493 lines
18 KiB
Python
Raw Normal View History

# Copyright 2012-2013 OpenStack Foundation
# Copyright 2015 Dean Troyer
2012-04-19 22:41:44 -05:00
#
# 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
2012-04-19 22:41:44 -05:00
#
# http://www.apache.org/licenses/LICENSE-2.0
2012-04-19 22:41:44 -05:00
#
# 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.
2012-04-19 22:41:44 -05:00
#
"""Command-line interface to the OpenStack APIs"""
2012-04-19 22:41:44 -05:00
import argparse
import getpass
arguments are not locale decoded into Unicode When the openstackclient in Python2 passes command line arguments to a subcommand it fails to pass the arguments as text (e.g. Unicode). Instead it passes the arguments as binary data encoded using the current locales encoding. An easy way to see this is trying to pass a username with a non-ASCII character. % openstack user delete ñew No user with a name or ID of 'ñew' exists. What occurs internally is when the user data is retrieved it's it properly represented in a Unicode object. However the username pased from the command line is still a str object encoded in the locales encoding (typically UTF-8). A string comparison is attempted between the encoded data from the command line and the Unicode text found in the user representation. This seldom ends well, either the comparison fails to match or a codec error is raised. There is a hard and fast rule, all text data must be stored in Unicode objects and the conversion from binary encoded text to Unicode must occur as close to the I/O boundary as possible. Python3 enforces this behavior automatically but in Python2 it is the programmers job to do so. In the past there have been attempts to fix problems deep inside internal code by attempting to decode from UTF-8. There are two problems with this approach. First, internal code has no way to accurately know what encoding was used to encode the binary data. This is way it needs to be decoded as close to the I/O source as possible because that is the best place to know the actual encoding. Guessing UTF-8 is at best a heuristic. Second, there must be a canonical representation for data "inside" the program, you don't want dozens of individual modules, classes, methods, etc. performing conversions, instead they should be able to make the assumption in what format text is represented in, the format for text data must be Unicode. This is another reason to decode as close to the I/O as possible. In Python3 the argv strings are decoded from the locales encoding by the interpreter. By the time any Python3 code sees the argv strings they will be Unicode. However in Python2 there must be explicit code added to decode the argv strings into Unicode. The conversion of sys.argv into Unicode only occurs when argv is not passed to OpenStackShell.run(). If a caller of OpenStackShell.run() supplies their own arg it is their responsiblity to assure they are passing actual text objects. Consider this a requirement of the API. Note: This patch does not contain a unittest to exercise the behavior because it is difficult to construct a test that depends on command invocation from a shell. The general structure of the unit tests is to pass fake argv into OpenStackShell.run() as if it came from a shell. Because the new code only operates when argv is not passed and defaults to sys.argv it conflicts with the unittest design. Change-Id: I779d260744728eae8455ff9dedb6e5c09c165559 Closes-Bug: 1603494 Signed-off-by: John Dennis <jdennis@redhat.com>
2016-07-15 14:46:29 -04:00
import locale
2012-04-25 16:48:19 -05:00
import logging
arguments are not locale decoded into Unicode When the openstackclient in Python2 passes command line arguments to a subcommand it fails to pass the arguments as text (e.g. Unicode). Instead it passes the arguments as binary data encoded using the current locales encoding. An easy way to see this is trying to pass a username with a non-ASCII character. % openstack user delete ñew No user with a name or ID of 'ñew' exists. What occurs internally is when the user data is retrieved it's it properly represented in a Unicode object. However the username pased from the command line is still a str object encoded in the locales encoding (typically UTF-8). A string comparison is attempted