Support for Unions in schemas and validation #1227
Conversation
|
hi @karajan1001 thanks for the contribution! Check out the contribution docs to learn how to run all the linters and tests locally to make sure things are working: https://pandera.readthedocs.io/en/stable/CONTRIBUTING.html |
karajan1001
force-pushed
the
fix1152
branch
3 times, most recently
from
4453746 to
091335b
Compare
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #1227 /- ##
==========================================
- Coverage 97.23% 93.72% -3.51%
==========================================
Files 65 90 25
Lines 5066 6714 1648
==========================================
Hits 4926 6293 1367
- Misses 140 421 281 ☔ View full report in Codecov by Sentry. |
The coverage failure is what I mentioned in previous that I didn't find a place to add some negative test examples. |
|
Excuse me, where should I add the negative tests to increase the coverage? |
|
hi @karajan1001, you can find the parts of the code changes that were not covered in the tests in the "Files changed" tab, e.g. here: https://github.com/unionai-oss/pandera/pull/1227/files#diff-2691db6e54a8a776ed24713a7fa20bc1b86b6361e8a7bf7323722e301c359a76R1357 |
pandera/engines/pandas_engine.py
Outdated
| try: | ||
| pandera_dtype = Engine.dtype(pandera_dtype) | ||
| except TypeError: | ||
| return False |
There was a problem hiding this comment.
feel free to create new tests in the test_dtypes.py file to cover the cases that codecov is complaining about.
fix: unionai-oss#1152 I would like pandera to support Union Type. That is the validation of a Series/Column should allow multiple types. 1. Add a new PythonUnion type. 2. Add a new test to for the new UnionType. Signed-off-by: karajan1001 <[email protected]>
| if pandera_dtype not in map(Engine.dtype, pandas_types): | ||
| return False | ||
|
|
||
| if data_container is None: |
There was a problem hiding this comment.
@karajan1001 can we add a test case for when data_container is None?
There was a problem hiding this comment.
Hello, @cosmicBboy .I'm sorry I do not know how to create a test case in which data_container is None.
I tried
"""
pd.DataDrame{[]}
"""
But it just put an empty pd.Series in.
|
any update on this? |
fix: #1152
Make Pandera to support Union Type. That is the validation of a Series/Column should allow multiple types.
It matches all of the requires in the original #1152
And I'm not sure if the tests I added are enough because I didn't know where to add a negative test. Besides this
mypywill fail and I don't know how to handle it.