Source code for libcloud.common.base

# Licensed to the Apache Software Foundation (ASF) under one or more
# contributor license agreements.  See the NOTICE file distributed with
# this work for additional information regarding copyright ownership.
# The ASF licenses this file to You 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 os
import sys
import ssl
import socket
import copy
import binascii
import time

try:
    from lxml import etree as ET
except ImportError:
    from xml.etree import ElementTree as ET


try:
    import simplejson as json
except:
    import json

import requests

import libcloud

from libcloud.utils.py3 import PY25
from libcloud.utils.py3 import httplib
from libcloud.utils.py3 import urlparse
from libcloud.utils.py3 import urlencode

from libcloud.utils.misc import lowercase_keys, retry
from libcloud.common.exceptions import exception_from_message
from libcloud.common.types import LibcloudError, MalformedResponseError
from libcloud.http import LibcloudConnection, HttpLibResponseProxy

__all__ = [
    'RETRY_FAILED_HTTP_REQUESTS',

    'BaseDriver',

    'Connection',
    'PollingConnection',
    'ConnectionKey',
    'ConnectionUserAndKey',
    'CertificateConnection',

    'Response',
    'HTTPResponse',
    'JsonResponse',
    'XmlResponse',
    'RawResponse'
]

# Module level variable indicates if the failed HTTP requests should be retried
RETRY_FAILED_HTTP_REQUESTS = False


class LazyObject(object):
    """An object that doesn't get initialized until accessed."""

    @classmethod
    def _proxy(cls, *lazy_init_args, **lazy_init_kwargs):
        class Proxy(cls, object):
            _lazy_obj = None

            def __init__(self):
                # Must override the lazy_cls __init__
                pass

            def __getattribute__(self, attr):
                lazy_obj = object.__getattribute__(self, '_get_lazy_obj')()
                return getattr(lazy_obj, attr)

            def __setattr__(self, attr, value):
                lazy_obj = object.__getattribute__(self, '_get_lazy_obj')()
                setattr(lazy_obj, attr, value)

            def _get_lazy_obj(self):
                lazy_obj = object.__getattribute__(self, '_lazy_obj')
                if lazy_obj is None:
                    lazy_obj = cls(*lazy_init_args, **lazy_init_kwargs)
                    object.__setattr__(self, '_lazy_obj', lazy_obj)
                return lazy_obj

        return Proxy()

    @classmethod
    def lazy(cls, *lazy_init_args, **lazy_init_kwargs):
        """Create a lazily instantiated instance of the subclass, cls."""
        return cls._proxy(*lazy_init_args, **lazy_init_kwargs)


