-
-
Notifications
You must be signed in to change notification settings - Fork 18.7k
Closed
Description
Was playing around with the new string functions and noticed this. Don't really know if it's a bug or expected
np.version.full_version
#'1.6.1'
df = pandas.DataFrame(['jack and jill','jesse and frank'])
split_df = df[0].str.split(r'\s+and\s+')
split_df.to_csv('test')
#<snip>
# 75 shape = values.shape
# 76 result = np.empty(shape, dtype=bool)
#---> 77 vec = lib.isnullobj(values.ravel())
# 78 result[:] = vec.reshape(shape)
# 79
#
#/usr/local/lib/python2.7/dist-packages/pandas-0.8.2.dev_3942eb0-py2.7-linux-x86_64.egg/pandas/lib.so in pandas.lib.isnullobj (pandas/src/tseries.c:5355)()
#
#ValueError: Does not understand character buffer dtype format string ('s')
from pandas import lib
lib.isnullobj(np.array(split_df.ix[0]))
#<snip>
#
#ValueError: Does not understand character buffer dtype format string ('s')