Routing is the act of matching a request to a given controller.
Typically, routing will examine the request URI, and attempt to match the URI path segment against provided constraints. If the constraints match, a set of "matches" are returned, one of which should be the controller name to execute. Routing can utilize other portions of the request URI or environment as well -- for example, the host or scheme, query parameters, headers, request method, and more.
Routing has been written from the ground up for Zend Framework 2.0. Execution is quite similar, but the internal workings are more consistent, performant, and often simpler.
Note
If you are a developer with knowledge of the routing system in Zend Framework 1.x, you should know that some of the old terminology does not apply in Zend Framework 2.x. In the new routing system we don't have a router as such, as every route can match and assemble URIs by themselves, which makes them routers, too.
That said, in most cases the developer does not need to worry about this, because Zend Framework 2.x will take
care of this "under the hood". The work of the router will be done by Zend\Mvc\Router\SimpleRouteStack
or Zend\Mvc\Router\Http\TreeRouteStack
.
The base unit of routing is a Route
:
namespace Zend\Mvc\Router;
use Zend\Stdlib\RequestInterface as Request;
interface RouteInterface
{
public static function factory(array $options = array());
public function match(Request $request);
public function assemble(array $params = array(), array $options = array());
}
A Route
accepts a Request
, and determines if it matches. If so, it returns a RouteMatch
object:
namespace Zend\Mvc\Router;
class RouteMatch
{
public function __construct(array $params);
public function setMatchedRouteName($name);
public function getMatchedRouteName();
public function setParam($name, $value);
public function getParams();
public function getParam($name, $default = null);
}
Typically, when a Route
matches, it will define one or more parameters. These are passed into the
RouteMatch
, and objects may query the RouteMatch
for their values.
$id = $routeMatch->getParam('id', false);
if (!$id) {
throw new Exception('Required identifier is missing!');
}
$entity = $resource->get($id);
Usually you will have multiple routes you wish to test against. In order to facilitate this, you will use a route
aggregate, usually implementing RouteStack
:
namespace Zend\Mvc\Router;
interface RouteStackInterface extends RouteInterface
{
public function addRoute($name, $route, $priority = null);
public function addRoutes(array $routes);
public function removeRoute($name);
public function setRoutes(array $routes);
}
Typically, routes should be queried in a LIFO order, and hence the reason behind the name RouteStack
. Zend
Framework provides two implementations of this interface, SimpleRouteStack
and TreeRouteStack
. In each, you
register routes either one at a time using addRoute()
, or in bulk using addRoutes()
.
// One at a time:
$route = Literal::factory(array(
'route' => '/foo',
'defaults' => array(
'controller' => 'foo-index',
'action' => 'index',
),
));
$router->addRoute('foo', $route);
// In bulk:
$router->addRoutes(array(
// using already instantiated routes:
'foo' => $route,
// providing configuration to allow lazy-loading routes:
'bar' => array(
'type' => 'literal',
'options' => array(
'route' => '/bar',
'defaults' => array(
'controller' => 'bar-index',
'action' => 'index',
),
),
),
));
Two routers are provided, the SimpleRouteStack
and TreeRouteStack
. Each works with the above interface, but
utilize slightly different options and execution paths. By default, the Zend\Mvc
uses the TreeRouteStack
as
the router.
This router simply takes individual routes that provide their full matching logic in one go, and loops through them
in LIFO order until a match is found. As such, routes that will match most often should be registered last, and
least common routes first. Additionally, you will need to ensure that routes that potentially overlap are
registered such that the most specific match will match first (i.e., register later). Alternatively, you can set
priorities by giving the priority as third parameter to the addRoute()
method, specifying the priority in the
route specifications or setting the priority property within a route instance before adding it to the route stack.
Zend\Mvc\Router\Http\TreeRouteStack
provides the ability to register trees of routes, and will use a B-tree
algorithm to match routes. As such, you register a single route with many children.
A TreeRouteStack
will consist of the following configuration:
- A base "route", which describes the base match needed, the root of the tree.
- An optional "route_plugins", which is a configured
Zend\Mvc\Router\RoutePluginManager
that can lazy-load routes. - The option "may_terminate", which hints to the router that no other segments will follow it.
- An optional "child_routes" array, which contains additional routes that stem from the base "route" (i.e., build
from it). Each child route can itself be a
TreeRouteStack
if desired; in fact, thePart
route works exactly this way.
When a route matches against a TreeRouteStack
, the matched parameters from each segment of the tree will be
returned.
A TreeRouteStack
can be your sole route for your application, or describe particular path segments of the
application.