[docs]class HTTPResponse(httplib.HTTPResponse): # On python 2.6 some calls can hang because HEAD isn't quite properly # supported. # In particular this happens on S3 when calls are made to get_object to # objects that don't exist. # This applies the behaviour from 2.7, fixing the hangs.
[docs] def read(self, amt=None): if self.fp is None: return '' if self._method == 'HEAD': self.close() return '' return httplib.HTTPResponse.read(self, amt)
[docs]class Response(object): """ A base Response class to derive from. """ status = httplib.OK # Response status code headers = {} # Response headers body = None # Raw response body object = None # Parsed response body error = None # Reason returned by the server. connection = None # Parent connection class parse_zero_length_body = False def __init__(self, response, connection): """ :param response: HTTP response object. (optional) :type response: :class:`httplib.HTTPResponse` :param connection: Parent connection object. :type connection: :class:`.Connection` """ self.connection = connection # http.client In Python 3 doesn't automatically lowercase the header # names self.headers = lowercase_keys(dict(response.headers)) self.error = response.reason self.status = response.status_code self.request = response.request self.iter_content = response.iter_content self.body = response.text.strip() \ if response.text is not None and hasattr(response.text, 'strip') \ else '' if not self.success(): raise exception_from_message(code=self.status, message=self.parse_error()) self.object = self.parse_body()
[docs] def parse_body(self): """ Parse response body. Override in a provider's subclass. :return: Parsed body. :rtype: ``str`` """ return self.body if self.body is not None else ''
[docs] def parse_error(self): """ Parse the error messages. Override in a provider's subclass. :return: Parsed error. :rtype: ``str`` """ return self.body
[docs] def success(self): """ Determine if our request was successful. The meaning of this can be arbitrary; did we receive OK status? Did the node get created? Were we authenticated? :rtype: ``bool`` :return: ``True`` or ``False`` """ # pylint: disable=E1101 return self.status in [requests.codes.ok, requests.codes.created, httplib.OK, httplib.CREATED, httplib.ACCEPTED]
[docs]class JsonResponse(Response): """ A Base JSON Response class to derive from. """
[docs] def parse_body(self): if len(self.body) == 0 and not self.parse_zero_length_body: return self.body try: body = json.loads(self.body) except: raise MalformedResponseError( 'Failed to parse JSON', body=self.body, driver=self.connection.driver) return body
parse_error = parse_body
[docs]class XmlResponse(Response): """ A Base XML Response class to derive from. """
[docs] def parse_body(self): if len(self.body) == 0 and not self.parse_zero_length_body: return self.body try: try: body = ET.XML(self.body) except ValueError: # lxml wants a bytes and tests are basically hard-coded to str body = ET.XML(self.body.encode('utf-8')) except: raise MalformedResponseError('Failed to parse XML', body=self.body, driver=self.connection.driver) return body
parse_error = parse_body
[docs]class RawResponse(Response): def __init__(self, connection, response=None): """ :param connection: Parent connection object. :type connection: :class:`.Connection` """ self._status = None self._response = None self._headers = {} self._error = None self._reason = None self.connection = connection if response is not None: self.headers = lowercase_keys(dict(response.headers)) self.error = response.reason self.status = response.status_code self.request = response.request self.iter_content = response.iter_content
[docs] def success(self): """ Determine if our request was successful. The meaning of this can be arbitrary; did we receive OK status? Did the node get created? Were we authenticated? :rtype: ``bool`` :return: ``True`` or ``False`` """ # pylint: disable=E1101 return self.status in [requests.codes.ok, requests.codes.created, httplib.OK, httplib.CREATED, httplib.ACCEPTED]
@property def response(self): if not self._response: response = self.connection.connection.getresponse() self._response = HttpLibResponseProxy(response) self.body = response.text if not self.success(): self.parse_error() return self._response @property def reason(self): if not self._reason: self._reason = self.response.reason return self._reason
[docs]class Connection(object): """ A Base Connection class to derive from. """ conn_class = LibcloudConnection responseCls = Response rawResponseCls = RawResponse connection = None host = '127.0.0.1' port = 443 timeout = None secure = 1 driver = None action = None cache_busting = False backoff = None retry_delay = None allow_insecure = True def __init__(self, secure=True, host=None, port=None, url=None, timeout=None, proxy_url=None, retry_delay=None, backoff=None): self.secure = secure and 1 or 0 self.ua = [] self.context = {} if not self.allow_insecure and not secure: # TODO: We should eventually switch to whitelist instead of # blacklist approach raise ValueError('Non https connections are not allowed (use ' 'secure=True)') self.request_path = '' if host: self.host = host if port is not None: self.port = port else: if self.secure == 1: self.port = 443 else: self.port = 80 if url: (self.host, self.port, self.secure, self.request_path) = self._tuple_from_url(url) self.timeout = timeout or self.timeout self.retry_delay = retry_delay self.backoff = backoff self.proxy_url = proxy_url
