# Copyright 2012 OpenStack Foundation
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
import abc
import codecs
import os.path
import random
import re
import sys
import typing as ty
import uuid
import weakref
import ldap.controls
import ldap.filter
import ldappool
from oslo_log import log
from oslo_utils import reflection
from keystone.common import driver_hints
from keystone import exception
from keystone.i18n import _
from keystone.identity.backends.ldap import models
LOG = log.getLogger(__name__)
LDAP_VALUES = {'TRUE': True, 'FALSE': False}
LDAP_SCOPES = {'one': ldap.SCOPE_ONELEVEL, 'sub': ldap.SCOPE_SUBTREE}
LDAP_DEREF = {
'always': ldap.DEREF_ALWAYS,
'default': None,
'finding': ldap.DEREF_FINDING,
'never': ldap.DEREF_NEVER,
'searching': ldap.DEREF_SEARCHING,
}
LDAP_TLS_CERTS = {
'never': ldap.OPT_X_TLS_NEVER,
'demand': ldap.OPT_X_TLS_DEMAND,
'allow': ldap.OPT_X_TLS_ALLOW,
}
# RFC 4511 (The LDAP Protocol) defines a list containing only the OID '1.1' to
# indicate that no attributes should be returned besides the DN.
DN_ONLY = ['1.1']
_utf8_encoder = codecs.getencoder('utf-8')
# FIXME(knikolla): This enables writing to the LDAP backend
# Only enabled during tests and unsupported
WRITABLE = False
[docs]
def utf8_encode(value):
"""Encode a basestring to UTF-8.
If the string is unicode encode it to UTF-8, if the string is
str then assume it's already encoded. Otherwise raise a TypeError.
:param value: A basestring
:returns: UTF-8 encoded version of value
:raises TypeError: If value is not basestring
"""
if isinstance(value, str):
return _utf8_encoder(value)[0]
elif isinstance(value, bytes):
return value
else:
value_cls_name = reflection.get_class_name(
value, fully_qualified=False
)
raise TypeError("value must be basestring, not %s" % value_cls_name)
_utf8_decoder = codecs.getdecoder('utf-8')
[docs]
def utf8_decode(value):
"""Decode a from UTF-8 into unicode.
If the value is a binary string assume it's UTF-8 encoded and decode
it into a unicode string. Otherwise convert the value from its
type into a unicode string.
:param value: value to be returned as unicode
:returns: value as unicode
:raises UnicodeDecodeError: for invalid UTF-8 encoding
"""
if isinstance(value, bytes):
try:
return _utf8_decoder(value)[0]
except UnicodeDecodeError:
# NOTE(lbragstad): We could be dealing with a UUID in byte form,
# which some LDAP implementations use.
uuid_byte_string_length = 16
if len(value) == uuid_byte_string_length:
return str(uuid.UUID(bytes_le=value))
else:
raise
return str(value)
[docs]
def py2ldap(val):
"""Type convert a Python value to a type accepted by LDAP (unicode).
The LDAP API only accepts strings for values therefore convert
the value's type to a unicode string. A subsequent type conversion
will encode the unicode as UTF-8 as required by the python-ldap API,
but for now we just want a string representation of the value.
:param val: The value to convert to a LDAP string representation
:returns: unicode string representation of value.
"""
if isinstance(val, bool):
return 'TRUE' if val else 'FALSE'
else:
return str(val)
[docs]
def enabled2py(val):
"""Similar to ldap2py, only useful for the enabled attribute."""
try:
return LDAP_VALUES[val]
except KeyError: # nosec
# It wasn't a boolean value, will try as an int instead.
pass
try:
return int(val)
except ValueError: # nosec
# It wasn't an int either, will try as utf8 instead.
pass
return utf8_decode(val)
[docs]
def ldap2py(val):
"""Convert an LDAP formatted value to Python type used by OpenStack.
Virtually all LDAP values are stored as UTF-8 encoded strings.
OpenStack prefers values which are unicode friendly.
:param val: LDAP formatted value
:returns: val converted to preferred Python type
"""
return utf8_decode(val)
[docs]
def convert_ldap_result(ldap_result):
"""Convert LDAP search result to Python types used by OpenStack.
Each result tuple is of the form (dn, attrs), where dn is a string
containing the DN (distinguished name) of the entry, and attrs is
a dictionary containing the attributes associated with the
entry. The keys of attrs are strings, and the associated values
are lists of strings.
OpenStack wants to use Python types of its choosing. Strings will
be unicode, truth values boolean, whole numbers int's, etc. DN's are
represented as text in python-ldap by default for Python 3 and when
bytes_mode=False for Python 2, and therefore do not require decoding.
:param ldap_result: LDAP search result
:returns: list of 2-tuples containing (dn, attrs) where dn is unicode
and attrs is a dict whose values are type converted to
OpenStack preferred types.
"""
py_result = []
at_least_one_referral = False
for dn, attrs in ldap_result:
ldap_attrs = {}
if dn is None:
# this is a Referral object, rather than an Entry object
at_least_one_referral = True
continue
for kind, values in attrs.items():
try:
val2py = enabled2py if kind == 'enabled' else ldap2py
ldap_attrs[kind] = [val2py(x) for x in values]
except UnicodeDecodeError:
LOG.debug('Unable to decode value for attribute %s', kind)
py_result.append((dn, ldap_attrs))
if at_least_one_referral:
LOG.debug(
'Referrals were returned and ignored. Enable referral '
'chasing in keystone.conf via [ldap] chase_referrals'
)
return py_result
[docs]
def safe_iter(attrs):
if attrs is None:
return
elif isinstance(attrs, list):
yield from attrs
else:
yield attrs
[docs]
def parse_deref(opt):
try:
return LDAP_DEREF[opt]
except KeyError:
raise ValueError(
_(
'Invalid LDAP deref option: %(option)s. '
'Choose one of: %(options)s'
)
% {
'option': opt,
'options': ', '.join(LDAP_DEREF.keys()),
}
)
[docs]
def parse_tls_cert(opt):
try:
return LDAP_TLS_CERTS[opt]
except KeyError:
raise ValueError(
_(
'Invalid LDAP TLS certs option: %(option)s. '
'Choose one of: %(options)s'
)
% {'option': opt, 'options': ', '.join(LDAP_TLS_CERTS.keys())}
)
[docs]
def ldap_scope(scope):
try:
return LDAP_SCOPES[scope]
except KeyError:
raise ValueError(
_('Invalid LDAP scope: %(scope)s. Choose one of: %(options)s')
% {'scope': scope, 'options': ', '.join(LDAP_SCOPES.keys())}
)
[docs]
def prep_case_insensitive(value):
"""Prepare a string for case-insensitive comparison.
This is defined in RFC4518. For simplicity, all this function does is
lowercase all the characters, strip leading and trailing whitespace,
and compress sequences of spaces to a single space.
"""
value = re.sub(r'\s+', ' ', value.strip().lower())
return value
[docs]
def is_ava_value_equal(attribute_type, val1, val2):
"""Return True if and only if the AVAs are equal.
When comparing AVAs, the equality matching rule for the attribute type
should be taken into consideration. For simplicity, this implementation
does a case-insensitive comparison.
Note that this function uses prep_case_insenstive so the limitations of
that function apply here.
"""
return prep_case_insensitive(val1) == prep_case_insensitive(val2)
[docs]
def is_rdn_equal(rdn1, rdn2):
"""Return True if and only if the RDNs are equal.
* RDNs must have the same number of AVAs.
* Each AVA of the RDNs must be the equal for the same attribute type. The
order isn't significant. Note that an attribute type will only be in one
AVA in an RDN, otherwise the DN wouldn't be valid.
* Attribute types aren't case sensitive. Note that attribute type
comparison is more complicated than implemented. This function only
compares case-insentive. The code should handle multiple names for an
attribute type (e.g., cn, commonName, and 2.5.4.3 are the same).
Note that this function uses is_ava_value_equal to compare AVAs so the
limitations of that function apply here.
"""
if len(rdn1) != len(rdn2):
return False
for attr_type_1, val1, dummy in rdn1:
found = False
for attr_type_2, val2, dummy in rdn2:
if attr_type_1.lower() != attr_type_2.lower():
continue
found = True
if not is_ava_value_equal(attr_type_1, val1, val2):
return False
break
if not found:
return False
return True
[docs]
def is_dn_equal(dn1, dn2):
"""Return True if and only if the DNs are equal.
Two DNs are equal if they've got the same number of RDNs and if the RDNs
are the same at each position. See RFC4517.