An example of a TreeRouteStack
is provided in the documentation of the Part
route.
Zend Framework 2.0 ships with the following HTTP route types.
The Hostname
route attempts to match the hostname registered in the request against specific criteria.
Typically, this will be in one of the following forms:
- "subdomain.domain.tld"
- ":subdomain.domain.tld"
In the above, the second route would return a "subdomain" key as part of the route match.
For any given hostname segment, you may also provide a constraint. As an example, if the "subdomain" segment needed to match only if it started with "fw" and contained exactly 2 digits following, the following route would be needed:
$route = Hostname::factory(array(
'route' => ':subdomain.domain.tld',
'constraints' => array(
'subdomain' => 'fw\d{2}',
),
));
In the above example, only a "subdomain" key will be returned in the RouteMatch
. If you wanted to also provide
other information based on matching, or a default value to return for the subdomain, you need to also provide
defaults.
$route = Hostname::factory(array(
'route' => ':subdomain.domain.tld',
'constraints' => array(
'subdomain' => 'fw\d{2}',
),
'defaults' => array(
'type' => 'json',
),
));
When matched, the above will return two keys in the RouteMatch
, "subdomain" and "type".
The Literal
route is for doing exact matching of the URI path. Configuration therefore is solely the path you
want to match, and the "defaults", or parameters you want returned on a match.
$route = Literal::factory(array(
'route' => '/foo',
'defaults' => array(
'controller' => 'Application\Controller\IndexController',
'action' => 'foo',
),
));
The above route would match a path "/foo", and return the key "action" in the RouteMatch
, with the value
"foo".
The Method
route is used to match the http method or 'verb' specified in the request (See RFC 2616 Sec. 5.1.1).
It can optionally be configured to match against multiple methods by providing a comma-separated list of method
tokens.
$route = Method::factory(array(
'verb' => 'post,put',
'defaults' => array(
'controller' => 'Application\Controller\IndexController',
'action' => 'form-submit',
),
));
The above route would match an http "POST" or "PUT" request and return a RouteMatch
object containing a key
"action" with a value of "form-submit".
A Part
route allows crafting a tree of possible routes based on segments of the URI path. It actually extends
the TreeRouteStack
.
Part
routes are difficult to describe, so we'll simply provide a sample one here.
$route = Part::factory(array(
'route' => array(
'type' => 'literal',
'options' => array(
'route' => '/',
'defaults' => array(
'controller' => 'Application\Controller\IndexController',
'action' => 'index',
),
),
),
'route_plugins' => $routePlugins,
'may_terminate' => true,
'child_routes' => array(
'blog' => array(
'type' => 'literal',
'options' => array(
'route' => '/blog',
'defaults' => array(
'controller' => 'Application\Controller\BlogController',
'action' => 'index',
),
),
'may_terminate' => true,
'child_routes' => array(
'rss' => array(
'type' => 'literal',
'options' => array(
'route' => '/rss',
'defaults' => array(
'action' => 'rss',
)
),
'may_terminate' => true,
'child_routes' => array(
'subrss' => array(
'type' => 'literal',
'options' => array(
'route' => '/sub',
'defaults' => array(
'action' => 'subrss',
),
),
),
),
),
),
),
'forum' => array(
'type' => 'literal',
'options' => array(
'route' => 'forum',
'defaults' => array(
'controller' => 'Application\Controller\ForumController',
'action' => 'index',
),
),
),
),
));
The above would match the following:
- "/" would load the "Index" controller, "index" action.
- "/blog" would load the "Blog" controller, "index" action.
- "/blog/rss" would load the "Blog" controller, "rss" action.
- "/blog/rss/sub" would load the "Blog" controller, "subrss" action.
- "/forum" would load the "Forum" controller, "index" action.
You may use any route type as a child route of a Part
route.
Note
Part
routes are not meant to be used directly. When you add definitions for child_routes
to any route
type, that route will become a Part
route. As already said, describing Part
routes with words is
difficult, so hopefully the additional :ref:`examples at the end <zend.mvc.routing.http-route-types.examples>`
will provide further insight.
Note
In the above example, the $routePlugins
is an instance of Zend\Mvc\Router\RoutePluginManager
.
