Sample code showing how to do snapshot testing in Data Science Pipelines and Systems

sample_snapshot_test_in_data_science.py

 import os
 import pickle

 # ...... more code
 # ......

 results = some_calculation()

 expected_result_file_path = 'path_to_the_snapshot_file_of_choice'
 # the snapshot file should not exist at the beginning
 
 if not os.path.isfile(expected_result_file_path):
    # create the snapshot on the first run or anytime it is deleted manually
    with open(expected_result_file_path, 'wb') as expected_results_file:
        pickle.dump(results, expected_results_file)

with open(expected_result_file_path, 'rb') as expected_results_file:
    # the actual test to compare against snapshot
    expected_results = pickle.loads(expected_results_file)

    # here is an assertion basing on a unittest test
    self.assertEqual(results, expected_results)