Add type annotations to Index and MultiIndex #1882
Conversation
| index_map[i], index_map[j], = index_map[j], index_map[i] | ||
| result = DataFrame(self._kdf._internal.copy(index_map=OrderedDict(index_map))).index | ||
| return result | ||
| return result # type: ignore |
There was a problem hiding this comment.
Otherwise, Mypy expects "Index" as the return type.
| index_map = OrderedDict(zip(sdf.columns, names)) | ||
| internal = InternalFrame(spark_frame=sdf, index_map=index_map) | ||
| return DataFrame(internal).index | ||
| return DataFrame(internal).index # type: ignore |
There was a problem hiding this comment.
Otherwise, Mypy expects "Index" as the return type.
There was a problem hiding this comment.
Oh, so maybe it seems that mypy requires us to add a kind of abstract methods to the Index class ?
Maybe can we ignore that rule like we did for inplace parameter ?? 😲
There was a problem hiding this comment.
Thanks @itholic, this is a good idea!
Unfortunately, I didn't find a parameter in https://mypy.readthedocs.io/en/stable/config_file.html can ignore this rule.
I do agree this should a common problem for other projects. Let me do more search and see if there is a better solution.
Codecov Report
@@ Coverage Diff @@
## master #1882 +/- ##
==========================================
- Coverage 94.20% 91.31% -2.89%
==========================================
Files 40 40
Lines 9915 9833 -82
==========================================
- Hits 9340 8979 -361
- Misses 575 854 +279
Continue to review full report at Codecov.
|
databricks/koalas/indexes.py
Outdated
| return isinstance(self.spark.data_type, IntegralType) | ||
|
|
||
| def item(self): | ||
| def item(self) -> Union[Scalar, Tuple]: |
There was a problem hiding this comment.
Actually using Scalar is strictly not correct because this can return a pandas instance too. For exmaple, pd.Timestamp which is not in the Scalar from my cursory reading. But I think it's okay - we can fix it later separately.
There was a problem hiding this comment.
This makes sense. How about we define a pandas Scalar separately? @HyukjinKwon
|
Looks pretty good |
xinrong-meng
force-pushed
the
typeIndex
branch
from
4ef7df9 to
dda6744
Compare
4 participants
No description provided.