between the encoded data from the command line and the Unicode text found in the user representation. This seldom ends well, either the comparison fails to match or a codec error is raised. There is a hard and fast rule, all text data must be stored in Unicode objects and the conversion from binary encoded text to Unicode must occur as close to the I/O boundary as possible. Python3 enforces this behavior automatically but in Python2 it is the programmers job to do so. In the past there have been attempts to fix problems deep inside internal code by attempting to decode from UTF-8. There are two problems with this approach. First, internal code has no way to accurately know what encoding was used to encode the binary data. This is way it needs to be decoded as close to the I/O source as possible because that is the best place to know the actual encoding. Guessing UTF-8 is at best a heuristic. Second, there must be a canonical representation for data "inside" the program, you don't want dozens of individual modules, classes, methods, etc. performing conversions, instead they should be able to make the assumption in what format text is represented in, the format for text data must be Unicode. This is another reason to decode as close to the I/O as possible. In Python3 the argv strings are decoded from the locales encoding by the interpreter. By the time any Python3 code sees the argv strings they will be Unicode. However in Python2 there must be explicit code added to decode the argv strings into Unicode. The conversion of sys.argv into Unicode only occurs when argv is not passed to OpenStackShell.run(). If a caller of OpenStackShell.run() supplies their own arg it is their responsiblity to assure they are passing actual text objects. Consider this a requirement of the API. Note: This patch does not contain a unittest to exercise the behavior because it is difficult to construct a test that depends on command invocation from a shell. The general structure of the unit tests is to pass fake argv into OpenStackShell.run() as if it came from a shell. Because the new code only operates when argv is not passed and defaults to sys.argv it conflicts with the unittest design. Change-Id: I779d260744728eae8455ff9dedb6e5c09c165559 Closes-Bug: 1603494 Signed-off-by: John Dennis <jdennis@redhat.com>
2016-07-15 14:46:29 -04:00
import six
2012-04-19 22:41:44 -05:00
import sys
import traceback
2012-04-19 22:41:44 -05:00
from cliff import app
from cliff import command
from cliff import complete
from cliff import help
from osc_lib.cli import client_config as cloud_config
from osc_lib.command import timing
from osc_lib import exceptions as exc
from osc_lib import logs
from osc_lib import utils
from oslo_utils import importutils
from oslo_utils import strutils
2012-04-25 16:48:19 -05:00
import openstackclient
from openstackclient.common import clientmanager
from openstackclient.common import commandmanager
from openstackclient.i18n import _
2012-04-19 22:41:44 -05:00
osprofiler_profiler = importutils.try_import("osprofiler.profiler")
2012-04-19 22:41:44 -05:00
DEFAULT_DOMAIN = 'default'
2012-04-25 16:48:19 -05:00
def prompt_for_password(prompt=None):
"""Prompt user for a password
Prompt for a password if stdin is a tty.
"""
if not prompt:
prompt = 'Password: '
pw = None
# If stdin is a tty, try prompting for the password
if hasattr(sys.stdin, 'isatty') and sys.stdin.isatty():
# Check for Ctl-D
try:
pw = getpass.getpass(prompt)
except EOFError:
pass
# No password because we did't have a tty or nothing was entered
if not pw:
raise exc.CommandError(_("No password entered, or found via"
" --os-password or OS_PASSWORD"),)
return pw
class OpenStackShell(app.App):
2012-04-25 16:48:19 -05:00
CONSOLE_MESSAGE_FORMAT = '%(levelname)s: %(name)s %(message)s'
2012-04-25 16:48:19 -05:00
log = logging.getLogger(__name__)
timing_data = []
2012-04-25 16:48:19 -05:00
def __init__(self):
# Patch command.Command to add a default auth_required = True
command.Command.auth_required = True
# Some commands do not need authentication
help.HelpCommand.auth_required = False
complete.CompleteCommand.auth_required = False
# Slight change to the meaning of --debug
self.DEFAULT_DEBUG_VALUE = None
self.DEFAULT_DEBUG_HELP = 'Set debug logging and traceback on errors.'
