Handling requests with controllers in ASP.NET Core MVC
By Steve Smith and Scott Addie
Controllers, actions, and action results are a fundamental part of how developers build apps using ASP.NET Core MVC.
What is a Controller?
A controller is used to define and group a set of actions. An action (or action method) is a method on a controller which handles requests. Controllers logically group similar actions together. This aggregation of actions allows common sets of rules, such as routing, caching, and authorization, to be applied collectively. Requests are mapped to actions through (xref:)routing.
By convention, controller classes: * Reside in the project’s root-level Controllers folder * Inherit from Microsoft.AspNetCore.Mvc.Controller
A controller is an instantiable class in which at least one of the following conditions is true: * The class name is suffixed with “Controller” * The class inherits from a class whose name is suffixed with “Controller” * The class is decorated with the [Controller]
attribute
A controller class must not have an associated [NonController]
attribute.
Controllers should follow the Explicit Dependencies Principle. There are a couple approaches to implementing this principle. If multiple controller actions require the same service, consider using (xref:)constructor injection to request those dependencies. If the service is needed by only a single action method, consider using (xref:)Action Injection to request the dependency.
Within the Model-View-Controller pattern, a controller is responsible for the initial processing of the request and instantiation of the model. Generally, business decisions should be performed within the model.
The controller takes the result of the model’s processing (if any) and returns either the proper view and its associated view data or the result of the API call. Learn more at (xref:)Overview of ASP.NET Core MVC and (xref:)Getting started with ASP.NET Core MVC and Visual Studio.
The controller is a UI-level abstraction. Its responsibilities are to ensure request data is valid and to choose which view (or result for an API) should be returned. In well-factored apps, it does not directly include data access or business logic. Instead, the controller delegates to services handling these responsibilities.
Defining Actions
Public methods on a controller, except those decorated with the [NonAction]
attribute, are actions. Parameters on actions are bound to request data and are validated using (xref:)model binding. Model validation occurs for everything that’s model-bound. The ModelState.IsValid
property value indicates whether model binding and validation succeeded.
Action methods should contain logic for mapping a request to a business concern. Business concerns should typically be represented as services that the controller accesses through (xref:)dependency injection. Actions then map the result of the business action to an application state.
Actions can return anything, but frequently return an instance of IActionResult
(or Task<IActionResult>
for async methods) that produces a response. The action method is responsible for choosing what kind of response. The action result does the responding.
Controller Helper Methods
Controllers usually inherit from Controller, although this is not required. Deriving from Controller
provides access to three categories of helper methods:
1. Methods resulting in an empty response body
No Content-Type
HTTP response header is included, since the response body lacks content to describe.
There are two result types within this category: Redirect and HTTP Status Code.
HTTP Status Code
This type returns an HTTP status code. A couple helper methods of this type are
BadRequest
,NotFound
, andOk
. For example,return BadRequest();
produces a 400 status code when executed. When methods such asBadRequest
,NotFound
, andOk
are overloaded, they no longer qualify as HTTP Status Code responders, since content negotiation is taking place.Redirect
This type returns a redirect to an action or destination (using
Redirect
,LocalRedirect
,RedirectToAction
, orRedirectToRoute
). For example,return RedirectToAction("Complete", new {id = 123});
redirects toComplete
, passing an anonymous object.The Redirect result type differs from the HTTP Status Code type primarily in the addition of a
Location
HTTP response header.
2. Methods resulting in a non-empty response body with a predefined content type
Most helper methods in this category include a ContentType
property, allowing you to set the Content-Type
response header to describe the response body.
There are two result types within this category: (xref:)View and (xref:)Formatted Response.
View
This type returns a view which uses a model to render HTML. For example,
return View(customer);
passes a model to the view for data-binding.Formatted Response
This type returns JSON or a similar data exchange format to represent an object in a specific manner. For example,
return Json(customer);
serializes the provided object into JSON format.Other common methods of this type include
File
,PhysicalFile
, andVirtualFile
. For example,return PhysicalFile(customerFilePath, "text/xml");
returns an XML file described by aContent-Type
response header value of “text/xml”.
3. Methods resulting in a non-empty response body formatted in a content type negotiated with the client
This category is better known as Content Negotiation. (xref:)Content negotiation applies whenever an action returns an ObjectResult type or something other than an IActionResult implementation. An action that returns a non-IActionResult
implementation (for example, object
) also returns a Formatted Response.
Some helper methods of this type include BadRequest
, CreatedAtRoute
, and Ok
. Examples of these methods include return BadRequest(modelState);
, return CreatedAtRoute("routename", values, newobject);
, and return Ok(value);
, respectively. Note that BadRequest
and Ok
perform content negotiation only when passed a value; without being passed a value, they instead serve as HTTP Status Code result types. The CreatedAtRoute
method, on the other hand, always performs content negotiation since its overloads all require that a value be passed.
Cross-Cutting Concerns
Applications typically share parts of their workflow. Examples include an app that requires authentication to access the shopping cart, or an app that caches data on some pages. To perform logic before or after an action method, use a filter. Using (xref:)Filters on cross-cutting concerns can reduce duplication, allowing them to follow the Don’t Repeat Yourself (DRY) principle.
Most filter attributes, such as [Authorize]
, can be applied at the controller or action level depending upon the desired level of granularity.
Error handling and response caching are often cross-cutting concerns: * (xref:)Error handling * (xref:)Response Caching
Many cross-cutting concerns can be handled using filters or custom (xref:)middleware.
|