# Copyright 2010 Jacob Kaplan-Moss
# Copyright 2011 OpenStack Foundation
# Copyright 2013 OpenStack Foundation
# All Rights Reserved.
#
# 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.
"""Base utilities to build API operation managers and objects on top of."""
import abc
import copy
import functools
import warnings
from keystoneauth1 import exceptions as ksa_exceptions
from keystoneauth1 import plugin
from oslo_utils import strutils
import six
from six.moves import urllib
from keystoneclient import exceptions as ksc_exceptions
from keystoneclient.i18n import _
class Response(object):
def __init__(self, http_response, data):
self.request_ids = []
if isinstance(http_response, list):
# http_response is a list of <requests.Response> in case
# of pagination
for resp_obj in http_response:
# Extract 'x-openstack-request-id' from headers
self.request_ids.append(resp_obj.headers.get(
'x-openstack-request-id'))
else:
self.request_ids.append(http_response.headers.get(
'x-openstack-request-id'))
self.data = data
def getid(obj):
"""Return id if argument is a Resource.
Abstracts the common pattern of allowing both an object or an object's ID
(UUID) as a parameter when dealing with relationships.
"""
if getattr(obj, 'uuid', None):
return obj.uuid
else:
return getattr(obj, 'id', obj)
def filter_none(**kwargs):
"""Remove any entries from a dictionary where the value is None."""
return dict((k, v) for k, v in kwargs.items() if v is not None)
def filter_kwargs(f):
@functools.wraps(f)
def func(*args, **kwargs):
new_kwargs = {}
for key, ref in kwargs.items():
if ref is None:
# drop null values
continue
id_value = getid(ref)
if id_value != ref:
# If an object with an id was passed, then use the id, e.g.:
# user: user(id=1) becomes user_id: 1
key = '%s_id' % key
new_kwargs[key] = id_value
return f(*args, **new_kwargs)
return func
class Manager(object):
"""Basic manager type providing common operations.
Managers interact with a particular type of API (servers, flavors, images,
etc.) and provide CRUD operations for them.
:param client: instance of BaseClient descendant for HTTP requests
"""
resource_class = None
def __init__(self, client):
super(Manager, self).__init__()
self.client = client
@property
def api(self):
"""The client.
.. warning::
This property is deprecated as of the 1.7.0 release in favor of
:meth:`client` and may be removed in the 2.0.0 release.
"""
warnings.warn(
'api is deprecated as of the 1.7.0 release in favor of client and '
'may be removed in the 2.0.0 release', DeprecationWarning)
return self.client
def _prepare_return_value(self, http_response, data):
if self.client.include_metadata:
return Response(http_response, data)
return data
def _list(self, url, response_key, obj_class=None, body=None, **kwargs):
"""List the collection.
:param url: a partial URL, e.g., '/servers'
:param response_key: the key to be looked up in response dictionary,
e.g., 'servers'
:param obj_class: class for constructing the returned objects
(self.resource_class will be used by default)
:param body: data that will be encoded as JSON and passed in POST
request (GET will be sent by default)
:param kwargs: Additional arguments will be passed to the request.
"""
if body:
resp, body = self.client.post(url, body=body, **kwargs)
else:
resp, body = self.client.get(url, **kwargs)
if obj_class is None:
obj_class = self.resource_class
data = body[response_key]
# NOTE(ja): keystone returns values as list as {'values': [ ... ]}
# unlike other services which just return the list...
try:
data = data['values']
except (KeyError, TypeError): # nosec(cjschaef): keystone data values
# not as expected (see comment above), assumption is that values
# are already returned in a list (so simply utilize that list)
pass
return self._prepare_return_value(
resp, [obj_class(self, res, loaded=True) for res in data if res])
def _get(self, url, response_key, **kwargs):
"""Get an object from collection.
:param url: a partial URL, e.g., '/servers'
:param response_key: the key to be looked up in response dictionary,
e.g., 'server'
:param kwargs: Additional arguments will be passed to the request.
"""
resp, body = self.client.get(url, **kwargs)
return self._prepare_return_value(
resp, self.resource_class(self, body[response_key], loaded=True))
def _head(self, url, **kwargs):
"""Retrieve request headers for an object.