2012-04-25 16:48:19 -05:00
super(OpenStackShell, self).__init__(
description=__doc__.strip(),
version=openstackclient.__version__,
command_manager=commandmanager.CommandManager('openstack.cli'),
deferred_help=True)
2012-04-25 16:48:19 -05:00
self.api_version = {}
# Until we have command line arguments parsed, dump any stack traces
self.dump_stack_trace = True
# Assume TLS host certificate verification is enabled
self.verify = True
self.client_manager = None
self.command_options = None
self.do_profile = False
def configure_logging(self):
"""Configure logging for the app."""
self.log_configurator = logs.LogConfigurator(self.options)
self.dump_stack_trace = self.log_configurator.dump_trace
def run(self, argv):
ret_val = 1
self.command_options = argv
try:
ret_val = super(OpenStackShell, self).run(argv)
return ret_val
except Exception as e:
if not logging.getLogger('').handlers:
logging.basicConfig()
if self.dump_stack_trace:
self.log.error(traceback.format_exc())
else:
self.log.error('Exception raised: ' + str(e))
return ret_val
finally:
self.log.info("END return value: %s", ret_val)
def init_profile(self):
# NOTE(dtroyer): Remove this 'if' block when the --profile global
# option is removed
if osprofiler_profiler and self.options.old_profile:
self.log.warning(
'The --profile option is deprecated, '
'please use --os-profile instead'
)
if not self.options.profile:
self.options.profile = self.options.old_profile
self.do_profile = osprofiler_profiler and self.options.profile
if self.do_profile:
osprofiler_profiler.init(self.options.profile)
def close_profile(self):
if self.do_profile:
trace_id = osprofiler_profiler.get().get_base_id()
# NOTE(dbelova): let's use warning log level to see these messages
# printed. In fact we can define custom log level here with value
# bigger than most big default one (CRITICAL) or something like
# that (PROFILE = 60 for instance), but not sure we need it here.
self.log.warning("Trace ID: %s" % trace_id)
self.log.warning("Display trace with command:\n"
"osprofiler trace show --html %s " % trace_id)
def run_subcommand(self, argv):
self.init_profile()
try:
ret_value = super(OpenStackShell, self).run_subcommand(argv)
finally:
self.close_profile()
return ret_value
def interact(self):
self.init_profile()
try:
ret_value = super(OpenStackShell, self).interact()
finally:
self.close_profile()
return ret_value
2012-04-25 16:48:19 -05:00
def build_option_parser(self, description, version):
parser = super(OpenStackShell, self).build_option_parser(
description,
version)
# service token auth argument
parser.add_argument(
'--os-cloud',
metavar='<cloud-config-name>',
dest='cloud',
default=utils.env('OS_CLOUD'),
help=_('Cloud name in clouds.yaml (Env: OS_CLOUD)'),
)
# Global arguments
parser.add_argument(
'--os-region-name',
metavar='<auth-region-name>',
dest='region_name',
default=utils.env('OS_REGION_NAME'),
help=_('Authentication region name (Env: OS_REGION_NAME)'),
)
parser.add_argument(
'--os-cacert',
metavar='<ca-bundle-file>',
dest='cacert',
default=utils.env('OS_CACERT'),
help=_('CA certificate bundle file (Env: OS_CACERT)'),
)
parser.add_argument(
'--os-cert',
metavar='<certificate-file>',
dest='cert',
default=utils.env('OS_CERT'),
help=_('Client certificate bundle file (Env: OS_CERT)'),
)
parser.add_argument(
'--os-key',
metavar='<key-file>',
dest='key',
default=utils.env('OS_KEY'),
help=_('Client certificate key file (Env: OS_KEY)'),
)
verify_group = parser.add_mutually_exclusive_group()
verify_group.add_argument(
'--verify',
action='store_true',
default=None,
help=_('Verify server certificate (default)'),
)
verify_group.add_argument(
'--insecure',
action='store_true',
default=None,
help=_('Disable server certificate verification'),
)
parser.add_argument(
'--os-default-domain',
metavar='<auth-domain>',
dest='default_domain',
default=utils.env(
'OS_DEFAULT_DOMAIN',
default=DEFAULT_DOMAIN),
help=_('Default domain ID, default=%s. '
'(Env: OS_DEFAULT_DOMAIN)') % DEFAULT_DOMAIN,
)
parser.add_argument(
'--os-interface',
metavar='<interface>',
dest='interface',
choices=['admin', 'public', 'internal'],
default=utils.env('OS_INTERFACE'),
help=_('Select an interface type.'
' Valid interface types: [admin, public, internal].'
