Path: | lib/active_ldap.rb |
Last Update: | Fri Jun 20 09:30:02 +0000 2008 |
"Ruby/ActiveLdap" Copyright (C) 2004,2005 Will Drewry will@alum.bu.edu
Ruby/ActiveLdap is a novel way of interacting with LDAP. Most interaction with LDAP is done using clunky LDIFs, web interfaces, or with painful APIs that required a thick reference manual nearby. Ruby/ActiveLdap aims to fix that. Inspired by ActiveRecord, Ruby/ActiveLdap provides an object oriented interface to LDAP entries.
The target audience is system administrators and LDAP users everywhere that need quick, clean access to LDAP in Ruby.
LDAP stands for "Lightweight Directory Access Protocol." Basically this means that it is the protocol used for accessing LDAP servers. LDAP servers lightweight directories. An LDAP server can contain anything from a simple digital phonebook to user accounts for computer systems. More and more frequently, it is being used for the latter. My examples in this text will assume some familiarity with using LDAP as a centralized authentication and authorization server for Unix systems. (Unfortunately, I‘ve yet to try this against Microsoft‘s ActiveDirectory, despite what the name implies.)
Further reading:
Well if you like to fumble around in the dark, dank innards of LDAP, you can quit reading now. However, if you‘d like a cleaner way to integrate LDAP in to your existing code, hopefully that‘s why you‘ll want to use Ruby/ActiveLdap.
Using LDAP directly (even with the excellent Ruby/LDAP), leaves you bound to the world of the predefined LDAP API. While this API is important for many reasons, having to extract code out of LDAP search blocks and create huge arrays of LDAP.mod entries make code harder to read, less intuitive, and just less fun to write. Hopefully, Ruby/ActiveLdap will remedy all of these problems!
Ruby/ActiveLdap does have some overhead when you get started. You must not only install the package and all of it‘s requirements, but you must also make customizations that will let it work in your environment.
Assuming all the requirements are installed, you can install by grabbing the latest tgz file from the download site.
The following steps will get the Ruby/ActiveLdap installed in no time!
$ tar -xzvf ruby-activeldap-current.tgz $ cd ruby-activeldap-VERSION
Edit lib/active_ldap/configuration.rb replacing values to match what will work with your LDAP servers. Please note that those variables are required, but can be overridden in any program as detailed later in this document. Also make sure that "ROOT" stays all upcase.
Now run:
$ (as root) rake install
Now as a quick test, you can run:
$ irb irb> require 'active_ldap' => true irb> exit
If the require returns false or an exception is raised, there has been a problem with the installation. You may need to customize what setup.rb does on install.
Now that Ruby/ActiveLdap is installed and working, we still have a few more steps to make it useful for programming.
Let‘s say that you are writing a Ruby program for managing user and group accounts in LDAP. I will use this as the running example throughout the document.
You will want to make a directory called ‘ldapadmin’ wherever is convenient. Under this directory, you‘ll want to make sure you have a ‘lib’ directory.
$ cd ~ $ mkdir ldapadmin $ cd ldapadmin $ mkdir lib $ cd lib
The lib directory is where we‘ll be making customizations. You can, of course, make this changes somewhere in Ruby‘s default search path to make this accessible to every Ruby scripts. Enough of my babbling, I‘m sure you‘d like to know what we‘re going to put in lib/.
We‘re going to put extension classes in there. What are extension classes you say . . .
This section covers using Ruby/ActiveLdap from writing extension classes to writing applications that use them.
Just to give a taste of what‘s to come, here is a quick example using irb:
irb> require 'active_ldap'
Here‘s an extension class that maps to the LDAP Group objects:
irb> class Group < ActiveLdap::Base irb* ldap_mapping irb* end
Here is the Group class in use:
irb> all_groups = Group.find(:all, '*').collect {|group| group.cn} => ["root", "daemon", "bin", "sys", "adm", "tty", ..., "develop"] irb> group = Group.find("develop") => #<Group:0x..........> irb> group.members.collect {|member| member.uid} => ["drewry"] irb> group.cn => "develop" irb> group.gid_number => "1003"
That‘s it! No let‘s get back in to it.
