BoxLang 🚀 A New JVM Dynamic Language Learn More...
cbguard
lets us lock down methods to logged in users and users with specific permissions using one annotation — secured
. Just sticking the secured annotation on a handler or action is enough to require a user to log in before executing those events.
Here's an example of how to lock down an entire handler:
component secured {
function index( event, rc, prc ) {
// ...
}
function show( event, rc, prc ) {
// ...
}
}
You can be more specific and lock down only specific actions using the same annotation:
component {
function create( event, rc, prc ) secured {
// ...
}
}
You can further lock down handlers and actions to a list of specific permissions. If specified, the logged in user must have one of the permissions in the list specified.
component secured="admin" {
function index( event, rc, prc ) {
// ...
}
function show( event, rc, prc ) {
// ...
}
}
In the above component, the user must have the admin
permission to access the actions in this handler.
component {
function show( event, rc, prc ) secured="admin,reviews_posts" {
// ...
}
}
Individual actions can be secured in the same way. Above, the show
action requires the logged in user to have either the admin
or the reviews_posts
permission.
These two approaches can be combined and both handler and actions can be secured together:
component secured {
function index( event, rc, prc ) {
// ...
}
function new( event, rc, prc ) secured="create_posts" {
// ...
}
}
While the user needs to be logged in to interact at all with this handler, they also need the create_posts
permission to interact with the new
action.
cbguard
also allows you to check for authorization at any point in the request lifecycle using the Guard@cbguard
component.
component secured {
property name="guard" inject="@cbguard";
function update( event, rc, prc ) {
var post = getInstance( "Post" ).findOrFail( rc.post );
// this will throw a `NotAuthorized` exception if the user cannot update the post
guard.authorize( "update-post", { "post": post } );
// update the post as normal...
}
}
The methods available to you on the Guard
component are as follows:
public boolean function allows( required any permissions, struct additionalArgs = {} );
public boolean function denies( required any permissions, struct additionalArgs = {} );
public boolean function all( required any permissions, struct additionalArgs = {} );
public boolean function none( required any permissions, struct additionalArgs = {} );
public void function authorize( required any permissions, struct additionalArgs = {}, string errorMessage );
In all cases permissions
can be either a string, a list of strings, or an array of strings.
In the case of authorize
the errorMessage
replaces the thrown error message
in the NotAuthorized
. exception. It can also be a closure that takes the following shape:
string function errorMessage( array permissions, any user, struct additionalArgs );
While handling all of your guard clauses inside the hasPermission
method on your user
works fine, you may want to define a different way to handle permissions. You
can do this by declaring custom guards using the guard.define
method. Here's the signature:
public Guard function define( required string name, required any callback );
The name
will match against a permission name. If it matches, the guard is
called instead of calling hasPermission
on the User
model. (You can always
call hasPermission
on the User
inside your guard callback if you need.)
The callback can be: a closure or UDF, a component with an authorize
function,
or a WireBox mapping to a component with an authorize
function. Please note
that the authorize function must be explicitly defined and public (No onMissingMethod
).
This authorize
function is called with two parameters: the user
being authorized
and a struct of additionalArgs
and must return a boolean
, like so:
public boolean function authorize( required any user, struct additionalArgs = {} );
Using this approach, you can define custom guards anywhere in your application:
config/ColdBox.cfc
, ModuleConfig.cfc
of your custom modules, etc. The
Guard
component is registered as a singleton, so it will keep track of all the
guards registered, even from different sources.
If you have a need to remove a guard definition you can do so with the removeDefinition
method:
public Guard function removeDefinition( required string name );
When a user is denied access to a action, an event of your choosing is executed instead. There are four keys that can be set in the moduleSettings
struct that all come with good defaults.
authenticationOverrideEvent
(Default: Main.onAuthenticationFailure
)This is the event that is executed when the user is not logged in and is attempting to execute a secured action, whether or not that handler or action has permissions.
authorizationOverrideEvent
(Default: same as authenticationOverrideEvent
)This is the event that is executed when the user is logged in and is attempting to execute a secured action but does not have the requisite permissions.
authenticationAjaxOverrideEvent
(Default: Main.onAuthenticationFailure
)This is the event that is executed when the user is not logged in and is attempting to execute a secured action via ajax (event.isAjax()
), whether or not that handler or action has permissions. By default, this will execute the same action that is configured for authenticationOverrideEvent
.
authorizationAjaxOverrideEvent
(Default: same as authorizationOverrideEvent
)This is the event that is executed when the user is logged in and is attempting to execute a secured action via ajax (event.isAjax()
) but does not have the requisite permissions. By default, this will execute the same action that is configured for authorizationOverrideEvent
.
