This file explans about the unittesting done to this utility using pytest. The test coverage for this utility is at 99%.
Testing is done as per two levels of this tool.
test_project2.py tests the pipeline of the complete cuisine prediction and similar dishes finding process.
The file consists of the following tests.
test_project tests the entire workflow of loading the models (creating the models, if they do not pre-exist as well), making cuisine based prediction and gathering given number of similar dishes from the complete yummly.json catlog.
test_model_builders.py tests the functionalities of normalizing ingredients, loading Cusine Predictor, Neighbors finder and Label Encoder models.
The file consists of the following tests and fixtures.
- tests_root
- dummy_raw_data_file
- test_load_raw_data
- test_normalize_ingreds
- dummy_models_folder
- test_fit_dump_le
- test_fit_dump_finder_models
tests_root is a module level fixture, which returns the root directory of tests. This is later used to create and automatically delete models after they are tested.
dummy_raw_data_file is a fixture, which gets the sample yummly.json file's path I have created for testing purpose. This file is a much smaller than the original file so it does not effect the testing performance.
This function tests load_raw_data from the sub-module model_builders.py. This function tests if the loaded dataframe is of the correct shape and contains id, cuisine and ingredients columns.
test_normalize_ingreds tests normalize_ingreds from the sub-module model_builders.py. This function tests if the passed ingredients are normalized in the expected way.
dummy_models_folder is a module level fixture, which setups a folder to store the created and tested models. After the models are tested, they are automatically deleted by this fixture itself.
test_fit_dump_le tests fit_dump_le from the sub-module model_builders.py. This function tests if the Label Encoder is created, fitted and dumped in the expected location or not.
test_fit_dump_finder_models is a parameterized test. It tests fit_dump_cf and fit_dump_nf functions from the sub-module model_builders.py. This function tests if the Cusine Finder and Similar Dishes finder are created and stored in correct location or not. This test also takes care of deleting what ever files it created during testing process.