Extension classes are classes that are subclassed from ActiveLdap::Base. They are used to represent objects in your LDAP server abstractly.
Extension classes are what make Ruby/ActiveLdap "active"! They do all the background work to make easy-to-use objects by mapping the LDAP object‘s attributes on to a Ruby class.
I will briefly talk about each of the methods you can use when defining an extension class. In the above example, I only made one special method call inside the Group class. More than likely, you will want to more than that.
ldap_mapping is the only required method to setup an extension class for use with Ruby/ActiveLdap. It must be called inside of a subclass as shown above.
Below is a much more realistic Group class:
class Group < ActiveLdap::Base ldap_mapping :dn_attribute => 'cn', :prefix => 'ou=Groups', :classes => ['top', 'posixGroup'] :scope => :one end
As you can see, this method is used for defining how this class maps in to LDAP. Let‘s say that my LDAP tree looks something like this:
* dc=dataspill,dc=org |- ou=People,dc=dataspill,dc=org |+ ou=Groups,dc=dataspill,dc=org # |- cn=develop,ou=Groups,dc=dataspill,dc=org |- cn=root,ou=Groups,dc=dataspill,dc=org |- ...
Under ou=People I store user objects, and under ou=Groups, I store group objects. What |ldap_mapping| has done is mapped the class in to the LDAP tree abstractly. With the given :dn_attributes and :prefix, it will only work for entries under ou=Groups,dc=dataspill,dc=org using the primary attribute ‘cn’ as the beginning of the distinguished name.
Just for clarity, here‘s how the arguments map out:
cn=develop,ou=Groups,dc=dataspill,dc=org ^^ ^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^ :dn_attribute | | :prefix | :base from configuration.rb
:scope tells ActiveLdap to only search under ou=Groups, and not to look deeper for dn_attribute matches. (e.g. cn=develop,ou=DevGroups,ou=Groups,dc=dataspill,dc=org)
Something‘s missing: :classes. :classes is used to tell Ruby/ActiveLdap what the minimum requirement is when creating a new object. LDAP uses objectClasses to define what attributes a LDAP object may have. Ruby/ActiveLdap needs to know what classes are required when creating a new object. Of course, you can leave that field out to default to [‘top’] only. Then you can let each application choose what objectClasses their objects should have by calling the method e.g. Group#add_class(*values).
Note that is can be very important to define the default :classes value. Due to implementation choices with most LDAP servers, once an object is created, its structural objectclasses may not be removed (or replaced). Setting a sane default may help avoid programmer error later.
:classes isn‘t the only optional argument. If :dn_attribute is left off, it defaults to underscored class name or ‘cn’. If :prefix is left off, it will default to ‘ou=PLURALIZED_CLASSNAME’. In this case, it would be ‘ou=Groups’.
:classes should be an Array. :dn_attribute should be a String and so should :prefix.
This method allows an extension class to make use of other extension classes tying objects together across the LDAP tree. Often, user objects will be members of, or belong_to, Group objects.
* dc=dataspill,dc=org |+ ou=People,dc=dataspill,dc=org # |- uid=drewry,ou=People,dc=dataspill,dc=org |- ou=Groups,dc=dataspill,dc=org
In the above tree, one such example would be user ‘drewry’ who is a part of the group ‘develop’. You can see this by looking at the ‘memberUid’ field of ‘develop’.
irb> develop = Group.find('develop') => ... irb> develop.memberUid => ['drewry', 'builder']
If we look at the LDAP entry for ‘drewry’, we do not see any references to group ‘develop’. In order to remedy that, we can use belongs_to
irb> class User < ActiveLdap::Base irb* ldap_mapping :dn_attribute => 'uid', :prefix => 'People', :classes => ['top','account'] irb* belongs_to :groups, :class => 'Group', :many => 'memberUid', :foreign_key => 'uid' irb* end
Now, class User will have a method called ‘groups’ which will retrieve all Group objects that a user is in.
irb> me = User.find('drewry') irb> me.groups => [#<Group:0x000001 ...>, #<Group:0x000002 ...>, ...] irb> me.groups.each { |group| p group.cn };nil "cdrom" "audio" "develop" => nil (Note: nil is just there to make the output cleaner...)
