aartgoossens / goldencheetahlib Goto Github PK

requirements.txt contains a lot of packages that are not really required (e.g. jupyter) or required by another package (e.g. numpy that is required by pandas). This file should only contain the bare minimum of packages that are required to use goldencheetahlib.
Also: Requirements need a version range instead of a fixed version. For most requirements this will be until the next major release. E.g. if current version of pandas is 1.18.1, the requirements.txt entry should be pandas>=1.18.1,<0.2.

Setup coveralls.io to track coverage of testing and add status image to readme.

All requests to the GC REST API now use the default Content-Type but because there are multiple options, this should be explicit, also to not break when defaults change in the future.

Data that is not in ACTIVITY_COLUMN_ORDER is now excluded from the activity data by this line:

return activity[[i for i in ACTIVITY_COLUMN_ORDER if i in activity.columns]]

(client.py line 117)

This trick to force ordering of the columns should be deleted or fixed to include all other columns.

Travis CI setup to automate testing and deployment to PyPi.

• Setup testing in .travis.yml
• Add PyPi deployment
• Add Travis CI status image to readme

When using Panda's read_csv without parameters some of the column headers get extra whitespace at the start. This is now fixed in client.py line 87:

activity_list.rename(columns=lambda x: x.lstrip().lower(), inplace=True)

But probably can be fixed in a nicer way by sending sep=', ' (or similar) to read_csv

GC REST API responses are currently not validated and therefore could result in unexpected or unclear errors and exceptions. All responses should be validated and appropriate exceptions should be raised accordingly.
Validations:

• status codes (!=2xx)
• response body (matches requested content-type)
• Other?...