Interview Alexa
A Python Testing Framework for Alexa using ASK CLI
Installation:
pip install interview-alexaTypical usage:
import unittest
from interview_alexa import say
class TestingAllTheThings(unittest.TestCase):
@say('open my skill')
def test_launch_intent(self, result):
self.assertEqual(result, 'My expected result')
if __name__ == '__main__':
unittest.main()Prerequisites
-
Install
node -
Install and initialize
ask-cli-
https://developer.amazon.com/docs/smapi/quick-start-alexa-skills-kit-command-line-interface.html
npm install -g ask-cli ask init
-
-
Follow the
ask-cliprompts -
Export environment variables:
export SKILL_ID=amzn1.ask.skill.12345
export ASK_DEFAULT_DEVICE_LOCALE=en-US- Clone your existing Alexa skill:
ask clone echo SKILL_IDUsage
Inside the directory of your cloned skill, create a file called tests.py
project
│
└───My_Cloned_Skill
│
└───tests.pyWriting Tests:
import unittest
from interview_alexa import say
class TestingAllTheThings(unittest.TestCase):
@say('open my skill')
def test_launch_intent(self, result):
self.assertEqual(result, "My expected result")
if __name__ == '__main__':
unittest.main()Options:
debug=True
@say('open my skill', debug=True)
def test_launch_intent(self, result):
self.assertEqual(result, "My expected result")
# => will produce a verbose output from ask-cliSimple Testing:
cd My_Skill
python tests.py--
A passing test would output something like:
✓ Simulation created for simulation id: 1234-5679-910112-abc-123
◠ Waiting for simulation response.
----------------------------------------------------------------------
Ran 1 test in 5.848s
OKDialog Testing
Since the Python test runner executes tests alphabetical by test name, you'll want to ensure that any tests that simulate dialog are named alphabetically.
@say('open my skill')
def test_aa__begin_dialog(self, result):
self.assertEqual(result, "My expected result")
@say('do something with my skill')
def test_ab__continue_dialog(self, result):
self.assertEqual(result, "My expected result")If the expected result is a delegated dialog, your response may not include any output speech. In that case, you may want to pass:
@say('do something with my skill')
def test_ac__delegated_dialog(self, result):
passLocal Testing
This package takes advantage of a another great package called python-lambda-local to run tests locally.
pip install python-lambda-localIn order to do so, we use ask-cli to record your request events, and python-lambda-local to test against recorded events.
First, make sure to import the record and localize functions. Then run record in your tests module's setUp method:
import unittest
from interview_alexa import say, record, localize
class TestingAllTheThings(unittest.TestCase):
def setUp(self):
record(self)
@say('open my skill')
def test_aa__launch_intent(self, result):
self.assertEqual(result, 'My expected result')
...Once you've run your test with record mode on, you should see a tmp folder in your working directory with some JSON files with the same names as your tests.
project
│
└───My_Cloned_Skill
│
└───tests.py
│
└───tmp
│
└───__main__.TestingAllTheThings.test_aa.jsonNow that you have some events recorded locally, you can run your tests in localize mode, and run your tests again with python tests.py as you normally would.
import unittest
from interview_alexa import say, record, localize
class TestingAllTheThings(unittest.TestCase):
def setUp(self):
# record(self)
localize(self, 'path/to/lambda/handler.py') # Default is lambda/custom/handler.py
...
