pd.to_datetime raises AttributeError with specific inputs when errors='ignore'

[aktiur]

I'm trying to import a csv file into a PostgreSQL table using odo. During import, odo tries to automatically detect date columns using pd.to_datetime with errors='ignore'. However, with one of my columns (that isn't a date column, it's some kind of zip code), pd.to_datetime raises an AttributeError.

I have been able to narrow down the problem to this small snippet:

>>> pd.to_datetime(pd.Series(['01210',  np.nan]), errors='ignore')
Traceback (most recent call last):
  File "pandas\tslib.pyx", line 1952, in pandas.tslib.array_to_datetime (pandas\tslib.c:35219)
  File "pandas\tslib.pyx", line 1432, in pandas.tslib._check_dts_bounds (pandas\tslib.c:26809)
pandas.tslib.OutOfBoundsDatetime: Out of bounds nanosecond timestamp: 1210-01-01 00:00:00

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "pandas\tslib.pyx", line 1981, in pandas.tslib.array_to_datetime (pandas\tslib.c:35724)
  File "pandas\tslib.pyx", line 1975, in pandas.tslib.array_to_datetime (pandas\tslib.c:35602)
  File "pandas\tslib.pyx", line 1228, in pandas.tslib.convert_to_tsobject (pandas\tslib.c:23563)
  File "pandas\tslib.pyx", line 1432, in pandas.tslib._check_dts_bounds (pandas\tslib.c:26809)
pandas.tslib.OutOfBoundsDatetime: Out of bounds nanosecond timestamp: 1210-01-01 00:00:00

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "C:\Miniconda3\envs\py35\lib\site-packages\pandas\util\decorators.py", line 89, in wrapper
    return func(*args, **kwargs)
  File "C:\Miniconda3\envs\py35\lib\site-packages\pandas\tseries\tools.py", line 276, in to_datetime
    unit=unit, infer_datetime_format=infer_datetime_format)
  File "C:\Miniconda3\envs\py35\lib\site-packages\pandas\tseries\tools.py", line 390, in _to_datetime
    values = _convert_listlike(arg._values, False, format)
  File "C:\Miniconda3\envs\py35\lib\site-packages\pandas\tseries\tools.py", line 372, in _convert_listlike
    require_iso8601=require_iso8601)
  File "pandas\tslib.pyx", line 1847, in pandas.tslib.array_to_datetime (pandas\tslib.c:37155)
  File "pandas\tslib.pyx", line 2005, in pandas.tslib.array_to_datetime (pandas\tslib.c:36116)
AttributeError: 'float' object has no attribute 'view'

output of pd.show_versions()

INSTALLED VERSIONS
------------------
commit: None
python: 3.5.1.final.0
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 61 Stepping 4, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None

pandas: 0.17.1
nose: None
pip: 8.0.2
setuptools: 19.6.2
Cython: None
numpy: 1.10.1
scipy: 0.16.0
statsmodels: None
IPython: 4.0.1
sphinx: 1.3.1
patsy: None
dateutil: 2.4.2
pytz: 2015.7
blosc: None
bottleneck: None
tables: None
numexpr: None
matplotlib: 1.5.0
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml: None
bs4: None
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: None
pymysql: None
psycopg2: 2.6.1 (dt dec pq3 ext lo64)
Jinja2: 2.8

[jorisvandenbossche]

On master, the out of bounds error is not shown anymore, but the AttributeError: 'float' object has no attribute 'view' is still there.

It seems this error is only triggered when there is a NaN and the other string is not valid:

In [7]: pd.to_datetime(pd.Series(['2012-01-01', np.nan]), errors='ignore')
Out[7]:
0   2012-01-01
1          NaT
dtype: datetime64[ns]

In [8]: pd.to_datetime(pd.Series([np.nan]), errors='ignore')
Out[8]:
0   NaT
dtype: datetime64[ns]

In [9]: pd.to_datetime(pd.Series(['1012-01-01', np.nan]), errors='ignore')
AttributeError: 'float' object has no attribute 'view'

In [10]: pd.to_datetime(pd.Series(['1012-01-01']), errors='ignore')
Out[10]:
0    1012-01-01
dtype: object

In [11]: pd.to_datetime(pd.Series(['1012-01-01', '2012-01-01']), errors='ignore'
Out[11]:
0    1012-01-01
1    2012-01-01
dtype: object