7.5. Der Standard Router: Zend_Controller_Router_Rewrite

7.5.1. Einführung

Zend_Controller_Router_Rewrite ist der Standard Router des Frameworks. Routing ist der Prozess der Übernahme und Zerteilung einer URI, um zu ermitteln, welches Modul, welcher Controller und welche Aktion des Controllers die Anfrage erhalten soll. Die Definition des Moduls, des Controllers, der Aktion sowie weiterer Parameter wird in einem Objekt mit Namen Zend_Controller_Dispatcher_Token gekapselt, das dann vom Zend_Controller_Dispatcher_Standard verarbeitet wird. Das Routing geschieht nur einmal: wenn zu Beginn die Anfrage erhalten wird und bevor der erste Controller aufgerufen wird.

Zend_Controller_Router_Rewrite wurde entwickelt, um mit reinen PHP Strukturen eine mod_rewrite ähnliche Funktionalität zu erlauben. Es richtet sich sehr frei nach dem Ruby on Rails Routing und benötigt kein tieferes Wissen über URL Weiterleitung des Webservers. Es wurde entwickelt, um mit einer einzigen mod_rewrite Regel zu arbeiten.

RewriteEngine on
RewriteRule !\.(js|ico|gif|jpg|png|css)$ index.php

or:

RewriteEngine on
RewriteCond %{SCRIPT_FILENAME} !-f
RewriteCond %{SCRIPT_FILENAME} !-d
RewriteRule ^(.*)$ index.php/$1 

Der Rewrite Router kann auch mit dem IIS Webserver verwendet werden, wenn Isapi_Rewrite als Isapi Erweiterung installiert wurde und folgende Umschreibungsregel verwendet wird:

RewriteRule ^[\w/\%]*(?:\.(?!(?:js|ico|gif|jpg|png|css)$)[\w\%]*$)? /index.php [I]
[Anmerkung] IIS Isapi_Rewrite

Bei Verwenung von IIS, wird $_SERVER['REQUEST_URI'] entweder nicht vorhanden sein oder auf einen leeren String gesetzt sein. In diesem Fall wird Zend_Controller_Request_Http versuchen, den durch die Isapi_Rewrite Erweiterung gesetzten Wert $_SERVER['HTTP_X_REWRITE_URL'] zu verwenden.

Bei der Vrwendung von Lighttpd, ist folgende Umschreibungsregel gültig:

url.rewrite-once = ( ".*\.(js|ico|gif|jpg|png|css)$" => "$0", "" => "/index.php")

7.5.2. Einen Router verwenden

Um den Rewrite Router richtig zu verwenden, mußt du ihn instanziieren, einige benutzerdefinierte Routen hinzufügen und in den Controller einbinden. Der folgende Code veranschaulicht die Vorgehensweise:

<?php
/* Erstelle einen Router */

$router = $ctrl->getRouter(); // gibt standardmäßig einen Rewrite Router zurück
$router->addRoute(
    'user',
    new Zend_Controller_Router_Route('user/:username', array('controller' => 'user', 'action' => 'info'))
);

7.5.3. Basic Rewrite Router operation

The heart of the RewriteRouter is the definition of user defined routes. Routes are added by calling the addRoute method of RewriteRouter and passing in a new instance of a class implementing Zend_Controller_Router_Route_Interface. Eg.:

<?php
$router->addRoute('user', new Zend_Controller_Router_Route('user/:username'));

Rewrite Router comes with four basic types of routes (one of which is special):

Routes may be used numerous times to create a chain or user defined application routing schema. You may use any number of routes in any configuration, with the exception of the Module route, which should rather be used once and probably as the most generic route (i.e., as a default). Each route will be described in greater detail later on.

The first parameter to addRoute is the name of the route. It is used as a handle for getting the routes out of the router (e.g., for URL generation purposes). The second parameter being the route itself.

[Anmerkung] Anmerkung

The most common use of the route name is through the means of Zend_View url helper:

<a href="<?= $this->url('user', array('username' => 'martel')) ?>">Martel</a>

Which would result in the href: user/martel.