:param url: a partial URL, e.g., '/servers'
:param kwargs: Additional arguments will be passed to the request.
"""
resp, body = self.client.head(url, **kwargs)
return self._prepare_return_value(resp, resp.status_code == 204)
def _post(self, url, body, response_key, return_raw=False, **kwargs):
"""Create an object.
:param url: a partial URL, e.g., '/servers'
:param body: data that will be encoded as JSON and passed in POST
request (GET will be sent by default)
:param response_key: the key to be looked up in response dictionary,
e.g., 'servers'
:param return_raw: flag to force returning raw JSON instead of
Python object of self.resource_class
:param kwargs: Additional arguments will be passed to the request.
"""
resp, body = self.client.post(url, body=body, **kwargs)
if return_raw:
return body[response_key]
return self._prepare_return_value(
resp, self.resource_class(self, body[response_key]))
def _put(self, url, body=None, response_key=None, **kwargs):
"""Update an object with PUT method.
:param url: a partial URL, e.g., '/servers'
:param body: data that will be encoded as JSON and passed in POST
request (GET will be sent by default)
:param response_key: the key to be looked up in response dictionary,
e.g., 'servers'
:param kwargs: Additional arguments will be passed to the request.
"""
resp, body = self.client.put(url, body=body, **kwargs)
# PUT requests may not return a body
if body is not None:
if response_key is not None:
return self._prepare_return_value(
resp, self.resource_class(self, body[response_key]))
else:
return self._prepare_return_value(
resp, self.resource_class(self, body))
def _patch(self, url, body=None, response_key=None, **kwargs):
"""Update an object with PATCH method.
:param url: a partial URL, e.g., '/servers'
:param body: data that will be encoded as JSON and passed in POST
request (GET will be sent by default)
:param response_key: the key to be looked up in response dictionary,
e.g., 'servers'
:param kwargs: Additional arguments will be passed to the request.
"""
resp, body = self.client.patch(url, body=body, **kwargs)
if response_key is not None:
return self._prepare_return_value(
resp, self.resource_class(self, body[response_key]))
else:
return self._prepare_return_value(
resp, self.resource_class(self, body))
def _delete(self, url, **kwargs):
"""Delete an object.
:param url: a partial URL, e.g., '/servers/my-server'
:param kwargs: Additional arguments will be passed to the request.
"""
resp, body = self.client.delete(url, **kwargs)
return resp, self._prepare_return_value(resp, body)
def _update(self, url, body=None, response_key=None, method="PUT",
**kwargs):
methods = {"PUT": self.client.put,
"POST": self.client.post,
"PATCH": self.client.patch}
try:
resp, body = methods[method](url, body=body,
**kwargs)
except KeyError:
raise ksc_exceptions.ClientException(_("Invalid update method: %s")
% method)
# PUT requests may not return a body
if body:
return self._prepare_return_value(
resp, self.resource_class(self, body[response_key]))
else:
return self._prepare_return_value(resp, body)
@six.add_metaclass(abc.ABCMeta)
class ManagerWithFind(Manager):
"""Manager with additional `find()`/`findall()` methods."""
@abc.abstractmethod
def list(self):
pass # pragma: no cover
def find(self, **kwargs):
"""Find a single item with attributes matching ``**kwargs``.
This isn't very efficient: it loads the entire list then filters on
the Python side.
"""
rl = self.findall(**kwargs)
if self.client.include_metadata:
base_response = rl
rl = rl.data
base_response.data = rl[0]
if len(rl) == 0:
msg = _("No %(name)s matching %(kwargs)s.") % {
'name': self.resource_class.__name__, 'kwargs': kwargs}
raise ksa_exceptions.NotFound(404, msg)
elif len(rl) > 1:
raise ksc_exceptions.NoUniqueMatch
else:
return base_response if self.client.include_metadata else rl[0]
def findall(self, **kwargs):
"""Find all items with attributes matching ``**kwargs``.
This isn't very efficient: it loads the entire list then filters on
the Python side.