Note that this function uses is_rdn_equal to compare RDNs so the
limitations of that function apply here.
:param dn1: Either a string DN or a DN parsed by ldap.dn.str2dn.
:param dn2: Either a string DN or a DN parsed by ldap.dn.str2dn.
"""
if not isinstance(dn1, list):
dn1 = ldap.dn.str2dn(dn1)
if not isinstance(dn2, list):
dn2 = ldap.dn.str2dn(dn2)
if len(dn1) != len(dn2):
return False
for rdn1, rdn2 in zip(dn1, dn2):
if not is_rdn_equal(rdn1, rdn2):
return False
return True
[docs]
def dn_startswith(descendant_dn, dn):
"""Return True if and only if the descendant_dn is under the dn.
:param descendant_dn: Either a string DN or a DN parsed by ldap.dn.str2dn.
:param dn: Either a string DN or a DN parsed by ldap.dn.str2dn.
"""
if not isinstance(descendant_dn, list):
descendant_dn = ldap.dn.str2dn(descendant_dn)
if not isinstance(dn, list):
dn = ldap.dn.str2dn(dn)
if len(descendant_dn) <= len(dn):
return False
# Use the last len(dn) RDNs.
return is_dn_equal(descendant_dn[-len(dn) :], dn)
[docs]
class LDAPHandler(metaclass=abc.ABCMeta):
"""Abstract class which defines methods for a LDAP API provider.
Native Keystone values cannot be passed directly into and from the
python-ldap API. Type conversion must occur at the LDAP API
boundary, examples of type conversions are:
* booleans map to the strings 'TRUE' and 'FALSE'
* integer values map to their string representation.
* unicode strings are encoded in UTF-8
Note, in python-ldap some fields (DNs, RDNs, attribute names, queries)
are represented as text (str on Python 3, unicode on Python 2 when
bytes_mode=False). For more details see:
http://www.python-ldap.org/en/latest/bytes_mode.html#bytes-mode
In addition to handling type conversions at the API boundary we
have the requirement to support more than one LDAP API
provider. Currently we have:
* python-ldap, this is the standard LDAP API for Python, it
requires access to a live LDAP server.
* Fake LDAP which emulates python-ldap. This is used for
testing without requiring a live LDAP server.
To support these requirements we need a layer that performs type
conversions and then calls another LDAP API which is configurable
(e.g. either python-ldap or the fake emulation).
We have an additional constraint at the time of this writing due to
limitations in the logging module. The logging module is not
capable of accepting UTF-8 encoded strings, it will throw an
encoding exception. Therefore all logging MUST be performed prior
to UTF-8 conversion. This means no logging can be performed in the
ldap APIs that implement the python-ldap API because those APIs
are defined to accept only UTF-8 strings. Thus the layer which
performs type conversions must also do the logging. We do the type
conversions in two steps, once to convert all Python types to
unicode strings, then log, then convert the unicode strings to
UTF-8.
There are a variety of ways one could accomplish this, we elect to
use a chaining technique whereby instances of this class simply
call the next member in the chain via the "conn" attribute. The
chain is constructed by passing in an existing instance of this
class as the conn attribute when the class is instantiated.
Here is a brief explanation of why other possible approaches were
not used:
subclassing
To perform the wrapping operations in the correct order
the type conversion class would have to subclass each of
the API providers. This is awkward, doubles the number of
classes, and does not scale well. It requires the type
conversion class to be aware of all possible API
providers.
decorators
Decorators provide an elegant solution to wrap methods and
would be an ideal way to perform type conversions before
calling the wrapped function and then converting the
values returned from the wrapped function. However
decorators need to be aware of the method signature, it
has to know what input parameters need conversion and how
to convert the result. For an API like python-ldap which
has a large number of different method signatures it would
require a large number of specialized
decorators. Experience has shown it's very easy to apply
the wrong decorator due to the inherent complexity and
tendency to cut-n-paste code. Another option is to
parameterize the decorator to make it "smart". Experience
has shown such decorators become insanely complicated and
difficult to understand and debug. Also decorators tend to
hide what's really going on when a method is called, the
operations being performed are not visible when looking at
the implemation of a decorated method, this too experience
has shown leads to mistakes.
Chaining simplifies both wrapping to perform type conversion as
well as the substitution of alternative API providers. One simply
creates a new instance of the API interface and insert it at the
front of the chain. Type conversions are explicit and obvious.
If a new method needs to be added to the API interface one adds it
to the abstract class definition. Should one miss adding the new
method to any derivations of the abstract class the code will fail
to load and run making it impossible to forget updating all the
derived classes.
"""
def __init__(self, conn=None):
self.conn = conn
[docs]
@abc.abstractmethod
def connect(
self,
url,
page_size=0,
alias_dereferencing=None,
use_tls=False,
tls_cacertfile=None,
tls_cacertdir=None,
tls_req_cert=ldap.OPT_X_TLS_DEMAND,
chase_referrals=None,
debug_level=None,
conn_timeout=None,
use_pool=None,
pool_size=None,
pool_retry_max=None,
pool_retry_delay=None,
pool_conn_timeout=None,
pool_conn_lifetime=None,
):
raise exception.NotImplemented() # pragma: no cover
[docs]
@abc.abstractmethod
def set_option(self, option, invalue):
raise exception.NotImplemented() # pragma: no cover
[docs]
@abc.abstractmethod
def get_option(self, option):
raise exception.NotImplemented() # pragma: no cover
[docs]
@abc.abstractmethod
def simple_bind_s(
self, who='', cred='', serverctrls=None, clientctrls=None
):
raise exception.NotImplemented() # pragma: no cover
[docs]
@abc.abstractmethod
def unbind_s(self):
raise exception.NotImplemented() # pragma: no cover
[docs]
@abc.abstractmethod
def add_s(self, dn, modlist):
raise exception.NotImplemented() # pragma: no cover
[docs]
@abc.abstractmethod
def search_s(
self,
base,
scope,
filterstr='(objectClass=*)',
attrlist=None,
attrsonly=0,
):
raise exception.NotImplemented() # pragma: no cover
[docs]
@abc.abstractmethod
def search_ext(
self,
base,
scope,
filterstr='(objectClass=*)',
attrlist=None,
attrsonly=0,
serverctrls=None,
clientctrls=None,
timeout=-1,
sizelimit=0,
):
raise exception.NotImplemented() # pragma: no cover
[docs]
@abc.abstractmethod
def result3(
self, msgid=ldap.RES_ANY, all=1, timeout=None, resp_ctrl_classes=None
):
raise exception.NotImplemented() # pragma: no cover
[docs]
@abc.abstractmethod
def modify_s(self, dn, modlist):
raise exception.NotImplemented() # pragma: no cover
[docs]
class PythonLDAPHandler(LDAPHandler):
"""LDAPHandler implementation which calls the python-ldap API.
Note, the python-ldap API requires all string attribute values to be UTF-8
encoded.
Note, in python-ldap some fields (DNs, RDNs, attribute names, queries)
are represented as text (str on Python 3, unicode on Python 2 when
bytes_mode=False). For more details see:
http://www.python-ldap.org/en/latest/bytes_mode.html#bytes-mode
The KeystoneLDAPHandler enforces this prior to invoking the methods in this
class.
"""
[docs]
def connect(
self,
url,
page_size=0,
alias_dereferencing=None,
use_tls=False,
tls_cacertfile=None,
tls_cacertdir=None,
tls_req_cert=ldap.OPT_X_TLS_DEMAND,
chase_referrals=None,
debug_level=None,
conn_timeout=None,
use_pool=None,
pool_size=None,
pool_retry_max=None,
pool_retry_delay=None,
pool_conn_timeout=None,
pool_conn_lifetime=None,
):
_common_ldap_initialization(
url=url,
use_tls=use_tls,
tls_cacertfile=tls_cacertfile,
tls_cacertdir=tls_cacertdir,
tls_req_cert=tls_req_cert,
debug_level=debug_level,
timeout=conn_timeout,
)
self.conn = ldap.initialize(url)
self.conn.protocol_version = ldap.VERSION3
if alias_dereferencing is not None:
self.conn.set_option(ldap.OPT_DEREF, alias_dereferencing)
self.page_size = page_size
if use_tls:
self.conn.start_tls_s()
if chase_referrals is not None:
self.conn.set_option(ldap.OPT_REFERRALS, int(chase_referrals))
[docs]
def set_option(self, option, invalue):
return self.conn.set_option(option, invalue)
[docs]
def get_option(self, option):
return self.conn.get_option(option)
[docs]
def simple_bind_s(
self, who='', cred='', serverctrls=None, clientctrls=None
):
return self.conn.simple_bind_s(who, cred, serverctrls, clientctrls)
[docs]
def unbind_s(self):
return self.conn.unbind_s()
[docs]
def add_s(self, dn, modlist):
return self.conn.add_s(dn, modlist)
[docs]
def search_s(
self,
base,
scope,
filterstr='(objectClass=*)',
attrlist=None,
attrsonly=0,
):
return self.conn.search_s(base, scope, filterstr, attrlist, attrsonly)
[docs]
def search_ext(
self,
base,
scope,
filterstr='(objectClass=*)',
attrlist=None,
attrsonly=0,
serverctrls=None,
clientctrls=None,
timeout=-1,
sizelimit=0,
):
return self.conn.search_ext(
base,
scope,
filterstr,
attrlist,
attrsonly,
serverctrls,
clientctrls,
timeout,
sizelimit,
)
[docs]
def result3(
self, msgid=ldap.RES_ANY, all=1, timeout=None, resp_ctrl_classes=None
):
