Author image Florian Ragwitz
and 1 contributors


Catalyst::Authentication::Realm::Adaptor - Adjust parameters of authentication processes on the fly


Version 0.02


The Catalyst::Authentication::Realm::Adaptor allows for modification of authentication parameters within the catalyst application. It's basically a filter used to adjust authentication parameters globally within the application or to adjust user retrieval parameters provided by the credential in order to be compatible with a different store. It provides for better control over interaction between credentials and stores. This is particularly useful when working with external authentication such as OpenID or OAuth.

    'Plugin::Authentication' => {
            'default' => {
                class => 'Adaptor'
                credential => {
                    class => 'Password',
                    password_field => 'secret',
                    password_type  => 'hashed',
                    password_hash_type => 'SHA-1',
                store => {
                    class      => 'DBIx::Class',
                    user_class => 'Schema::Person',
                store_adaptor => {
                    method => 'merge_hash',
                    merge_hash => {
                        status => [ 'temporary', 'active' ]

The above example ensures that no matter how $c->authenticate() is called within your application, the key 'status' is added to the authentication hash. This allows you to, among other things, set parameters that should always be applied to your authentication process or modify the parameters to better connect a credential and a store that were not built to work together. In the above example, we are making sure that the user search is restricted to those with a status of either 'temporary' or 'active.'

This realm works by intercepting the original authentication information between the time $c->authenticate($authinfo) is called and the time the realm's $realm->authenticate($c,$authinfo) method is called, allowing for the $authinfo parameter to be modified or replaced as your application requires. It can also operate after the call to the credential's authenticate() method but before the call to the store's find_user method.

If you don't know what the above means, you probably do not need this module.


The configuration for this module goes within your realm configuration alongside your credential and store options.

This module can operate in two points during authentication processing. The first is prior the realm's authenticate call (immediately after the call to $c->authenticate().) To operate here, your filter options should go in a hash under the key credential_adaptor.

The second point is after the call to credential's authenticate method but immediately before the call to the user store's find_user method. To operate prior to find_user, your filter options should go in a hash under the key store_adaptor.

The filtering options for both points are the same, and both the store_adaptor and credential_adaptor can be used simultaneously in a single realm.


There are four ways to configure your filters. You specify which one you want by setting the method configuration option to one of the following: merge_hash, new_hash, code, or action. You then provide the additional information based on which method you have chosen. The different options are described below.

 credential_adaptor => {
     method => 'merge_hash',
     merge_hash => {
         status => [ 'temporary', 'active' ]

This causes the original authinfo hash to be merged with a hash provided by the realm configuration under the key merge_hash key. This is a deep merge and in the case of a conflict, the hash specified by merge_hash takes precedence over what was passed into the authenticate or find_user call. The method of merging is described in detail in the "HASH MERGING" section below.

 store_adaptor => {
     method => 'new_hash',
     new_hash => {
         username => '+(user)',  # this sets username to the value of $originalhash{user}
         user_source => 'openid'

This causes the original authinfo hash to be set aside and replaced with a new hash provided under the new_hash key. The new hash can grab portions of the original hash. This can be used to remap the authinfo into a new format. See the "HASH MERGING" section for information on how to do this.

 store_adaptor => {
     method => 'code',
     code => sub {
         my ($realmname, $original_authinfo, $hashref_to_config ) = @_;
         my $newauthinfo = {};
         ## do something
         return $newauthinfo;

The code method allows for more complex filtering by executing code provided as a subroutine reference in the code key. The realm name, original auth info and the portion of the config specific to this filter are passed as arguments to the provided subroutine. In the above example, it would be the entire store_adaptor hash. If you were using a code ref in a credential_adaptor, you'd get the credential_adapter config instead.

 credential_adaptor => {
     method => 'action',
     controller => 'UserProcessing',
     action => 'FilterCredentials'

The action method causes the adaptor to delegate filtering to a Catalyst action. This is similar to the code ref above, except that instead of simply calling the routine, the action specified is called via <$c-forward>>. The arguments passed to the action are the same as the code method as well, namely the realm name, the original authinfo hash and the config for the adaptor.