"""
found = []
searches = kwargs.items()
def _extract_data(objs, response_data):
for obj in objs:
try:
if all(getattr(obj, attr) == value
for (attr, value) in searches):
response_data.append(obj)
except AttributeError:
continue
return response_data
objs = self.list()
if self.client.include_metadata:
# 'objs' is the object of 'Response' class.
objs.data = _extract_data(objs.data, found)
return objs
return _extract_data(objs, found)
class CrudManager(Manager):
"""Base manager class for manipulating Keystone entities.
Children of this class are expected to define a `collection_key` and `key`.
- `collection_key`: Usually a plural noun by convention (e.g. `entities`);
used to refer collections in both URL's (e.g. `/v3/entities`) and JSON
objects containing a list of member resources (e.g. `{'entities': [{},
{}, {}]}`).
- `key`: Usually a singular noun by convention (e.g. `entity`); used to
refer to an individual member of the collection.
"""
collection_key = None
key = None
base_url = None
def build_url(self, dict_args_in_out=None):
"""Build a resource URL for the given kwargs.
Given an example collection where `collection_key = 'entities'` and
`key = 'entity'`, the following URL's could be generated.
By default, the URL will represent a collection of entities, e.g.::
/entities
If kwargs contains an `entity_id`, then the URL will represent a
specific member, e.g.::
/entities/{entity_id}
If a `base_url` is provided, the generated URL will be appended to it.
If a 'tail' is provided, it will be appended to the end of the URL.
"""
if dict_args_in_out is None:
dict_args_in_out = {}
url = dict_args_in_out.pop('base_url', None) or self.base_url or ''
url += '/%s' % self.collection_key
# do we have a specific entity?
entity_id = dict_args_in_out.pop('%s_id' % self.key, None)
if entity_id is not None:
url += '/%s' % entity_id
if dict_args_in_out.get('tail'):
url += dict_args_in_out['tail']
return url
@filter_kwargs
def create(self, **kwargs):
url = self.build_url(dict_args_in_out=kwargs)
return self._post(
url,
{self.key: kwargs},
self.key)
@filter_kwargs
def get(self, **kwargs):
return self._get(
self.build_url(dict_args_in_out=kwargs),
self.key)
@filter_kwargs
def head(self, **kwargs):
return self._head(self.build_url(dict_args_in_out=kwargs))
def _build_query(self, params):
if params is None:
return ''
else:
# NOTE(spilla) Since the manager cannot take in a hyphen as a
# key in the kwarg, it is passed in with a _. This needs to be
# replaced with a proper hyphen for the URL to work properly.
tags_params = ('tags_any', 'not_tags', 'not_tags_any')
for tag_param in tags_params:
if tag_param in params:
params[tag_param.replace('_', '-')] = params.pop(tag_param)
return '?%s' % urllib.parse.urlencode(params, doseq=True)
def build_key_only_query(self, params_list):
"""Build a query that does not include values, just keys.
The Identity API has some calls that define queries without values,
this can not be accomplished by using urllib.parse.urlencode(). This
method builds a query using only the keys.
"""
return '?%s' % '&'.join(params_list) if params_list else ''
@filter_kwargs
def list(self, fallback_to_auth=False, **kwargs):
def return_resp(resp, include_metadata=False):
base_response = None
list_data = resp
if include_metadata:
base_response = resp
list_data = resp.data
base_response.data = list_data
return base_response if include_metadata else list_data
if 'id' in kwargs.keys():
# Ensure that users are not trying to call things like
# ``domains.list(id='default')`` when they should have used
# ``[domains.get(domain_id='default')]`` instead. Keystone supports
# ``GET /v3/domains/{domain_id}``, not ``GET
# /v3/domains?id={domain_id}``.
raise TypeError(
_("list() got an unexpected keyword argument 'id'. To "
"retrieve a single object using a globally unique "
"identifier, try using get() instead."))