# The resp_ctrl_classes parameter is a recent addition to the
# API. It defaults to None. We do not anticipate using it.
# To run with older versions of python-ldap we do not pass it.
return self.conn.result3(msgid, all, timeout)
[docs]
def modify_s(self, dn, modlist):
return self.conn.modify_s(dn, modlist)
def _common_ldap_initialization(
url,
use_tls=False,
tls_cacertfile=None,
tls_cacertdir=None,
tls_req_cert=None,
debug_level=None,
timeout=None,
):
"""LDAP initialization for PythonLDAPHandler and PooledLDAPHandler."""
LOG.debug('LDAP init: url=%s', url)
LOG.debug(
'LDAP init: use_tls=%s tls_cacertfile=%s tls_cacertdir=%s '
'tls_req_cert=%s tls_avail=%s',
use_tls,
tls_cacertfile,
tls_cacertdir,
tls_req_cert,
ldap.TLS_AVAIL,
)
if debug_level is not None:
ldap.set_option(ldap.OPT_DEBUG_LEVEL, debug_level)
using_ldaps = url.lower().startswith("ldaps")
if timeout is not None and timeout > 0:
# set network connection timeout
ldap.set_option(ldap.OPT_NETWORK_TIMEOUT, timeout)
if use_tls and using_ldaps:
raise AssertionError(_('Invalid TLS / LDAPS combination'))
# The certificate trust options apply for both LDAPS and TLS.
if use_tls or using_ldaps:
if not ldap.TLS_AVAIL:
raise ValueError(
_('Invalid LDAP TLS_AVAIL option: %s. TLS not available')
% ldap.TLS_AVAIL
)
if not tls_cacertfile and not tls_cacertdir:
raise ValueError(
_(
'You need to set tls_cacertfile or '
'tls_cacertdir if use_tls is true or '
'url uses ldaps: scheme.'
)
)
if tls_cacertfile:
# NOTE(topol)
# python ldap TLS does not verify CACERTFILE or CACERTDIR
# so we add some extra simple sanity check verification
# Also, setting these values globally (i.e. on the ldap object)
# works but these values are ignored when setting them on the
# connection
if not os.path.isfile(tls_cacertfile):
raise OSError(
_("tls_cacertfile %s not found or is not a file")
% tls_cacertfile
)
ldap.set_option(ldap.OPT_X_TLS_CACERTFILE, tls_cacertfile)
elif tls_cacertdir:
# NOTE(topol)
# python ldap TLS does not verify CACERTFILE or CACERTDIR
# so we add some extra simple sanity check verification
# Also, setting these values globally (i.e. on the ldap object)
# works but these values are ignored when setting them on the
# connection
if not os.path.isdir(tls_cacertdir):
raise OSError(
_("tls_cacertdir %s not found or is not a directory")
% tls_cacertdir
)
ldap.set_option(ldap.OPT_X_TLS_CACERTDIR, tls_cacertdir)
if tls_req_cert in list(LDAP_TLS_CERTS.values()):
ldap.set_option(ldap.OPT_X_TLS_REQUIRE_CERT, tls_req_cert)
else:
LOG.debug(
'LDAP TLS: invalid TLS_REQUIRE_CERT Option=%s', tls_req_cert
)
[docs]
class AsynchronousMessage:
"""A container for handling asynchronous LDAP responses.
Some LDAP APIs, like `search_ext`, are asynchronous and return a message ID
when the server successfully initiates the operation. Clients can use this
message ID and the original connection to make the request to fetch the
results using `result3`.
This object holds the message ID, the original connection, and a callable
weak reference Finalizer that cleans up context managers specific to the
connection associated to the message ID.
:param message_id: The message identifier (str).
:param connection: The connection associated with the message identifier
(ldappool.StateConnector).
The `clean` attribute is a callable that cleans up the context manager used
to create or return the connection object (weakref.finalize).
"""
def __init__(self, message_id, connection, context_manager):
self.id = message_id
self.connection = connection
self.clean = weakref.finalize(
self, self._cleanup_connection_context_manager, context_manager
)
def _cleanup_connection_context_manager(self, context_manager):
context_manager.__exit__(None, None, None)
[docs]
def use_conn_pool(func):
"""Use this only for connection pool specific ldap API.
This adds connection object to decorated API as next argument after self.
"""
def wrapper(self, *args, **kwargs):
# assert isinstance(self, PooledLDAPHandler)
with self._get_pool_connection() as conn:
self._apply_options(conn)
return func(self, conn, *args, **kwargs)
return wrapper
[docs]
class PooledLDAPHandler(LDAPHandler):
"""LDAPHandler implementation which uses pooled connection manager.
Pool specific configuration is defined in [ldap] section.
All other LDAP configuration is still used from [ldap] section
Keystone LDAP authentication logic authenticates an end user using its DN
and password via LDAP bind to establish supplied password is correct.
This can fill up the pool quickly (as pool re-uses existing connection
based on its bind data) and would not leave space in pool for connection
re-use for other LDAP operations.
Now a separate pool can be established for those requests when related flag
'use_auth_pool' is enabled. That pool can have its own size and
connection lifetime. Other pool attributes are shared between those pools.
If 'use_pool' is disabled, then 'use_auth_pool' does not matter.
If 'use_auth_pool' is not enabled, then connection pooling is not used for
those LDAP operations.
Note, the python-ldap API requires all string attribute values to be UTF-8
encoded. The KeystoneLDAPHandler enforces this prior to invoking the
methods in this class.
Note, in python-ldap some fields (DNs, RDNs, attribute names, queries)
are represented as text (str on Python 3, unicode on Python 2 when
bytes_mode=False). For more details see:
http://www.python-ldap.org/en/latest/bytes_mode.html#bytes-mode
"""
# Added here to allow override for testing
Connector = ldappool.StateConnector
auth_pool_prefix = 'auth_pool_'
connection_pools: dict = {} # static connector pool dict
def __init__(self, conn=None, use_auth_pool=False):
super().__init__(conn=conn)
self.who = ''
self.cred = ''
self.conn_options = {} # connection specific options
self.page_size = None
self.use_auth_pool = use_auth_pool
self.conn_pool = None
[docs]
def connect(
self,
url,
page_size=0,
alias_dereferencing=None,
use_tls=False,
tls_cacertfile=None,
tls_cacertdir=None,
tls_req_cert=ldap.OPT_X_TLS_DEMAND,
chase_referrals=None,
debug_level=None,
conn_timeout=None,
use_pool=None,
pool_size=None,
pool_retry_max=None,
pool_retry_delay=None,
pool_conn_timeout=None,
pool_conn_lifetime=None,
):
_common_ldap_initialization(
url=url,
use_tls=use_tls,
tls_cacertfile=tls_cacertfile,
tls_cacertdir=tls_cacertdir,
tls_req_cert=tls_req_cert,
debug_level=debug_level,
timeout=pool_conn_timeout,
)
