Issue 4: Feature/property based testing

[frank113]

Summary

This PR partially addresses issue #4 by adding property-based tests utilizing hypothesis and pytest.

Issues Addressed

• 4

Changes Made

1. Add hypothesis in both poetry.lock and under dev in pyproject.toml
2. Add custom property generation strategies in tests/custom_strategies.py
3. Add documentation for added tests
4. Add property based tests for:
   • iso3c to iso3n conversion. This replaces the prior test_numeric unit test
   • iso3c to fips conversion
   • Tests to ensure that when an invalid code is supplied the countrycode function will return None
5. Test to ensure that the codelist dataframe has dimensions of 291 by 624. This test is included to protect against small, accidentical changes to the underlying csv file.

Considerations

• The extent of our property-based framework depends on the number of scenarios we wish to test. As of now we can test most conversions, such as iso3c to iso3n using the provided tests as a boilerplate. Given the repetitive nature of the property-based tests these tests can be present, albeit not exhaustive.
• Edge cases that do not correspond to the codelist can be accounted for within this framework

Other

Looking forward to learning more about the plot functionality you mentioned in an issue thread!

[vincentarelbundock]

This looks fantastic! Thanks a ton for taking the time to write this. I really appreciate it!

Ooops, sounds like miscommunication. Adrien and I are not working on plots for pycountrycode. We are improving the Python port of my marginaleffects package for R. Eventually, the online "book" will be fully bilingual R/Python, like this page:

https://marginaleffects.com/articles/marginaleffects.html

[frank113]

The R package is quite robust, thanks for the heads up! The bilingual approach seems to be increasingly common in the data world -- curious to see how the split settles.