A PyPI package containing shared code for ShipChain's Python/Django projects
When shipchain-common is installed, a pytest plugin named json_asserter is automatically registered. This plugin is
designed for writing concise pytest cases that make json_asserter about responses from a Django Rest Framework API. Most
of the functionality is tailored to the application/vnd.api+json response type, but should still be usable for
plain application/json responses.
The json_asserter fixture exposes several methods for testing specific HTTP Status codes as well as a class for
building consistent entity references that must be found within the responses.
This is the default when utilizing the json_asserter. If the response does not conform to the
JSON Api standard, the assertions will fail.
To assert that a given response must have an error status, there are several 400-level response methods. With the exception of the HTTP_400 method, each of these include the default error message for ease of use.
The following will assert that the response status was 403 and that the default error message ("You do not have permission to perform this action") is present.
response = api_client.get(self.detail_url)
json_asserter.HTTP_403(response)If a different error message should exist, or when checking the error of a 400 response, the specific error may be provided as an argument.
response = api_client.get(self.detail_url)
json_asserter.HTTP_400(response, error='Specific error message that should be in the respose')To assert that a given response must have status 200, call the HTTP_200 method with only the Response object:
response = api_client.get(self.detail_url)
json_asserter.HTTP_200(response)While this is valid, it is very strongly recommended to include additional details about the data present in the response. There are two ways to provide the data; however only one way can be used at a time in a given invocation.
For simple responses, the easiest way to specify required data in the responses is by directly specifying the
Resource Type resource, the Resource Identifier pk, as well as any specific Attributes of the resource
attributes.
response = api_client.get(self.detail_url)
json_asserter.HTTP_200(response,
resource='User',
pk='4b56399d-3155-4fe5-ba4a-9718289a78b7',
attributes={'username': 'example_user'})This will throw an assertion if the response is not for the resource type User with id
4b56399d-3155-4fe5-ba4a-9718289a78b7 and with at least the attribute username example_user. If the response
includes additional attributes that are not listed in the call to the json_asserter method, they are ignored. The
methods check partial objects and do not require that every attribute in the response must be defined in the
assertion.
It is also possible to assert only on the resource type and id without providing attributes. This is useful if you are testing a response that generates content for the fields that may not be known prior to obtaining the response. Additionally, providing only the attributes and not the type and id will check only that an object in the response has those attributes, regardless of resource type or id.
For responses where the associated Relationship and any extra Included resources are important, those can be included in the assertion.
response = api_client.get(self.detail_url)
json_asserter.HTTP_200(response,
entity_refs=json_asserter.EntityRef(
resource='User',
pk='4b56399d-3155-4fe5-ba4a-9718289a78b7',
attributes={'username': 'example_user'},
relationships={
'manager': json_asserter.EntityRef(
resource='User',
pk='88e38305-9775-4b34-95d0-4e935bb7156c')}),
included=json_asserter.EntityRef(
resource='User',
pk='88e38305-9775-4b34-95d0-4e935bb7156c',
attributes={'username': 'manager_user'}))This requires the same original record in the response, but now also requires that there be at least one relationship
named manager with the associated User and that User must be present (with at least the one attribute) in the
included property of the response.
The above example utilizes the EntityRef exposed via the json_asserter fixture. This is a reference to a single
entity defined by a combination of: ResourceType, ResourceID, Attributes, and Relationships. When providing the
entity_refs argument to an assertion, you cannot provide any of the following arguments to the assertion directly:
resource, pk, attributes, or relationships.
When providing included json_asserter, you can provide either a single EntityRef or a list of EntityRef instances. If
a list is provided, all referenced entities must be present in the included property of the response. As they do
for the simple usage above, The same assertion rules apply here regarding providing a combination of resource,
pk, and attributes.
The entity_refs parameter can be a list of EntityRef instances as well. However, this is only valid for List
responses. If a list of entity_refs is provided for a non-list response, an assertion will occur. To assert that a
response is a list, the parameter is_list=True must be provided. You can provide either a single EntityRef or a
list of EntityRef instances. If a list is provided, all referenced entities must be present in the list of
returned data.
Support is included for making assertions on plain JSON responses with json_asserter. To ignore the JSON API specific
assertions, you must provide the vnd=False parameter. Only the attributes parameter is valid as there are no
relationships or included properties in a plain json response.
Given this response:
{
"id": "07b374c3-ed9b-4811-901a-d0c5d746f16a",
"name": "example 1",
"field_1": 1,
"owner": {
"username": "user1"
}
}Asserting the top level attributes as well as nested attributes is possible using the following call:
response = api_client.get(self.detail_url)
json_asserter.HTTP_200(response,
vnd=False,
attributes={
'id': '07b374c3-ed9b-4811-901a-d0c5d746f16a',
'owner': {
'username': 'user1'
}
})For a list response:
[{
"username": "user1",
"is_active": False
},
{
"username": "user2",
"is_active": False
},
{
"username": "user3",
"is_active": False
}]It is possible to assert that one or many sets of attributes exist in the response:
response = api_client.get(self.detail_url)
json_asserter.HTTP_200(response,
vnd=False,
is_list=True,
attributes=[{
"username": "user1",
"is_active": False
}, {
"username": "user3",
"is_active": False
}])If there is a class where every test may wish to use the json_asserter, than it may be easier to use to the JsonAsserterMixin found in shipchain_common.test_utils.
This will automatically add the json_asserter and set it as a class attribute before the tests are run.
This allows you to just call self.json_asserter, allowing for cleaner unit tests imports.
When mocking calls, this can help in ensuring all calls, and only those, were made as expected, with the desired parameters.
In order to use, simply import the modified_responses fixture from test_utils.
Then, you just need to register the uris for the calls you want to mock:
@pytest.fixture
def responses_list_mocking(modified_responses):
modified_responses.register_uri(modified_responses.POST, 'http://google.com/path', status=status.HTTP_200_OK)
modified_responses.register_uri(modified_responses.POST, 'http://google.com/other_path',
status=status.HTTP_200_OK)
modified_responses.register_uri(modified_responses.POST, 'http://bing.com/bing_path',
status=status.HTTP_200_OK)
return modified_responsesIn a test that you want to check the calls on, you simply need to use the mocking fixture and call .assert_calls(assertions) on the fixture.
These assertions will be a list of details that the call should have made. An example assertion is this:
{
'path': '/path',
'body': {
'integer': 1
},
'query': {
'query_param_1': 1
},
'host': 'google.com',
}Only the path and the host are required parameters for the assertion. The body and query can be left out, but if included will be tested against. If there is a difference between the amount of calls made and the amount of assertions, no assertion will be made and instead an error will return.