BUG: Convert non-dates in xls date cells to number #13042
Conversation
|
pls add some tests |
| @@ -329,29 +329,38 @@ def _parse_cell(cell_contents, cell_typ): | |||
| appropriate object""" | |||
|
|
|||
| if cell_typ == XL_CELL_DATE: | |||
| if xlrd_0_9_3: | |||
| # Use the newer xlrd datetime handling. | |||
| cell_contents = xldate.xldate_as_datetime(cell_contents, | |||
There was a problem hiding this comment.
rather than a giant try: except: can you narrow down its placement
kordek
force-pushed
the
#10001
branch
2 times, most recently
from
d4ebfdb to
515a5fb
Compare
Current coverage is 84.14%@@ master #13042 diff @@
==========================================
Files 137 137
Lines 50214 50220 +6
Methods 0 0
Messages 0 0
Branches 0 0
==========================================
+ Hits 42253 42256 +3
- Misses 7961 7964 +3
Partials 0 0
|
kordek
force-pushed
the
#10001
branch
3 times, most recently
from
77951ed to
c3523db
Compare
|
Prevoiusly failed tests on the lines that now appear as uncovered according to codecov, I'm not getting this. |
|
if you want to rebase I think I fixed codecov |
|
Thanks, seems ok now |
|
can u post a screen shot of what this excel file looks like |
|
Hashed cell is this 10^n number |
| # GH 10001 : pandas.ExcelFile ignore parse_dates=False | ||
| expected_value = 100000000000000000000 | ||
| act_series = self.get_exceldf('testdateoverflow')['DateColWithBigInt'] | ||
| act_value = act_series.iloc[2] |
There was a problem hiding this comment.
construct the expected frame and use assert_frame_equal
|
Same thing with codecov |
|
you need to rebase on master
you are not picking up the change I made recently. |
kordek
force-pushed
the
#10001
branch
2 times, most recently
from
2441fcb to
05683f6
Compare
|
check now pls |
| # GH 10001 : pandas.ExcelFile ignore parse_dates=False | ||
| refdf = pd.DataFrame([[pd.Timestamp('2016-03-12'), 'Marc Johnson'], | ||
| [pd.Timestamp('2016-03-16'), 'Jack Black'], | ||
| [1e+20, 'Timothy Brown']], |
There was a problem hiding this comment.
I am think this would be converted as an integer, right? (or is it actually a float)
There was a problem hiding this comment.
In the current implementation it is a float (simply cell_contents is returned). Or such way would be more desired ?:
val = int(cell_contents)
if val == cell_contents:
cell_contents = valcloses GH10001 If there is a date column in excel in which there are cells with some big integers, that during parsing to date cause int/long overflow, issue a warning and convert the value to int or float.
|
thanks @kordek |
git diff upstream/master | flake8 --diffcloses GH10001