TIP: If you weren‘t sure what the distinguished name attribute was for Group, you could also do the following:
irb> me.groups.each { |group| p group.id };nil "cdrom" "audio" "develop" => nil
Now let‘s talk about the arguments. The first argument is the name of the method you wish to create. In this case, we created a method called groups using the symbol :groups. The next collection of arguments are actually a Hash (as with ldap_mapping). :class should be a string that has the name of a class you‘ve already included. If you class is inside of a module, be sure to put the whole name, e.g. :class => "MyLdapModule::Group". :primary_key tells belongs_to what attribute Group objects have that match the :many. :many is the name of the local attribute whose value should be looked up in Group under the primary key. If :foreign_key is left off of the argument list, it is assumed to be the dn_attribute. With this in mind, the above definition could become:
irb> class User < ActiveLdap::Base irb* ldap_mapping :dn_attribute => 'uid', :prefix => 'People', :classes => ['top','account'] irb* belongs_to :groups, :class => 'Group', :many => 'memberUid' irb* end
In addition, you can do simple membership tests by doing the following:
irb> me.groups.member? 'root' => false irb> me.groups.member? 'develop' => true
This method is the opposite of belongs_to. Instead of checking other objects in other parts of the LDAP tree to see if you belong to them, you have multiple objects from other trees listed in your object. To show this, we can just invert the example from above:
class Group < ActiveLdap::Base ldap_mapping :dn_attribute => 'cn', :prefix => 'ou=Groups', :classes => ['top', 'posixGroup'] has_many :members, :class => "User", :wrap => "memberUid", :primary_key => 'uid' end
Now we can see that group develop has user ‘drewry’ as a member, and it can even return all responses in object form just like belongs_to methods.
irb> develop = Group.find('develop') => ... irb> develop.members => [#<User:0x000001 ...>, #<User:...>]
The arguments for has_many follow the exact same idea that belongs_to‘s arguments followed. :wrap‘s contents are used to search for matching :primary_key content. If :primary_key is not specified, it defaults to the dn_attribute of the specified :class.
These new classes have many method calls. Many of them are automatically generated to provide access to the LDAP object‘s attributes. Other were defined during class creation by special methods like belongs_to. There are a few other methods that do not fall in to these categories.
.find is a class method that is accessible from any subclass of Base that has ‘ldap_mapping’ called. When called it returns the first match of the given class.
irb> Group.find('*').cn => "root"
In this simple example, Group.find took the search string of ‘deve*’ and searched for the first match in Group where the dn_attribute matched the query. This is the simplest example of .find.
irb> Group.find(:all, '*').collect {|group| group.cn} => ["root", "daemon", "bin", "sys", "adm", "tty", ..., "develop"]
Here .find(:all) returns all matches to the same query. Both .find and .find(:all) also can take more expressive arguments:
irb> Group.find(:all, :attribute => 'gidNumber', :value => '1003').collect {|group| group.cn} => ["develop"]
So it is pretty clear what :attribute and :value do - they are used to query as :attribute=:value.
If :attribute is unspecified, it defaults to the dn_attribute.
It is also possible to override :attribute and :value by specifying :filter. This argument allows the direct specification of a LDAP filter to retrieve objects by.
.search is a class method that is accessible from any subclass of Base, and Base. It lets the user perform an arbitrary search against the current LDAP connection irrespetive of LDAP mapping data. This is meant to be useful as a utility method to cover 80% of the cases where a user would want to use Base.connection directly.
irb> Base.search(:base => 'dc=example,dc=com', :filter => '(uid=roo*)', :scope => :sub, :attributes => ['uid', 'cn']) => [["uid=root,ou=People,dc=dataspill,dc=org",{"cn"=>["root"], "uidNumber"=>["0"]}]
You can specify the :filter, :base, :scope, and :attributes, but they all have defaults —
* :filter defaults to objectClass=* - usually this isn't what you want * :base defaults to the base of the class this is executed from (as set in ldap_mapping) * :scope defaults to :sub. Usually you won't need to change it * :attributes defaults to [] and is the list of attributes you want back. Empty means all of them.
valid? is a method that verifies that all attributes that are required by the objects current objectClasses are populated.
save is a method that writes any changes to an object back to the LDAP server. It automatically handles the addition of new objects, and the modification of existing ones.
exists? is a simple method which returns true is the current object exists in LDAP, or false if it does not.
irb> User.exists?("dshadsadsa") => false
ActiveLdap::Base has come up a number of times in the examples above. Every time, it was being used as the super class for the wrapper objects. While this is it‘s main purpose, it also handles quite a bit more in the background.
