keystone/keystone/resource/backends/sql.py

409 lines
17 KiB
Python

# 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.
from oslo_log import log
from six import text_type
from sqlalchemy import orm
from sqlalchemy.sql import expression
from keystone.common import driver_hints
from keystone.common import sql
from keystone import exception
from keystone.resource.backends import base
LOG = log.getLogger(__name__)
class Resource(base.ResourceDriverBase):
# TODO(morgan): Merge all of this code into the manager, Resource backend
# is only SQL. There is no configurable driver.
def _encode_domain_id(self, ref):
if 'domain_id' in ref and ref['domain_id'] is None:
new_ref = ref.copy()
new_ref['domain_id'] = base.NULL_DOMAIN_ID
return new_ref
else:
return ref
def _is_hidden_ref(self, ref):
return ref.id == base.NULL_DOMAIN_ID
def _get_project(self, session, project_id):
project_ref = session.query(Project).get(project_id)
if project_ref is None or self._is_hidden_ref(project_ref):
raise exception.ProjectNotFound(project_id=project_id)
return project_ref
def get_project(self, project_id):
with sql.session_for_read() as session:
return self._get_project(session, project_id).to_dict()
def get_project_by_name(self, project_name, domain_id):
with sql.session_for_read() as session:
query = session.query(Project)
query = query.filter_by(name=project_name)
if domain_id is None:
query = query.filter_by(
domain_id=base.NULL_DOMAIN_ID)
else:
query = query.filter_by(domain_id=domain_id)
try:
project_ref = query.one()
except sql.NotFound:
raise exception.ProjectNotFound(project_id=project_name)
if self._is_hidden_ref(project_ref):
raise exception.ProjectNotFound(project_id=project_name)
return project_ref.to_dict()
@driver_hints.truncated
def list_projects(self, hints):
# If there is a filter on domain_id and the value is None, then to
# ensure that the sql filtering works correctly, we need to patch
# the value to be NULL_DOMAIN_ID. This is safe to do here since we
# know we are able to satisfy any filter of this type in the call to
# filter_limit_query() below, which will remove the filter from the
# hints (hence ensuring our substitution is not exposed to the caller).
for f in hints.filters:
if (f['name'] == 'domain_id' and f['value'] is None):
f['value'] = base.NULL_DOMAIN_ID
with sql.session_for_read() as session:
query = session.query(Project)
query = query.filter(Project.id != base.NULL_DOMAIN_ID)
project_refs = sql.filter_limit_query(Project, query, hints)
return [project_ref.to_dict() for project_ref in project_refs]
def list_projects_from_ids(self, ids):
if not ids:
return []
else:
with sql.session_for_read() as session:
query = session.query(Project)
query = query.filter(Project.id.in_(ids))
return [project_ref.to_dict() for project_ref in query.all()
if not self._is_hidden_ref(project_ref)]
def list_project_ids_from_domain_ids(self, domain_ids):
if not domain_ids:
return []
else:
with sql.session_for_read() as session:
query = session.query(Project.id)
query = (
query.filter(Project.domain_id.in_(domain_ids)))
return [x.id for x in query.all()
if not self._is_hidden_ref(x)]
def list_projects_in_domain(self, domain_id):
with sql.session_for_read() as session:
try:
self._get_project(session, domain_id)
except exception.ProjectNotFound:
raise exception.DomainNotFound(domain_id=domain_id)
query = session.query(Project)
project_refs = query.filter(Project.domain_id == domain_id)
return [project_ref.to_dict() for project_ref in project_refs]
def list_projects_acting_as_domain(self, hints):
hints.add_filter('is_domain', True)
return self.list_projects(hints)
def _get_children(self, session, project_ids, domain_id=None):
query = session.query(Project)
query = query.filter(Project.parent_id.in_(project_ids))
project_refs = query.all()
return [project_ref.to_dict() for project_ref in project_refs]
def list_projects_in_subtree(self, project_id):
with sql.session_for_read() as session:
children = self._get_children(session, [project_id])
subtree = []
examined = set([project_id])
while children:
children_ids = set()
for ref in children:
if ref['id'] in examined:
msg = ('Circular reference or a repeated '
'entry found in projects hierarchy - '
'%(project_id)s.')
LOG.error(msg, {'project_id': ref['id']})
return
children_ids.add(ref['id'])
examined.update(children_ids)
subtree += children
children = self._get_children(session, children_ids)
return subtree
def list_project_parents(self, project_id):
with sql.session_for_read() as session:
project = self._get_project(session, project_id).to_dict()
parents = []
examined = set()
while project.get('parent_id') is not None:
if project['id'] in examined:
msg = ('Circular reference or a repeated '
'entry found in projects hierarchy - '
'%(project_id)s.')
