node.ext.ldap

Overview

node.ext.ldap is a LDAP convenience library for LDAP communication based on python-ldap (version 2.4 or later) and node.

The package contains base configuration and communication objects, a LDAP node object and a LDAP node based user and group management implementation utilizing node.ext.ugm.

This package is the successor of bda.ldap.

Contents

• node.ext.ldap
• Overview
• API changes compared to 0.9.x
• Usage
  • LDAP Properties
  • LDAP Connection
  • LDAP Communication
  • LDAP Session
  • LDAP Nodes
  • Searching LDAP
  • JSON Serialization
  • User and Group management
  • Character Encoding
  • Caching Support
  • Dependencies
• Contributors

API changes compared to 0.9.x

• LDAPNode instances cannot have direct children of subtree any longer. This was a design flaw because of possible duplicate RDN's.
• LDAPNode.search returns DN's instead of RDN's by default.
• Secondary keys and alternative key attribute features have been removed entirely from LDAPNode.
• LDAPProps.check_duplicates setting has been removed.

Usage

LDAP Properties

To define connection properties for LDAP use node.ext.ldap.LDAPProps object:

>>> from node.ext.ldap import LDAPProps

>>> props = LDAPProps(
...     uri='ldap://localhost:12345/',
...     user='cn=Manager,dc=my-domain,dc=com',
...     password='secret',
...     cache=False
... )

Test server connectivity with node.ext.ldap.testLDAPConnectivity:

>>> from node.ext.ldap import testLDAPConnectivity

>>> assert testLDAPConnectivity(props=props) == 'success'

LDAP Connection

For handling LDAP connections, node.ext.ldap.LDAPConnector is used. It expects a LDAPProps instance in the constructor. Normally there is no need to instantiate this object directly, this happens during creation of higher abstractions, see below:

>>> from node.ext.ldap import LDAPConnector
>>> import ldap

>>> connector = LDAPConnector(props=props)

Calling bind creates and returns the LDAP connection:

>>> conn = connector.bind()
>>> assert isinstance(conn, ldap.ldapobject.ReconnectLDAPObject)

Calling unbind destroys the connection:

>>> connector.unbind()

LDAP Communication

For communicating with an LDAP server, node.ext.ldap.LDAPCommunicator is used. It provides all the basic functions needed to search and modify the directory.

LDAPCommunicator expects a LDAPConnector instance at creation time:

>>> from node.ext.ldap import LDAPCommunicator

>>> communicator = LDAPCommunicator(connector)

Bind to server:

>>> communicator.bind()

Adding directory entry:

>>> communicator.add(
...     'cn=foo,ou=demo,dc=my-domain,dc=com',
...     {
...         'cn': 'foo',
...         'sn': 'Mustermann',
...         'userPassword': 'secret',
...         'objectClass': ['person'],
...     }
... )

Set default search DN:

>>> communicator.baseDN = 'ou=demo,dc=my-domain,dc=com'

Search in directory:

>>> import node.ext.ldap

>>> res = communicator.search(
...     '(objectClass=person)',
...     node.ext.ldap.SUBTREE
... )

>>> assert res == [(
...     'cn=foo,ou=demo,dc=my-domain,dc=com',
...     {
...         'objectClass': ['person'],
...         'userPassword': ['secret'],
...         'cn': ['foo'],
...         'sn': ['Mustermann']
...     }
... )]

Modify directory entry:

>>> from ldap import MOD_REPLACE

>>> communicator.modify(
...     'cn=foo,ou=demo,dc=my-domain,dc=com',
...     [(MOD_REPLACE, 'sn', 'Musterfrau')]
... )

>>> res = communicator.search(
...     '(objectClass=person)',
...     node.ext.ldap.SUBTREE,
...     attrlist=['cn']
... )

>>> assert res == [('cn=foo,ou=demo,dc=my-domain,dc=com', {'cn': ['foo']})]

Change the password of a directory entry which represents a user:

>>> communicator.passwd(
...     'cn=foo,ou=demo,dc=my-domain,dc=com',
...     'secret',
...     '12345'
... )

>>> res = communicator.search(
...     '(objectClass=person)',
...     node.ext.ldap.SUBTREE,
...     attrlist=['userPassword']
... )