self.page_size = page_size
# Following two options are not added in common initialization as they
# need to follow a sequence in PythonLDAPHandler code.
if alias_dereferencing is not None:
self.set_option(ldap.OPT_DEREF, alias_dereferencing)
if chase_referrals is not None:
self.set_option(ldap.OPT_REFERRALS, int(chase_referrals))
if self.use_auth_pool: # separate pool when use_auth_pool enabled
pool_url = self.auth_pool_prefix + url
else:
pool_url = url
try:
self.conn_pool = self.connection_pools[pool_url]
except KeyError:
self.conn_pool = ldappool.ConnectionManager(
url,
size=pool_size,
retry_max=pool_retry_max,
retry_delay=pool_retry_delay,
timeout=pool_conn_timeout,
connector_cls=self.Connector,
use_tls=use_tls,
max_lifetime=pool_conn_lifetime,
)
self.connection_pools[pool_url] = self.conn_pool
[docs]
def set_option(self, option, invalue):
self.conn_options[option] = invalue
[docs]
def get_option(self, option):
value = self.conn_options.get(option)
# if option was not specified explicitly, then use connection default
# value for that option if there.
if value is None:
with self._get_pool_connection() as conn:
value = conn.get_option(option)
return value
def _apply_options(self, conn):
# if connection has a lifetime, then it already has options specified
if conn.get_lifetime() > 30:
return
for option, invalue in self.conn_options.items():
conn.set_option(option, invalue)
def _get_pool_connection(self):
return self.conn_pool.connection(self.who, self.cred)
[docs]
def simple_bind_s(
self, who='', cred='', serverctrls=None, clientctrls=None
):
# Not using use_conn_pool decorator here as this API takes cred as
# input.
self.who = who
self.cred = cred
with self._get_pool_connection() as conn:
self._apply_options(conn)
[docs]
def unbind_s(self):
# After connection generator is done `with` statement execution block
# connection is always released via finally block in ldappool.
# So this unbind is a no op.
pass
[docs]
@use_conn_pool
def add_s(self, conn, dn, modlist):
return conn.add_s(dn, modlist)
[docs]
@use_conn_pool
def search_s(
self,
conn,
base,
scope,
filterstr='(objectClass=*)',
attrlist=None,
attrsonly=0,
):
return conn.search_s(base, scope, filterstr, attrlist, attrsonly)
[docs]
def search_ext(
self,
base,
scope,
filterstr='(objectClass=*)',
attrlist=None,
attrsonly=0,
serverctrls=None,
clientctrls=None,
timeout=-1,
sizelimit=0,
):
"""Return an AsynchronousMessage instance, it asynchronous API.
The AsynchronousMessage instance can be safely used in a call to
`result3()`.
To work with `result3()` API in predictable manner, the same LDAP
connection is needed which originally provided the `msgid`. So, this
method wraps the existing connection and `msgid` in a new
`AsynchronousMessage` instance. The connection associated with
`search_ext()` is released after `result3()` fetches the data
associated with `msgid`.
"""
conn_ctxt = self._get_pool_connection()
conn = conn_ctxt.__enter__()
try:
msgid = conn.search_ext(
base,
scope,
filterstr,
attrlist,
attrsonly,
serverctrls,
clientctrls,
timeout,
sizelimit,
)
except Exception:
conn_ctxt.__exit__(*sys.exc_info())
raise
return AsynchronousMessage(msgid, conn, conn_ctxt)
[docs]
def result3(self, message, all=1, timeout=None, resp_ctrl_classes=None):
"""Wait for and return the result to an asynchronous message.
This method returns the result of an operation previously initiated by
one of the LDAP asynchronous operation routines (e.g., `search_ext()`).
The `search_ext()` method in python-ldap returns an invocation
identifier, or a message ID, upon successful initiation of the
operation by the LDAP server.
The `message` is expected to be instance of class
`AsynchronousMessage`, which contains the message ID and the connection
used to make the original request.
The connection and context manager associated with `search_ext()` are
cleaned up when message.clean() is called.
"""
# message.connection.result3 might throw an exception
# so the code must ensure that message.clean() is invoked
# regardless of the result3's result. Otherwise, the
# connection will be marked as active forever, which
# ultimately renders the pool unusable, causing a DoS.
try:
results = message.connection.result3(message.id, all, timeout)
finally:
# Now that we have the results from the LDAP server for
# the message, we don't need the the context manager used
# to create the connection.
message.clean()
return results
[docs]
@use_conn_pool
def modify_s(self, conn, dn, modlist):
return conn.modify_s(dn, modlist)
[docs]
class KeystoneLDAPHandler(LDAPHandler):
"""Convert data types and perform logging.
This LDAP interface wraps the python-ldap based interfaces. The
python-ldap interfaces require string values encoded in UTF-8 with
the exception of [1]. The OpenStack logging framework at the time
of this writing is not capable of accepting strings encoded in
UTF-8, the log functions will throw decoding errors if a non-ascii
character appears in a string.
[1] In python-ldap, some fields (DNs, RDNs, attribute names,
queries) are represented as text (str on Python 3, unicode on
Python 2 when bytes_mode=False). For more details see:
http://www.python-ldap.org/en/latest/bytes_mode.html#bytes-mode
Prior to the call Python data types are converted to a string
representation as required by the LDAP APIs.
Then logging is performed so we can track what is being
sent/received from LDAP. Also the logging filters security
sensitive items (i.e. passwords).
Then the string values are encoded into UTF-8.
Then the LDAP API entry point is invoked.
Data returned from the LDAP call is converted back from UTF-8
encoded strings into the Python data type used internally in
OpenStack.
"""
def __init__(self, conn=None):
super().__init__(conn=conn)
self.page_size = 0
def __enter__(self):
"""Enter runtime context."""
return self
def _disable_paging(self):
# Disable the pagination from now on
self.page_size = 0
[docs]
def connect(
self,
url,
page_size=0,
alias_dereferencing=None,
use_tls=False,
tls_cacertfile=None,
tls_cacertdir=None,
tls_req_cert=ldap.OPT_X_TLS_DEMAND,
chase_referrals=None,
debug_level=None,
conn_timeout=None,
use_pool=None,
pool_size=None,
pool_retry_max=None,
pool_retry_delay=None,
pool_conn_timeout=None,
pool_conn_lifetime=None,
):
self.page_size = page_size
return self.conn.connect(
url,
page_size,
alias_dereferencing,
use_tls,
tls_cacertfile,
tls_cacertdir,
tls_req_cert,
chase_referrals,
debug_level=debug_level,
conn_timeout=conn_timeout,
use_pool=use_pool,
pool_size=pool_size,
pool_retry_max=pool_retry_max,
pool_retry_delay=pool_retry_delay,
pool_conn_timeout=pool_conn_timeout,
pool_conn_lifetime=pool_conn_lifetime,
)
[docs]
def set_option(self, option, invalue):
return self.conn.set_option(option, invalue)
[docs]
def get_option(self, option):
return self.conn.get_option(option)
[docs]
def simple_bind_s(
self, who='', cred='', serverctrls=None, clientctrls=None
):
LOG.debug('LDAP bind: who=%s', who)
return self.conn.simple_bind_s(
who, cred, serverctrls=serverctrls, clientctrls=clientctrls
)
[docs]
def unbind_s(self):
LOG.debug('LDAP unbind')
return self.conn.unbind_s()
[docs]
def add_s(self, dn, modlist):
ldap_attrs = [
(kind, [py2ldap(x) for x in safe_iter(values)])
for kind, values in modlist
]
logging_attrs = [
(kind, values if kind != 'userPassword' else ['****'])
for kind, values in ldap_attrs
]
LOG.debug('LDAP add: dn=%s attrs=%s', dn, logging_attrs)
ldap_attrs_utf8 = [
(kind, [utf8_encode(x) for x in safe_iter(values)])
for kind, values in ldap_attrs
]
return self.conn.add_s(dn, ldap_attrs_utf8)
[docs]
def search_s(
self,
base,
scope,
filterstr='(objectClass=*)',
attrlist=None,
attrsonly=0,
):
