A few posts ago, we published a blog about how to use the Jira API. We did not write any unit tests for the application we wrote and that is exactly what we are going to do now. More specifically, we will focus on how we can unit test a Rest API.
1. Why Unit Tests Anyway?
Our main focus when writing software is building new features, fixing bugs. Of course, we need to test what we built, but we get the most joyful moment when our new developed feature works. Next step is to write unit tests… But, we already know it is working, so why spend much effort in it? It is much more fun to start with the next feature, right? So, we skip writing unit tests. But what happens when we need to extend the feature we wrote with some new functionality? How are we going to ensure we did not break the original feature? What if we want to refactor the code we have written? These are the moments we are glad when we notice someone took the effort to write unit tests. Unit tests will give us the confidence not breaking anything when changing the code (assuming the unit tests are well written and provide enough code coverage). It is therefore also common practice to run the unit tests as part of your CI/CD pipeline. When you do not like writing unit tests afterwards, you can also consider writing your unit tests first, let them fail and when you implement your code, the unit tests will pass (TDD: Test Driven Development). This way, when you have reached your uppermost moment of joy (It works!), you also have your unit tests passed (All the unit tests passed!). That doubles your moment of joy 🙂 .
2. Create Your First Unit Test
We will build upon the sources of the Jira time report generator. We are using Python 3.7 and PyCharm as IDE. First, let’s create a test
directory and right-click the directory in PyCharm. Choose New - Python File
and Python unit test
. This creates the following default file:
import unittest class MyTestCase(unittest.TestCase): def test_something(self): self.assertEqual(True, False) if __name__ == '__main__': unittest.main()
Running this unit test obviously fails (True does not equals False), but we do have set up the basics for writing our own unit tests now.
3. Mocking a Rest API
We want to unit test the get_updated_issues
function and this provides us a first challenge: the get_updated_issues
function contains a call to the Jira Rest API. We do not want our unit test to be dependent of a third party service and therefore we need a way to mock the Rest API. There are several options to mock a Rest API, but we will make use of the requests-mock Python library which fits our needs.
Install the requests-mock
Python library:
pip install requests_mock
3.1 Test Single Page Response
The get_updated_issues
function will request the issues which are updated in a certain time period. In our unit test, we will verify the behavior when one page with results is retrieved (the Rest API supports pagination, but that is something for a next unit test):
def test_get_updated_issues_one_page(self): with open("issues_one_page.json", "r") as issues_file: mock_response = issues_file.read() expected_result = [ {'expand': 'operations,versionedRepresentations,editmeta,changelog,renderedFields', 'id': '10005', 'self': 'https://jira_url/rest/api/2/issue/10005', 'key': 'MYB-5'}, {'expand': 'operations,versionedRepresentations,editmeta,changelog,renderedFields', 'id': '10004', 'self': 'https://jira_url/rest/api/2/issue/10004', 'key': 'MYB-4'}] with requests_mock.Mocker() as m: m.register_uri('GET', '/rest/api/2/search', text=mock_response) response = jiratimereport.get_updated_issues("https://jira_url", "user_name", "api_token", "MYB", "2020-01-10", "2020-01-20", "") self.assertEqual(expected_result, response)
Let’s take a closer look at what is happening here. We have defined the Jira JSON response in file issues_one_page.json
. At line 2 and 3, we read the contents of the file into variable mock_response
. In line 5, we define the expected result with variable expected_result
when the function get_updated_issues
returns. At lines 11 up to 13 the magic happens. We register the URI we call from within the get_updated_issues
function with the Mocker
we defined. The third parameter of register_uri
defines the response which should be returned from the mocked API call. At line 15 we call the get_updated_issues
function and at the end we verify whether the response equals the expected result.
3.2 Test Paginated Response
The Jira API supports pagination. We added some functionality to the get_updated_issues
function in order to handle paginated responses. The JSON response contains three fields for this:
startAt
: indicates from which result the page should be retrieved;maxResults
: the number of maximum results which are returned within one response;total
: the total number of results.