[docs] def set_http_proxy(self, proxy_url): """ Set a HTTP proxy which will be used with this connection. :param proxy_url: Proxy URL (e.g. http://<hostname>:<port> without authentication and http://<username>:<password>@<hostname>:<port> for basic auth authentication information. :type proxy_url: ``str`` """ self.proxy_url = proxy_url
[docs] def set_context(self, context): if not isinstance(context, dict): raise TypeError('context needs to be a dictionary') self.context = context
[docs] def reset_context(self): self.context = {}
def _tuple_from_url(self, url): secure = 1 port = None (scheme, netloc, request_path, param, query, fragment) = urlparse.urlparse(url) if scheme not in ['http', 'https']: raise LibcloudError('Invalid scheme: %s in url %s' % (scheme, url)) if scheme == "http": secure = 0 if ":" in netloc: netloc, port = netloc.rsplit(":") port = int(port) if not port: if scheme == "http": port = 80 else: port = 443 host = netloc port = int(port) return (host, port, secure, request_path)
[docs] def connect(self, host=None, port=None, base_url=None, **kwargs): """ Establish a connection with the API server. :type host: ``str`` :param host: Optional host to override our default :type port: ``int`` :param port: Optional port to override our default :returns: A connection """ # prefer the attribute base_url if its set or sent connection = None secure = self.secure if getattr(self, 'base_url', None) and base_url is None: (host, port, secure, request_path) = \ self._tuple_from_url(getattr(self, 'base_url')) elif base_url is not None: (host, port, secure, request_path) = self._tuple_from_url(base_url) else: host = host or self.host port = port or self.port # Make sure port is an int port = int(port) if not hasattr(kwargs, 'host'): kwargs.update({'host': host}) if not hasattr(kwargs, 'port'): kwargs.update({'port': port}) if not hasattr(kwargs, 'secure'): kwargs.update({'secure': self.secure}) if not hasattr(kwargs, 'key_file') and hasattr(self, 'key_file'): kwargs.update({'key_file': getattr(self, 'key_file')}) if not hasattr(kwargs, 'cert_file') and hasattr(self, 'cert_file'): kwargs.update({'cert_file': getattr(self, 'cert_file')}) # Timeout is only supported in Python 2.6 and later # http://docs.python.org/library/httplib.html#httplib.HTTPConnection if self.timeout and not PY25: kwargs.update({'timeout': self.timeout}) if self.proxy_url: kwargs.update({'proxy_url': self.proxy_url}) connection = self.conn_class(**kwargs) # You can uncoment this line, if you setup a reverse proxy server # which proxies to your endpoint, and lets you easily capture # connections in cleartext when you setup the proxy to do SSL # for you # connection = self.conn_class("127.0.0.1", 8080) self.connection = connection
def _user_agent(self): user_agent_suffix = ' '.join(['(%s)' % x for x in self.ua]) if self.driver: user_agent = 'libcloud/%s (%s) %s' % ( libcloud.__version__, self.driver.name, user_agent_suffix) else: user_agent = 'libcloud/%s %s' % ( libcloud.__version__, user_agent_suffix) return user_agent
[docs] def user_agent_append(self, token): """ Append a token to a user agent string. Users of the library should call this to uniquely identify their requests to a provider. :type token: ``str`` :param token: Token to add to the user agent. """ self.ua.append(token)
[docs] def request(self, action, params=None, data=None, headers=None, method='GET', raw=False, stream=False): """ Request a given `action`. Basically a wrapper around the connection object's `request` that does some helpful pre-processing. :type action: ``str`` :param action: A path. This can include arguments. If included, any extra parameters are appended to the existing ones. :type params: ``dict`` :param params: Optional mapping of additional parameters to send. If None, leave as an empty ``dict``. :type data: ``unicode`` :param data: A body of data to send with the request. :type headers: ``dict`` :param headers: Extra headers to add to the request None, leave as an empty ``dict``. :type method: ``str`` :param method: An HTTP method such as "GET" or "POST". :type raw: ``bool`` :param raw: True to perform a "raw" request aka only send the headers and use the rawResponseCls class. This is used with storage API when uploading a file. :type stream: ``bool`` :param stream: True to return an iterator in Response.iter_content and allow streaming of the response data (for downloading large files) :return: An :class:`Response` instance. :rtype: :class:`Response` instance """ if params is None: params = {} else: params = copy.copy(params) if headers is None: headers = {} else: headers = copy.copy(headers) retry_enabled = os.environ.get('LIBCLOUD_RETRY_FAILED_HTTP_REQUESTS', False) or RETRY_FAILED_HTTP_REQUESTS action = self.morph_action_hook(action) self.action = action self.method = method self.data = data # Extend default parameters params = self.add_default_params(params) # Add cache busting parameters (if enabled) if self.cache_busting and method == 'GET': params = self._add_cache_busting_to_params(params=params) # Extend default headers headers = self.add_default_headers(headers) # We always send a user-agent header headers.update({'User-Agent': self._user_agent()}) # Indicate that we support gzip and deflate compression headers.update({'Accept-Encoding': 'gzip,deflate'}) port = int(self.port) if port not in (80, 443): headers.update({'Host': "%s:%d" % (self.host, port)}) else: headers.update({'Host': self.host}) if data: data = self.encode_data(data) params, headers = self.pre_connect_hook(params, headers) if params: if '?' in action: url = '&'.join((action, urlencode(params, doseq=True))) else: url = '?'