Routing is a simple process of iterating through all provided routes and matching its definitions to current request URI. When a positive match is found, variable values are returned from the Route instance and are injected into the Zend_Controller_Request object for later use in the dispatcher as well as in user created controllers. On a negative match result, the next route in the chain is checked.

[Anmerkung] Reverse matching

Routes are matched in reverse order so make sure your most generic routes are defined first.

[Anmerkung] Returned values

Values returned from routing come from URL parameters or user defined route defaults. These variables are later accessible through the Zend_Controller_Request::getParam() or Zend_Controller_Action::_getParam() methods.

There are three special variables which can be used in your routes - 'module', 'controller' and 'action'. These special variables are used by Zend_Controller_Dispatcher to find a controller and action to dispatch to.

[Anmerkung] Special variables

The names of these special variables may be different if you choose to alter the defaults in Zend_Controller_Request_Http by means of the setControllerKey and setActionKey methods.

7.5.4. Default routes

Zend_Controller_Router_Rewrite comes preconfigured with a default route, which will match URIs in the shape of controller/action. Additionally, a module name may be specified as the first path element, allowing URIs of the form module/controller/action. Finally, it will also match any additional parameters appended to the URI by default - controller/action/var1/value1/var2/value2.

Some examples of how such routes are matched:

// Assuming the following:
$ctrl->setControllerDirectory(
    array(
        'default' => '/path/to/default/controllers',
        'news'    => '/path/to/blog/controllers',
        'blog'    => '/path/to/blog/controllers'
    )
);

Module only:
http://example/news
    module == news

Invalid module maps to controller name:
http://example/foo
    controller == foo

Module + controller:
http://example/blog/archive
    module     == blog
    controller == archive

Module + controller + action:
http://example/blog/archive/list
    module     == blog
    controller == archive
    action     == list

Module + controller + action + params:
http://example/blog/archive/list/sort/alpha/date/desc
    module     == blog
    controller == archive
    action     == list
    sort       == alpha
    date       == desc

The default route is simply a Zend_Controller_Router_Route_Module object stored under the name (index) of 'default' in RewriteRouter. It's created more-or-less like below:

<?php
$compat = new Zend_Controller_Router_Route_Module(array(), $dispatcher, $request);
$this->addRoute('default', $compat);

If you do not want this particular default route in your routing schema, you may override it by creating your own 'default' route (i.e., storing it under the name of 'default') or removing it altogether by using removeDefaultRoutes():

<?php
// Remove any default routes
$router->removeDefaultRoutes();

7.5.5. Base URL and subdirectories

The rewrite router can be used in subdirectories (e.g., http://domain.com/~user/application-root/>) in which case the base URL of the application (/~user/application-root) should be automatically detected by Zend_Controller_Request_Http and used accordingly.

Should the base URL be detected incorrectly you can override it with your own base path by using Zend_Controller_Request_Http and calling the setBaseUrl() method (see Abschnitt 7.4.2.2, „Basis Url und Unterverzeichnisse“):

<?php
$request->setBaseUrl(/~user/application-root/);

7.5.6. Route Types

7.5.6.1. Zend_Controller_Router_Route

Zend_Controller_Router_Route is the standard framework route. It combines ease of use with flexible route definition. Each route consists primarily of URL mapping (of static and dynamic parts (variables)) and may be initialized with defaults as well as with variable requirements.

Let's imagine our fictional application will need some informational page about the content authors. We want to be able to point our web browsers to http://domain.com/author/martel to see the information about this "martel" guy. And the route for such functionality could look like:

<?php
$route = new Zend_Controller_Router_Route(
    'author/:username', 
    array(
        'controller' => 'profile', 
        'action'     => 'userinfo'
    )
);

$router->addRoute('user', $route);

The first parameter in the Zend_Controller_Router_Route constructor is a route definition that will be matched to a URL. Route definitions consist of static and dynamic parts separated by the slash ('/') character. Static parts are just simple text: author. Dynamic parts, called variables, are marked by prepending a colon to the variable name: :username.