# NOTE(morganfainberg): Remove "None" singletons from this list, which
# allows us to set mapped attributes to "None" as defaults in config.
# Without this filtering, the ldap query would raise a TypeError since
# attrlist is expected to be an iterable of strings.
if attrlist is not None:
attrlist = [attr for attr in attrlist if attr is not None]
LOG.debug(
'LDAP search: base=%s scope=%s filterstr=%s '
'attrs=%s attrsonly=%s',
base,
scope,
filterstr,
attrlist,
attrsonly,
)
if self.page_size:
ldap_result = self._paged_search_s(
base, scope, filterstr, attrlist
)
else:
try:
ldap_result = self.conn.search_s(
base, scope, filterstr, attrlist, attrsonly
)
except ldap.SIZELIMIT_EXCEEDED:
raise exception.LDAPSizeLimitExceeded()
py_result = convert_ldap_result(ldap_result)
return py_result
[docs]
def search_ext(
self,
base,
scope,
filterstr='(objectClass=*)',
attrlist=None,
attrsonly=0,
serverctrls=None,
clientctrls=None,
timeout=-1,
sizelimit=0,
):
if attrlist is not None:
attrlist = [attr for attr in attrlist if attr is not None]
LOG.debug(
'LDAP search_ext: base=%s scope=%s filterstr=%s '
'attrs=%s attrsonly=%s '
'serverctrls=%s clientctrls=%s timeout=%s sizelimit=%s',
base,
scope,
filterstr,
attrlist,
attrsonly,
serverctrls,
clientctrls,
timeout,
sizelimit,
)
return self.conn.search_ext(
base,
scope,
filterstr,
attrlist,
attrsonly,
serverctrls,
clientctrls,
timeout,
sizelimit,
)
def _paged_search_s(self, base, scope, filterstr, attrlist=None):
res = []
use_old_paging_api = False
# The API for the simple paged results control changed between
# python-ldap 2.3 and 2.4. We need to detect the capabilities
# of the python-ldap version we are using.
if hasattr(ldap, 'LDAP_CONTROL_PAGE_OID'):
use_old_paging_api = True
lc = ldap.controls.SimplePagedResultsControl(
controlType=ldap.LDAP_CONTROL_PAGE_OID,
criticality=True,
controlValue=(self.page_size, ''),
)
page_ctrl_oid = ldap.LDAP_CONTROL_PAGE_OID
else:
lc = ldap.controls.libldap.SimplePagedResultsControl(
criticality=True, size=self.page_size, cookie=''
)
page_ctrl_oid = ldap.controls.SimplePagedResultsControl.controlType
message = self.conn.search_ext(
base, scope, filterstr, attrlist, serverctrls=[lc]
)
# Endless loop request pages on ldap server until it has no data
while True:
# Request to the ldap server a page with 'page_size' entries
rtype, rdata, rmsgid, serverctrls = self.conn.result3(message)
# Receive the data
res.extend(rdata)
pctrls = [c for c in serverctrls if c.controlType == page_ctrl_oid]
if pctrls:
# LDAP server supports pagination
if use_old_paging_api:
est, cookie = pctrls[0].controlValue
lc.controlValue = (self.page_size, cookie)
else:
cookie = lc.cookie = pctrls[0].cookie
if cookie:
# There is more data still on the server
# so we request another page
message = self.conn.search_ext(
base, scope, filterstr, attrlist, serverctrls=[lc]
)
else:
# Exit condition no more data on server
break
else:
LOG.warning(
'LDAP Server does not support paging. '
'Disable paging in keystone.conf to '
'avoid this message.'
)
self._disable_paging()
break
return res
[docs]
def result3(
self, msgid=ldap.RES_ANY, all=1, timeout=None, resp_ctrl_classes=None
):
ldap_result = self.conn.result3(msgid, all, timeout, resp_ctrl_classes)
LOG.debug(
'LDAP result3: msgid=%s all=%s timeout=%s '
'resp_ctrl_classes=%s ldap_result=%s',
msgid,
all,
timeout,
resp_ctrl_classes,
ldap_result,
)
# ldap_result returned from result3 is a tuple of
# (rtype, rdata, rmsgid, serverctrls). We don't need use of these,
# except rdata.
rtype, rdata, rmsgid, serverctrls = ldap_result
py_result = convert_ldap_result(rdata)
return py_result
[docs]
def modify_s(self, dn, modlist):
ldap_modlist = [
(
op,
kind,
(
None
if values is None
else [py2ldap(x) for x in safe_iter(values)]
),
)
for op, kind, values in modlist
]
logging_modlist = [
(op, kind, (values if kind != 'userPassword' else ['****']))
for op, kind, values in ldap_modlist
]
LOG.debug('LDAP modify: dn=%s modlist=%s', dn, logging_modlist)
ldap_modlist_utf8 = [
(
op,
kind,
(
None
if values is None
else [utf8_encode(x) for x in safe_iter(values)]
),
)
for op, kind, values in ldap_modlist
]
return self.conn.modify_s(dn, ldap_modlist_utf8)
def __exit__(self, exc_type, exc_val, exc_tb):
"""Exit runtime context, unbind LDAP."""
self.unbind_s()
_HANDLERS = {}
[docs]
def register_handler(prefix, handler):
_HANDLERS[prefix] = handler
def _get_connection(conn_url, use_pool=False, use_auth_pool=False):
for prefix, handler in _HANDLERS.items():
if conn_url.startswith(prefix):
return handler()
if use_pool:
return PooledLDAPHandler(use_auth_pool=use_auth_pool)
else:
return PythonLDAPHandler()
[docs]
def filter_entity(entity_ref):
"""Filter out private items in an entity dict.
:param entity_ref: the entity dictionary. The 'dn' field will be removed.
'dn' is used in LDAP, but should not be returned to the user. This
value may be modified.
:returns: entity_ref
"""
if entity_ref:
entity_ref.pop('dn', None)
return entity_ref
[docs]
class BaseLdap:
DEFAULT_OU: str
DEFAULT_STRUCTURAL_CLASSES: list[str]
DEFAULT_ID_ATTR: str = 'cn'
DEFAULT_OBJECTCLASS: str
DEFAULT_FILTER: ty.Optional[str] = None
DEFAULT_EXTRA_ATTR_MAPPING: list[str] = []
NotFound: ty.Type[exception.Error]
notfound_arg: ty.Optional[str] = None
options_name: ty.Optional[str] = None
model: ty.Type[models.Model]
attribute_options_names: dict[str, str] = {}
immutable_attrs: list[str] = []
attribute_ignore: list[str] = []
tree_dn: ty.Optional[str] = None
def __init__(self, conf):
if conf.ldap.randomize_urls:
urls = re.split(r'[\s,]+', conf.ldap.url)
random.shuffle(urls)
self.LDAP_URL = ','.join(urls)
else:
self.LDAP_URL = conf.ldap.url
self.LDAP_USER = conf.ldap.user
self.LDAP_PASSWORD = conf.ldap.password
self.LDAP_SCOPE = ldap_scope(conf.ldap.query_scope)
self.alias_dereferencing = parse_deref(conf.ldap.alias_dereferencing)
self.page_size = conf.ldap.page_size
self.use_tls = conf.ldap.use_tls
self.tls_cacertfile = conf.ldap.tls_cacertfile
self.tls_cacertdir = conf.ldap.tls_cacertdir
self.tls_req_cert = parse_tls_cert(conf.ldap.tls_req_cert)
self.attribute_mapping = {}
self.chase_referrals = conf.ldap.chase_referrals
self.debug_level = conf.ldap.debug_level
self.conn_timeout = conf.ldap.connection_timeout
# LDAP Pool specific attribute
self.use_pool = conf.ldap.use_pool
self.pool_size = conf.ldap.pool_size
self.pool_retry_max = conf.ldap.pool_retry_max
self.pool_retry_delay = conf.ldap.pool_retry_delay
self.pool_conn_timeout = conf.ldap.pool_connection_timeout
self.pool_conn_lifetime = conf.ldap.pool_connection_lifetime
# End user authentication pool specific config attributes
self.use_auth_pool = self.use_pool and conf.ldap.use_auth_pool
self.auth_pool_size = conf.ldap.auth_pool_size
self.auth_pool_conn_lifetime = conf.ldap.auth_pool_connection_lifetime
if self.options_name is not None:
self.tree_dn = (
getattr(conf.ldap, '%s_tree_dn' % self.options_name)
or f'{self.DEFAULT_OU},{conf.ldap.suffix}'
)
idatt = '%s_id_attribute' % self.options_name
self.id_attr = getattr(conf.ldap, idatt) or self.DEFAULT_ID_ATTR
objclass = '%s_objectclass' % self.options_name
self.object_class = (
getattr(conf.ldap, objclass) or self.DEFAULT_OBJECTCLASS
)
for k, v in self.attribute_options_names.items():
v = f'{self.options_name}_{v}_attribute'
self.attribute_mapping[k] = getattr(conf.ldap, v)
attr_mapping_opt = (
'%s_additional_attribute_mapping' % self.options_name
)
attr_mapping = (
getattr(conf.ldap, attr_mapping_opt)
or self.DEFAULT_EXTRA_ATTR_MAPPING
)
self.extra_attr_mapping = self._parse_extra_attrs(attr_mapping)
ldap_filter = '%s_filter' % self.options_name
self.ldap_filter = (
getattr(conf.ldap, ldap_filter) or self.DEFAULT_FILTER
)
member_attribute = '%s_member_attribute' % self.options_name
self.member_attribute = getattr(conf.ldap, member_attribute, None)
self.structural_classes = self.DEFAULT_STRUCTURAL_CLASSES
if self.notfound_arg is None:
self.notfound_arg = self.options_name + '_id'
attribute_ignore = '%s_attribute_ignore' % self.options_name
self.attribute_ignore = getattr(conf.ldap, attribute_ignore)
def _not_found(self, object_id):
if self.NotFound is None:
return exception.NotFound(target=object_id)
else:
return self.NotFound(**{self.notfound_arg: object_id})
@staticmethod
def _parse_extra_attrs(option_list):
mapping = {}
for item in option_list:
try:
ldap_attr, attr_map = item.split(':')
except ValueError:
LOG.warning(
'Invalid additional attribute mapping: "%s". '
'Format must be <ldap_attribute>:<keystone_attribute>',
item,
)
continue
mapping[ldap_attr] = attr_map
return mapping
[docs]
def get_connection(self, user=None, password=None, end_user_auth=False):
use_pool = self.use_pool
pool_size = self.pool_size
pool_conn_lifetime = self.pool_conn_lifetime
if end_user_auth:
if not self.use_auth_pool:
use_pool = False
else:
pool_size = self.auth_pool_size
pool_conn_lifetime = self.auth_pool_conn_lifetime
conn = _get_connection(
self.LDAP_URL, use_pool, use_auth_pool=end_user_auth
)
conn = KeystoneLDAPHandler(conn=conn)
