another repository out there with a similar aim. While I didn't draw inspiration from the repo, it was first and deserves mention. TomyDurazno's PipeExtensions

winstonpuckett / WinstonPuckett.PipeExtensions