>>> assert res == [(
...     'cn=foo,ou=demo,dc=my-domain,dc=com',
...     {'userPassword': ['{SSHA}...']}
... )]

Delete directory entry:

>>> communicator.delete('cn=foo,ou=demo,dc=my-domain,dc=com')

>>> res = communicator.search(
...     '(objectClass=person)',
...     node.ext.ldap.SUBTREE
... )

>>> assert res == []

Close connection:

>>> communicator.unbind()

LDAP Session

A more convenient way for dealing with LDAP is provided by node.ext.ldap.LDAPSession. It basically provides the same functionality as LDAPCommunicator, but automatically creates the connectivity objects and checks the connection state before performing actions.

Instantiate LDAPSession object. Expects LDAPProps instance:

>>> from node.ext.ldap import LDAPSession

>>> session = LDAPSession(props)

LDAP session has a convenience to check given properties:

>>> res = session.checkServerProperties()

>>> assert res == (True, 'OK')

Set default search DN for session:

>>> session.baseDN = 'ou=demo,dc=my-domain,dc=com'

Search in directory:

>>> res = session.search()

>>> assert res == [
...     ('ou=demo,dc=my-domain,dc=com',
...     {
...         'objectClass': ['top', 'organizationalUnit'],
...         'ou': ['demo'],
...         'description': ['Demo organizational unit']
...     }
... )]

Add directory entry:

>>> session.add(
...     'cn=foo,ou=demo,dc=my-domain,dc=com',
...     {
...         'cn': 'foo',
...         'sn': 'Mustermann',
...         'userPassword': 'secret',
...         'objectClass': ['person'],
...     }
... )

Change the password of a directory entry which represents a user:

>>> session.passwd('cn=foo,ou=demo,dc=my-domain,dc=com', 'secret', '12345')

Authenticate a specific user:

>>> res = session.authenticate('cn=foo,ou=demo,dc=my-domain,dc=com', '12345')

>>> assert res is True

Modify directory entry:

>>> session.modify(
...     'cn=foo,ou=demo,dc=my-domain,dc=com',
...     [(MOD_REPLACE, 'sn', 'Musterfrau')]
... )

>>> res = session.search(
...     '(objectClass=person)',
...     node.ext.ldap.SUBTREE,
...     attrlist=['cn']
... )

>>> assert res == [(
...     'cn=foo,ou=demo,dc=my-domain,dc=com',
...     {'cn': ['foo']}
... )]

Delete directory entry:

>>> session.delete('cn=foo,ou=demo,dc=my-domain,dc=com')

>>> res = session.search('(objectClass=person)', node.ext.ldap.SUBTREE)

>>> assert res == []

Close session:

>>> session.unbind()

LDAP Nodes

One can deal with LDAP entries as node objects. Therefor node.ext.ldap.LDAPNode is used. To get a clue of the complete node API, see node package.

Create a LDAP node. The root Node expects the base DN and a LDAPProps instance:

>>> from node.ext.ldap import LDAPNode

>>> root = LDAPNode('ou=demo,dc=my-domain,dc=com', props=props)

Every LDAP node has a DN and a RDN:

>>> root.DN
u'ou=demo,dc=my-domain,dc=com'

>>> root.rdn_attr
u'ou'

Check whether created node exists in the database:

>>> root.exists
True

Directory entry has no children yet:

>>> root.keys()
[]

Add children to root node:

>>> person = LDAPNode()
>>> person.attrs['objectClass'] = ['person', 'inetOrgPerson']
>>> person.attrs['sn'] = 'Mustermann'
>>> person.attrs['userPassword'] = 'secret'
>>> root['cn=person1'] = person

>>> person = LDAPNode()
>>> person.attrs['objectClass'] = ['person', 'inetOrgPerson']
>>> person.attrs['sn'] = 'Musterfrau'
>>> person.attrs['userPassword'] = 'secret'
>>> root['cn=person2'] = person

If the RDN attribute was not set during node creation, it is computed from node key and set automatically:

>>> person.attrs['cn']
u'person2'

Fetch children DN by key from LDAP node:

>>> root.child_dn('cn=person1')
u'cn=person1,ou=demo,dc=my-domain,dc=com'

Have a look at the tree:

>>> root.printtree()
<ou=demo,dc=my-domain,dc=com - True>
  <cn=person2,ou=demo,dc=my-domain,dc=com:cn=person2 - True>
  <cn=person1,ou=demo,dc=my-domain,dc=com:cn=person1 - True>

The entries have not been written to the directory yet. When modifying a LDAP node tree, everything happens im memory. Persisting is done by calling the tree, or a part of it. You can check sync state of a node with its changed flag. If changed is True it means either that the node attributes or node children has changed:

>>> root.changed
True

>>> root()
>>> root.changed
False

Modify a LDAP node:

>>> person = root['cn=person1']

Modify existing attribute:

>>> person.attrs['sn'] = 'Mustermensch'

Add new attribute:

>>> person.attrs['description'] = 'Mustermensch description'
>>> person()

Delete an attribute:

>>> del person.attrs['description']
>>> person()

Delete LDAP node:

>>> del root['cn=person2']
>>> root()
>>> root.printtree()
<ou=demo,dc=my-domain,dc=com - False>
  <cn=person1,ou=demo,dc=my-domain,dc=com:cn=person1 - False>

Searching LDAP

Add some users and groups we'll search for:

>>> for i in range(2, 6):
...     node = LDAPNode()
...     node.attrs['objectClass'] = ['person', 'inetOrgPerson']
...     node.attrs['sn'] = 'Surname %s' % i
...     node.attrs['userPassword'] = 'secret%s' % i
...     node.attrs['description'] = 'description%s' % i
...     node.attrs['businessCategory'] = 'group1'
...     root['cn=person%s' % i] = node

>>> node = LDAPNode()
>>> node.attrs['objectClass'] = ['groupOfNames']
>>> node.attrs['member'] = [
...     root.child_dn('cn=person1'),
...     root.child_dn('cn=person2'),
... ]
... node.attrs['description'] = 'IT'
>>> root['cn=group1'] = node

>>> node = LDAPNode()
>>> node.attrs['objectClass'] = ['groupOfNames']
>>> node.attrs['member'] = [
...     root.child_dn('cn=person4'),
...     root.child_dn('cn=person5'),
... ]
>>> root['cn=group2'] = node

>>> root()
>>> root.printtree()
<ou=demo,dc=my-domain,dc=com - False>
  <cn=person1,ou=demo,dc=my-domain,dc=com:cn=person1 - False>
  <cn=person2,ou=demo,dc=my-domain,dc=com:cn=person2 - False>
  <cn=person3,ou=demo,dc=my-domain,dc=com:cn=person3 - False>
  <cn=person4,ou=demo,dc=my-domain,dc=com:cn=person4 - False>
  <cn=person5,ou=demo,dc=my-domain,dc=com:cn=person5 - False>
  <cn=group1,ou=demo,dc=my-domain,dc=com:cn=group1 - False>
  <cn=group2,ou=demo,dc=my-domain,dc=com:cn=group2 - False>

For defining search criteria LDAP filters are used, which can be combined by bool operators '&' and '|':

>>> from node.ext.ldap import LDAPFilter

>>> filter = LDAPFilter('(objectClass=person)')
>>> filter |= LDAPFilter('(objectClass=groupOfNames)')

>>> res = sorted(root.search(queryFilter=filter))

>>> assert res == [
...     u'cn=group1,ou=demo,dc=my-domain,dc=com',
...     u'cn=group2,ou=demo,dc=my-domain,dc=com',
...     u'cn=person1,ou=demo,dc=my-domain,dc=com',
...     u'cn=person2,ou=demo,dc=my-domain,dc=com',
...     u'cn=person3,ou=demo,dc=my-domain,dc=com',
...     u'cn=person4,ou=demo,dc=my-domain,dc=com',
...     u'cn=person5,ou=demo,dc=my-domain,dc=com'
... ]

Define multiple criteria LDAP filter:

>>> from node.ext.ldap import LDAPDictFilter

>>> filter = LDAPDictFilter({
...     'objectClass': ['person'],
...     'cn': 'person1'
... })

>>> res = root.search(queryFilter=filter)

>>> assert res == [u'cn=person1,ou=demo,dc=my-domain,dc=com']

Define a relation LDAP filter. In this case we build a relation between group 'cn' and person 'businessCategory':

