Two points learned from compiling a new release of the country converter coco.
First and as always, the new developments for the country converter can be found in the CHANGELOG file.
Just to highlight to vaguely related things:
Using pytest for testing administrative package content
I tend to forget to update the version in the version.py file and the one in the CHANGELOG.rst . Most of the time these are not aligned which occasionally leads to some headache for the release on the pypi test server.
So I finally decided to add a test case for this stuff.
By no means something complicated to do once I released that I can use pytest for this task.
So the tests in country converter now include a function which parse the first version number found in the CHANGELOG.rst file and compares it to the version given in the version.py file:
import os import re import sys TESTPATH = os.path.dirname(os.path.abspath(__file__)) sys.path.append(os.path.join(TESTPATH, "..")) import country_converter as coco # noqa CHANGELOG_FILE = os.path.join(TESTPATH, "..", "CHANGELOG.rst") def test_version_consistency(): """Test CHANGELOG.rst latest version consistency with module version""" version_match = re.compile(r"\d+\.\d+\.\d+[a-zA-Z0-9_.]*") with open(CHANGELOG_FILE, "r") as cf: for line in cf: pot_match = re.match(version_match, line) if pot_match: version_changelog = pot_match.group() break else: raise ValueError("No version information found in the CHANGELOG file") assert ( coco.__version__ == version_changelog ), "Version module - CHANGELOG.rst do not match"
For more details see the actual test file here.
Same example script for parsing data for a new classification
The main obstacle for adding a new classification to the country converter is to parse the classification from some given country data file. For version 0.7.3 I did that for the Global Burden of Disease country codes. You can find the gist of it here. For most cases this should be easily adaptable to other classification schemes.
Thats all folks, happy converting!