.join((action, urlencode(params, doseq=True))) else: url = action # IF connection has not yet been established if self.connection is None: self.connect() try: # @TODO: Should we just pass File object as body to request method # instead of dealing with splitting and sending the file ourselves? if raw: self.connection.prepared_request( method=method, url=url, body=data, headers=headers, stream=stream) else: if retry_enabled: retry_request = retry(timeout=self.timeout, retry_delay=self.retry_delay, backoff=self.backoff) retry_request(self.connection.request)(method=method, url=url, body=data, headers=headers, stream=stream) else: self.connection.request(method=method, url=url, body=data, headers=headers, stream=stream) except socket.gaierror: e = sys.exc_info()[1] message = str(e) errno = getattr(e, 'errno', None) if errno == -5: # Throw a more-friendly exception on "no address associated # with hostname" error. This error could simpli indicate that # "host" Connection class attribute is set to an incorrect # value class_name = self.__class__.__name__ msg = ('%s. Perhaps "host" Connection class attribute ' '(%s.connection) is set to an invalid, non-hostname ' 'value (%s)?' % (message, class_name, self.host)) raise socket.gaierror(msg) self.reset_context() raise e except ssl.SSLError: e = sys.exc_info()[1] self.reset_context() raise ssl.SSLError(str(e)) if raw: responseCls = self.rawResponseCls kwargs = {'connection': self, 'response': self.connection.getresponse()} else: responseCls = self.responseCls kwargs = {'connection': self, 'response': self.connection.getresponse()} try: response = responseCls(**kwargs) finally: # Always reset the context after the request has completed self.reset_context() return response
[docs] def morph_action_hook(self, action): url = urlparse.urljoin(self.request_path.lstrip('/').rstrip('/') + '/', action.lstrip('/')) if not url.startswith('/'): return '/' + url else: return url
[docs] def add_default_params(self, params): """ Adds default parameters (such as API key, version, etc.) to the passed `params` Should return a dictionary. """ return params
[docs] def add_default_headers(self, headers): """ Adds default headers (such as Authorization, X-Foo-Bar) to the passed `headers` Should return a dictionary. """ return headers
[docs] def pre_connect_hook(self, params, headers): """ A hook which is called before connecting to the remote server. This hook can perform a final manipulation on the params, headers and url parameters. :type params: ``dict`` :param params: Request parameters. :type headers: ``dict`` :param headers: Request headers. """ return params, headers
[docs] def encode_data(self, data): """ Encode body data. Override in a provider's subclass. """ return data
def _add_cache_busting_to_params(self, params): """ Add cache busting parameter to the query parameters of a GET request. Parameters are only added if "cache_busting" class attribute is set to True. Note: This should only be used with *naughty* providers which use excessive caching of responses. """ cache_busting_value = binascii.hexlify(os.urandom(8)).decode('ascii') if isinstance(params, dict): params['cache-busting'] = cache_busting_value else: params.append(('cache-busting', cache_busting_value)) return params
[docs]class PollingConnection(Connection): """ Connection class which can also work with the async APIs. After initial requests, this class periodically polls for jobs status and waits until the job has finished. If job doesn't finish in timeout seconds, an Exception thrown. """ poll_interval = 0.5 timeout = 200 request_method = 'request'
[docs] def async_request(self, action, params=None, data=None, headers=None, method='GET', context=None): """ Perform an 'async' request to the specified path. Keep in mind that this function is *blocking* and 'async' in this case means that the hit URL only returns a job ID which is the periodically polled until the job has completed. This function works like this: - Perform a request to the specified path. Response should contain a 'job_id'. - Returned 'job_id' is then used to construct a URL which is used for retrieving job status. Constructed URL is then periodically polled until the response indicates that the job has completed or the timeout of 'self.timeout' seconds has been reached. :type action: ``str`` :param action: A path :type params: ``dict`` :param params: Optional mapping of additional parameters to send. If None, leave as an empty ``dict``. :type data: ``unicode`` :param data: A body of data to send with the request. :type headers: ``dict`` :param headers: Extra headers to add to the request None, leave as an empty ``dict``. :type method: ``str`` :param method: An HTTP method such as "GET" or "POST". :type context: ``dict`` :param context: Context dictionary which is passed to the functions which construct initial and poll URL. :return: An :class:`Response` instance. :rtype: :class:`Response` instance """ request = getattr(self, self.request_method) kwargs = self.get_request_kwargs(action=action, params=params, data=data, headers=headers, method=method, context=context) response = request(**kwargs) kwargs = self.get_poll_request_kwargs(response=response, context=context, request_kwargs=kwargs) end = time.time() + self.timeout completed = False while time.time() < end and not completed: response = request(**kwargs) completed = self.has_completed(response=response) if not completed: time.sleep(self.poll_interval) if not completed: raise LibcloudError('Job did not complete in %s seconds' % (self.timeout)) return response
[docs] def get_request_kwargs(self, action, params=None, data=None, headers=None, method='GET', context=None): """ Arguments which are passed to the initial request() call inside async_request. """ kwargs = {'action': action, 'params': params, 'data': data, 'headers': headers, 'method': method} return kwargs
[docs] def get_poll_request_kwargs(self, response, context, request_kwargs): """ Return keyword arguments which are passed to the request() method when polling for the job status. :param response: Response object returned by poll request. :type response: :class:`HTTPResponse` :param request_kwargs: Kwargs previously used to initiate the poll request. :type response: ``dict`` :return ``dict`` Keyword arguments """ raise NotImplementedError('get_poll_request_kwargs not implemented')
[docs] def has_completed(self, response): """ Return job completion status. :param response: Response object returned by poll request. :type response: :class:`HTTPResponse` :return ``bool`` True if the job has completed, False otherwise. """ raise NotImplementedError('has_completed not implemented')
[docs]class ConnectionKey(Connection): """ Base connection class which accepts a single ``key`` argument. """ def __init__(self, key, secure=True, host=None, port=None, url=None, timeout=None, proxy_url=None, backoff=None, retry_delay=None): """ Initialize `user_id` and `key`; set `secure` to an ``int`` based on passed value. """ super(ConnectionKey, self).__init__(secure=secure, host=host, port=port, url=url, timeout=timeout, proxy_url=proxy_url, backoff=backoff, retry_delay=retry_delay) self.key = key
[docs]class CertificateConnection(Connection): """ Base connection class which accepts a single ``cert_file`` argument. """ def __init__(self, cert_file, secure=True, host=None, port=None, url=None, proxy_url=None, timeout=None, backoff=None, retry_delay=None): """ Initialize `cert_file`; set `secure` to an ``int`` based on passed value. """ super(CertificateConnection, self).__init__(secure=secure, host=host, port=port, url=url, timeout=timeout, backoff=backoff, retry_delay=retry_delay, proxy_url=proxy_url) self.cert_file = cert_file
[docs]class ConnectionUserAndKey(ConnectionKey): """ Base connection class which accepts a ``user_id`` and ``key`` argument. """ user_id = None def __init__(self, user_id, key, secure=True, host=None, port=None, url=None, timeout=None, proxy_url=None, backoff=None, retry_delay=None): super(ConnectionUserAndKey, self).__init__(key, secure=secure, host=host, port=port, url=url, timeout=timeout, backoff=backoff, retry_delay=retry_delay, proxy_url=proxy_url) self.user_id = user_id
[docs]class BaseDriver(object): """ Base driver class from which other classes can inherit from. """ connectionCls = ConnectionKey def __init__(self, key, secret=None, secure=True, host=None, port=None, api_version=None, region=None, **kwargs): """ :param key: API key or username to be used (required) :type key: ``str`` :param secret: Secret password to be used (required) :type secret: ``str`` :param secure: Whether to use HTTPS or HTTP. Note: Some providers only support HTTPS, and it is on by default. :type secure: ``bool`` :param host: Override hostname used for connections. :type host: ``str`` :param port: Override port used for connections. :type port: ``int`` :param api_version: Optional API version. Only used by drivers which support multiple API versions. :type api_version: ``str`` :param region: Optional driver region. Only used by drivers which support multiple regions. :type region: ``str`` :rtype: ``None`` """ self.key = key self.secret = secret self.secure = secure args = [self.key] if self.secret is not None: args.append(self.secret) args.append(secure) if host is not None: args.append(host) if port is not None: args.append(port) self.api_version = api_version self.region = region conn_kwargs = self._ex_connection_class_kwargs() conn_kwargs.update({'timeout': kwargs.pop('timeout', None), 'retry_delay': kwargs.pop('retry_delay', None), 'backoff': kwargs.pop('backoff', None), 'proxy_url': kwargs.pop('proxy_url', None)}) self.connection = self.connectionCls(*args, **conn_kwargs) self.connection.driver = self self.connection.connect() def _ex_connection_class_kwargs(self): """ Return extra connection keyword arguments which are passed to the Connection class constructor. """ return {}