[Anmerkung] Character usage

The current implementation allows you to use any character (except a slash) as a variable identifier, but it is strongly recommended that one uses only characters that are valid for PHP variable identifiers. Future implementations may alter this behaviour, which could result in hidden bugs in your code.

This example route should be matched when you point your browser to http://domain.com/author/martel, in which case all its variables will be injected to the Zend_Controller_Request object and will be accessible in your ProfileController. Variables returned by this example may be represented as an array of the following key and value pairs:

<?php
$values = array(
    'username'   => 'martel',
    'controller' => 'profile',
    'action'     => 'userinfo'
);

Later on, Zend_Controller_Dispatcher_Standard should invoke the userinfoAction() method of your ProfileController class (in the default module) based on these values. There you will be able to access all variables by means of the Zend_Controller_Action::_getParam() or Zend_Controller_Request::getParam() methods:

<?php
public function userinfoAction() 
{
    $request = $this->getRequest();
    $username = $request->getParam('username');

    $username = $this->_getParam('username');
}

Route definition can contain one more special character - a wildcard - represented by '*' symbol. It is used to gather parameters similarly to the default Module route (var => value pairs defined in the URI). The following route more-or-less mimics the Module route behavior:

<?php
$route = new Zend_Controller_Router_Route(
    ':module/:controller/:action/*', 
    array('module' => 'default')
);
$router->addRoute('default', $route);
7.5.6.1.1. Variable defaults

Every variable in the route can have a default and this is what the second parameter of the Zend_Controller_Router_Route constructor is used for. This parameter is an array with keys representing variable names and with values as desired defaults:

<?php
$route = new Zend_Controller_Router_Route(
    'archive/:year', 
    array('year' => 2006)
);
$router->addRoute('archive', $route);

The above route will match URLs like http://domain.com/archive/2005 and http://example.com/archive. In the latter case the variable year will have an initial default value of 2006.

This example will result in injecting a year variable to the request object. Since no routing information is present (no controller and action parameters are defined), the application will be dispatched to the default controller and action method (which are both defined in Zend_Controller_Dispatcher_Abstract). To make it more usable, you have to provide a valid controller and a valid action as the route's defaults:

<?php
$route = new Zend_Controller_Router_Route(
    'archive/:year', 
    array(
        'year'       => 2006, 
        'controller' => 'archive',
        'action'     => 'show'
    )
);
$router->addRoute('archive', $route);

This route will then result in dispatching to the method showAction() of the class ArchiveController.

7.5.6.1.2. Variable requirements

One can add a third parameter to the Zend_Controller_Router_Route constructor where variable requirements may be set. These are defined as parts of a regular expression:

<?php
$route = new Zend_Controller_Router_Route(
    'archive/:year', 
    array(
        'year'       => 2006, 
        'controller' => 'archive',
        'action'     => 'show'
    ),
    array('year' => '\d+')
);
$router->addRoute('archive', $route);

With a route defined like above, the router will match it only when the year variable will contain numeric data, eg. http://domain.com/archive/2345. A URL like http://example.com/archive/test will not be matched and control will be passed to the next route in the chain instead.

7.5.6.2. Zend_Controller_Router_Route_Static

The examples above all use dynamic routes -- routes that contain patterns to match against. Sometimes, however, a particular route is set in stone, and firing up the regular expression engine would be an overkill. The answer to this situation is to use static routes:

<?php
$route = new Zend_Controller_Router_Route_Static(
    'login', 
    array('controller' => 'auth', 'action' => 'login')
);
$router->addRoute('login', $route);

Above route will match a URL of http://domain.com/login, and dispatch to AuthController::loginAction().

7.5.6.3. Zend_Controller_Router_Route_Regex

In addition to the default and static route types, a Regular Expression route type is available. This route offers more power and flexibility over the others, but at a slight cost of complexity. At the same time, it should be faster than the standard Route.