# The LDAP server may be down or a connection may not
# exist. If that is the case, the bind attempt will
# fail with a server down exception.
try:
conn.connect(
self.LDAP_URL,
page_size=self.page_size,
alias_dereferencing=self.alias_dereferencing,
use_tls=self.use_tls,
tls_cacertfile=self.tls_cacertfile,
tls_cacertdir=self.tls_cacertdir,
tls_req_cert=self.tls_req_cert,
chase_referrals=self.chase_referrals,
debug_level=self.debug_level,
conn_timeout=self.conn_timeout,
use_pool=use_pool,
pool_size=pool_size,
pool_retry_max=self.pool_retry_max,
pool_retry_delay=self.pool_retry_delay,
pool_conn_timeout=self.pool_conn_timeout,
pool_conn_lifetime=pool_conn_lifetime,
)
if user is None:
user = self.LDAP_USER
if password is None:
password = self.LDAP_PASSWORD
# not all LDAP servers require authentication, so we don't bind
# if we don't have any user/pass
if user and password:
conn.simple_bind_s(user, password)
else:
conn.simple_bind_s()
return conn
except ldap.INVALID_CREDENTIALS:
raise exception.LDAPInvalidCredentialsError()
except ldap.SERVER_DOWN:
raise exception.LDAPServerConnectionError(url=self.LDAP_URL)
def _id_to_dn_string(self, object_id):
return '{}={},{}'.format(
self.id_attr,
ldap.dn.escape_dn_chars(str(object_id)),
self.tree_dn,
)
def _id_to_dn(self, object_id):
if self.LDAP_SCOPE == ldap.SCOPE_ONELEVEL:
return self._id_to_dn_string(object_id)
with self.get_connection() as conn:
search_result = conn.search_s(
self.tree_dn,
self.LDAP_SCOPE,
'(&(%(id_attr)s=%(id)s)(objectclass=%(objclass)s))'
% {
'id_attr': self.id_attr,
'id': ldap.filter.escape_filter_chars(str(object_id)),
'objclass': self.object_class,
},
attrlist=DN_ONLY,
)
if search_result:
dn, attrs = search_result[0]
return dn
else:
return self._id_to_dn_string(object_id)
def _dn_to_id(self, dn):
# Check if the naming attribute in the DN is the same as keystone's
# configured 'id' attribute'. If so, extract the ID value from the DN
if self.id_attr == ldap.dn.str2dn(dn)[0][0][0].lower():
return ldap.dn.str2dn(dn)[0][0][1]
else:
# The 'ID' attribute is NOT in the DN, so we need to perform an
# LDAP search to look it up from the user entry itself.
with self.get_connection() as conn:
search_result = conn.search_s(dn, ldap.SCOPE_BASE)
if search_result:
try:
id_list = search_result[0][1][self.id_attr]
except KeyError:
message = (
'ID attribute %(id_attr)s not found in LDAP '
'object %(dn)s.'
) % ({'id_attr': self.id_attr, 'dn': search_result})
LOG.warning(message)
raise exception.NotFound(message=message)
if len(id_list) > 1:
message = (
'In order to keep backward compatibility, in '
'the case of multivalued ids, we are '
'returning the first id %(id_attr)s in the '
'DN.'
) % ({'id_attr': id_list[0]})
LOG.warning(message)
return id_list[0]
else:
message = _('DN attribute %(dn)s not found in LDAP') % (
{'dn': dn}
)
raise exception.NotFound(message=message)
def _ldap_res_to_model(self, res):
# LDAP attribute names may be returned in a different case than
# they are defined in the mapping, so we need to check for keys
# in a case-insensitive way. We use the case specified in the
# mapping for the model to ensure we have a predictable way of
# retrieving values later.
lower_res = {k.lower(): v for k, v in res[1].items()}
id_attrs = lower_res.get(self.id_attr.lower())
if not id_attrs:
message = _(
'ID attribute %(id_attr)s not found in LDAP object %(dn)s'
) % ({'id_attr': self.id_attr, 'dn': res[0]})
raise exception.NotFound(message=message)
if len(id_attrs) > 1:
# FIXME(gyee): if this is a multi-value attribute and it has
# multiple values, we can't use it as ID. Retain the dn_to_id
# logic here so it does not potentially break existing
# deployments. We need to fix our read-write LDAP logic so
# it does not get the ID from DN.
message = (
'ID attribute %(id_attr)s for LDAP object %(dn)s '
'has multiple values and therefore cannot be used '
'as an ID. Will get the ID from DN instead'
) % ({'id_attr': self.id_attr, 'dn': res[0]})
LOG.warning(message)
id_val = self._dn_to_id(res[0])
else:
id_val = id_attrs[0]
obj = self.model(id=id_val)
for k in obj.known_keys:
if k in self.attribute_ignore:
continue
try:
map_attr = self.attribute_mapping.get(k, k)
if map_attr is None:
# Ignore attributes that are mapped to None.
continue
v = lower_res[map_attr.lower()]
except KeyError: # nosec
# Didn't find the attr, so don't add it.
pass
else:
try:
value = v[0]
except IndexError:
value = None
# NOTE(xek): Some LDAP servers return bytes data type
# We convert it to string here, so that it is consistent with
# the other (SQL) backends.
# Bytes data type caused issues in the past, because it could
# be cached and then passed into str() method to be used as
# LDAP filters, which results in an unexpected b'...' prefix.
if isinstance(value, bytes):
try:
value = value.decode('utf-8')
except UnicodeDecodeError:
LOG.error(
"Error decoding value %r (object id %r).",
value,
res[0],
)
raise
obj[k] = value
return obj
[docs]
def affirm_unique(self, values):
if values.get('name') is not None:
try:
self.get_by_name(values['name'])
except exception.NotFound: # nosec
# Didn't find it so it's unique, good.
pass
else:
raise exception.Conflict(
type=self.options_name,
details=_('Duplicate name, %s.') % values['name'],
)
if values.get('id') is not None:
try:
self.get(values['id'])
except exception.NotFound: # nosec
# Didn't find it, so it's unique, good.
pass
else:
raise exception.Conflict(
type=self.options_name,
details=_('Duplicate ID, %s.') % values['id'],
)
[docs]
def create(self, values):
self.affirm_unique(values)
object_classes = self.structural_classes + [self.object_class]
attrs = [('objectClass', object_classes)]
for k, v in values.items():
if k in self.attribute_ignore:
continue
if k == 'id':
# no need to check if v is None as 'id' will always have
# a value
attrs.append((self.id_attr, [v]))
elif v is not None:
attr_type = self.attribute_mapping.get(k, k)
if attr_type is not None:
attrs.append((attr_type, [v]))
extra_attrs = [
attr
for attr, name in self.extra_attr_mapping.items()
if name == k
]
for attr in extra_attrs:
attrs.append((attr, [v]))
with self.get_connection() as conn:
conn.add_s(self._id_to_dn(values['id']), attrs)