When an override event is used, the url the user was trying to access is stored in the flash
scope as _securedUrl
.
You can make use of this in your login actions to send the user back where they intended after logging in.
cbguard
requires a bit of setup to function properly.
First, there are two interfaces that must be followed:
interface {
/**
* Must return an object that conforms to `HasPermissionInterface`.
* (This may be an implicit implements.)
*/
public HasPermissionInterface function getUser();
/**
* Returns true if the user is logged in.
*/
public boolean function isLoggedIn();
}
interface {
/**
* Returns true if the user has the specified permission.
* Any additional arguments may be passed in as the second argument.
* This allows you to check if a user can access a specific resource,
* rather than just a generic check.
*/
public boolean function hasPermission( required string permission, struct additionalArgs );
}
Note: These interfaces are not enforced at compile time to give you maximum flexibility.
To configure the AuthenticationService, set the value of authenticationService
in your moduleSettings
to a WireBox mapping:
moduleSettings = {
cbguard = {
authenticationService = "SecurityService@myapp"
}
};
The default authenticationService
for cbguard
is AuthenticationService@cbauth
. cbauth
follows the AuthenticationServiceInterface
out of the box.
You can change the method names called on the AuthenticationService
and the returned User
if you need to. We highly discourage this use case, as it makes it harder to utilize the cbguard
conventions across projects. However, should the need arise, you can modify the method names as follows:
moduleSettings = {
cbguard = {
methodNames = {
isLoggedIn = "getIsLoggedIn",
getUser = "retrieveUser",
hasPermission = "checkPermission"
}
}
};
Additionally, you can modify the override action for each of the event types:
moduleSettings = {
cbguard = {
overrideActions = {
authenticationOverrideEvent = "relocate",
authenticationAjaxOverrideEvent = "override",
authorizationOverrideEvent = "relocate",
authorizationAjaxOverrideEvent = "override"
}
}
};
relocate
refers to calling relocate
on the controller. The user will be redirected to the new page.
override
refers to event.overrideEvent
. This will not redirect but simply change the running event.
All of the cbguard
settings can be overriden inside a module. This allows modules, such as an API module, to provide
their own authentication services as well as redirect events.
To specify some overrides, create a cbguard
struct in your desired module's settings
in that module's ModuleConfig.cfc
.
component {
this.name = "myModule";
function configure() {
settings = {
"cbguard" = {
"authenticationOverrideEvent" = "myModule:Main.onAuthenticationFailure",
"authorizationOverrideEvent" = "myModule:Main.onAuthorizationFailure"
}
};
}
}
If an onAuthenticationFailure
or onAuthorizationFailure
method exists on the handler being
secured, it will be used in the case of an authentication or authorization failure event,
respectively.
// handlers/Admin.cfc
component secured {
function index( event, rc, prc ) {
event.setView( "admin/index" );
}
function secret( event, rc, prc ) secured="superadmin" {
event.setView( "admin/secret" );
}
function onAuthenticationFailure( event, rc, prc ) {
relocate( "/login" );
}
function onAuthenticationFailure( event, rc, prc ) {
flash.put( "authorizationError", "You don't have the correct permissions to access that resource." );
redirectBack(); // from the redirectBack module
}
}
cbguard will process your authorization and authentication failures in the following order:
onAuthenticationFailure
& onAuthorizationFailure
within your handlers).modules_app/api/ModuleConfig.cfc
when the handler is in the module, i.e. modules_app/api/handlers/Main.cfc
.)config/ColdBox.cfc
using moduleSettings
.autoRegisterInterceptor
If you need more control over the order of your interceptors you can
disable the automatic loading of the SecuredEventInterceptor
interceptor.
If you do this you will need to register it yourself
(most likely in config/ColdBox.cfc
) as cbguard.interceptors.SecuredEventInterceptor
.
$
box install cbguard