# Copyright 2012-2014 Amazon.com, Inc. or its affiliates. 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. A copy of

# the License is located at

#

# http://aws.amazon.com/apache2.0/

#

# or in the "license" file accompanying this file. This file 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.

"""Test utilities for the AWS CLI.

This module includes various classes/functions that help in writing

CLI unit/integration tests. This module should not be imported by

any module **except** for test code. This is included in the CLI

package so that code that is not part of the CLI can still take

advantage of all the testing utilities we provide.

"""

import os

import sys

import copy

import shutil

import time

import json

import random

import logging

import tempfile

import platform

import contextlib

from pprint import pformat

from subprocess import Popen, PIPE

try:

import mock

except ImportError as e:

# In the off chance something imports this module

# that's not suppose to, we should not stop the CLI

# by raising an ImportError. Now if anything actually

# *uses* this module that isn't suppose to, that's s

# different story.

mock = None

from awscli.compat import six

from botocore.hooks import HierarchicalEmitter

from botocore.session import Session

import botocore.loaders

from botocore.vendored import requests

import awscli.clidriver

from awscli.plugin import load_plugins

from awscli.clidriver import CLIDriver

from awscli import EnvironmentVariables

# The unittest module got a significant overhaul

# in 2.7, so if we're in 2.6 we can use the backported

# version unittest2.

if sys.version_info[:2] == (2, 6):

import unittest2 as unittest

else:

import unittest

# In python 3, order matters when calling assertEqual to

# compare lists and dictionaries with lists. Therefore,

# assertItemsEqual needs to be used but it is renamed to

# assertCountEqual in python 3.

if six.PY2:

unittest.TestCase.assertCountEqual = unittest.TestCase.assertItemsEqual

_LOADER = botocore.loaders.Loader()

INTEG_LOG = logging.getLogger('awscli.tests.integration')

AWS_CMD = None

def create_clidriver():

driver = awscli.clidriver.create_clidriver()

session = driver.session

data_path = session.get_config_variable('data_path')

_LOADER.data_path = data_path or ''

session.register_component('data_loader', _LOADER)

return driver

def get_aws_cmd():

global AWS_CMD

import awscli

if AWS_CMD is None:

# Try <repo>/bin/aws

repo_root = os.path.dirname(os.path.abspath(awscli.__file__))

aws_cmd = os.path.join(repo_root, 'bin', 'aws')

if not os.path.isfile(aws_cmd):

aws_cmd = _search_path_for_cmd('aws')

if aws_cmd is None:

raise ValueError('Could not find "aws" executable. Either '

'make sure it is on your PATH, or you can '

'explicitly set this value using '

'"set_aws_cmd()"')

AWS_CMD = aws_cmd

return AWS_CMD

def _search_path_for_cmd(cmd_name):

for path in os.environ.get('PATH', '').split(os.pathsep):

full_cmd_path = os.path.join(path, cmd_name)

if os.path.isfile(full_cmd_path):

return full_cmd_path

return None

def set_aws_cmd(aws_cmd):

global AWS_CMD

AWS_CMD = aws_cmd

@contextlib.contextmanager

def temporary_file(mode):

"""This is a cross platform temporary file creation.

tempfile.NamedTemporary file on windows creates a secure temp file

that can't be read by other processes and can't be opened a second time.

For tests, we generally *want* them to be read multiple times.

The test fixture writes the temp file contents, the test reads the

temp file.

"""

temporary_directory = tempfile.mkdtemp()

basename = 'tmpfile-%s-%s' % (int(time.time()), random.randint(1, 1000))

full_filename = os.path.join(temporary_directory, basename)

open(full_filename, 'w').close()

try:

with open(full_filename, mode) as f:

yield f

finally:

shutil.rmtree(temporary_directory)

class BaseCLIDriverTest(unittest.TestCase):

"""Base unittest that use clidriver.

This will load all the default plugins as well so it

will simulate the behavior the user will see.