>>> from node.ext.ldap import LDAPRelationFilter

>>> filter = LDAPRelationFilter(root['cn=group1'], 'cn:businessCategory')

>>> res = root.search(queryFilter=filter)

>>> assert res == [
...     u'cn=person2,ou=demo,dc=my-domain,dc=com',
...     u'cn=person3,ou=demo,dc=my-domain,dc=com',
...     u'cn=person4,ou=demo,dc=my-domain,dc=com',
...     u'cn=person5,ou=demo,dc=my-domain,dc=com'
... ]

Different LDAP filter types can be combined:

>>> filter &= LDAPFilter('(cn=person2)')
>>> str(filter)
'(&(businessCategory=group1)(cn=person2))'

The following keyword arguments are accepted by LDAPNode.search. If multiple keywords are used, combine search criteria with '&' where appropriate.

If attrlist is given, the result items consists of 2-tuples with a dict containing requested attributes at position 1:

queryFilter

Either a LDAP filter instance or a string. If given argument is string type, a LDAPFilter instance is created.

criteria

A dictionary containing search criteria. A LDAPDictFilter instance is created.

attrlist

List of attribute names to return. Special attributes rdn and dn are allowed.

relation

Either LDAPRelationFilter instance or a string defining the relation. If given argument is string type, a LDAPRelationFilter instance is created.

relation_node

In combination with relation argument, when given as string, use relation_node instead of self for filter creation.

exact_match

Flag whether 1-length result is expected. Raises an error if empty result or more than one entry found.

or_search

In combination with criteria, this parameter is passed to the creation of LDAPDictFilter. This flag controls whether to combine criteria keys and values with '&' or '|'.

or_keys

In combination with criteria, this parameter is passed to the creation of LDAPDictFilter. This flag controls whether criteria keys are combined with '|' instead of '&'.

or_values

In combination with criteria, this parameter is passed to the creation of LDAPDictFilter. This flag controls whether criteria values are combined with '|' instead of '&'.

page_size

Used in conjunction with cookie for querying paged results.

cookie

Used in conjunction with page_size for querying paged results.

get_nodes

If True result contains LDAPNode instances instead of DN's

You can define search defaults on the node which are always considered when calling search on this node. If set, they are always '&' combined with any (optional) passed filters.

Define the default search scope:

>>> from node.ext.ldap import SUBTREE

>>> root.search_scope = SUBTREE

Define default search filter, could be of type LDAPFilter, LDAPDictFilter, LDAPRelationFilter or string:

>>> root.search_filter = LDAPFilter('objectClass=groupOfNames')

>>> res = root.search()

>>> assert res == [
...     u'cn=group1,ou=demo,dc=my-domain,dc=com',
...     u'cn=group2,ou=demo,dc=my-domain,dc=com'
... ]

>>> root.search_filter = None

Define default search criteria as dict:

>>> root.search_criteria = {'objectClass': 'person'}

>>> res = root.search()

>>> assert res == [
...     u'cn=person1,ou=demo,dc=my-domain,dc=com',
...     u'cn=person2,ou=demo,dc=my-domain,dc=com',
...     u'cn=person3,ou=demo,dc=my-domain,dc=com',
...     u'cn=person4,ou=demo,dc=my-domain,dc=com',
...     u'cn=person5,ou=demo,dc=my-domain,dc=com'
... ]

Define default search relation:

>>> root.search_relation = LDAPRelationFilter(
...     root['cn=group1'],
...     'cn:businessCategory'
... )

>>> res = root.search()

>>> assert res == [
...     u'cn=person2,ou=demo,dc=my-domain,dc=com',
...     u'cn=person3,ou=demo,dc=my-domain,dc=com',
...     u'cn=person4,ou=demo,dc=my-domain,dc=com',
...     u'cn=person5,ou=demo,dc=my-domain,dc=com'
... ]

Again, like with the keyword arguments, multiple defined defaults are '&' combined:

# empty result, there are no groups with group 'cn' as 'description'
>>> root.search_criteria = {'objectClass': 'group'}

>>> res = root.search()

>>> assert res == []

JSON Serialization

Serialize and deserialize LDAP nodes:

>>> root = LDAPNode('ou=demo,dc=my-domain,dc=com', props=props)

