Timber is a request router with regular expression support, high performance, and a developer-friendly, human-readable API

2.3.1 2021-02-25 10:18 UTC


PHP Version Build Status

Timber is a request router with regular expression support, high performance, and a developer-friendly, human-readable API.

Originally a fork of TreeRoute by Vadim Baryshev, the API and feature-set quickly grew into something else entirely. What does carry over from the original fork, is great performance.


Install the latest version with composer require mindplay/timber


The core router system is independent of some other optional features - this includes a Dispatcher and a UrlHelper, both of which work great together with the router, but the router itself is not dependent on either of these components.


In the examples below, we assume that handlers (attached using get(), post(), etc.) are controller class-names - the included Dispatcher (see below) uses a class-per-action strategy, which may not be everyone's cup of tea, but the point is that, the meaning or interpretation of the handler string isn't baked into the router; it's supported by the Dispatcher only, and as such, you can easily implement any convention or strategy you like, either by doing your own dispatch (without using our Dispatcher) or by extending the Dispatcher and overriding the toCallable() function.

Note the use of the magic ::class constant in the following examples - the library is compatible with PHP 5.4, but this syntax is only supported by PHP 5.5 and up.

Basic usage of the router, without Dispatcher, goes a little something like this:

use mindplay\timber\Router;

require __DIR__ . '/vendor/autoload.php';

$router = new Router();

// Defining route for one HTTP method

// Defining route for several HTTP methods

// Defining a route with regular expression param
$news_route = $router->route('news/<id:^[0-9]+$>')->get(ShowNews::class);

// Defining another route with symbolic param

// Defining static route that conflicts with previous route, but static routes have high priority

// Defining a wildcard route, matching e.g. "categories/foo", "categories/foo/bar", etc.:

// Resolve HTTP method and URL:

$method = 'GET';
$url = '/news/1';

$result = $router->resolve($method, $url);

if (!$result->error) {
    $handler = $result->handler;
    $params = $result->params;
    // Do something with handler and params
} else {
    switch ($result->error->code) {
        case 404 :
            // Not found handler here
        case 405 :
            // Method not allowed handler here
            $allowedMethods = $result->error->allowed;
            if ($method == 'OPTIONS') {
                // OPTIONS method handler here

If you're building a set of routes under the same parent route, you can continue routing from by keeping a temporary reference to a parent route - for example:

$admin = $router->route('admin')->get(AdminMenu::class);


This example will route /admin to AdminMenu, and /admin/users to AdminUserList, etc.

Continuing from a parent route avoids the overhead of repeatedly parsing the same parent path, and also enables modular reuse of route definitions - for example:

$build_comment_routes = function (Route $parent) {


This example creates two identical sets of routes for displaying and posting comments for two different parent routes.


We also provide a Dispatcher class and Controller interface, following a simple convention of handler names being class-names, and a class-per-action strategy:

use mindplay\timber\Router;
use mindplay\timber\Controller;
use mindplay\timber\Dispatcher;

class ShowNews implements Controller
    public function run($id) {
        // load and display news article...

$router = new Router();
$router->route('news/<id:int>', ShowNews::class);

$dispatcher = new Dispatcher($router);

$dispatcher->run('GET', '/news');

The named parameter in this example will be converted to an integer and provided to the run() method.

URL Creation

In addition, we provide a base class for URL creation helpers - this has no direct relationship with the router as such, it's provided for convenience, to support our overall strategy of creating testable URL helpers.


You can save and restore the defined routes:

use mindplay\timber\Router;

require __DIR__ . '/vendor/autoload.php';

$router = new Router();

$routes = $router->getRoutes();

$anotherRouter = new Router();

$method = 'GET';
$url = '/news';

$result = $anotherRouter->resolve($method, $url);

The point is, you can serialize/unserialize the routes that have been built, and store them in a cache somewhere, to avoid the initialization overhead. For most projects, this would be considered a micro-optimization - the overhead of building an extremely high number of routes (hundreds or thousands) may make this worthwhile in a very large modular project.


To run the test-suite, navigate to the project root folder and type:

php test/test.php

Design Notes

During the development of this library, a design problem was identified, which required us to make a trade-off. This library did at one point have URL creation as a feature, but after carefully weighing the pros and cons, it was decided to forego this feature, in favor of simpler implementation and support for caching.

Use-cases for three different approaches were explored and evaluated - our whiteboard summarizes the pros and cons as we saw them, and the approach without URL creation was unanimously our favorite, as it leads to the greatest simplicity, both in the library and in the use-case, and supports caching.

The first trade-off is that we don't get to use closures (which can't be serialized) and thereby do not get any direct static coupling between the route and controller/action/params - we do get static coupling to the controller class-name, by using the ::class constant.

The other trade-off is that we can't have a URL creation feature within the router itself, as this leads to either complexity (with the addition of a named route registry as per case 1) or prevents caching (as per case 2 - after some discussion, we decided URL creation provides only a small benefit, guaranteeing that URL creation is consistent with defined patterns; but also, we value the freedom to fully customize URL creation on a case-by-case basis using simpler code (as per case 3) and as such the absence of URL creation can actually be seen as a benefit.