如何使用Python连接ldap
好多使用ldap认证的软件都是Python的，比如superset和airflow， 好吧，他们都是airbnb家的。在配置ldap的时候可能会出现认证失败，你不知道是因为什么导致配置失败的。所以，就要
跟踪源码，看看内部怎么认证实现的。
ldap介绍和使用安装参见： https://www.cnblogs.com/woshimrf/p/ldap.html
具体来实现如下：
为了模拟环境，我们使用docker-python。基于Debian Python3: https://github.com/Ryan-Miao/docker-china-source/tree/master/docker-python
启动
docker run -it ryan/python:3 /bin/bash
下载ldap3
pip install ldap3
测试连接
root@5edee218d962:/# python
Python 3.7.4 (default, Jul 13 2019, 14:20:24)
[GCC 6.3.0 20170516] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from ldap3 import Server, Connection, ALL, NTLM, ALL_ATTRIBUTES, LEVEL, SUBTREE
>>> server = Server('172.17.0.2', get_info=ALL)
>>> conn = Connection(server, 'cn=admin,dc=demo,dc=com', 'admin', auto_bind=True)
>>> conn.extend.standard.who_am_i()
'dn:cn=admin,dc=demo,dc=com'
测试登录部分
登录源码如下：
@staticmethod
def try_login(username, password):
conn = get_ldap_connection(configuration.conf.get("ldap", "bind_user"),
configuration.conf.get("ldap", "bind_password"))
search_filter = "(&({0})({1}={2}))".format(
configuration.conf.get("ldap", "user_filter"),
configuration.conf.get("ldap", "user_name_attr"),
username
)
search_scope = LEVEL
if configuration.conf.has_option("ldap", "search_scope"):
if configuration.conf.get("ldap", "search_scope") == "SUBTREE":
search_scope = SUBTREE
else:
search_scope = LEVEL
# todo: BASE or ONELEVEL?
res = conn.search(configuration.conf.get("ldap", "basedn"), search_filter, search_scope=search_scope)
# todo: use list or result?
if not res:
log.info("Cannot find user %s", username)
raise AuthenticationError("Invalid username or password")
entry = conn.response[0]
conn.unbind()
if 'dn' not in entry:
# The search filter for the user did not return any values, so an
# invalid user was used for credentials.
raise AuthenticationError("Invalid username or password")
try:
conn = get_ldap_connection(entry['dn'], password)
except KeyError:
log.error("""
Unable to parse LDAP structure. If you're using Active Directory
and not specifying an OU, you must set search_scope=SUBTREE in airflow.cfg.
%s
""", traceback.format_exc())
raise LdapException(
"Could not parse LDAP structure. "
"Try setting search_scope in airflow.cfg, or check logs"
)
if not conn:
log.info("Password incorrect for user %s", username)
raise AuthenticationError("Invalid username or password")
第一步： 获取连接
from ldap3 import Server, Connection, ALL, NTLM, ALL_ATTRIBUTES, LEVEL, SUBTREE
server = Server('172.17.0.2:389', get_info=ALL)
conn = Connection(server, 'cn=admin,dc=demo,dc=com', 'admin', auto_bind=True)
conn.extend.standard.who_am_i()
第二步： 根据filter search用户。 这里我们的配置文件如下：
[ldap]
# set this to ldaps://<your.ldap.server>:<port>
uri = ldap://172.17.0.2:389
user_filter = objectClass=inetOrgPerson
user_name_attr = sn
group_member_attr = memberOf
superuser_filter =
data_profiler_filter =
bind_user = cn=admin,dc=demo,dc=com
bind_password = admin
basedn = dc=demo,dc=com
cacert =
search_scope = SUBTREE
源码就是拼接filter， 最后变成(&(objectClass=inetOrgPerson)(sn=ryanmiao)), 然后search scope.
>>> with Connection(server, 'cn=admin,dc=demo,dc=com', 'admin') as conn:
... conn.search('dc=demo,dc=com', '(&(objectClass=inetOrgPerson)(cn=hr-ryan))', search_scope=SUBTREE)
... entry = conn.entries[0]
...
True
>>> entry
DN: cn=hr-ryan,ou=HR,ou=People,dc=demo,dc=com - STATUS: Read - READ TIME: 2019-08-19T12:58:34.966181
第三步： 从上一步得到dn，然后根据用户输入的密码，再次连接, 不抛异常就证明密码正确
>>> conn = Connection(server, 'cn=hr-ryan,ou=HR,ou=People,dc=demo,dc=com', '123456', auto_bind=True)
>>>
>>> conn = Connection(server, 'cn=hr-ryan,ou=HR,ou=People,dc=demo,dc=com', '1234567', auto_bind=True)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/local/lib/python3.7/site-packages/ldap3/core/connection.py", line 325, in __init__
self.do_auto_bind()
File "/usr/local/lib/python3.7/site-packages/ldap3/core/connection.py", line 353, in do_auto_bind
raise LDAPBindError(self.last_error)
ldap3.core.exceptions.LDAPBindError: automatic bind not successful - invalidCredentials
测试分组
ldap提供了分组，配置见前文。我们采用ldap统一登录之后，还要把用户放入不同的group里来区分权限。比如splunk-users, splunk-admin, gitlab-admin, gitlab-user等。
根据用户名和密码，我们实现了用户登录密码验证，接下来需要取到用户所属于的group。
源码如下：
def groups_user(conn, search_base, user_filter, user_name_att, username):
search_filter = "(&({0})({1}={2}))".format(user_filter, user_name_att, username)
try:
memberof_attr = configuration.conf.get("ldap", "group_member_attr")
except Exception:
memberof_attr = "memberOf"
res = conn.search(search_base, search_filter, attributes=[memberof_attr])
if not res:
log.info("Cannot find user %s", username)
raise AuthenticationError("Invalid username or password")
if conn.response and memberof_attr not in conn.response[0]["attributes"]:
log.warning("""Missing attribute "%s" when looked-up in Ldap database.
The user does not seem to be a member of a group and therefore won't see any dag
if the option filter_by_owner=True and owner_mode=ldapgroup are set""",
memberof_attr)
return []
user_groups = conn.response[0]["attributes"][memberof_attr]
regex = re.compile("cn=([^,]*).*", re.IGNORECASE)
groups_list = []
try:
groups_list = [regex.search(i).group(1) for i in user_groups]
except IndexError:
log.warning("Parsing error when retrieving the user's group(s)."
" Check if the user belongs to at least one group"
" or if the user's groups name do not contain special characters")
return groups_list
同样，第一步，连接，第二步search，但需要返回字段memberof。注意我们前面配置了group_member_attr=memberof
>>> with Connection(server, 'cn=admin,dc=demo,dc=com', 'admin') as conn:
... conn.search('dc=demo,dc=com', '(&(objectClass=inetOrgPerson)(sn=hr-ryan))', attributes=['memberof'])
... entry = conn.entries[0]
... entry
...
True
DN: cn=hr-ryan,ou=HR,ou=People,dc=demo,dc=com - STATUS: Read - READ TIME: 2019-08-19T13:05:20.592805
memberOf: cn=g-admin,ou=Group,dc=demo,dc=com
cn=g-users,ou=Group,dc=demo,dc=com
cn=g-a,ou=Group,dc=demo,dc=com
如果取不到group会报错。
以上就差不多是airflow的ldap配置原理了。其他雷同，不一样的地方也许是在filter的地方，我们找对应软件的源码look一下就ok了。