Flask brand API endpoint unit test

Question

# unit tests for the /brands endpoint

from flask.ext import restful
import unittest
import os
import json
import sys

# tells sql_models to use the testing database
os.environ['API_TESTING'] = '1'
sys.path.append('/opt/backend_api/')

import api
from models import sql_models
from config import settings

class TestBrandsEndpoint(unittest.TestCase):

    def setUp(self):
        ''' fire up a test instance of the flask app '''
        self.app = api.APP.test_client()
        self.app.testing = True
        self.api = api.API
        self.test_brand_name = 'Momcorp'

    def tearDown(self):
        ''' 
            tear down app, reverse any changes made 
            - remove the brand inserted in test_post()
        '''
        brand_object = sql_models.Brand.get_by_name(self.test_brand_name)
        if brand_object is not None:
            sql_models.Brand.delete_brand(brand_object)
        pass

    def test_post(self):
        '''
            insert a new brand into the database
            - only given field for test is the mandatory brand name but could(should?) be extended for the rest
            - test that the response status code == 201 and the returned brand object has the correct name
        '''
        result = self.app.post( '/' + settings.BASE_API_VERSION + '/brands', 
                                headers=[('user-api-key', '<redacted>')],
                                data={'name':self.test_brand_name})
        # print(json.loads(result.data))
        self.assertEqual(result.status_code, 201)
        self.assertEqual(json.loads(result.data)['brand']['name'], self.test_brand_name)

    def test_get(self):
        ''' 
            get all brands and all currently running offers
            - sets the brand's 'offers_running' status
            - returns serialized brands like so: {'brands':[serialized_brands]}
            - test status code + name, id fields are not null
        '''
        result = self.app.get('/' + settings.BASE_API_VERSION + '/brands', headers=[('user-api-key','<redacted>')])
        self.assertEqual(result.status_code, 200)
        for brand in json.loads(result.data)['brands']:
            self.assertIsNotNone(brand['name'])
            self.assertIsNotNone(brand['id'])


if __name__ == '__main__':
    unittest.main()

Looking for advice about the 'right way' to do this. Not sure if I should have a separate file and class for each endpoint or if all endpoints+request types should just be their own methods in a larger TestAllEndpoints(unittest.TestCase) class. Any advice or opinions are appreciated.

Answer 1

• Move the os.environ and sys.path calls into a new setUpClass method instead of have them at the global level?
• pass at the end of setUp should be removed, similarly the commented out print.
• I'd put the constants for the URL and the headers into fields or methods (self.url() / self.url) to reduce duplication.
• The docstrings are a bit much. Ideally the code would be very self explanatory so that it almost spells out what you currently have (duplicated) in the docstring.
• I'd put things that belong together in the same file; so same endpoint and requests definitely in the same file and probably in the same test class as well. I'd go for practicality first - if you need to have convoluted setup because you want to handle many different tests, then split it into multiple classes such that the tests are again easy to read.