return values
# NOTE(prashkre): Filter ldap search results on an attribute to ensure
# that attribute has a value set on ldap. This keeps keystone away
# from entities that don't have attribute value set on ldap.
# for e.g. In ldap configuration, if user_name_attribute = personName
# then it will ignore ldap users who don't have 'personName' attribute
# value set on user.
def _filter_ldap_result_by_attr(self, ldap_result, ldap_attr_name):
attr = self.attribute_mapping[ldap_attr_name]
# To ensure that ldap attribute value is not empty in ldap config.
if not attr:
attr_name = '{}_{}_attribute'.format(
self.options_name,
self.attribute_options_names[ldap_attr_name],
)
raise ValueError(
'"%(attr)s" is not a valid value for'
' "%(attr_name)s"' % {'attr': attr, 'attr_name': attr_name}
)
# consider attr = "cn" and
# ldap_result = [{'uid': ['fake_id1']}, , 'cN': ["name"]}]
# doing lower case on both user_name_attribute and ldap users
# attribute
result = []
# consider attr = "cn" and
# ldap_result = [(u'cn=fake1,o=ex_domain', {'uid': ['fake_id1']}),
# (u'cn=fake2,o=ex_domain', {'uid': ['fake_id2'],
# 'cn': [' ']}),
# (u'cn=fake3,o=ex_domain', {'uid': ['fake_id3'],
# 'cn': ['']}),
# (u'cn=fake4,o=ex_domain', {'uid': ['fake_id4'],
# 'cn': []}),
# (u'cn=fake5,o=ex_domain', {'uid': ['fake_id5'],
# 'cn': ["name"]})]
for obj in ldap_result:
# ignore ldap object(user/group entry) which has no attr set
# in it or whose value is empty list.
ldap_res_low_keys_dict = {k.lower(): v for k, v in obj[1].items()}
result_attr_vals = ldap_res_low_keys_dict.get(attr.lower())
# ignore ldap object whose attr value has empty strings or
# contains only whitespaces.
if result_attr_vals:
if result_attr_vals[0] and result_attr_vals[0].strip():
result.append(obj)
# except {'uid': ['fake_id5'], 'cn': ["name"]}, all entries
# will be ignored in ldap_result
return result
def _ldap_get(self, object_id, ldap_filter=None):
query = (
'(&(%(id_attr)s=%(id)s)'
'%(filter)s'
'(objectClass=%(object_class)s))'
% {
'id_attr': self.id_attr,
'id': ldap.filter.escape_filter_chars(str(object_id)),
'filter': (ldap_filter or self.ldap_filter or ''),
'object_class': self.object_class,
}
)
with self.get_connection() as conn:
try:
attrs = list(
set(
[self.id_attr]
+ list(self.attribute_mapping.values())
+ list(self.extra_attr_mapping.keys())
)
)
res = conn.search_s(
self.tree_dn, self.LDAP_SCOPE, query, attrs
)
except ldap.NO_SUCH_OBJECT:
return None
# TODO(prashkre): add functional testing for missing name attibute
# on ldap entities.
# NOTE(prashkre): Filter ldap search result to keep keystone away from
# entities that don't have names. We can also do the same by appending
# a condition '(!(!(self.attribute_mapping.get('name')=*))' to ldap
# search query but the repsonse time of the query is pretty slow when
# compared to explicit filtering by 'name' through ldap result.
try:
return self._filter_ldap_result_by_attr(res[:1], 'name')[0]
except IndexError:
return None
def _ldap_get_limited(self, base, scope, filterstr, attrlist, sizelimit):
with self.get_connection() as conn:
try:
control = ldap.controls.libldap.SimplePagedResultsControl(
criticality=True, size=sizelimit, cookie=''
)
msgid = conn.search_ext(
base, scope, filterstr, attrlist, serverctrls=[control]
)
rdata = conn.result3(msgid)
return rdata
except ldap.NO_SUCH_OBJECT:
return []
@driver_hints.truncated
def _ldap_get_all(self, hints, ldap_filter=None):
query = '(&{}(objectClass={})({}=*))'.format(
ldap_filter or self.ldap_filter or '',
self.object_class,
self.id_attr,
)
sizelimit = 0
attrs = list(
set(
[self.id_attr]
+ list(self.attribute_mapping.values())
+ list(self.extra_attr_mapping.keys())
)
)
if hints.limit:
sizelimit = hints.limit['limit']
res = self._ldap_get_limited(
self.tree_dn, self.LDAP_SCOPE, query, attrs, sizelimit
)
else:
with self.get_connection() as conn:
try:
res = conn.search_s(
self.tree_dn, self.LDAP_SCOPE, query, attrs
)
except ldap.NO_SUCH_OBJECT:
return []
# TODO(prashkre): add functional testing for missing name attribute
# on ldap entities.
# NOTE(prashkre): Filter ldap search result to keep keystone away from
# entities that don't have names. We can also do the same by appending
# a condition '(!(!(self.attribute_mapping.get('name')=*))' to ldap
# search query but the repsonse time of the query is pretty slow when
# compared to explicit filtering by 'name' through ldap result.
return self._filter_ldap_result_by_attr(res, 'name')
def _ldap_get_list(
self, search_base, scope, query_params=None, attrlist=None
):
query = '(objectClass=%s)' % self.object_class
if query_params:
def calc_filter(attrname, value):
val_esc = ldap.filter.escape_filter_chars(value)
return f'({attrname}={val_esc})'
query = '(&{}{})'.format(
query,
''.join([calc_filter(k, v) for k, v in query_params.items()]),
)
with self.get_connection() as conn:
return conn.search_s(search_base, scope, query, attrlist)
[docs]
def get(self, object_id, ldap_filter=None):
res = self._ldap_get(object_id, ldap_filter)
if res is None:
raise self._not_found(object_id)
else:
return self._ldap_res_to_model(res)
[docs]
def get_by_name(self, name, ldap_filter=None):
query = '({}={})'.format(
self.attribute_mapping['name'],
ldap.filter.escape_filter_chars(str(name)),
)
res = self.get_all(query)
try:
return res[0]
except IndexError:
raise self._not_found(name)
[docs]
def get_all(self, ldap_filter=None, hints=None):
hints = hints or driver_hints.Hints()
return [
self._ldap_res_to_model(x)
for x in self._ldap_get_all(hints, ldap_filter)
]
[docs]
def update(self, object_id, values, old_obj=None):
if old_obj is None:
old_obj = self.get(object_id)
modlist = []
for k, v in values.items():
if k == 'id':
# id can't be modified.
continue
if k in self.attribute_ignore:
# Handle 'enabled' specially since can't disable if ignored.
if k == 'enabled' and (not v):
action = _(
"Disabling an entity where the 'enable' "
"attribute is ignored by configuration."
)
raise exception.ForbiddenAction(action=action)
continue
# attribute value has not changed
if k in old_obj and old_obj[k] == v:
continue
if k in self.immutable_attrs:
msg = _("Cannot change %(option_name)s %(attr)s") % {
'option_name': self.options_name,
'attr': k,
}
raise exception.ValidationError(msg)
if v is None:
if old_obj.get(k) is not None:
modlist.append(
(
ldap.MOD_DELETE,
self.attribute_mapping.get(k, k),
None,
)
)
continue
current_value = old_obj.get(k)
if current_value is None:
op = ldap.MOD_ADD
modlist.append((op, self.attribute_mapping.get(k, k), [v]))
elif current_value != v:
op = ldap.MOD_REPLACE
modlist.append((op, self.attribute_mapping.get(k, k), [v]))
if modlist:
with self.get_connection() as conn:
try:
conn.modify_s(self._id_to_dn(object_id), modlist)
except ldap.NO_SUCH_OBJECT:
raise self._not_found(object_id)
return self.get(object_id)
[docs]
def add_member(self, member_dn, member_list_dn):
"""Add member to the member list.
:param member_dn: DN of member to be added.
:param member_list_dn: DN of group to which the
member will be added.
:raises keystone.exception.Conflict: If the user was already a member.
:raises self.NotFound: If the group entry didn't exist.
"""
with self.get_connection() as conn:
try:
mod = (ldap.MOD_ADD, self.member_attribute, member_dn)
conn.modify_s(member_list_dn, [mod])
except ldap.TYPE_OR_VALUE_EXISTS:
raise exception.Conflict(
_(
'Member %(member)s '
'is already a member'
' of group %(group)s'
)
% {'member': member_dn, 'group': member_list_dn}
)
except ldap.NO_SUCH_OBJECT:
raise self._not_found(member_list_dn)
[docs]
def filter_query(self, hints, query=None):
"""Apply filtering to a query.
:param hints: contains the list of filters, which may be None,
indicating that there are no filters to be applied.