Like the standard route, this route has to be initialized with a route definition and some defaults. Let's create an archive route as an example, similar to the previoulsy defined one, only using the Regex route this time:

$route = new Zend_Controller_Router_Route_Regex(
    'archive/(\d+)', 
    array(
        'controller' => 'archive',
        'action'     => 'show'
    )
);
$router->addRoute('archive', $route);

Every defined regex subpattern will be injected to the request object. With our above example, after successful matching http://domain.com/archive/2006, the resulting value array may look like:

$values = array(
    1            => '2006',
    'controller' => 'archive',
    'action'     => 'show'
);
[Anmerkung] Anmerkung

Leading and trailing slashes are trimmed from the URL in the Router prior to a match. As a result, matching the URL http://domain.com/foo/bar/, would involve a regex of foo/bar, and not /foo/bar.

[Anmerkung] Anmerkung

Line start and line end anchors ('^' and '$', respectively) are automatically pre- and appended to all expressions. Thus, you should not use these in your regular expressions, and you should match the entire string.

[Anmerkung] Anmerkung

This route class uses the # character for a delimiter. This means that you will need to escape hash characters ('#') but not forward slashes ('/') in your route definitions. Since the '#' character (named anchor) is rarely passed to the webserver, you will rarely need to use that character in your regex.

You can get the contents of the defined subpatterns the usual way:

public function showAction() 
{
    $request = $this->getRequest();
    $year    = $request->getParam(1); // $year = '2006';
}
[Anmerkung] Anmerkung

Notice the key is an integer (1) instead of a string ('1').

This route will not yet work exactly the same as its standard route counterpart since the default for 'year' is not yet set. And what may not yet be evident is that we will have a problem with a trailing slash even if we declare a default for the year and make the subpattern optional. The solution is to make the whole year part optional along with the slash but catch only the numeric part:

$route = new Zend_Controller_Router_Route_Regex(
    'archive(?:/(\d+))?', 
    array(
        'year'       => '2006',
        'controller' => 'archive',
        'action'     => 'show'
    )
);
$router->addRoute('archive', $route);

Now let's get to the problem you have probably noticed on your own by now. Using integer based keys for parameters is not an easily manageable solution and may be potentially problematic in the long run. And that's where the third parameter comes in. This parameter is an associative array that represents a map of regex subpatterns to parameter named keys. Let's work on our easier example:

$route = new Zend_Controller_Router_Route_Regex(
    'archive/(\d+)', 
    array(
        'controller' => 'archive',
        'action' => 'show'
    ),
    array(
        1 => 'year'
    )
);
$router->addRoute('archive', $route);

This will result in following values injected into Request:

$values = array(
    'year'       => '2006',
    'controller' => 'archive',
    'action'     => 'show'
);

The map may be defined in either direction to make it work in any environment. Keys may contain variable names or subpattern indexes:

$route = new Zend_Controller_Router_Route_Regex(
    'archive/(\d+)', 
    array( ... ),
    array(1 => 'year')
);

// OR
        
$route = new Zend_Controller_Router_Route_Regex(
    'archive/(\d+)', 
    array( ... ),
    array('year' => 1)
);       
[Anmerkung] Anmerkung

Subpattern keys have to be represented by integers.

Notice that the numeric index in Request values is now gone and a named variable is shown in its place. Of course you can mix numeric and named variables if you wish:

$route = new Zend_Controller_Router_Route_Regex(
    'archive/(\d+)/page/(\d+)', 
    array( ... ),
    array('year' => 1)
);       

Which will result in mixed values available in the Request. As an example, the URL http://domain.com/archive/2006/page/10 will result in following values:

$values = array(
    'year'       => '2006',
    2            => 10,
    'controller' => 'archive',
    'action'     => 'show'
);

Since regex patterns are not easily reversed, you will need to prepare a reverse URL if you wish to use a URL helper or even an assemble method of this class. This reversed path is represented by a string parsable by sprintf() and is defined as a fourth construct parameter:

$route = new Zend_Controller_Router_Route_Regex(
    'archive/(\d+)', 
    array( ... ),
    array('year' => 1),
    'archive/%s'
);       

All of this is something which was already possible by the means of a standard route object, so where's the benefit in using the Regex route, you ask? Primarily, it allows you to describe any type of URL without any restrictions. Imagine you have a blog and wish to create URLs like: http://domain.com/blog/archive/01-Using_the_Regex_Router.html, and have it decompose the last path element, 01-Using_the_Regex_Router.html, into an article ID and article title/description; this is not possible with the standard route. With the Regex route, you can do something like the following solution:

$route = new Zend_Controller_Router_Route_Regex(
    'blog/archive/(\d+)-(.+)\.html',
    array(
        'controller' => 'blog', 
        'action'     => 'view'
    ), 
    array(
        1 => 'id', 
        2 => 'description'
    ),
    'blog/archive/%d-%s.html'
);
$router->addRoute('blogArchive', $route);

As you can see, this adds a tremendous amount of flexibility over the standard route.