ActiveLdap::Base is the heart of Ruby/ActiveLdap. It does all the schema parsing for validation and attribute-to-method mangling as well as manage the connection to LDAP.
Base.establish_connection takes many (optional) arguments and is used to connect to the LDAP server. Sometimes you will want to connect anonymously and other times over TLS with user credentials. Base.establish_connection is here to do all of that for you.
By default, if you call any subclass of Base, such as Group, it will call Base.establish_connection() if these is no active LDAP connection. If your server allows anonymous binding, and you only want to access data in a read-only fashion, you won‘t need to call Base.establish_connection. Here is a fully parameterized call:
Base.establish_connection( :host => 'ldap.dataspill.org', :port => 389, :base => 'dc=dataspill,dc=org', :logger => logger_object, :bind_dn => "uid=drewry,ou=People,dc=dataspill,dc=org", :password_block => Proc.new { 'password12345' }, :allow_anonymous => false, :try_sasl => false )
There are quite a few arguments, but luckily many of them have safe defaults:
Most of these are obvious, but I‘ll step through them for completeness:
See lib/configuration.rb for defaults for each option
Base.establish_connection both connects and binds in one step. It follows roughly the following approach:
On connect, the configuration options passed in are stored in an internal class variable @configuration which is used to cache the information without ditching the defaults passed in from configuration.rb
Base.connection returns the ActiveLdap::Connection object.
There are a few custom exceptions used in Ruby/ActiveLdap. They are detailed below.
This exception is raised when delete fails. It will include LDAP error information that was passed up during the error.
This exception is raised when there is a problem in save updating or creating an LDAP entry. Often the error messages are cryptic. Looking at the server logs or doing an Ethereal dump of the connection will often provide better insight.
This exception is raised during Base.establish_connection if no valid authentication methods succeeded.
This exception is raised during Base.establish_connection if no valid connection to the LDAP server could be created. Check you configuration.rb, Base.establish_connection arguments, and network connectivity! Also check your LDAP server logs to see if it ever saw the request.
This exception is raised when an object class is used that is not defined in the schema.
Other exceptions may be raised by the Ruby/LDAP module, or by other subsystems. If you get one of these exceptions and think it should be wrapped, write me an email and let me know where it is and what you expected. For faster results, email a patch!
Now that all of the components of Ruby/ActiveLdap have been covered, it‘s time to put it all together! The rest of this section will show the steps to setup example user and group management scripts for use with the LDAP tree described above.
All of the scripts here are in the package‘s examples/ directory.
In ldapadmin/lib/ create the file user.rb:
cat <<EOF class User < ActiveLdap::Base ldap_mapping :dn_attribute => 'uid', :prefix => 'ou=People', :classes => ['top', 'account', 'posixAccount'] belongs_to :groups, :class => 'Group', :wrap => 'memberUid' end EOF
In ldapadmin/lib/ create the file group.rb:
cat <<EOF class Group < ActiveLdap::Base ldap_mapping :classes => ['top', 'posixGroup'], :prefix => 'ou=Group' has_many :members, :class => "User", :many => "memberUid" has_many :primary_members, :class => 'User', :foreign_key => 'gidNumber', :primary_key => 'gidNumber' end # Group EOF
Now, we can write some small scripts to do simple management tasks.