' (Env: OS_INTERFACE)'),
)
parser.add_argument(
'--timing',
default=False,
action='store_true',
help=_("Print API call timing info"),
)
parser.add_argument(
'--os-beta-command',
action='store_true',
help=_("Enable beta commands which are subject to change"),
)
2012-04-19 22:41:44 -05:00
# osprofiler HMAC key argument
if osprofiler_profiler:
parser.add_argument(
'--os-profile',
metavar='hmac-key',
dest='profile',
help=_('HMAC key for encrypting profiling context data'),
)
# NOTE(dtroyer): This global option should have been named
# --os-profile as --profile interferes with at
# least one existing command option. Deprecate
# --profile and remove after Apr 2017.
parser.add_argument(
'--profile',
metavar='hmac-key',
dest='old_profile',
help=argparse.SUPPRESS,
)
return clientmanager.build_plugin_option_parser(parser)
2012-04-19 22:41:44 -05:00
def initialize_app(self, argv):
"""Global app init bits:
* set up API versions
* validate authentication info
* authenticate against Identity if requested
"""
# Parent __init__ parses argv into self.options
super(OpenStackShell, self).initialize_app(argv)
self.log.info("START with options: %s",
strutils.mask_password(self.command_options))
self.log.debug("options: %s",
strutils.mask_password(self.options))
# Set the default plugin to token_endpoint if url and token are given
if (self.options.url and self.options.token):
# Use service token authentication
auth_type = 'token_endpoint'
else:
auth_type = 'password'
project_id = getattr(self.options, 'project_id', None)
project_name = getattr(self.options, 'project_name', None)
tenant_id = getattr(self.options, 'tenant_id', None)
tenant_name = getattr(self.options, 'tenant_name', None)
# Save default domain
self.default_domain = self.options.default_domain
# handle some v2/v3 authentication inconsistencies by just acting like
# both the project and tenant information are both present. This can
# go away if we stop registering all the argparse options together.
if project_id and not tenant_id:
self.options.tenant_id = project_id
if project_name and not tenant_name:
self.options.tenant_name = project_name
if tenant_id and not project_id:
self.options.project_id = tenant_id
if tenant_name and not project_name:
self.options.project_name = tenant_name
# Do configuration file handling
# Ignore the default value of interface. Only if it is set later
# will it be used.
try:
cc = cloud_config.OSC_Config(
override_defaults={
'interface': None,
'auth_type': auth_type,
},
)
except (IOError, OSError):
self.log.critical("Could not read clouds.yaml configuration file")
self.print_help_if_requested()
raise
# TODO(thowe): Change cliff so the default value for debug
# can be set to None.
if not self.options.debug:
self.options.debug = None
self.cloud = cc.get_one_cloud(
cloud=self.options.cloud,
argparse=self.options,
)
self.log_configurator.configure(self.cloud)
self.dump_stack_trace = self.log_configurator.dump_trace
self.log.debug("defaults: %s", cc.defaults)
self.log.debug("cloud cfg: %s",
strutils.mask_password(self.cloud.config))
# Set up client TLS
# NOTE(dtroyer): --insecure is the non-default condition that
# overrides any verify setting in clouds.yaml
# so check it first, then fall back to any verify
# setting provided.
self.verify = not self.cloud.config.get(
'insecure',
not self.cloud.config.get('verify', True),
)