"""

def setUp(self):

self.environ = {

'AWS_DATA_PATH': os.environ['AWS_DATA_PATH'],

'AWS_DEFAULT_REGION': 'us-east-1',

'AWS_ACCESS_KEY_ID': 'access_key',

'AWS_SECRET_ACCESS_KEY': 'secret_key',

'AWS_CONFIG_FILE': '',

}

self.environ_patch = mock.patch('os.environ', self.environ)

self.environ_patch.start()

emitter = HierarchicalEmitter()

session = Session(EnvironmentVariables, emitter, loader=_LOADER)

load_plugins({}, event_hooks=emitter)

driver = CLIDriver(session=session)

self.session = session

self.driver = driver

def tearDown(self):

self.environ_patch.stop()

class BaseAWSHelpOutputTest(BaseCLIDriverTest):

def setUp(self):

super(BaseAWSHelpOutputTest, self).setUp()

self.renderer_patch = mock.patch('awscli.help.get_renderer')

self.renderer_mock = self.renderer_patch.start()

self.renderer = CapturedRenderer()

self.renderer_mock.return_value = self.renderer

def tearDown(self):

super(BaseAWSHelpOutputTest, self).tearDown()

self.renderer_patch.stop()

def assert_contains(self, contains):

if contains not in self.renderer.rendered_contents:

self.fail("The expected contents:\n%s\nwere not in the "

"actual rendered contents:\n%s" % (

contains, self.renderer.rendered_contents))

def assert_not_contains(self, contents):

if contents in self.renderer.rendered_contents:

self.fail("The contents:\n%s\nwere not suppose to be in the "

"actual rendered contents:\n%s" % (

contents, self.renderer.rendered_contents))

def assert_text_order(self, *args, **kwargs):

# First we need to find where the SYNOPSIS section starts.

starting_from = kwargs.pop('starting_from')

args = list(args)

contents = self.renderer.rendered_contents

self.assertIn(starting_from, contents)

start_index = contents.find(starting_from)

arg_indices = [contents.find(arg, start_index) for arg in args]

previous = arg_indices[0]

for i, index in enumerate(arg_indices[1:], 1):

if index == -1:

self.fail('The string %r was not found in the contents: %s'

% (args[index], contents))

if index < previous:

self.fail('The string %r came before %r, but was suppose to come '

'after it.\n%s' % (args[i], args[i - 1], contents))

previous = index

class CapturedRenderer(object):

def __init__(self):

self.rendered_contents = ''

def render(self, contents):

self.rendered_contents = contents.decode('utf-8')

class BaseAWSCommandParamsTest(unittest.TestCase):

maxDiff = None

def setUp(self):

self.last_params = {}

# awscli/__init__.py injects AWS_DATA_PATH at import time

# so that we can find cli.json. This might be fixed in the

# future, but for now we just grab that value out of the real

# os.environ so the patched os.environ has this data and

# the CLI works.

self.environ = {

'AWS_DATA_PATH': os.environ['AWS_DATA_PATH'],

'AWS_DEFAULT_REGION': 'us-east-1',

'AWS_ACCESS_KEY_ID': 'access_key',

'AWS_SECRET_ACCESS_KEY': 'secret_key',

}

self.environ_patch = mock.patch('os.environ', self.environ)

self.environ_patch.start()

self.http_response = requests.models.Response()

self.http_response.status_code = 200

self.parsed_response = {}

self.make_request_patch = mock.patch('botocore.endpoint.Endpoint.make_request')

self.make_request_is_patched = False

self.operations_called = []

self.parsed_responses = None

self.driver = create_clidriver()

def tearDown(self):

# This clears all the previous registrations.

self.environ_patch.stop()

if self.make_request_is_patched:

self.make_request_patch.stop()

def before_call(self, params, **kwargs):

self._store_params(params)

def _store_params(self, params):

self.last_request_dict = params

self.last_params = params['body']

def patch_make_request(self):

make_request_patch = self.make_request_patch.start()

if self.parsed_responses is not None:

make_request_patch.side_effect = lambda *args, **kwargs: \

(self.http_response, self.parsed_responses.pop(0))

else:

make_request_patch.return_value = (self.http_response, self.parsed_response)

self.make_request_is_patched = True

def assert_params_for_cmd(self, cmd, params=None, expected_rc=0,

stderr_contains=None, ignore_params=None):

stdout, stderr, rc = self.run_cmd(cmd, expected_rc)

if stderr_contains is not None:

self.assertIn(stderr_contains, stderr)

if params is not None:

last_params = self.last_params

if isinstance(last_params, dict):

last_params = copy.copy(self.last_params)

extra_params_to_ignore = ['Action', 'Version']

if ignore_params is None:

ignore_params = extra_params_to_ignore

else:

ignore_params.extend(extra_params_to_ignore)

for key in ignore_params:

try:

del last_params[key]

except KeyError:

pass

if params != last_params:

self.fail("Actual params did not match expected params.\n"