LOG.error(msg, {'project_id': project['id']})
return
examined.add(project['id'])
parent_project = self._get_project(
session, project['parent_id']).to_dict()
parents.append(parent_project)
project = parent_project
return parents
def is_leaf_project(self, project_id):
with sql.session_for_read() as session:
project_refs = self._get_children(session, [project_id])
return not project_refs
def list_projects_by_tags(self, filters):
filtered_ids = []
with sql.session_for_read() as session:
query = session.query(ProjectTag)
if 'tags' in filters.keys():
filtered_ids += self._filter_ids_by_tags(
query, filters['tags'].split(','))
if 'tags-any' in filters.keys():
any_tags = filters['tags-any'].split(',')
subq = query.filter(ProjectTag.name.in_(any_tags))
any_tags = [ptag['project_id'] for ptag in subq]
if 'tags' in filters.keys():
any_tags = set(any_tags) & set(filtered_ids)
filtered_ids = any_tags
if 'not-tags' in filters.keys():
blacklist_ids = self._filter_ids_by_tags(
query, filters['not-tags'].split(','))
filtered_ids = self._filter_not_tags(session,
filtered_ids,
blacklist_ids)
if 'not-tags-any' in filters.keys():
any_tags = filters['not-tags-any'].split(',')
subq = query.filter(ProjectTag.name.in_(any_tags))
blacklist_ids = [ptag['project_id'] for ptag in subq]
if 'not-tags' in filters.keys():
filtered_ids += blacklist_ids
else:
filtered_ids = self._filter_not_tags(session,
filtered_ids,
blacklist_ids)
if not filtered_ids:
return []
query = session.query(Project)
query = query.filter(Project.id.in_(filtered_ids))
return [project_ref.to_dict() for project_ref in query.all()
if not self._is_hidden_ref(project_ref)]
def _filter_ids_by_tags(self, query, tags):
filtered_ids = []
subq = query.filter(ProjectTag.name.in_(tags))
for ptag in subq:
subq_tags = query.filter(ProjectTag.project_id ==
ptag['project_id'])
result = map(lambda x: x['name'], subq_tags.all())
if set(tags) <= set(result):
filtered_ids.append(ptag['project_id'])
return filtered_ids
def _filter_not_tags(self, session, filtered_ids, blacklist_ids):
subq = session.query(Project)
valid_ids = [q['id'] for q in subq if q['id'] not in blacklist_ids]
if filtered_ids:
valid_ids = list(set(valid_ids) & set(filtered_ids))
return valid_ids
# CRUD
@sql.handle_conflicts(conflict_type='project')
def create_project(self, project_id, project):
new_project = self._encode_domain_id(project)
with sql.session_for_write() as session:
project_ref = Project.from_dict(new_project)
session.add(project_ref)
return project_ref.to_dict()
@sql.handle_conflicts(conflict_type='project')
def update_project(self, project_id, project):
update_project = self._encode_domain_id(project)
with sql.session_for_write() as session:
project_ref = self._get_project(session, project_id)
old_project_dict = project_ref.to_dict()
for k in update_project:
old_project_dict[k] = update_project[k]
# When we read the old_project_dict, any "null" domain_id will have
# been decoded, so we need to re-encode it
old_project_dict = self._encode_domain_id(old_project_dict)
new_project = Project.from_dict(old_project_dict)
for attr in Project.attributes:
if attr != 'id':
setattr(project_ref, attr, getattr(new_project, attr))
project_ref.extra = new_project.extra
return project_ref.to_dict(include_extra_dict=True)
@sql.handle_conflicts(conflict_type='project')
def delete_project(self, project_id):
with sql.session_for_write() as session:
project_ref = self._get_project(session, project_id)
session.delete(project_ref)
@sql.handle_conflicts(conflict_type='project')
def delete_projects_from_ids(self, project_ids):
if not project_ids:
return
with sql.session_for_write() as session:
query = session.query(Project).filter(Project.id.in_(
project_ids))
project_ids_from_bd = [p['id'] for p in query.all()]
for project_id in project_ids:
if (project_id not in project_ids_from_bd or
project_id == base.NULL_DOMAIN_ID):
LOG.warning('Project %s does not exist and was not '
'deleted.', project_id)
query.delete(synchronize_session=False)
def check_project_depth(self, max_depth):
with sql.session_for_read() as session:
obj_list = []
