DOC: slice_indexer correction + examples #17829
Conversation
jreback
added
Docs
Indexing
topper-123
force-pushed
the
MultiIndex.slice_indexer
branch
from
ccc34b2 to
bb8050a
Compare
Codecov Report
@@ Coverage Diff @@
## master #17829 +/- ##
=========================================
Coverage ? 91.22%
=========================================
Files ? 163
Lines ? 49994
Branches ? 0
=========================================
Hits ? 45606
Misses ? 4388
Partials ? 0
Continue to review full report at Codecov.
|
Codecov Report
@@ Coverage Diff @@
## master #17829 +/- ##
==========================================
- Coverage 91.26% 91.22% -0.05%
==========================================
Files 163 163
Lines 50099 50099
==========================================
- Hits 45724 45703 -21
- Misses 4375 4396 +21
Continue to review full report at Codecov.
|
pandas/core/indexes/base.py
Outdated
| @@ -3412,7 +3412,7 @@ def _get_string_slice(self, key, use_lhs=True, use_rhs=True): | |||
|
|
|||
| def slice_indexer(self, start=None, end=None, step=None, kind=None): | |||
| """ | |||
| For an ordered Index, compute the slice indexer for input labels and | |||
| For an ordered index, compute the slice indexer for input labels and | |||
There was a problem hiding this comment.
I am not fully sure it has to be ordered? When there are no duplicates, I think it doesn't need to be sorted:
In [9]: pd.Index([1, 3, 4, 2, 5]).slice_indexer(1, 2)
Out[9]: slice(0, 4, None)
There was a problem hiding this comment.
Maybe you can add a "Raises" section to specify when a KeyError is raised:
In [10]: pd.Index([1, 3, 4, 2, 5, 2]).slice_indexer(1, 2)
...
KeyError: 'Cannot get right slice bound for non-unique label: 2'
In [11]: pd.Index([1, 3, 4, 2, 5, 2]).slice_indexer(1, 4)
Out[11]: slice(0, 3, None)
pandas/core/indexes/base.py
Outdated
| @@ -3426,11 +3426,23 @@ def slice_indexer(self, start=None, end=None, step=None, kind=None): | |||
|
|
|||
| Returns | |||
| ------- | |||
| indexer : ndarray or slice | |||
| indexer : slice | |||
|
|
|||
| Notes | |||
| ----- | |||
| This function assumes that the data is sorted, so use at your own peril | |||
There was a problem hiding this comment.
Although there is this note .. Not fully sure what it means ..
topper-123
force-pushed
the
MultiIndex.slice_indexer
branch
from
0652130 to
e78f3cd
Compare
|
Ok, I've changed according to the suggestions. I think the warning is ok, because I'm a bit surprised that it's even possible to slice a non-ordered unique index; maybe you'd want that some times, but I bet 9 out of 10 times, someone who slices a index expects it to be ordered. |
|
@topper-123 Thanks! |
The stated return value of
Index.slice_indexerin the doc string is wrong, the method can only return slices. Also added some examples.In addition, some very minor corrections on the user guide text on
IntervalIndex(moved the versionadded to the chapter top + a spelling correction). EDIT: These are not related to the.slice_indexerchanges, but are attached to this PR as the changes are very small.