BUG: astype(Int64) raises AttributeError #22869
Conversation
|
Hello @sinhrks! Thanks for submitting the PR.
|
pandas/core/internals/blocks.py
Outdated
| @@ -671,15 +671,6 @@ def _astype(self, dtype, copy=False, errors='raise', values=None, | |||
| raise | |||
| newb = self.copy() if copy else self | |||
|
|
|||
| if newb.is_numeric and self.is_numeric: | |||
| if newb.shape != self.shape: | |||
There was a problem hiding this comment.
Does anyone know what case meets the condition?
There was a problem hiding this comment.
I think .astype('i1') on an int64 might hit this
There was a problem hiding this comment.
Thx. It looks work (might depending on OS?)
pd.DataFrame([1]).astype('i1')
To keep the error condition for safety, remove itemsize from the error message? (or add .itemsize to extension dtypes)
pd.Series([1], dtype='int64').dtype.itemsize
# 8
pd.Series([1], dtype='Int64').dtype.itemsize
# AttributeError: 'Int64Dtype' object has no attribute 'itemsize'
|
looks good. can you add on the issue to the whatsnew for integer array (just add to the list), and rebase. |
| @@ -675,11 +675,11 @@ def _astype(self, dtype, copy=False, errors='raise', values=None, | |||
| if newb.shape != self.shape: | |||
| raise TypeError( | |||
| "cannot set astype for copy = [{copy}] for dtype " | |||
| "({dtype} [{itemsize}]) with smaller itemsize than " | |||
| "current ({newb_dtype} [{newb_size}])".format( | |||
| "({dtype} [{shape}]) to different shape " | |||
There was a problem hiding this comment.
Changed the message not to raise AttributeError.
|
Thx for the review. Updated and rebased. |
|
seems there is a lint error. |
Codecov Report
@@ Coverage Diff @@
## master #22869 +/- ##
=======================================
Coverage 92.19% 92.19%
=======================================
Files 169 169
Lines 50911 50911
=======================================
Hits 46939 46939
Misses 3972 3972
Continue to review full report at Codecov.
|
|
Sorry, fixed. |
|
thanks! |
git diff upstream/master -u -- "*.py" | flake8 --diff