Serialize children:

>>> from node.serializer import serialize

>>> json_dump = serialize(root.values())

Clear and persist root:

>>> root.clear()

>>> root()

Deserialize JSON dump:

>>> from node.serializer import deserialize

>>> deserialize(json_dump, root=root)
[<cn=person1,ou=demo,dc=my-domain,dc=com:cn=person1 - True>,
<cn=person2,ou=demo,dc=my-domain,dc=com:cn=person2 - True>,
<cn=person3,ou=demo,dc=my-domain,dc=com:cn=person3 - True>,
<cn=person4,ou=demo,dc=my-domain,dc=com:cn=person4 - True>,
<cn=person5,ou=demo,dc=my-domain,dc=com:cn=person5 - True>,
<cn=group1,ou=demo,dc=my-domain,dc=com:cn=group1 - True>,
<cn=group2,ou=demo,dc=my-domain,dc=com:cn=group2 - True>]

Since root has been given, created nodes were added:

>>> root()
>>> root.printtree()
<ou=demo,dc=my-domain,dc=com - False>
  <cn=person1,ou=demo,dc=my-domain,dc=com:cn=person1 - False>
  <cn=person2,ou=demo,dc=my-domain,dc=com:cn=person2 - False>
  <cn=person3,ou=demo,dc=my-domain,dc=com:cn=person3 - False>
  <cn=person4,ou=demo,dc=my-domain,dc=com:cn=person4 - False>
  <cn=person5,ou=demo,dc=my-domain,dc=com:cn=person5 - False>
  <cn=group1,ou=demo,dc=my-domain,dc=com:cn=group1 - False>
  <cn=group2,ou=demo,dc=my-domain,dc=com:cn=group2 - False>

Non simple vs simple mode. Create container with children:

>>> container = LDAPNode()
>>> container.attrs['objectClass'] = ['organizationalUnit']
>>> root['ou=container'] = container

>>> person = LDAPNode()
>>> person.attrs['objectClass'] = ['person', 'inetOrgPerson']
>>> person.attrs['sn'] = 'Mustermann'
>>> person.attrs['userPassword'] = 'secret'
>>> container['cn=person1'] = person

>>> root()

Serialize in default mode contains type specific information. Thus JSON dump can be deserialized later:

>>> serialized = serialize(container)

>>> assert serialized == (
...     '{'
...         '"__node__": {'
...             '"attrs": {'
...                 '"objectClass": ["organizationalUnit"], '
...                 '"ou": "container"'
...             '}, '
...             '"children": [{'
...                 '"__node__": {'
...                     '"attrs": {'
...                         '"objectClass": ["person", "inetOrgPerson"], '
...                         '"userPassword": "secret", '
...                         '"sn": "Mustermann", '
...                         '"cn": "person1"'
...                     '},'
...                     '"class": "node.ext.ldap._node.LDAPNode", '
...                     '"name": "cn=person1"'
...                 '}'
...             '}], '
...             '"class": "node.ext.ldap._node.LDAPNode", '
...             '"name": "ou=container"'
...         '}'
...     '}'
... )

Serialize in simple mode is better readable, but not deserialzable any more:

>>> serialized = serialize(container, simple_mode=True)

>>> assert serialized == (
...     '{'
...         '"attrs": {'
...             '"objectClass": ["organizationalUnit"], '
...             '"ou": "container"'
...         '}, '
...         '"name": "ou=container", '
...         '"children": [{'
...             '"name": "cn=person1", '
...             '"attrs": {'
...                 '"objectClass": ["person", "inetOrgPerson"], '
...                 '"userPassword": "secret", '
...                 '"sn": "Mustermann", '
...                 '"cn": "person1"'
...             '}'
...         '}]'
...     '}'
... )

User and Group management

LDAP is often used to manage Authentication, thus node.ext.ldap provides an API for User and Group management. The API follows the contract of node.ext.ugm:

>>> from node.ext.ldap import ONELEVEL
>>> from node.ext.ldap.ugm import UsersConfig
>>> from node.ext.ldap.ugm import GroupsConfig
>>> from node.ext.ldap.ugm import RolesConfig
>>> from node.ext.ldap.ugm import Ugm

