Fixes the test failure happening on the SUSE Leap build in the nitrogen
branch.
Version can be an empty dictionary and pass the check, but then we hit
a key error on the next line. Let's avoid that.
This failure was caused by the changes in #41086.
Fixes https://github.com/saltstack/salt-jenkins/issues/332
We need to mock the calls to `salt.client.get_local_client` in the
tests.unit.test_auth.py file. Without these mocks, there are many tests
that fail when the entire set of unit tests runs. (Specifically, these fail
with both the module and state boto_apigateway unit tests run.)
This mocked patch is similar to the change in PR #41082 and fixes all 16 test
failures reported in https://github.com/saltstack/salt-jenkins/issues/326.
The previous way of calling `str.split(path, '/')` was causing a
TypeError in the test runs of the boto_apigateway state and
module unit tests. Switching this to `path.split('/')` instead
avoids this type of error.
```
Traceback (most recent call last):
File "/testing/tests/unit/states/test_boto_apigateway.py", line 584, in test_present_when_stage_is_to_associate_to_new_deployment
**conn_parameters)
File "/testing/salt/states/boto_apigateway.py", line 310, in present
authorization_type=authorization_type)
File "/testing/salt/states/boto_apigateway.py", line 1660, in deploy_resources
**self._common_aws_args)
File "/testing/salt/modules/boto_apigateway.py", line 358, in create_api_resources
path_parts = str.split(path, '/')
TypeError: descriptor 'split' requires a 'str' object but received a 'unicode'
```
I mixed up the test that was actually failing for Arch. I originally
submitted #41074 to skip the test_salt_documentation test in the
shell matcher tests. This is the wrong test to skip.
I reverted the previous commit, and applied the skipTest to the
correct test that is failing on Arch, which is the
test_salt_documentation_arguments_not_assumed test.
The "head_ref" variable is not used anywhere in the code, and should
be "head_rev" instead. Without setting "head_rev" to "None", there is
a potential to hit an UnboundLocalError further down in the file.
This was caught by a test in the nitrogen branch, but the bug is present
on 2016.11 as well. Here's the output from the integration test failure:
```
-> integration.states.test_git.GitTest.test_latest_updated_remote_rev .........................................
Traceback (most recent call last):
File "/testing/tests/integration/states/test_git.py", line 365, in test_latest_updated_remote_rev
self.assertSaltTrueReturn(ret)
File "/testing/tests/support/mixins.py", line 547, in assertSaltTrueReturn
**(next(six.itervalues(ret)))
AssertionError: False is not True. Salt Comment:
An exception occurred in this state: Traceback (most recent call last):
File "/testing/salt/state.py", line 1822, in call
**cdata['kwargs'])
File "/testing/salt/loader.py", line 1727, in wrapper
return f(*args, **kwargs)
File "/testing/salt/states/git.py", line 727, in latest
if head_rev is not None:
UnboundLocalError: local variable 'head_rev' referenced before assignment
```
