Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Support for writing PSC authorization plugins in ruby
Java Ruby
branch: master

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.

caBIG CTMS Suite authorization in ruby

Patient Study Calendar 2.10 and later supports an pluggable authorization scheme defined by an OSGi bundle exporting a service with a particular interface. Other caBIG CTMS Suite applications may be modified to support this scheme in the future.

This library, suite_authorization_source.rb, provides a plugin for this scheme which allows you to implement this interface using a lightly-coupled ruby script.

The interface

The ruby interface this plugin expects is similar to the java SuiteAuthorizationSource it adapts. The design difference is that it uses ruby-native simple values for arguments and return values instead of the strongly-typed objects used in the java version. While the java objects would be available via JRuby, using simple native types should make it possible to test adapters using MRI and make it easier for developers without specific JRuby experience to write ruby authorization sources.

The user hash

Most of the native types used in the ruby interface are described below alongside the methods. One structure is shared among all four methods: the hash representing a single user. An example user hash:

  :username => 'superuser',
  :id => 1,
  :first_name => 'Sue',
  :last_name => 'User',
  :email_address => '',
  :roles => {
    :system_administrator => true,
    :user_administrator => { :sites => true }
  :account_end_date =>, 3, 9)

This hash describes a user with username "superuser" and two roles. Here's what the attributes mean.

Attribute Mandatory Description
:username Yes The desired username for the user in the suite apps.
:id Yes A stable numeric ID for this user. This ID will be used to associate domain information with the user in the CTMS applications, so it should never change.
:first_name and :last_name Yes The first and last name for the user.
:email_address Yes A working e-mail address for the user.
:account_end_date No The date after which the user should no longer have access to the system. If the current date is later than this date, any authorization attempts for the user will fail. However, the user will still show up in user lists in the CTMS Suite apps (which may be desirable).
:roles Yes The suite roles this user should have. See the next section for more detail.

All mandatory attributes must be present, non-nil, and not blank.


In the suite

The caBIG CTMS Suite shares a set of 23 roles across all applications. Roles generally have non-overlapping capabilities and each user may have several roles assigned.

Some roles are scopeable by site or by site and study. Such roles may be associated with one or more specific sites (and studies, as applicable), but may also be scoped to the concept "all sites" (or "all studies"). The suite documentation contains a list of all the roles, the scopes that apply to each, and brief descriptions of their capabilities in each suite application.

In the user hash

The :roles key in the user mapping result must point to a hash whose keys are the names of the roles of which the user is a member. The name must be a symbol and must be a downcased and underscored version of one of the role names from the suite docs.

Each one of these role name keys may point to one of a couple of values:

  • true, meaning that the user has the role to the maximum extent possible. If the role is scoped, true here means that the user has all sites and, if applicable, all studies scope.
  • A hash providing scoping information. This should be of the form { :sites => %w(IL034 MN070), :studies => true }. More technically, the two scopes are specified with the symbol keys :sites and :studies. The values may be either:
    • An array of strings indicating the assigned identifiers for the related domain objects under which the user's role membership is scoped, or
    • true, indicating that the user has the "all" scope for that scope type.

As noted above, each role has one of three possible required scopes: site & study, site, and unscoped. If you do not provide scope information for each type of scope that applies to the role, the role won't take effect. (The plugin will detect and log this situation.)

The interface (for real this time)

As noted above, the interface your script must provide is very similar to the SuiteAuthorizationSource java interface natively supported by PSC. It has four methods:

get_user_by_username(username, role_detail_level)

Corresponding java method: getUser(String, SuiteUserRoleLevel).

Must return the user hash for the single user with exactly the given username, or nil if there is no such user.

role_detail_level indicates how much detail about the user's roles will be read from the returned user. It may be :none, :roles, or :roles_and_scopes. Your implementation may choose to only include the level of detail indicated if that's desirable. (E.g., if determining role scope is particularly expensive, you could exclude it unless the level is :roles_and_scopes.) Unless you have a specific need to optimize, it is a good idea to ignore this parameter and return all role detail at all times (since this will make your code simpler).

get_user_by_id(id, role_detail_level)

Corresponding java method: getUser(long, SuiteUserRoleLevel).

Must return the user hash for the single user with the given numeric ID, or nil if there is no such user. The numeric ID must be in the range of a 32-bit signed integer.

See get_user_by_username for a description of role_detail_level.


Corresponding java method: getUsersByRole(SuiteRole).

Must return an array of user hashes describing the users with the given named role, regardless of scope. role_name will be a symbol following the same construction rules as the keys in the user hash's role hash (see above).

If there are no such users, it may return nil or an empty array.

The returned user hashes may omit role data if desired. (For example, if not computing it will improve performance.)


Corresponding java method: searchUsers(SuiteUserSearchOptions).

Must return an array of user hashes describing the users matching the given criteria. criteria will be a hash with zero or more of the following keys:

  • :username_substring: a case-insensitive substring of the username.
  • :first_name_substring: a case-insensitive substring of the first name.
  • :last_name_substring: definition left as an exercise for the reader.

If criteria includes multiple keys, the conditions should be ORed together.

If criteria is an empty hash, this method must return all the available users.

The returned user hashes may omit role data if desired. (For example, if not computing it will improve performance.)

Configure the plugin

This plugin requires that you specify (via a configuration property) the filename of a ruby script which implements the interface described above. The result of evaling this script must be that the global variable $suite_authorization_source refers to an object that responds to the four specified methods in the way described. All methods on the object returned must be re-entrant.

The script may refer to other files using the usual ruby methods require and load. It may refer to rubygems which have been packaged into JARs and which JARs are required in the script using their absolute paths. buildr with buildr-gemjar is an easy way to package a set of gems with its dependencies into a single JAR.

Deploying in PSC

All versions of PSC that support pluggable authorization also support dynamically loading OSGi bundles and configuration from a special directory structure.

The way this library is structured, you will want to deploy the bundles (i.e., the JARs) before creating the configuration files.


You'll need to deploy the JARs produced and used by this project, dividing them up like so:

  • libraries
  • plugins
    • ctms-auth-ruby-source-{version}.jar (get it from the downloads page)

Note that any gemjars that your authorization source depends on should not be deployed in either of these directories. See the next section.


In PSC's bundle configurations directory, you'll need to put at least a file named ctmssuite.authorization.ruby-{any_name}.cfg. This file should contain a single line of the following form:


You may also put your authorization source script and any of its gemjar or other dependencies in the configurations directory. PSC ignores any file in that directory that does not end with .cfg.

Project information


suite_authorization_source.rb is copyright 2012 Rhett Sutphin and was built at and for NUBIC. It is made available under the MIT license; see LICENSE alongside this file for details.

Something went wrong with that request. Please try again.