Instantiate users, groups and roles configuration. They are based on PrincipalsConfig class and expect this settings:

baseDN

Principals container base DN.

attrmap

Principals Attribute map as odict.odict. This object must contain the mapping between reserved keys and the real LDAP attribute, as well as mappings to all accessible attributes for principal nodes if instantiated in strict mode, see below.

scope

Search scope for principals.

queryFilter

Search Query filter for principals

objectClasses

Object classes used for creation of new principals. For some objectClasses default value callbacks are registered, which are used to generate default values for mandatory attributes if not already set on principal vessel node.

defaults

Dict like object containing default values for principal creation. A value could either be static or a callable accepting the principals node and the new principal id as arguments. This defaults take precedence to defaults detected via set object classes.

strict

Define whether all available principal attributes must be declared in attmap, or only reserved ones. Defaults to True.

memberOfSupport

Flag whether to use 'memberOf' attribute (AD) or memberOf overlay (openldap) for Group membership resolution where appropriate.

Reserved attrmap keys for Users, Groups and roles:

id

The attribute containing the user id (mandatory).

rdn

The attribute representing the RDN of the node (mandatory) XXX: get rid of, should be detected automatically

Reserved attrmap keys for Users:

login

Alternative login name attribute (optional)

Create config objects:

>>> ucfg = UsersConfig(
...     baseDN='ou=demo,dc=my-domain,dc=com',
...     attrmap={
...         'id': 'cn',
...         'rdn': 'cn',
...         'login': 'sn',
...     },
...     scope=ONELEVEL,
...     queryFilter='(objectClass=person)',
...     objectClasses=['person'],
...     defaults={},
...     strict=False,
... )

>>> gcfg = GroupsConfig(
...     baseDN='ou=demo,dc=my-domain,dc=com',
...     attrmap={
...         'id': 'cn',
...         'rdn': 'cn',
...     },
...     scope=ONELEVEL,
...     queryFilter='(objectClass=groupOfNames)',
...     objectClasses=['groupOfNames'],
...     defaults={},
...     strict=False,
...     memberOfSupport=False,
... )

Roles are represented in LDAP like groups. Note, if groups and roles are mixed up in the same container, make sure that query filter fits. For our demo, different group object classes are used. Anyway, in real world it might be worth considering a seperate container for roles:

>>> rcfg = GroupsConfig(
...     baseDN='ou=demo,dc=my-domain,dc=com',
...     attrmap={
...         'id': 'cn',
...         'rdn': 'cn',
...     },
...     scope=ONELEVEL,
...     queryFilter='(objectClass=groupOfUniqueNames)',
...     objectClasses=['groupOfUniqueNames'],
...     defaults={},
...     strict=False,
... )

Instantiate Ugm object:

>>> ugm = Ugm(props=props, ucfg=ucfg, gcfg=gcfg, rcfg=rcfg)

The Ugm object has 2 children, the users container and the groups container. The are accessible via node API, but also on users respective groups attribute:

>>> ugm.keys()
['users', 'groups']

>>> ugm.users
<Users object 'users' at ...>

>>> ugm.groups
<Groups object 'groups' at ...>

Fetch user:

>>> user = ugm.users['person1']
>>> user
<User object 'person1' at ...>

User attributes. Reserved keys are available on user attributes:

>>> user.attrs['id']
u'person1'

>>> user.attrs['login']
u'Mustermensch'

'login' maps to 'sn':

>>> user.attrs['sn']
u'Mustermensch'

>>> user.attrs['login'] = u'Mustermensch1'
>>> user.attrs['sn']
u'Mustermensch1'

>>> user.attrs['description'] = 'Some description'
>>> user()

Check user credentials:

>>> user.authenticate('secret')
True

Change user password:

>>> user.passwd('secret', 'newsecret')
>>> user.authenticate('newsecret')
True

Groups user is member of:

>>> user.groups
[<Group object 'group1' at ...>]

Add new User:

>>> user = ugm.users.create('person99', sn='Person 99')
>>> user()

>>> res = ugm.users.keys()

>>> assert res == [
...     u'person1',
...     u'person2',
...     u'person3',
...     u'person4',
...     u'person5',
...     u'person99'
... ]

Delete User:

>>> del ugm.users['person99']
>>> ugm.users()

>>> res = ugm.users.keys()

>>> assert res == [
...     u'person1',
...     u'person2',
...     u'person3',
...     u'person4',
...     u'person5'
... ]

Fetch Group:

>>> group = ugm.groups['group1']

Group members:

>>> res = group.member_ids

>>> assert res == [u'person1', u'person2']

>>> group.users
[<User object 'person1' at ...>, <User object 'person2' at ...>]

Add group member:

>>> group.add('person3')

>>> member_ids = group.member_ids

>>> assert member_ids == [u'person1', u'person2', u'person3']

Delete group member:

>>> del group['person3']

>>> member_ids = group.member_ids

>>> assert member_ids == [u'person1', u'person2']

Group attribute manipulation works the same way as on user objects.

Manage roles for users and groups. Roles can be queried, added and removed via ugm or principal object. Fetch a user:

>>> user = ugm.users['person1']

Add role for user via ugm:

>>> ugm.add_role('viewer', user)

Add role for user directly:

>>> user.add_role('editor')

Query roles for user via ugm:

>>> roles = sorted(ugm.roles(user))

>>> assert roles == ['editor', 'viewer']

Query roles directly:

>>> roles = sorted(user.roles)

>>> assert roles == ['editor', 'viewer']

Call UGM to persist roles:

>>> ugm()

Delete role via ugm:

>>> ugm.remove_role('viewer', user)

>>> roles = user.roles

>>> assert roles == ['editor']

Delete role directly:

>>> user.remove_role('editor')

>>> roles = user.roles

>>> assert roles == []

Call UGM to persist roles:

>>> ugm()

Same with group. Fetch a group:

>>> group = ugm.groups['group1']

Add roles:

>>> ugm.add_role('viewer', group)

>>> group.add_role('editor')

>>> roles = sorted(ugm.roles(group))

>>> assert roles == ['editor', 'viewer']

>>> roles = sorted(group.roles)

>>> assert roles == ['editor', 'viewer']

>>> ugm()

Remove roles:

>>> ugm.remove_role('viewer', group)

>>> group.remove_role('editor')

>>> roles = group.roles

>>> assert roles == []

>>> ugm()

Character Encoding

LDAP (v3 at least, RFC 2251) uses utf-8 string encoding only. LDAPNode does the encoding for you. Consider it a bug, if you receive anything else than unicode from LDAPNode, except attributes configured as binary. The LDAPSession, LDAPConnector and LDAPCommunicator are encoding-neutral, they do no decoding or encoding.

Unicode strings you pass to nodes or sessions are automatically encoded as uft8 for LDAP, except if configured binary. If you feed them ordinary strings they are decoded as utf8 and reencoded as utf8 to make sure they are utf8 or compatible, e.g. ascii.

If you have an LDAP server that does not use utf8, monkey-patch node.ext.ldap._node.CHARACTER_ENCODING.

Caching Support

node.ext.ldap can cache LDAP searches using bda.cache. You need to provide a cache factory utility in you application in order to make caching work. If you don't, node.ext.ldap falls back to use bda.cache.NullCache, which does not cache anything and is just an API placeholder.

To provide a cache based on Memcached install memcached server and configure it. Then you need to provide the factory utility:

>>> from zope.interface import registry

>>> components = registry.Components('comps')

>>> from node.ext.ldap.cache import MemcachedProviderFactory

>>> cache_factory = MemcachedProviderFactory()

>>> components.registerUtility(cache_factory)

In case of multiple memcached backends on various IPs and ports initialization of the factory looks like this:

>>> components = registry.Components('comps')

>>> cache_factory = MemcachedProviderFactory(servers=[
...     '10.0.0.10:22122',
...     '10.0.0.11:22322'
... ])

>>> components.registerUtility(cache_factory)

Dependencies

• python-ldap
• passlib
• argparse
• plumber
• node
• node.ext.ugm
• bda.cache

Contributors

• Robert Niederreiter
• Florian Friesdorf
• Jens Klein
• Georg Bernhard
• Johannes Raggam
• Alexander Pilz
• Domen Kožar
• Daniel Widerin
• Asko Soukka
• Alex Milosz Sielicki
• Manuel Reinhardt
• Philip Bauer