# Using db table self outerjoin to find the project descendants.
#
# We'll only outerjoin the project table `max_depth` times to
# check whether current project tree exceed the max depth limit.
#
# For example:
#
# If max_depth is 2, we will take the outerjoin 2 times, then the
# SQL result may be like:
#
# +---- +-------------+-------------+-------------+
# | No. | project1_id | project2_id | project3_id |
# +--- -+-------------+-------------+-------------+
# | 1 | domain_x | | |
# +- ---+-------------+-------------+-------------+
# | 2 | project_a | | |
# +- ---+-------------+-------------+-------------+
# | 3 | domain_y | project_a | |
# +- ---+-------------+-------------+-------------+
# | 4 | project_b | project_c | |
# +- ---+-------------+-------------+-------------+
# | 5 | domain_y | project_b | project_c |
# +- ---+-------------+-------------+-------------+
#
# `project1_id` column is the root. It is a project or a domain.
# If `project1_id` is a project, there must exist a line that
# `project1` is its domain.
#
# We got 5 lines here. It includes three scenarios:
#
# 1). The No.1 line means there is a domain `domain_x` which has no
# children. The depth is 1.
#
# 2). The No.2 and No.3 lines mean project `project_a` has no child
# and its parent is domain `domain_y`. The depth is 2.
#
# 3). The No.4 and No.5 lines mean project `project_b` has a child
# `project_c` and its parent is domain `domain_y`. The depth is
# 3. This tree hit the max depth
#
# So we can see that if column "project3_id" has value, it means
# some trees hit the max depth limit.
for _ in range(max_depth + 1):
obj_list.append(orm.aliased(Project))
query = session.query(*obj_list)
for index in range(max_depth):
query = query.outerjoin(
obj_list[index + 1],
obj_list[index].id == obj_list[index + 1].parent_id)
exceeded_lines = query.filter(
obj_list[-1].id != expression.null())
if exceeded_lines:
return [line[max_depth].id for line in exceeded_lines]
class Project(sql.ModelBase, sql.ModelDictMixinWithExtras):
# NOTE(henry-nash): From the manager and above perspective, the domain_id
# is nullable. However, to ensure uniqueness in multi-process
# configurations, it is better to still use the sql uniqueness constraint.
# Since the support for a nullable component of a uniqueness constraint
# across different sql databases is mixed, we instead store a special value
# to represent null, as defined in NULL_DOMAIN_ID above.
def to_dict(self, include_extra_dict=False):
d = super(Project, self).to_dict(
include_extra_dict=include_extra_dict)
if d['domain_id'] == base.NULL_DOMAIN_ID:
d['domain_id'] = None
return d
__tablename__ = 'project'
attributes = ['id', 'name', 'domain_id', 'description', 'enabled',
'parent_id', 'is_domain', 'tags']
id = sql.Column(sql.String(64), primary_key=True)
name = sql.Column(sql.String(64), nullable=False)
domain_id = sql.Column(sql.String(64), sql.ForeignKey('project.id'),
nullable=False)
description = sql.Column(sql.Text())
enabled = sql.Column(sql.Boolean)
extra = sql.Column(sql.JsonBlob())
parent_id = sql.Column(sql.String(64), sql.ForeignKey('project.id'))
is_domain = sql.Column(sql.Boolean, default=False, nullable=False,
server_default='0')
_tags = orm.relationship(
'ProjectTag',
single_parent=True,
lazy='subquery',
cascade='all,delete-orphan',
backref='project',
primaryjoin='and_(ProjectTag.project_id==Project.id)'
)
# Unique constraint across two columns to create the separation
# rather than just only 'name' being unique
__table_args__ = (sql.UniqueConstraint('domain_id', 'name'),)
@property
def tags(self):
if self._tags:
return [tag.name for tag in self._tags]
return []
@tags.setter
def tags(self, values):
new_tags = []
for tag in values:
tag_ref = ProjectTag()
tag_ref.project_id = self.id
tag_ref.name = text_type(tag)
new_tags.append(tag_ref)
self._tags = new_tags
class ProjectTag(sql.ModelBase, sql.ModelDictMixin):
def to_dict(self):
d = super(ProjectTag, self).to_dict()
return d
__tablename__ = 'project_tag'
attributes = ['project_id', 'name']
project_id = sql.Column(
sql.String(64), sql.ForeignKey('project.id', ondelete='CASCADE'),
nullable=False, primary_key=True)
name = sql.Column(sql.Unicode(255), nullable=False, primary_key=True)
__table_args__ = (sql.UniqueConstraint('project_id', 'name'),)