Python unit testing and amazon web services

So you've wrote an awesome python automation tool that utilizes boto, what's next?


We will utilize the following Python libraries, each of them has fantastic documentation:

Setup Script

Each python project should include a, this file outlines the project's package dependencies, version and entry_points.

setup provides us with two dependency declarations:

  • install_requires
  • tests_require

install_requires specifies libraries required to install and unitize the project, for example this project is unable to run without boto3:

install_requires=[ 'boto3' ]

tests_require specifies libraries required to test your project, for example this project requires moto to test:

tests_require=[ 'moto' ]

The version is declared as a string, I suggest using a semver versioning pattern of major, minor, patch:


entry_points allows us to declare console_scripts, these allow us to write slick command line tools. This project provides two console_scripts:

    'console_scripts': [
        's3 = sample_project.s3:main',
        'route53 = sample_project.route53:main'


In this sample project I used boto3 to interface with two popular amazon web service offerings:

  • Simple Storage Service (s3)
  • Domain Name System (route53)

Each of these offerings has a python module in my projects sample_project directory:

Each module includes a main function that is declared as an entry_point within

The s3 command will print out each bucket, and it's content:

$ s3
[ static ]
 => style.css
 => style.js

The route command prints out each domain, and it's records:

$ route53
[ ]

Test Cases

In the I've declared my test_suite referencing the tests directory:


For each of my project modules, I define a test module with the suffix _test:

These modules contain a test case for each function found in the code they test, and are prefixed with test_.

For example, defines a function named list_s3_buckets, where defines a method named test_list_s3_buckets.

When writing your unittest you will be using the assertion methods, that is to say you will call the module function, then verify the returned results are what you expected.

Keep in mind the smaller and more concise a module function, the easier it is to write a test case.

Mocking Amazon Web Services

Rather than running your unittest against amazon web services directly, we will mock these requests. This allows us to run our test without network connectivity, or credentials to the underling amazon web services.

Moto is a fantastic library which mocks our boto3 service calls.

In I wrote a method that performs the following actions:

  • mock s3 bucket creation
  • mock s3 file upload
def __moto_setup(self):
    Simulate s3 file upload

    s3 = get_client()
    s3.put_object(Bucket=self.bucket, Key=self.key, Body=self.value)

Then in the test_list_s3_buckets method, I call list_s3_buckets function and assert the expected bucket is in the return:

def test_list_s3_buckets(self):
    check that our bucket shows as expected

    # setup s3 environment

    buckets = [b for b in list_s3_buckets()]
    self.assertTrue(self.bucket in buckets)

Test Coverage

The python coverage package is a tool for measuring the code coverage of your program.

Without changing anything in your existing tests, coverage will provide an insightful report on how well your unittest cover your application

Rather than running the traditional python test, we will run with coverage:

$ coverage run test
running test
running egg_info
writing requirements to sample_project.egg-info/requires.txt
writing sample_project.egg-info/PKG-INFO
writing top-level names to sample_project.egg-info/top_level.txt
writing dependency_links to sample_project.egg-info/dependency_links.txt
writing entry points to sample_project.egg-info/entry_points.txt
reading manifest file 'sample_project.egg-info/SOURCES.txt'
writing manifest file 'sample_project.egg-info/SOURCES.txt'
running build_ext
test_get_client (tests.route53_test.Route53TestCase) ... ok
test_list_route53_record_sets (tests.route53_test.Route53TestCase) ... ok
test_list_route53_zones (tests.route53_test.Route53TestCase) ... ok
test_main (tests.route53_test.Route53TestCase) ... ok
test_get_client (tests.s3_test.S3TestCase) ... ok
test_list_s3_buckets (tests.s3_test.S3TestCase) ... ok
test_list_s3_objects (tests.s3_test.S3TestCase) ... ok
test_main (tests.s3_test.S3TestCase) ... ok
test_read_s3_object (tests.s3_test.S3TestCase) ... ok

Ran 9 tests in 0.562s


You can then view the test results using the report option:

$ coverage report -m -i --omit=venv/*
Name                         Stmts   Miss  Cover   Missing
sample_project/       0      0   100%
sample_project/       21      0   100%
sample_project/            25      0   100%                         4      0   100%
tests/                0      0   100%
tests/           37      0   100%
tests/                38      0   100%
TOTAL                          125      0   100%

Or you can view the results in a detailed collection of html files:

$ coverage html --omit=venv/*

index.html gives you an entry point to each of your modules coverage.

$ ls -1 htmlcov/