# NOTE(dtroyer): Per bug https://bugs.launchpad.net/bugs/1447784
# --insecure now overrides any --os-cacert setting,
# where before --insecure was ignored if --os-cacert
# was set.
if self.verify and self.cloud.cacert:
self.verify = self.cloud.cacert
# Loop through extensions to get API versions
for mod in clientmanager.PLUGIN_MODULES:
default_version = getattr(mod, 'DEFAULT_API_VERSION', None)
option = mod.API_VERSION_OPTION.replace('os_', '')
version_opt = str(self.cloud.config.get(option, default_version))
if version_opt:
api = mod.API_NAME
self.api_version[api] = version_opt
# Add a plugin interface to let the module validate the version
# requested by the user
skip_old_check = False
mod_check_api_version = getattr(mod, 'check_api_version', None)
if mod_check_api_version:
# this throws an exception if invalid
skip_old_check = mod_check_api_version(version_opt)
mod_versions = getattr(mod, 'API_VERSIONS', None)
if not skip_old_check and mod_versions:
if version_opt not in mod_versions:
self.log.warning(
"%s version %s is not in supported versions %s"
% (api, version_opt,
', '.join(list(mod.API_VERSIONS.keys()))))
# Command groups deal only with major versions
version = '.v' + version_opt.replace('.', '_').split('_')[0]
cmd_group = 'openstack.' + api.replace('-', '_') + version
self.command_manager.add_command_group(cmd_group)
self.log.debug(
'%(name)s API version %(version)s, cmd group %(group)s',
{'name': api, 'version': version_opt, 'group': cmd_group}
)
# Commands that span multiple APIs
self.command_manager.add_command_group(
'openstack.common')
# This is the naive extension implementation referred to in
# blueprint 'client-extensions'
# Extension modules can register their commands in an
# 'openstack.extension' entry point group:
# entry_points={
# 'openstack.extension': [
# 'list_repo=qaz.github.repo:ListRepo',
# 'show_repo=qaz.github.repo:ShowRepo',
# ],
# }
self.command_manager.add_command_group(
'openstack.extension')
# call InitializeXxx() here
# set up additional clients to stuff in to client_manager??
# Handle deferred help and exit
self.print_help_if_requested()
self.client_manager = clientmanager.ClientManager(
cli_options=self.cloud,
api_version=self.api_version,
pw_func=prompt_for_password,
)
def prepare_to_run_command(self, cmd):
"""Set up auth and API versions"""
self.log.info(
'command: %s -> %s.%s',
getattr(cmd, 'cmd_name', '<none>'),
cmd.__class__.__module__,
cmd.__class__.__name__,
)
if cmd.auth_required:
self.client_manager.setup_auth()
if hasattr(cmd, 'required_scope') and cmd.required_scope:
# let the command decide whether we need a scoped token
self.client_manager.validate_scope()
# Trigger the Identity client to initialize
self.client_manager.auth_ref
2012-04-27 12:25:25 -05:00
2012-04-25 16:48:19 -05:00
def clean_up(self, cmd, result, err):
self.log.debug('clean_up %s: %s', cmd.__class__.__name__, err or '')
2012-04-25 16:48:19 -05:00
# Process collected timing data
if self.options.timing:
# Get session data
self.timing_data.extend(
self.client_manager.session.get_timings(),
)
# Use the Timing pseudo-command to generate the output
tcmd = timing.Timing(self, self.options)
tparser = tcmd.get_parser('Timing')
# If anything other than prettytable is specified, force csv
format = 'table'
# Check the formatter used in the actual command
if hasattr(cmd, 'formatter') \
and cmd.formatter != cmd._formatter_plugins['table'].obj:
format = 'csv'
sys.stdout.write('\n')
targs = tparser.parse_args(['-f', format])
tcmd.run(targs)
2012-04-25 16:48:19 -05:00
arguments are not locale decoded into Unicode When the openstackclient in Python2 passes command line arguments to a subcommand it fails to pass the arguments as text (e.g. Unicode). Instead it passes the arguments as binary data encoded using the current locales encoding. An easy way to see this is trying to pass a username with a non-ASCII character. % openstack user delete ñew No user with a name or ID of 'ñew' exists. What occurs internally is when the user data is retrieved it's it properly represented in a Unicode object. However the username pased from the command line is still a str object encoded in the locales encoding (typically UTF-8). A string comparison is attempted between the encoded data from the command line and the Unicode text found in the user representation. This seldom ends well, either the comparison fails to match or a codec error is raised. There is a hard and fast rule, all text data must be stored in Unicode objects and the conversion from binary encoded text to Unicode must occur as close to the I/O boundary as possible. Python3 enforces this behavior automatically but in Python2 it is the programmers job to do so. In the past there have been attempts to fix problems deep inside internal code by attempting to decode from UTF-8. There are two problems with this approach. First, internal code has no way to accurately know what encoding was used to encode the binary data. This is way it needs to be decoded as close to the I/O