The maxResults
for retrieving issues is, at the time of writing, 50. If we would like to test this against a real Jira server, we would have to create at least 51 issues. But for testing with our unit test, we can easily change the response in order that field maxResults
returns 2 for example, which makes it a lot easier to test. Our paginated unit test looks as follows:
def test_get_updated_issues_multiple_pages(self): with open("issues_multiple_first_page.json", "r") as issues_first_file: mock_response_first_page = issues_first_file.read() with open("issues_multiple_second_page.json", "r") as issues_second_file: mock_response_second_page = issues_second_file.read() expected_result = [ {'expand': 'operations,versionedRepresentations,editmeta,changelog,renderedFields', 'id': '10005', 'self': 'https://jira_url/rest/api/2/issue/10005', 'key': 'MYB-5'}, {'expand': 'operations,versionedRepresentations,editmeta,changelog,renderedFields', 'id': '10004', 'self': 'https://jira_url/rest/api/2/issue/10004', 'key': 'MYB-4'}, {'expand': 'operations,versionedRepresentations,editmeta,changelog,renderedFields', 'id': '10006', 'self': 'https://jira_url/rest/api/2/issue/10006', 'key': 'MYB-6'}] with requests_mock.Mocker() as m: m.register_uri('GET', '/rest/api/2/search', [{'text': mock_response_first_page}, {'text': mock_response_second_page}]) response = jiratimereport.get_updated_issues("https://jira_url", "user_name", "api_token", "MYB", "2020-01-10", "2020-01-20", "") self.assertEqual(expected_result, response)
The unit test looks quite similar to the one for the single page response. We defined two mock responses this time, because the Jira API will be called twice and we want to return two different responses. The expected_result
variable contains the combined result of both API calls. The real difference can be seen at line 17. Instead of defining a single mock response, we now define a list of mock responses. When the API is called more than the defined responses, the latest defined mock response is returned again.
3.3 Test Failed
The unit tests above all pass. But do they fail when something is wrong? We can therefore change e.g. the following line in the get_updated_issues
function:
issues_json.extend(response_json['issues'])
Change it with:
issues_json = response_json['issues']
This will ensure that only the response of the first API call will be added to the returned issues list, but not the response of succeeding API calls. Run both tests again. The test_get_updated_issues_one_page
passes, but the test_get_updated_issues_multiple_pages
fails:
AssertionError: Lists differ
3.4 Test Multiple URI’s
The Jira work logs are to be retrieved per issue. We therefore need to register more than one URI with different responses for each issue. The response of the get_work_logs
function returns a list of WorkLog
objects which we can assert.
def test_get_work_logs_one_page(self): with open("work_logs_first_issue_one_page.json", "r") as first_issue_file: mock_response_first_issue = first_issue_file.read() with open("work_logs_second_issue_one_page.json", "r") as second_issue_file: mock_response_second_issue = second_issue_file.read() issues_json = [ {'expand': 'operations,versionedRepresentations,editmeta,changelog,renderedFields', 'id': '10005', 'self': 'https://jira_url/rest/api/2/issue/10005', 'key': 'MYB-5'}, {'expand': 'operations,versionedRepresentations,editmeta,changelog,renderedFields', 'id': '10004', 'self': 'https://jira_url/rest/api/2/issue/10004', 'key': 'MYB-4'}] with requests_mock.Mocker() as m: m.register_uri('GET', '/rest/api/2/issue/MYB-5/worklog/', text=mock_response_first_issue) m.register_uri('GET', '/rest/api/2/issue/MYB-4/worklog/', text=mock_response_second_issue) work_logs = jiratimereport.get_work_logs("https://jira_url", "user_name", "api_token", "2020-01-10", "2020-01-20", "", issues_json) self.assertEqual(work_logs[0], WorkLog("MYB-5", datetime(2020, 1, 18), 3600, "John Doe")) self.assertEqual(work_logs[1], WorkLog("MYB-5", datetime(2020, 1, 18), 5400, "John Doe")) self.assertEqual(work_logs[2], WorkLog("MYB-4", datetime(2020, 1, 12), 3600, "John Doe"))
4. Conclusion
Writing unit tests is absolutely necessary when you want to develop software in a professional manner. In this post, we took a look at how to unit test a Rest API by means of the requests-mock
Python library. We only scratched the surface of what this library has to offer, but our first impressions are very good.
Your Way Of Explanation was good. You add some valuable points. Here I am sharing some information about python.
Python software engineers are called as Pythonists. Along these lines, trying Pythonists, we should plunge into an ocean of learning and gain proficiency with the significant highlights of Python programming language.
Highlights of Python Programming Language:
Coherence: Python has a basic and exquisite sentence structure. Lucidness keeps up a huge number of lines of code and you compose less code when contrasted with other programming dialects like C, Java, or C++.
Free and Open Source: Python is unreservedly accessible on practically all mainstream Operating Systems. On the off chance that you are utilizing Mac OS, Windows, or any Unix-based OS, for example, Linux or Ubuntu you have a Python mediator in your Operating System as a matter of course.
This is only one language that has power over any other language. It is called as All in one language. As python is much simpler and easy to code language than any other. python training institutes in Hyderabad help in programming, scripting and web development, application development etc. Python is a universally useful programming language planned and created by Guido Van Rossum. The name Python has its cause from Guido Van Rossum’s #1 TV program “Monty Python’s Flying Circus”. It doesn’t make any difference on the off chance that you are new to programming or have been modifying in different dialects. Python is not difficult to learn and code.
LikeLike