"Expected:\n\n"

"%s\n"

"Actual:\n\n%s\n" % (

pformat(params), pformat(last_params)))

return stdout, stderr, rc

def assert_params_for_cmd2(self, cmd, params=None, expected_rc=0,

stderr_contains=None, ignore_params=None):

# XXX: This has a terrible name because it's intended to be

# temporary. I want to switch everything off of

# assert_params_for_cmd and then I'll rename this to

# assert_params_for_cmd2. The difference between this command

# and the other one is that we verify the kwargs that are sent

# to botocore's Operation.call(), *not* the serialized parameters

# onto the HTTP request. We're one level up from that.

stdout, stderr, rc = self.run_cmd(cmd, expected_rc)

if stderr_contains is not None:

self.assertIn(stderr_contains, stderr)

if params is not None:

# The last kwargs of Operation.call() in botocore.

last_kwargs = copy.copy(self.last_kwargs)

if ignore_params is not None:

for key in ignore_params:

try:

del last_kwargs[key]

except KeyError:

pass

if params != last_kwargs:

self.fail("Actual params did not match expected params.\n"

"Expected:\n\n"

"%s\n"

"Actual:\n\n%s\n" % (

pformat(params), pformat(last_kwargs)))

return stdout, stderr, rc

def before_parameter_build(self, params, model, **kwargs):

self.last_kwargs = params

self.operations_called.append((model, params))

def run_cmd(self, cmd, expected_rc=0):

logging.debug("Calling cmd: %s", cmd)

self.patch_make_request()

self.driver.session.register('before-call', self.before_call)

self.driver.session.register('before-parameter-build',

self.before_parameter_build)

if not isinstance(cmd, list):

cmdlist = cmd.split()

else:

cmdlist = cmd

captured_stderr = six.StringIO()

captured_stdout = six.StringIO()

with mock.patch('sys.stderr', captured_stderr):

with mock.patch('sys.stdout', captured_stdout):

try:

rc = self.driver.main(cmdlist)

except SystemExit as e:

# We need to catch SystemExit so that we

# can get a proper rc and still present the

# stdout/stderr to the test runner so we can

# figure out what went wrong.

rc = e.code

stderr = captured_stderr.getvalue()

stdout = captured_stdout.getvalue()

self.assertEqual(

rc, expected_rc,

"Unexpected rc (expected: %s, actual: %s) for command: %s\n"

"stdout:\n%sstderr:\n%s" % (

expected_rc, rc, cmd, stdout, stderr))

return stdout, stderr, rc

class FileCreator(object):

def __init__(self):

self.rootdir = tempfile.mkdtemp()

def remove_all(self):

shutil.rmtree(self.rootdir)

def create_file(self, filename, contents, mtime=None, mode='w'):

"""Creates a file in a tmpdir

``filename`` should be a relative path, e.g. "foo/bar/baz.txt"

It will be translated into a full path in a tmp dir.

If the ``mtime`` argument is provided, then the file's

mtime will be set to the provided value (must be an epoch time).

Otherwise the mtime is left untouched.

``mode`` is the mode the file should be opened either as ``w`` or

`wb``.

Returns the full path to the file.

"""

full_path = os.path.join(self.rootdir, filename)

if not os.path.isdir(os.path.dirname(full_path)):

os.makedirs(os.path.dirname(full_path))

with open(full_path, mode) as f:

f.write(contents)

current_time = os.path.getmtime(full_path)

# Subtract a few years off the last modification date.

os.utime(full_path, (current_time, current_time - 100000000))

if mtime is not None:

os.utime(full_path, (mtime, mtime))

return full_path

def append_file(self, filename, contents):

"""Append contents to a file

``filename`` should be a relative path, e.g. "foo/bar/baz.txt"

It will be translated into a full path in a tmp dir.

Returns the full path to the file.