$routePlugins = new Zend\Mvc\Router\RoutePluginManager();
$plugins = array(
'hostname' => 'Zend\Mvc\Router\Http\Hostname',
'literal' => 'Zend\Mvc\Router\Http\Literal',
'part' => 'Zend\Mvc\Router\Http\Part',
'regex' => 'Zend\Mvc\Router\Http\Regex',
'scheme' => 'Zend\Mvc\Router\Http\Scheme',
'segment' => 'Zend\Mvc\Router\Http\Segment',
'wildcard' => 'Zend\Mvc\Router\Http\Wildcard',
'query' => 'Zend\Mvc\Router\Http\Query',
'method' => 'Zend\Mvc\Router\Http\Method',
);
$foreach ($plugins as $name => $class) {
$routePlugins->setInvokableClass($name, $class);
}
When using Zend\Mvc\Router\Http\TreeRouteStack
, the RoutePluginManager
is set up by default, and the
developer does not need to worry about the autoloading of standard HTTP routes.
A Regex
route utilizes a regular expression to match against the URI path. Any valid regular expression is
allowed; our recommendation is to use named captures for any values you want to return in the RouteMatch
.
Since regular expression routes are often complex, you must specify a "spec" or specification to use when
assembling URLs from regex routes. The spec is simply a string; replacements are identified using "%keyname%"
within the string, with the keys coming from either the captured values or named parameters passed to the
assemble()
method.
Just like other routes, the Regex
route can accept "defaults", parameters to include in the RouteMatch
when
successfully matched.
$route = Regex::factory(array(
'regex' => '/blog/(?<id>[a-zA-Z0-9_-]+)(\.(?<format>(json|html|xml|rss)))?',
'defaults' => array(
'controller' => 'Application\Controller\BlogController',
'action' => 'view',
'format' => 'html',
),
'spec' => '/blog/%id%.%format%',
));
The above would match "/blog/001-some-blog_slug-here.html", and return four items in the RouteMatch
, an "id",
the "controller", the "action", and the "format". When assembling a URL from this route, the "id" and "format"
values would be used to fill the specification.
The Scheme
route matches the URI scheme only, and must be an exact match. As such, this route, like the
Literal
route, simply takes what you want to match and the "defaults", parameters to return on a match.
$route = Scheme::factory(array(
'scheme' => 'https',
'defaults' => array(
'https' => true,
),
));
The above route would match the "https" scheme, and return the key "https" in the RouteMatch
with a boolean
true
value.
A Segment
route allows matching any segment of a URI path. Segments are denoted using a colon, followed by
alphanumeric characters; if a segment is optional, it should be surrounded by brackets. As an example,
"/:foo[/:bar]" would match a "/" followed by text and assign it to the key "foo"; if any additional "/" characters
are found, any text following the last one will be assigned to the key "bar".
The separation between literal and named segments can be anything. For example, the above could be done as "/:foo{-}[-:bar] as well. The {-} after the :foo parameter indicates a set of one or more delimiters, after which matching of the parameter itself ends.
Each segment may have constraints associated with it. Each constraint should simply be a regular expression expressing the conditions under which that segment should match.
Also, as you can in other routes, you may provide defaults to use; these are particularly useful when using optional segments.
As a complex example:
$route = Segment::factory(array(
'route' => '/:controller[/:action]',
'constraints' => array(
'controller' => '[a-zA-Z][a-zA-Z0-9_-]+',
'action' => '[a-zA-Z][a-zA-Z0-9_-]+',
),
'defaults' => array(
'controller' => 'Application\Controller\IndexController',
'action' => 'index',
),
));
This route part is deprecated since you can now add query parameters without a query route.
The Query
route part allows you to specify and capture query string parameters for a given route.
The intention of the Query
part is that you do not instantiate it in its own right but to use it as a child of
another route part.
An example of its usage would be
$route = Part::factory(array(
'route' => array(
'type' => 'literal',
'options' => array(
'route' => 'page',
'defaults' => array(
),
),
),
'may_terminate' => true,
'route_plugins' => $routePlugins,
'child_routes' => array(
'query' => array(
'type' => 'Query',
'options' => array(
'defaults' => array(
'foo' => 'bar',
),
),
),
),
));
As you can see, it's pretty straight forward to specify the query part. This then allows you to create query strings using the url view helper.
$this->url(
'page/query',
array(
'name' => 'my-test-page',
'format' => 'rss',
'limit' => 10,
)
);
As you can see above, you must add "/query" to your route name in order to append a query string. If you do not specify "/query" in the route name then no query string will be appended.
Our example "page" route has only one defined parameter of "name" ("/page[/:name]"), meaning that the remaining parameters of "format" and "limit" will then be appended as a query string.
The output from our example should then be "/page/my-test-page?format=rss&limit=10"
Most of the routing definitions will be done in module configuration files, so the following examples will show how to set up routes in config files.
