Routing

Routing is the core Ktor plugin for handling incoming requests in a server application. When the client makes a request to a specific URL (for example, /hello), the routing mechanism allows us to define how we want this request to be served.

Install Routing

The Routing plugin can be installed in the following way:

import io.ktor.server.routing.*

install(Routing) {
    // ...
}

Given the Routing plugin is so common in any application, there is a convenient routing function that makes it simpler to install routing. In the code snippet below, install(Routing) is replaced with the routing function:

import io.ktor.server.routing.*

routing {
    // ...
}

Define a route handler

After installing the Routing plugin, you can call the route function inside routing to define a route:

import io.ktor.server.routing.*
import io.ktor.http.*
import io.ktor.server.response.*

routing {
    route("/hello", HttpMethod.Get) {
        handle {
            call.respondText("Hello")
        }
    }
}

Ktor also provides a series of functions that make defining route handlers much easier and more concise. For example, you can replace the previous code with a get function that now only needs to take the URL and the code to handle the request:

import io.ktor.server.routing.*
import io.ktor.server.response.*

routing {
    get("/hello") {
        call.respondText("Hello")
    }
}

Similarly, Ktor provides functions for all the other verbs, that is put, post, head, and so on.

In summary, you need to specify the following settings to define a route:

HTTP verb
Choose the HTTP verb, such as GET, POST, PUT, and so on. The most convenient way is to use a dedicated verb function, such as get, post, put, and so on.
Path pattern
Specify a path pattern used to match a URL path, for example, /hello, /customer/{id}. You can pass a path pattern right to the get/post /etc. function, or you can use the route function to group route handlers and define nested routes.
Handler
Specify how to handle requests and responses. Inside the handler, you can get access to ApplicationCall, handle client requests, and send responses.

Specify a path pattern

A path pattern passed to the routing functions (route, get, post, etc.) is used to match a path component of the URL. A path can contain a sequence of path segments separated by a slash / character.

Below are several path examples:

/hello
A path containing a single path segment.
/order/shipment
A path containing several path segments. You can pass such a path to the route/get/etc. function as is or organize sub-routes by nesting several route functions.
/user/{login}
A path with the login path parameter, whose value can be accessed inside the route handler.
/user/*
A path with a wildcard character that matches any path segment.
/user/{...}
A path with a tailcard that matches all the rest of the URL path.
/user/{param...}
A path containing a path parameter with tailcard.
Regex("/.+/hello")
A path containing a regular expression that matches path segments up to and including the first occurrence of the /hello.

Wildcard

A wildcard (*) matches any path segment and can't be missing. For example, /user/* matches /user/john, but doesn't match /user.

Tailcard

A tailcard ({...}) matches all the rest of the URL path, can include several path segments, and can be empty. For example, /user/{...} matches /user/john/settings as well as /user.

Path parameter

A path parameter ({param}) matches a path segment and captures it as a parameter named param. This path segment is mandatory, but you can make it optional by adding a question mark: {param?}. For example:

/user/{login} matches /user/john, but doesn't match /user.
/user/{login?} matches /user/john as well as /user.
Note that optional path parameters {param?} can only be used at the end of the path.

To access a parameter value inside the route handler, use the call.parameters property. For example, call.parameters["login"] in the code snippet below will return admin for the /user/admin path:

get("/user/{login}") {
    if (call.parameters["login"] == "admin") {
        // ...
    }
}

Path parameter with tailcard

A path parameter with a tailcard ({param...}) matches all the rest of the URL path and puts multiple values for each path segment into parameters using param as a key. For example, /user/{param...} matches /user/john/settings. To access path segments' values inside the route handler, use call.parameters.getAll("param"). For the example above, the getAll function will return an array containing the john and settings values.

Regular expression

Regular expressions can be used with all defining route handlers functions: route, get, post, and so on.