If it's not None, then any filters satisfied here will be
removed so that the caller will know if any filters
remain to be applied.
:param query: LDAP query into which to include filters
:returns query: LDAP query, updated with any filters satisfied
"""
def build_filter(filter_):
"""Build a filter for the query.
:param filter_: the dict that describes this filter
:returns query: LDAP query term to be added
"""
ldap_attr = self.attribute_mapping[filter_['name']]
val_esc = ldap.filter.escape_filter_chars(filter_['value'])
if filter_['case_sensitive']:
# NOTE(henry-nash): Although dependent on the schema being
# used, most LDAP attributes are configured with case
# insensitive matching rules, so we'll leave this to the
# controller to filter.
return
if filter_['name'] == 'enabled':
# NOTE(henry-nash): Due to the different options for storing
# the enabled attribute (e,g, emulated or not), for now we
# don't try and filter this at the driver level - we simply
# leave the filter to be handled by the controller. It seems
# unlikley that this will cause a signifcant performance
# issue.
return
# TODO(henry-nash): Currently there are no booleans (other than
# 'enabled' that is handled above) on which you can filter. If
# there were, we would need to add special handling here to
# convert the booleans values to 'TRUE' and 'FALSE'. To do that
# we would also need to know which filter keys were actually
# booleans (this is related to bug #1411478).
if filter_['comparator'] == 'equals':
query_term = '({attr}={val})'.format(
attr=ldap_attr,
val=val_esc,
)
elif filter_['comparator'] == 'contains':
query_term = '({attr}=*{val}*)'.format(
attr=ldap_attr,
val=val_esc,
)
elif filter_['comparator'] == 'startswith':
query_term = '({attr}={val}*)'.format(
attr=ldap_attr,
val=val_esc,
)
elif filter_['comparator'] == 'endswith':
query_term = '({attr}=*{val})'.format(
attr=ldap_attr,
val=val_esc,
)
else:
# It's a filter we don't understand, so let the caller
# work out if they need to do something with it.
return
return query_term
if query is None:
# make sure query is a string so the ldap filter is properly
# constructed from filter_list later
query = ''
if hints is None:
return query
filter_list = []
satisfied_filters = []
for filter_ in hints.filters:
if filter_['name'] not in self.attribute_mapping:
continue
new_filter = build_filter(filter_)
if new_filter is not None:
filter_list.append(new_filter)
satisfied_filters.append(filter_)
if filter_list:
query = '(&{}{})'.format(query, ''.join(filter_list))
# Remove satisfied filters, then the caller will know remaining filters
for filter_ in satisfied_filters:
hints.filters.remove(filter_)
return query
[docs]
class EnabledEmuMixIn(BaseLdap):
"""Emulates boolean 'enabled' attribute if turned on.
Creates a group holding all enabled objects of this class, all missing
objects are considered disabled.
Options:
* $name_enabled_emulation - boolean, on/off
* $name_enabled_emulation_dn - DN of that group, default is
cn=enabled_${name}s,${tree_dn}
* $name_enabled_emulation_use_group_config - boolean, on/off
Where ${name}s is the plural of self.options_name ('users' or 'tenants'),
${tree_dn} is self.tree_dn.
"""
DEFAULT_GROUP_OBJECTCLASS = 'groupOfNames'
DEFAULT_MEMBER_ATTRIBUTE = 'member'
DEFAULT_GROUP_MEMBERS_ARE_IDS = False
def __init__(self, conf):
super().__init__(conf)
enabled_emulation = '%s_enabled_emulation' % self.options_name
self.enabled_emulation = getattr(conf.ldap, enabled_emulation)
enabled_emulation_dn = '%s_enabled_emulation_dn' % self.options_name
self.enabled_emulation_dn = getattr(conf.ldap, enabled_emulation_dn)
use_group_config = (
'%s_enabled_emulation_use_group_config' % self.options_name
)
self.use_group_config = getattr(conf.ldap, use_group_config)
if not self.use_group_config:
self.member_attribute = self.DEFAULT_MEMBER_ATTRIBUTE
self.group_objectclass = self.DEFAULT_GROUP_OBJECTCLASS
self.group_members_are_ids = self.DEFAULT_GROUP_MEMBERS_ARE_IDS
else:
self.member_attribute = conf.ldap.group_member_attribute
self.group_objectclass = conf.ldap.group_objectclass
self.group_members_are_ids = conf.ldap.group_members_are_ids
if not self.enabled_emulation_dn:
naming_attr_name = 'cn'
naming_attr_value = 'enabled_%ss' % self.options_name
sub_vals = (naming_attr_name, naming_attr_value, self.tree_dn)
self.enabled_emulation_dn = '%s=%s,%s' % sub_vals
naming_attr = (naming_attr_name, [naming_attr_value])
else:
# Extract the attribute name and value from the configured DN.
naming_dn = ldap.dn.str2dn(self.enabled_emulation_dn)
naming_rdn = naming_dn[0][0]
naming_attr = (naming_rdn[0], naming_rdn[1])
self.enabled_emulation_naming_attr = naming_attr
def _id_to_member_attribute_value(self, object_id):
"""Convert id to value expected by member_attribute."""
if self.group_members_are_ids:
return object_id
return self._id_to_dn(object_id)
def _is_id_enabled(self, object_id, conn):
member_attr_val = self._id_to_member_attribute_value(object_id)
return self._is_member_enabled(member_attr_val, conn)
def _is_member_enabled(self, member_attr_val, conn):
query = '({}={})'.format(
self.member_attribute,
ldap.filter.escape_filter_chars(member_attr_val),
)
try:
enabled_value = conn.search_s(
self.enabled_emulation_dn,
ldap.SCOPE_BASE,
query,
attrlist=DN_ONLY,
)
except ldap.NO_SUCH_OBJECT:
return False
else:
return bool(enabled_value)
def _add_enabled(self, object_id):
member_attr_val = self._id_to_member_attribute_value(object_id)
with self.get_connection() as conn:
if not self._is_member_enabled(member_attr_val, conn):
modlist = [
(ldap.MOD_ADD, self.member_attribute, [member_attr_val])
]
try:
conn.modify_s(self.enabled_emulation_dn, modlist)
except ldap.NO_SUCH_OBJECT:
attr_list = [
('objectClass', [self.group_objectclass]),
(self.member_attribute, [member_attr_val]),
self.enabled_emulation_naming_attr,
]
conn.add_s(self.enabled_emulation_dn, attr_list)
def _remove_enabled(self, object_id):
member_attr_val = self._id_to_member_attribute_value(object_id)
modlist = [(ldap.MOD_DELETE, self.member_attribute, [member_attr_val])]
with self.get_connection() as conn:
try:
conn.modify_s(self.enabled_emulation_dn, modlist)
except (ldap.NO_SUCH_OBJECT, ldap.NO_SUCH_ATTRIBUTE): # nosec
# It's already gone, good.
pass
[docs]
def create(self, values):
if self.enabled_emulation:
enabled_value = values.pop('enabled', True)
ref = super().create(values)
if 'enabled' not in self.attribute_ignore:
if enabled_value:
self._add_enabled(ref['id'])
ref['enabled'] = enabled_value
return ref
else:
return super().create(values)
[docs]
def get(self, object_id, ldap_filter=None):
with self.get_connection() as conn:
ref = super().get(object_id, ldap_filter)
if (
'enabled' not in self.attribute_ignore
and self.enabled_emulation
):
ref['enabled'] = self._is_id_enabled(object_id, conn)
return ref
[docs]
def get_all(self, ldap_filter=None, hints=None):
hints = hints or driver_hints.Hints()
if 'enabled' not in self.attribute_ignore and self.enabled_emulation:
# had to copy BaseLdap.get_all here to ldap_filter by DN
obj_list = [
self._ldap_res_to_model(x)
for x in self._ldap_get_all(hints, ldap_filter)
if x[0] != self.enabled_emulation_dn
]
with self.get_connection() as conn:
for obj_ref in obj_list:
obj_ref['enabled'] = self._is_id_enabled(
obj_ref['id'], conn
)
return obj_list
else:
return super().get_all(ldap_filter, hints)
[docs]
def update(self, object_id, values, old_obj=None):
if 'enabled' not in self.attribute_ignore and self.enabled_emulation:
data = values.copy()
enabled_value = data.pop('enabled', None)
ref = super().update(object_id, data, old_obj)
if enabled_value is not None:
if enabled_value:
self._add_enabled(object_id)
else:
self._remove_enabled(object_id)
ref['enabled'] = enabled_value
return ref
else:
return super().update(object_id, values, old_obj)
