Added support to download example datasets #24
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. Additional details and impacted files@@ Coverage Diff @@
## master #24 +/- ##
=========================================
Coverage 100.00% 100.00%
=========================================
Files 23 24 +1
Lines 1462 1471 +9
=========================================
+ Hits 1462 1471 +9 ☔ View full report in Codecov by Sentry. |
concepts/examples.py
Outdated
|
|
||
|
|
||
| # inspired by https://github.com/mwaskom/seaborn/blob/master/seaborn/utils.py#L524 | ||
| def load_dataset(name: str, data_src: typing.Optional[str] = DATASET_SOURCE, |
There was a problem hiding this comment.
nit: How about keyword-only arguments for eveything except name (and maybe that one even positional-only)?
There was a problem hiding this comment.
Good idea. I implemented this in the latest commits.
concepts/examples.py
Outdated
|
|
||
| # TODO: implement caching here? | ||
|
|
||
| return Context.fromstring(urlopen(url).read().decode(encoding), 'cxt') |
There was a problem hiding this comment.
Maybe we should use a context-manager for urlopen()?
tests/test_examples.py
Outdated
| @@ -0,0 +1,10 @@ | |||
| import pytest | |||
|
|
||
|
|
||
| def test_load_dataset(): | ||
| context = concepts.load_dataset('livingbeings_en') |
There was a problem hiding this comment.
We should probbaly not depend on internet connectivity in the test.
How about having a mocked test and another one doing the actual thing that would be opt-in with a flag?
Something mildy related here: https://github.com/xflr6/graphviz/blob/e5578d39009469df2b7c6743458970643e228226/tests/conftest.py#L5
There was a problem hiding this comment.
I agree but I did not manage to implement this. The examples I found are mainly targeting unittest and not pytest (or pytest with the requests module).
|
Thanks for the PR.
Of course :) In recent times, I did not have much time for my open source projects but it would still be nice to make at least some small improvements. |
Dear Sebastian,
Inspired by other libraries (e.g., Seaborn and scikit-learn) I have added support to download exemplary contexts from the repository https://github.com/fcatools/contexts. For more context, please have a look at this post on the FCA mailing list.
We can discuss details on how to properly handle, curate and implement this but I hope you like the idea. I'd happily make you (co)owner of the conexts repo or https://github.com/orgs/fcatools/ such that you have control over what can be loaded.