Let's write a route that matches any path that ends with /hello.

import io.ktor.server.routing.*
import io.ktor.server.response.*

routing {
    get(Regex(".+/hello")) {
        call.respondText("Hello")
    }
}

With this route definition, any incoming request to a path ending with /hello, such as /foo/hello, /bar/baz/hello, and so on, will be matched.

Accessing path parts in handler

In regular expressions, named groups are a way to capture a specific part of a string that matches a pattern and assign it a name. The syntax (?<name>pattern) is used to define named groups, where name is the name of a group and pattern is a regular expression pattern that matches the group.

By defining a named group in a route function, you can capture a part of the path, and then in the handler function, you can access the captured parameter using the call.parameters object.

For example, you can define a route that matches requests to a path that includes an integer identifier followed by /hello.

import io.ktor.server.routing.*
import io.ktor.server.response.*

routing {
    get(Regex("""(?<id>\d+)/hello""")) {
        val id = call.parameters["id"]!!
        call.respondText(id)
    }
}

In the code below, the (?<id>\d+) named group is used to capture the integer identifier id from a requested path, and the call.parameters property is used to access the captured id parameter in the handler function.

Unnamed groups can't be accessed inside a regex route handler, but you can use them to match the path. For example, path hello/world will be matched while hello/World not:

import io.ktor.server.routing.*
import io.ktor.server.response.*

routing {
    get(Regex("hello/([a-z]+)")) {
        call.respondText("Hello")
    }
}

Also, the whole path segment needs to be consumed by regex. For example, path pattern get(Regex("[a-z]+")) will not match the path "hello1" but will match the part hello of the path hello/1 and leave /1 for the next route.

Define multiple route handlers

Group routes by verb functions

If you want to define multiple route handlers, which of course is the case for any application, you can just add them to the routing function:

routing {
    get("/customer/{id}") {

    }
    post("/customer") {

    }
    get("/order") {

    }
    get("/order/{id}") {
    
    }
}

In this case, each route has its own function and responds to the specific endpoint and HTTP verb.

Group routes by paths

An alternative way is to group these by paths, whereby you define the path and then place the verbs for that path as nested functions, using the route function:

routing {
    route("/customer") {
        get {

        }
        post {

        }
    }
    route("/order") {
        get {

        }
        get("/{id}") {

        }
    }
}

Nested routes

Independently of how you do the grouping, Ktor also allows you to have sub-routes as parameters to route functions. This can be useful to define resources that are logically children of other resources. The following example shows us how to respond to GET and POST requests to /order/shipment:

routing {
    route("/order") {
        route("/shipment") {
            get {
                
            }
            post {
                
            }
        }
    }
}

So, each route call generates a separate path segment.

Route extension functions

A common pattern is to use extension functions on the Route type to define the actual routes, allowing us easy access to the verbs and remove clutter of having all routes in a single routing function. You can apply this pattern independently of how you decide to group routes. As such, the first example could be represented in a cleaner way:

routing {
    listOrdersRoute()
    getOrderRoute()
    totalizeOrderRoute()
}

fun Route.listOrdersRoute() {
    get("/order") {

    }
}

fun Route.getOrderRoute() {
    get("/order/{id}") {
        
    }
}

fun Route.totalizeOrderRoute() {
    get("/order/{id}/total") {
        
    }
}

You can find the full example demonstrating this approach here: tutorial-website-interactive.

Trace routes

With logging configured, Ktor enables route tracing that helps you determine why some routes are not being executed. For example, if you run the application and make a request to a specified endpoint, the application's output might look as follows:

TRACE Application - Trace for [missing-page]
/, segment:0 -> SUCCESS @ /
  /, segment:0 -> SUCCESS @ /
    /(method:GET), segment:0 -> FAILURE "Not all segments matched" @ /(method:GET)
Matched routes:
  No results
Route resolve result:
  FAILURE "No matched subtrees found" @ /

Last modified: 19 April 2023