"""

full_path = os.path.join(self.rootdir, filename)

if not os.path.isdir(os.path.dirname(full_path)):

os.makedirs(os.path.dirname(full_path))

with open(full_path, 'a') as f:

f.write(contents)

return full_path

def full_path(self, filename):

"""Translate relative path to full path in temp dir.

f.full_path('foo/bar.txt') -> /tmp/asdfasd/foo/bar.txt

"""

return os.path.join(self.rootdir, filename)

class ProcessTerminatedError(Exception):

pass

class Result(object):

def __init__(self, rc, stdout, stderr, memory_usage=None):

self.rc = rc

self.stdout = stdout

self.stderr = stderr

INTEG_LOG.debug("rc: %s", rc)

INTEG_LOG.debug("stdout: %s", stdout)

INTEG_LOG.debug("stderr: %s", stderr)

if memory_usage is None:

memory_usage = []

self.memory_usage = memory_usage

@property

def json(self):

return json.loads(self.stdout)

def _escape_quotes(command):

# For windows we have different rules for escaping.

# First, double quotes must be escaped.

command = command.replace('"', '\\"')

# Second, single quotes do nothing, to quote a value we need

# to use double quotes.

command = command.replace("'", '"')

return command

def aws(command, collect_memory=False, env_vars=None,

wait_for_finish=True, input_data=None, input_file=None):

"""Run an aws command.

This help function abstracts the differences of running the "aws"

command on different platforms.

If collect_memory is ``True`` the the Result object will have a list

of memory usage taken at 2 second intervals. The memory usage

will be in bytes.

If env_vars is None, this will set the environment variables

to be used by the aws process.

If wait_for_finish is False, then the Process object is returned

to the caller. It is then the caller's responsibility to ensure

proper cleanup. This can be useful if you want to test timeout's

or how the CLI responds to various signals.

:type input_data: string

:param input_data: This string will be communicated to the process through

the stdin of the process. It essentially allows the user to

avoid having to use a file handle to pass information to the process.

Note that this string is not passed on creation of the process, but

rather communicated to the process.

:type input_file: a file handle

:param input_file: This is a file handle that will act as the

the stdin of the process immediately on creation. Essentially

any data written to the file will be read from stdin of the

process. This is needed if you plan to stream data into stdin while

collecting memory.

"""

if platform.system() == 'Windows':

command = _escape_quotes(command)

if 'AWS_TEST_COMMAND' in os.environ:

aws_command = os.environ['AWS_TEST_COMMAND']

else:

aws_command = 'python %s' % get_aws_cmd()

full_command = '%s %s' % (aws_command, command)

stdout_encoding = get_stdout_encoding()

if isinstance(full_command, six.text_type) and not six.PY3:

full_command = full_command.encode(stdout_encoding)

INTEG_LOG.debug("Running command: %s", full_command)

env = os.environ.copy()

env['AWS_DEFAULT_REGION'] = "us-east-1"

if env_vars is not None:

env = env_vars

if input_file is None:

input_file = PIPE

process = Popen(full_command, stdout=PIPE, stderr=PIPE, stdin=input_file,

shell=True, env=env)

if not wait_for_finish:

return process

memory = None

if not collect_memory:

kwargs = {}

if input_data:

kwargs = {'input': input_data}

stdout, stderr = process.communicate(**kwargs)

else:

stdout, stderr, memory = _wait_and_collect_mem(process)

return Result(process.returncode,

stdout.decode(stdout_encoding),

stderr.decode(stdout_encoding),

memory)

def get_stdout_encoding():

encoding = getattr(sys.__stdout__, 'encoding', None)

if encoding is None:

encoding = 'utf-8'

return encoding

def _wait_and_collect_mem(process):

# We only know how to collect memory on mac/linux.

if platform.system() == 'Darwin':

get_memory = _get_memory_with_ps

elif platform.system() == 'Linux':

get_memory = _get_memory_with_ps

else:

raise ValueError(

"Can't collect memory for process on platform %s." %

platform.system())

memory = []

while process.poll() is None:

try:

current = get_memory(process.pid)

except ProcessTerminatedError:

# It's possible the process terminated between .poll()

# and get_memory().

break

memory.append(current)

stdout, stderr = process.communicate()

return stdout, stderr, memory

def _get_memory_with_ps(pid):

# It's probably possible to do with proc_pidinfo and ctypes on a Mac,

# but we'll do it the easy way with parsing ps output.

command_list = 'ps u -p'.split()

command_list.append(str(pid))

p = Popen(command_list, stdout=PIPE)

stdout = p.communicate()[0]

if not p.returncode == 0:

raise ProcessTerminatedError(str(pid))

else:

# Get the RSS from output that looks like this:

# USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND

# user 47102 0.0 0.1 2437000 4496 s002 S+ 7:04PM 0:00.12 python2.6

return int(stdout.splitlines()[1].split()[5]) * 1024