source as possible because that is the best place to know the actual encoding. Guessing UTF-8 is at best a heuristic. Second, there must be a canonical representation for data "inside" the program, you don't want dozens of individual modules, classes, methods, etc. performing conversions, instead they should be able to make the assumption in what format text is represented in, the format for text data must be Unicode. This is another reason to decode as close to the I/O as possible. In Python3 the argv strings are decoded from the locales encoding by the interpreter. By the time any Python3 code sees the argv strings they will be Unicode. However in Python2 there must be explicit code added to decode the argv strings into Unicode. The conversion of sys.argv into Unicode only occurs when argv is not passed to OpenStackShell.run(). If a caller of OpenStackShell.run() supplies their own arg it is their responsiblity to assure they are passing actual text objects. Consider this a requirement of the API. Note: This patch does not contain a unittest to exercise the behavior because it is difficult to construct a test that depends on command invocation from a shell. The general structure of the unit tests is to pass fake argv into OpenStackShell.run() as if it came from a shell. Because the new code only operates when argv is not passed and defaults to sys.argv it conflicts with the unittest design. Change-Id: I779d260744728eae8455ff9dedb6e5c09c165559 Closes-Bug: 1603494 Signed-off-by: John Dennis <jdennis@redhat.com>
2016-07-15 14:46:29 -04:00
def main(argv=None):
if argv is None:
argv = sys.argv[1:]
if six.PY2:
# Emulate Py3, decode argv into Unicode based on locale so that
# commands always see arguments as text instead of binary data
encoding = locale.getpreferredencoding()
if encoding:
argv = map(lambda arg: arg.decode(encoding), argv)
return OpenStackShell().run(argv)
2012-04-19 22:41:44 -05:00
if __name__ == "__main__":
arguments are not locale decoded into Unicode When the openstackclient in Python2 passes command line arguments to a subcommand it fails to pass the arguments as text (e.g. Unicode). Instead it passes the arguments as binary data encoded using the current locales encoding. An easy way to see this is trying to pass a username with a non-ASCII character. % openstack user delete ñew No user with a name or ID of 'ñew' exists. What occurs internally is when the user data is retrieved it's it properly represented in a Unicode object. However the username pased from the command line is still a str object encoded in the locales encoding (typically UTF-8). A string comparison is attempted between the encoded data from the command line and the Unicode text found in the user representation. This seldom ends well, either the comparison fails to match or a codec error is raised. There is a hard and fast rule, all text data must be stored in Unicode objects and the conversion from binary encoded text to Unicode must occur as close to the I/O boundary as possible. Python3 enforces this behavior automatically but in Python2 it is the programmers job to do so. In the past there have been attempts to fix problems deep inside internal code by attempting to decode from UTF-8. There are two problems with this approach. First, internal code has no way to accurately know what encoding was used to encode the binary data. This is way it needs to be decoded as close to the I/O source as possible because that is the best place to know the actual encoding. Guessing UTF-8 is at best a heuristic. Second, there must be a canonical representation for data "inside" the program, you don't want dozens of individual modules, classes, methods, etc. performing conversions, instead they should be able to make the assumption in what format text is represented in, the format for text data must be Unicode. This is another reason to decode as close to the I/O as possible. In Python3 the argv strings are decoded from the locales encoding by the interpreter. By the time any Python3 code sees the argv strings they will be Unicode. However in Python2 there must be explicit code added to decode the argv strings into Unicode. The conversion of sys.argv into Unicode only occurs when argv is not passed to OpenStackShell.run(). If a caller of OpenStackShell.run() supplies their own arg it is their responsiblity to assure they are passing actual text objects. Consider this a requirement of the API. Note: This patch does not contain a unittest to exercise the behavior because it is difficult to construct a test that depends on command invocation from a shell. The general structure of the unit tests is to pass fake argv into OpenStackShell.run() as if it came from a shell. Because the new code only operates when argv is not passed and defaults to sys.argv it conflicts with the unittest design. Change-Id: I779d260744728eae8455ff9dedb6e5c09c165559 Closes-Bug: 1603494 Signed-off-by: John Dennis <jdennis@redhat.com>
2016-07-15 14:46:29 -04:00
sys.exit(main())