BUG: allow conversion of Timestamp and Timedelta to string in astype #9758
Conversation
|
this needs some discussion |
| # GH9757 | ||
| ts = Series([Timestamp('2010-01-04 00:00:00')]) | ||
| ts = ts.astype(str) | ||
| self.assertEqual(ts.dtype, np.object_) |
There was a problem hiding this comment.
this should be a more specific test. assert what this is actuallly converted.
jreback
added
Datetime
|
ok, seems reasonable. can you rebase / squash. ping on green. |
| @@ -66,3 +66,4 @@ Bug Fixes | |||
|
|
|||
|
|
|||
| - Bug in ``Series.quantile`` on empty Series of type ``Datetime`` or ``Timedelta`` (:issue:`9675`) | |||
| - Bug causing ``astype(str)`` to fail for dtype ``datetime64`` or ``timedelta64`` (:issue:`9757`) | |||
There was a problem hiding this comment.
And move this to enhancements (and change wording to say allow astype(str))
| @@ -5511,6 +5511,17 @@ def test_astype_str(self): | |||
| expec = s.map(compat.text_type) | |||
| assert_series_equal(res, expec) | |||
|
|
|||
| # GH9757 | |||
There was a problem hiding this comment.
could you add a test for converting a dataframe, ideally one with multiple dtypes?
Also, this would test for above astype(unicode) and astype(str) on Python 2.
|
Added a test for |
|
OK, this seems good to from my perspective. I'll merge unless there are any further objections.... |
| ns = Series(range(5)) | ||
| fs = Series([0.0, 0.2, 0.4, 0.6, 0.8]) | ||
|
|
||
| df = pd.DataFrame({ |
There was a problem hiding this comment.
just for consistency can u add an index with a timezone (and also do a test on a time stamp with a timezone separately)
|
I've totally borked this branch with an accidental force push. I'll fix it tonight. On the positive side, I finally learned what push.default=simple does :). |
|
Should be fixed. |
BUG: allow conversion of Timestamp and Timedelta to string in astype
|
Thanks @evanpw |
Fixes GH #9757