Now let‘s create a really dumb script for adding users - ldapadmin/useradd:
#!/usr/bin/ruby -W0 require 'active_ldap' require 'lib/user' require 'lib/group' require 'password' argv, opts, options = ActiveLdap::Command.parse_options do |opts, options| opts.banner += " USER_NAME CN UID" end if argv.size == 3 name, cn, uid = argv else $stderr.puts opts exit 1 end pwb = Proc.new do |user| ActiveLdap::Command.read_password("[#{user}] Password: ") end ActiveLdap::Base.establish_connection(:password_block => pwb, :allow_anonymous => false) if User.exists?(name) $stderr.puts("User #{name} already exists.") exit 1 end user = User.new(name) user.add_class('shadowAccount') user.cn = cn user.uid_number = uid user.gid_number = uid user.home_directory = "/home/#{name}" user.sn = "somesn" unless user.save puts "failed" puts user.errors.full_messages exit 1 end
Now let‘s create another dumb script for modifying users - ldapadmin/usermod:
#!/usr/bin/ruby -W0 require 'active_ldap' require 'lib/user' require 'lib/group' argv, opts, options = ActiveLdap::Command.parse_options do |opts, options| opts.banner += " USER_NAME CN UID" end if argv.size == 3 name, cn, uid = argv else $stderr.puts opts exit 1 end pwb = Proc.new do |user| ActiveLdap::Command.read_password("[#{user}] Password: ") end ActiveLdap::Base.establish_connection(:password_block => pwb, :allow_anonymous => false) unless User.exists?(name) $stderr.puts("User #{name} doesn't exist.") exit 1 end user = User.find(name) user.cn = cn user.uid_number = uid user.gid_number = uid unless user.save puts "failed" puts user.errors.full_messages exit 1 end
And finally, a dumb script for removing user - ldapadmin/userdel:
#!/usr/bin/ruby -W0 require 'active_ldap' require 'lib/user' require 'lib/group' argv, opts, options = ActiveLdap::Command.parse_options do |opts, options| opts.banner += " USER_NAME" end if argv.size == 1 name = argv.shift else $stderr.puts opts exit 1 end pwb = Proc.new do |user| ActiveLdap::Command.read_password("[#{user}] Password: ") end ActiveLdap::Base.establish_connection(:password_block => pwb, :allow_anonymous => false) unless User.exists?(name) $stderr.puts("User #{name} doesn't exist.") exit 1 end User.destroy(name)
Below are some situation tips and tricks to get the most out of Ruby/ActiveLdap.
Sometimes, you may want to store attributes with language specifiers, or perhaps in binary form. This is (finally!) fully supported. To do so, follow the examples below:
irb> user = User.new('drewry') => ... # This adds a cn entry in lang-en and whatever the server default is. irb> user.cn = [ 'wad', {'lang-en' => ['wad', 'foo']} ] => ... irb> user.cn => ["wad", {"lang-en-us" => ["wad", "Will Drewry"]}] # Now let's add a binary X.509 certificate (assume objectClass is correct) irb> user.user_certificate = File.read('example.der') => ... irb> user.save
So that‘s a lot to take in. Here‘s what is going on. I just set the LDAP object‘s cn to "wad" and cn:lang-en-us to ["wad", "Will Drewry"]. Anytime a LDAP subtype is required, you must encapsulate the data in a Hash.
But wait a minute, I just read in a binary certificate without wrapping it up. So any binary attribute _that requires ;binary subtyping_ will automagically get wrapped in {‘binary’ => value} if you don‘t do it. This keeps your writes from breaking, and my code from crying. For correctness, I could have easily done the following:
irb> user.user_certificate = {'binary' => File.read('example.der')}
You should note that some binary data does not use the binary subtype all the time. One example is jpegPhoto. You can use it as jpegPhoto;binary or just as jpegPhoto. Since the schema dictates that it is a binary value, Ruby/ActiveLdap will write it as binary, but the subtype will not be automatically appended as above. The use of the subtype on attributes like jpegPhoto is ultimately decided by the LDAP site policy and not by any programmatic means.
The only subtypes defined in LDAPv3 are lang-* and binary. These can be nested though:
irb> user.cn = [{'lang-JP-jp' => {'binary' => 'somejp'}}]
As I understand it, OpenLDAP does not support nested subtypes, but some documentation I‘ve read suggests that Netscape‘s LDAP server does. I only have access to OpenLDAP. If anyone tests this out, please let me know how it goes!
And that pretty much wraps up this section.
If you want this to cleanly integrate into your system-wide Ruby include path, you should put your extension classes inside a custom module.