7.5.7. Zend_Config mit dem RewriteRouter verwenden

Manchmal ist es praktischer, eine Konfigurationsdatei mit neuen Routen zu aktualisieren, als den Code zu ändern. Dies ist mit Hilfe der addConfig() Methode möglich. Im Wesentlichen kann man eine Zend_Config kompatible Konfiguration erstellen, in seinem Code einlesen und an den RewriteRouter übergeben:

Als Beispiel wird die folgende INI Datei angenommen:

[production]
routes.archive.route = "archive/:year/*"
routes.archive.defaults.controller = archive
routes.archive.defaults.action = show
routes.archive.defaults.year = 2000
routes.archive.reqs.year = "\d+"

routes.news.type = "Zend_Controller_Router_Route_Static"
routes.news.route = "news"
routes.news.defaults.controller = "news"
routes.news.defaults.action = "list"

routes.archive.type = "Zend_Controller_Router_Route_Regex"
routes.archive.route = "archive/(\d+)"
routes.archive.defaults.controller = "archive"
routes.archive.defaults.action = "show"
routes.archive.map.1 = "year" 
; OR: routes.archive.map.year = 1

Die oben angeführte INI Datei kann dann wie folgt in ein Zend_Config Objekt eingelesen werden:

$config = new Zend_Config_Ini('/path/to/config.ini', 'production');
$router = new Zend_Controller_Router_Rewrite();
$router->addConfig($config, 'routes');

Im oberen Beispiel teilen wir dem Router mit, den 'routes' Bereich der INI Datei für seine Routen zu verwenden. Jeder Schlüssel auf erster Ebene in diesem Bereich wird verwendet, um den Namen der Routen zu definieren; das obige Beispiel definiert die Routen 'archive' und 'news'. Jede Route erfordert dann mindestens einen 'route' Eintrag und einen oder mehrere 'defaults' Einträge; optional können eine oder mehrere 'reqs' (kurz für 'required', d.h. erforderlich) Einträge angegeben werden. Alles in allem entspricht dies den drei Argumenten, die an ein Zend_Controller_Router_Route_Interface Objekt übergeben werden. Ein Optionsschlüssel 'type' kann verwendet werden, um den Typ der Routenklasse für diese Route anzugeben; standardmäßig wird Zend_Controller_Router_Route verwendet. Im obigen Beispiel wird die 'news' Route definiert, um Zend_Controller_Router_Route_Static zu verwenden.

7.5.8. Subclassing the Router

The standard rewrite router should provide most functionality you may need; most often, you will only need to create a new route type in order to provide new or modified functionality over the provided routes.

That said, you may at some point find yourself wanting to use a different routing paradigm. The interface Zend_Controller_Router_Interface provides the minimal information required to create a router, and consists of a single method.

<?php
interface Zend_Controller_Router_Interface
{
  /**
   * @param  Zend_Controller_Request_Abstract $request
   * @throws Zend_Controller_Router_Exception
   * @return Zend_Controller_Request_Abstract
   */
  public function route(Zend_Controller_Request_Abstract $request);
}
?>

Das Routing findet nur einmal statt, wenn die Anfrage das erste Mal vom System erhalten wird. Der Zweck des Routers ist es, Controller, Aktion und optionale Parameter auf Basis der Anfrageumgebung zu ermitteln und im Request zu setzen. Das Request Objekt wird dann an den Dispatcher übergeben. Wenn es nicht möglich ist, eine Route auf einen Dispatch Token abzubilden, soll der Router nichts mit dem Request Objekt machen.