Simple example with two literal routes
return array(
'router' => array(
'routes' => array(
// Literal route named "home"
'home' => array(
'type' => 'literal',
'options' => array(
'route' => '/',
'defaults' => array(
'controller' => 'Application\Controller\IndexController',
'action' => 'index',
),
),
),
// Literal route named "contact"
'contact' => array(
'type' => 'literal',
'options' => array(
'route' => 'contact',
'defaults' => array(
'controller' => 'Application\Controller\ContactController',
'action' => 'form',
),
),
),
),
),
);
A complex example with child routes
return array(
'router' => array(
'routes' => array(
// Literal route named "home"
'home' => array(
'type' => 'literal',
'options' => array(
'route' => '/',
'defaults' => array(
'controller' => 'Application\Controller\IndexController',
'action' => 'index',
),
),
),
// Literal route named "blog", with child routes
'blog' => array(
'type' => 'literal',
'options' => array(
'route' => '/blog',
'defaults' => array(
'controller' => 'Application\Controller\BlogController',
'action' => 'index',
),
),
'may_terminate' => true,
'child_routes' => array(
// Segment route for viewing one blog post
'post' => array(
'type' => 'segment',
'options' => array(
'route' => '/[:slug]',
'constraints' => array(
'slug' => '[a-zA-Z0-9_-]+',
),
'defaults' => array(
'action' => 'view',
),
),
),
// Literal route for viewing blog RSS feed
'rss' => array(
'type' => 'literal',
'options' => array(
'route' => '/rss',
'defaults' => array(
'action' => 'rss',
),
),
),
),
),
),
),
);
When using child routes, naming of the routes follows the parent/child
pattern, so to use the child routes
from the above example:
echo $this->url('blog'); // gives "/blog"
echo $this->url('blog/post', array('slug' => 'my-post')); // gives "/blog/my-post"
echo $this->url('blog/rss'); // gives "/blog/rss"
An example with multiple Hostnames and subdomains within a single application
return array(
'router' => array(
'routes' => array(
'modules.zendframework.com' => array(
'type' => 'Zend\Mvc\Router\Http\Hostname',
'options' => array(
'route' => ':4th.[:3rd.]:2nd.:1st', // domain levels from right to left
'contraints' => array(
'4th' => 'modules',
'3rd' => '.*?', // optional 3rd level domain such as .ci, .dev or .test
'2nd' => 'zendframework',
'1st' => 'com',
),
// Purposely omit default controller and action
// to let the child routes control the route match
),
// child route controllers may span multiple modules as desired
'child_routes' => array(
'index' => array(
'type' => 'Zend\Mvc\Router\Http\Literal',
'options' => array(
'route' => '/',
'defaults' => array(
'controller' => 'Module\Controller\Index',
'action' = > 'index',
),
),
'may_terminate' => true,
),
),
),
'packages.zendframework.com' => array(
'type' => 'Zend\Mvc\Router\Http\Hostname',
'options' => array(
'route' => ':4th.[:3rd.]:2nd.:1st', // domain levels from right to left
'contraints' => array(
'4th' => 'packages',
'3rd' => '.*?', // optional 3rd level domain such as .ci, .dev or .test
'2nd' => 'zendframework',
'1st' => 'com',
),
// Purposely omit default controller and action
// to let the child routes control the route match
),
// child route controllers may span multiple modules as desired
'child_routes' => array(
'index' => array(
'type' => 'Zend\Mvc\Router\Http\Literal',
'options' => array(
'route' => '/',
'defaults' => array(
'controller' => 'Package\Controller\Index',
'action' = > 'index',
),
),
'may_terminate' => true,
),
),
),
),
),
);
The above would match the following:
modules.zendframework.com
would dispatch theIndex
controller'sindex
action of theModule
module.modules.ci.zendframework.com
would dispatch theIndex
controller'sindex
action of theModule
module.packages.zendframework.com
would dispatch theIndex
controller'sindex
action of thePackage
module.packages.dev.zendframework.com
would dispatch theIndex
controller'sindex
action of thePackage
module.
The Url
controller plugin or view helper may be used to generate URLs following the above example:
// reuse the route matched parameters to generate URLs
echo $this->url('modules.zendframework.com/index', array(), array(), true);
echo $this->url('packages.zendframework.com/index', array(), array(), true);
Warning
When defining child routes pay attention that the may_terminate
and child_routes
definitions
are in same level as the options
and type
definitions. A common pitfall is to have those two
definitions nested in options
, which will not result in the desired routes.
Zend Framework 2.0 also comes with routes for writing Console based applications, which is explained in the :ref:`Console routes and routing <zend.console.routes>` section.