Example:
./myldap.rb: require 'active_ldap' require 'myldap/user' require 'myldap/group' module MyLDAP end ./myldap/user.rb: module MyLDAP class User < ActiveLdap::Base ldap_mapping :dn_attribute => 'uid', :prefix => 'ou=People', :classes => ['top', 'account', 'posixAccount'] belongs_to :groups, :class => 'MyLDAP::Group', :many => 'memberUid' end end ./myldap/group.rb: module MyLDAP class Group < ActiveLdap::Base ldap_mapping :classes => ['top', 'posixGroup'], :prefix => 'ou=Group' has_many :members, :class => 'MyLDAP::User', :wrap => 'memberUid' has_many :primary_members, :class => 'MyLDAP::User', :foreign_key => 'gidNumber', :primary_key => 'gidNumber' end end
Now in your local applications, you can call
require 'myldap' MyLDAP::Group.new('foo') ...
and everything should work well.
Even though Ruby/ActiveLdap attempts to maintain programmatic ease by returning Array values only. By specifying ‘true’ as an argument to any attribute method you will get back a Array if it is single value. Here‘s an example:
irb> user = User.new('drewry') => ... irb> user.cn(true) => ["Will Drewry"]
If you use tab completion in irb, you‘ll notice that you /can/ tab complete the dynamic attribute methods. You can still see which methods are for attributes using Base#attribute_names:
irb> d = Group.new('develop') => ... irb> d.attribute_names => ["gidNumber", "cn", "memberUid", "commonName", "description", "userPassword", "objectClass"]
In the same vein as the last tip, you can use multiple LDAP connections by per class as follows:
irb> anon_class = Class.new(Base) => ... irb> anon_class.establish_connection => ... irb> auth_class = Class.new(Base) => ... irb> auth_class.establish_connection(:password_block => {'mypass'}) => ...
This can be useful for doing authentication tests and other such tricks.
If you have the Ruby/LDAP package with the SASL/GSSAPI patch from Ian MacDonald‘s web site, you can use Kerberos to bind to your LDAP server. By default, :try_sasl is false.
Also note that you must be using OpenLDAP 2.1.29 or higher to use SASL/GSSAPI due to some bugs in older versions of OpenLDAP.
Don‘t be afraid to add more methods to the extensions classes and to experiment. That‘s exactly how I ended up with this package. If you come up with something cool, please share it!
The internal structure of ActiveLdap::Base, and thus all its subclasses, is still in flux. I‘ve tried to minimize the changes to the overall API, but the internals are still rough around the edges.
When you call ldap_mapping, it overwrites several class methods inherited from Base:
* Base.base() * Base.required_classes() * Base.dn_attribute()
You can access these from custom class methods by calling MyClass.base(), or whatever. There are predefined instance methods for getting to these from any new instance methods you define:
* Base#base() * Base#required_classes() * Base#dn_attribute()
Well if you want to use the LDAP connection for anything, I‘d suggest still calling Base.connection to get it. There really aren‘t many other internals that need to be worried about. You could get the LDAP schema with Base.schema.
The only other useful tricks are dereferencing and accessing the stored data. Since LDAP attributes can have multiple names, e.g. cn or commonName, any methods you write might need to figure it out. I‘d suggest just calling self[attribname] to get the value, but if that‘s not good enough, you can call look up the stored name by to_real_attribute_name as follows:
irb> to_real_attribute_name('commonName') => 'cn'
This tells you the name the attribute is stored in behind the scenes (@data). Again, self[attribname] should be enough for most extensions, but if not, it‘s probably safe to dabble here.
Also, if you like to look up all aliases for an attribute, you can call the following:
irb> schema.attribute_aliases('cn') => ['cn','commonName']
This is discovered automagically from the LDAP server‘s schema.
Currently, Ruby/ActiveLdap could be faster. I have some recursive type checking going on which slows object creation down, and I‘m sure there are many, many other places optimizations can be done. Feel free to send patches, or just hang in there until I can optimize away the slowness.
Any and all feedback and patches are welcome. I am very excited about this package, and I‘d like to see it prove helpful to more people than just myself.
DEB_RAILS_PATH | = | "/usr/share/rails" | add Rails load path for Debian, until rails framework is split properly | |
ACTIVE_LDAP_CONNECTION_ADAPTERS | = | %w(ldap net_ldap jndi) |