url = self.build_url(dict_args_in_out=kwargs)
try:
query = self._build_query(kwargs)
url_query = '%(url)s%(query)s' % {'url': url, 'query': query}
list_resp = self._list(url_query, self.collection_key)
return return_resp(list_resp,
include_metadata=self.client.include_metadata)
except ksa_exceptions.EmptyCatalog:
if fallback_to_auth:
list_resp = self._list(url_query, self.collection_key,
endpoint_filter={
'interface': plugin.AUTH_INTERFACE})
return return_resp(
list_resp, include_metadata=self.client.include_metadata)
else:
raise
@filter_kwargs
def put(self, **kwargs):
return self._update(
self.build_url(dict_args_in_out=kwargs),
method='PUT')
@filter_kwargs
def update(self, **kwargs):
url = self.build_url(dict_args_in_out=kwargs)
return self._update(
url,
{self.key: kwargs},
self.key,
method='PATCH')
@filter_kwargs
def delete(self, **kwargs):
return self._delete(
self.build_url(dict_args_in_out=kwargs))
@filter_kwargs
def find(self, **kwargs):
"""Find a single item with attributes matching ``**kwargs``."""
url = self.build_url(dict_args_in_out=kwargs)
query = self._build_query(kwargs)
url_query = '%(url)s%(query)s' % {
'url': url,
'query': query
}
elements = self._list(
url_query,
self.collection_key)
if self.client.include_metadata:
base_response = elements
elements = elements.data
base_response.data = elements[0]
if not elements:
msg = _("No %(name)s matching %(kwargs)s.") % {
'name': self.resource_class.__name__, 'kwargs': kwargs}
raise ksa_exceptions.NotFound(404, msg)
elif len(elements) > 1:
raise ksc_exceptions.NoUniqueMatch
else:
return (base_response if self.client.include_metadata
else elements[0])
class Resource(object):
"""Base class for OpenStack resources (tenant, user, etc.).
This is pretty much just a bag for attributes.
"""
HUMAN_ID = False
NAME_ATTR = 'name'
def __init__(self, manager, info, loaded=False):
"""Populate and bind to a manager.
:param manager: BaseManager object
:param info: dictionary representing resource attributes
:param loaded: prevent lazy-loading if set to True
"""
self.manager = manager
self._info = info
self._add_details(info)
self._loaded = loaded
def __repr__(self):
"""Return string representation of resource attributes."""
reprkeys = sorted(k
for k in self.__dict__.keys()
if k[0] != '_' and k != 'manager')
info = ", ".join("%s=%s" % (k, getattr(self, k)) for k in reprkeys)
return "<%s %s>" % (self.__class__.__name__, info)
@property
def human_id(self):
"""Human-readable ID which can be used for bash completion."""
if self.HUMAN_ID:
name = getattr(self, self.NAME_ATTR, None)
if name is not None:
return strutils.to_slug(name)
return None
def _add_details(self, info):
for (k, v) in info.items():
try:
try:
setattr(self, k, v)
except UnicodeEncodeError:
# This happens when we're running with Python version that
# does not support Unicode identifiers (e.g. Python 2.7).
# In that case we can't help but not set this attrubute;
# it'll be available in a dict representation though
pass
self._info[k] = v
except AttributeError: # nosec(cjschaef): we already defined the
# attribute on the class
pass
def __getattr__(self, k):
"""Checking attrbiute existence."""
if k not in self.__dict__:
# NOTE(bcwaldon): disallow lazy-loading if already loaded once
if not self.is_loaded():
self.get()
return self.__getattr__(k)
raise AttributeError(k)
else:
return self.__dict__[k]
def get(self):
"""Support for lazy loading details.
Some clients, such as novaclient have the option to lazy load the
details, details which can be loaded with this function.
"""
# set_loaded() first ... so if we have to bail, we know we tried.
self.set_loaded(True)
if not hasattr(self.manager, 'get'):
return
new = self.manager.get(self.id)
if new:
self._add_details(new._info)
def __eq__(self, other):
"""Define equality for resources."""
if not isinstance(other, Resource):
return NotImplemented
# two resources of different types are not equal
if not isinstance(other, self.__class__):
return False
return self._info == other._info
def __ne__(self, other):
"""Define inequality for resources."""
return not self == other
def is_loaded(self):
return self._loaded
def set_loaded(self, val):
self._loaded = val
def to_dict(self):
return copy.deepcopy(self._info)
def delete(self):
return self.manager.delete(self)