Closed LinguList closed 4 years ago
Merging #35 (efbcd36) into master (956b072) will increase coverage by
0.07%
. The diff coverage is77.77%
.
@@ Coverage Diff @@
## master #35 +/- ##
==========================================
+ Coverage 93.78% 93.85% +0.07%
==========================================
Files 32 33 +1
Lines 1528 1563 +35
==========================================
+ Hits 1433 1467 +34
- Misses 95 96 +1
Impacted Files | Coverage Δ | |
---|---|---|
src/pyclts/commands/test_dataset.py | 71.42% <71.42%> (ø) |
|
src/pyclts/commands/make_dataset.py | 76.47% <72.41%> (+48.81%) |
:arrow_up: |
src/pyclts/commands/make_pkg.py | 91.89% <100.00%> (+4.85%) |
:arrow_up: |
src/pyclts/commands/map.py | 42.69% <100.00%> (-16.86%) |
:arrow_down: |
tests/test_cli.py | 100.00% <100.00%> (ø) |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 956b072...efbcd36. Read the comment docs.
@xrotwang, I have no idea why the tests fail: they pass on my system, and the error message is all but clear to me...
The make_dataset
command requires lingpy. So you'd need to add lingpy
to the test deps in setup.py
.
Ah, okay, then I remove it for now. Sorry, I didn't really this.
But you can leave the test in - just add lingpy
here: https://github.com/cldf-clts/pyclts/blob/956b07297b1402c39403b7c40e8f32d4eb24bae9/setup.py#L35-L40
Yes, but it was stupid from me, since lingpy is not even required in these lines, so I removed them now, and we have the advantage of having quicker tests. It was any way onlly dead code that I added there...
Okay, I just updated the test_data.tsv, which is now in concordance with the updates we made (including some changes in the features, and the like).
@xrotwang, I am adding one more test for clts data here, which I consider important, as it allows to check if a given transcription dataset in the
sources
has been properly explicitly coded. This adds one more step to the workflow:If an improper mapping is encountered (either not passing
valid_sound
or not yielding an unknown sound, this now also raises an error. The whole check routine for new transcription data was now added to one single function that is recycled inmake_pkg
which should only be used upon the final release. This yields more consistency, I hope.