The hash merging mechanism in Catalyst::Authentication::Realm::Adaptor is not a simple merge of two hashes. It has some niceties which allow for both re-mapping of existing keys, and a mechanism for removing keys from the original hash. When using the 'merge_hash' method above, the keys from the original hash and the keys for the merge hash are simply combined with the merge_hash taking precedence in the case of a key conflict. If there are sub-hashes they are merged as well.

If both the source and merge hash contain an array for a given hash-key, the values in the merge array are appended to the original array. Note that hashes within arrays will not be merged, and will instead simply be copied.

Simple values are left intact, and in the case of a key existing in both hashes, the value from the merge_hash takes precedence. Note that in the case of a key conflict where the values are of different types, the value from the merge_hash will be used and no attempt is made to merge or otherwise convert them.

Advanced merging

Whether you are using merge_hash or new_hash as the method, you have access to the values from the original authinfo hash. In your new or merged hash, you can use values from anywhere within the original hash. You do this by setting the value for the key you want to set to a special string indicating the key path in the original hash. The string is formatted as follows: <'+(key1.key2.key3)'> This will grab the hash associated with key1, retrieve the hash associated with key2, and finally obtain the value associated with key3. This is easier to show than to explain:

 my $originalhash = {
                        user => {
                                details => {
                                    age       => 27,
                                    haircolor => 'black',
                                    favoritenumbers => [ 17, 42, 19 ]

  my $newhash = {
                    # would result in a value of 'black'
                    haircolor => '+(user.details.haircolor)',

                    # bestnumber would be 42.
                    bestnumber => '+(user.details.favoritenumbers.1)'

Given the example above, the value for the userage key would be 27, (obtained via <'+(user.details.age)'>) and the value for bestnumber would be 42. Note that you can traverse both hashes and arrays using this method. This can be quite useful when you need the values that were passed in, but you need to put them under different keys.

When using the merge_hash method, you sometimes may want to remove an item from the original hash. You can do this by providing a key in your merge_hash at the same point, but setting it's value to '-()'. This will remove the key entirely from the resultant hash. This works better than simply setting the value to undef in some cases.


The authentication system for Catalyst is quite flexible. In most cases this module is not needed. Evidence of this fact is that the Catalyst auth system was substantially unchanged for 2+ years prior to this modules first release. If you are looking at this module, then there is a good chance your problem would be better solved by adjusting your credential or store directly.

That said, there are some areas where this module can be particularly useful. For example, this module allows for global application of additional arguments to authinfo for a certain realm via your config. It also allows for preliminary testing of alternate configs before you adjust every $c->authenticate() call within your application.

It is also useful when combined with the various external authentication modules available, such as OpenID, OAuth or Facebook. These modules expect to store their user information in the Hash provided by the Minimal user store. Often, however, you want to store user information locally in a database or other storage mechanism. Doing this lies somewhere between difficult and impossible normally. With the Adapter realm, you can massage the authinfo hash between the credential's verification and the creation of the local user, and instead use the information returned to look up a user instead.

Using the external auth mechanisms and the action method, you can actually trigger an action to create a user record on the fly when the user has authenticated via an external method. These are just some of the possibilities that Adaptor provides that would otherwise be very difficult to accomplish, even with Catalyst's flexible authentication system.

With all of that said, caution is warranted when using this module. It modifies the behavior of the application in ways that are not obvious and can therefore lead to extremely hard to track-down bugs. This is especially true when using the action filter method. When a developer calls $c->authenticate() they are not expecting any actions to be called before it returns.

If you use the action method, I strongly recommend that you use it only as a filter routine and do not do other catalyst dispatch related activities (such as further forwards, detach's or redirects). Also note that it is EXTREMELY DANGEROUS to call authentication routines from within a filter action. It is extremely easy to accidentally create an infinite recursion bug which can crash your Application. In short - DON'T DO IT.


Jay Kuri, <jayk at>


Please report any bugs or feature requests to bug-catalyst-authentication-realm-adaptor at, or through the web interface at I will be notified, and then you'll automatically be notified of progress on your bug as I make changes.


You can find documentation for this module with the perldoc command.

    perldoc Catalyst::Authentication::Realm::Adaptor

You can also look for information at:



Copyright 2009 Jay Kuri, all rights reserved.

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.