Skip to content

A Jupyter Spawner pre_spawn_hook for creating/retrieving LDAP DIT entries during prespawn

License

Notifications You must be signed in to change notification settings

ucphhpc/ldap_hooks

Repository files navigation

ldap_hooks

https://travis-ci.org/rasmunk/ldap_hooks.svg?branch=master

A Jupyter Spawner hook for creating LDAP DIT entries via pre_spawn_hook

Installation

Installation from pypi:

pip install ldap-hooks

Installation from local git repository:

cd ldap_hooks
pip install .

Configuration

You should edit your jupyterhub_config.py config file to set a particular pre_spawn_hook, E.g:

from ldap_hooks import hello_hook

c.Spawner.pre_spawn_hook = hello_hook

In addition to specifying the pre_spawn_hook, a set of connection parameters must be set in order for the JupyterHub server to be able to interact with the designated LDAP host:

from ldap_hooks import LDAP

LDAP.url = "openldap"
LDAP.user = "cn=admin,dc=example,dc=org"
LDAP.password = "dummyldap_password"
LDAP.base_dn = "dc=example,dc=org"

The user's permissions here depend on whether the hook is just extracting information, or is creating entries as well.

The hooks that this library provides can be found below.

By default, any of these hooks are called by the Spawner with the following syntax:

def hook(spawner):
    # Do stuff inside the hook
    return True

That is, the hook expects that the current spawner instance is passed to it, which it can subsequently use to access properties of it, such as the user instance.

setup_ldap_entry_hook

This hook enables that the Spawner will submit/create an LDAP entry before the spawner starts the notebook. It is activated by setting the following parameter in the JupyterHub config:

from ldap_hooks import setup_ldap_entry_hook

c.Spawner.pre_spawn_hook = setup_ldap_entry_hook

In addition, the hook requires a number of a parameters to be configured before it will work as intended.

Basic Configuration

First, to defined the following options, the LDAP class must be imported into the jupyterhub_config.py file:

from ldap_hooks import LDAP

With this completed, the submit_spawner_attribute must be set, this must point to the variable path in the spawner instance where it can find the Distinguished Name String (DN) value. This string value makes up the entry that is to be submitted to the LDAP DIT, E.g:

# Retrieve the Distinguished Name from the 'spawner.user.data' variable
LDAP.submit_spawner_attribute = 'user.data'

In addition if this variable is of a dictionary structure, a tuple row can be specified to define the set of keys that should be used to extract the Distinguished Name value. For instance, if the value is in the spawner.user.data['User']['DN'] structure:

# Extract the Distinguished Name string from the
# spawner.user.data['User']['DN'] path.
LDAP.submit_spawner_attribute = 'user.data'
LDAP.submit_spawner_attribute_keys = ('User', 'DN')

If this extracted string is formatted in a way that is incorrectly seperated, the replace_object_with parameter can be used to fix this, E.g.:

# Prepare LDAP DN object entry
LDAP.replace_object_with = {'/': '+'}
# Does the following replacement
# /telephoneNumber=23012303403/SN=My Surname/CN=a-new-user
# +telephoneNumber=23012303403+SN=My Surname+CN=a-new-user

By default the name_strip_chars parameter is defined to strip extra characters that are either pre or postfixed to the DN:

# Default value
LDAP.name_strip_chars = ['/', '+', '*', ',', '.', '!', ' ']

Which means that it will automatically strip the prefixed + from the replace_object_with output.

Before the hook can submit the prepared DN, it first has to know which Structural ObjectClass should be used to create the entry with. Beyond at least one required Structural ObjectClass, a list of additional Auxiliary ObjectClasses can be specified as well. All of which must be set via the object_classes parameter, E.g:

# Structural 'Person'
LDAP.object_classes = ['Person']

Any specified object class must be supported as part of the specified LDAP.url server schema.

Beyond the object_classes, the hook also provides a parameter to specify additional object attributes to submittet DN entry:

LDAP.object_attributes = {'description': 'A default person account',
                          'surname': 'MySurname'}

Duplicate entries can be default not exist in the LDAP DIT, therefore any duplicate DN submission will fail. By default the hook will search the DIT for an entry that matches every attribute of the DN string, if such an entry exists, the hook will simply stop before attempting to submit it. This behaviour can be customised via the unique_object_attributes parameter as shown in the "Extra Features" section.

Extra Features

unique_object_attributes

It is possible to specify special attributes that the hook should use for this search via the unique_object_attributes parameter:

# Optional parameter
LDAP.unique_object_attributes = ['surname']

Now the hook will search for if an entry with object_classes exists, if so it will stop the submission.

set_spawner_attributes

Use this to set JupyterHub Spawner attributes. For instance set an environment variable of the Spawned notebooks:

# Set Spawned Notebook environment vars
LDAP.set_spawner_attributes = {
    'environment': {'ENV_VAR': 'Hello from LDAP Hook'}
}

search_attribute_queries

Use this to define a list of LDAP search operations to extract a list of attributes from the existing DIT which can subsequntly be used to perform some subsequent operation on the extracted attributes, or share them with the set_spawner_attributes or object_attributes via the dynamic_attributes definition.

For instance, extract the uidNumber attribute from the LDAP DIT which has the x-nextUserIdentifier objectclass:

LDAP.search_attribute_queries = [
    {'search_base': LDAP.base_dn,
    'search_filter': '(objectclass=X-nextUserIdentifier)',
    'attributes': ['uidNumber']}
]

search_result_operations

Use this to perform an operation action on extracted attributes of the search_attribute_queries. The specific action must be defined as a LDAP.SEARCH_RESULT_OPERATION_ACTIONS. For instance, increment the value of the extracted uidNumber attribute by 1, for this particular action, it is required that the modify_dn key is also provided since it defines the Distinguished Name that should be used to select that attribute to be incremented in the DIT:

modify_dn = 'cn=uidNumber' + ',' + LDAP.base_dn
LDAP.search_result_operation = {'uidNumber': {'action': INCREMENT_ATTRIBUTE,
                                            'modify_dn': modify_dn}}

This will produce an atomic modify-increment to the value of the cn=uidNumber,dc=example,dc=org.

dynamic_attributes

To format set_spawner_attributes and object_attributes with dynamic attributes, such as the result of an LDAP.SEARCH_RESULT_OPERATION_ACTIONS or values provided by a submit_spawner_attribute dictionary. The dynamic_attributes can be used to format such attributes. For instance, if the set_spawner_attributes defines attributes that expects formatting of the 'emailAddress' and 'uidNumber':

LDAP.set_spawner_attributes = {
    'environment': {'NB_USER': '{emailAddress}',
                    'NB_UID': '{uidNumber}'},
}

The dynamic_attributes can provide these as follows:

LDAP.dynamic_attributes = {
    'emailAddress': SPAWNER_SUBMIT_DATA,
    'uidNumber': LDAP_SEARCH_ATTRIBUTE_QUERY
}

Where the values of the keys define how and where the attribute values should be extracted. The specified value must be defined as a LDAP.DYNAMIC_ATTRIBUTE_METHODS.

In addition these dynamic_attributes are made available to the defined object_attributes. For example:

LDAP.object_attributes = {'uidNumber': '{uidNumber}',
                          'homeDirectory': '/home/{emailAddress}'}