Series.str[...] should maybe default to NA if some list/tuple does not contain enough values

[wesm]

[wesm]

Anyone have an opinion on this? Here is an example:

In [8]: s = Series([(1, 2), (1,), (3,4,5)])

In [9]: s.str[0]
Out[9]: 
0    1
1    1
2    3
dtype: int64

In [10]: s.str[1]
---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-10-946f1b211e37> in <module>()
----> 1 s.str[1]

/Users/wesm/code/pandas/pandas/core/strings.pyc in __getitem__(self, key)
    660                               step=key.step)
    661         else:
--> 662             return self.get(key)
    663 
    664     def _wrap_result(self, result):

/Users/wesm/code/pandas/pandas/core/strings.pyc in get(self, i)
    678     @copy(str_get)
    679     def get(self, i):
--> 680         result = str_get(self.series, i)
    681         return self._wrap_result(result)
    682 

/Users/wesm/code/pandas/pandas/core/strings.pyc in str_get(arr, i)
    561     """
    562     f = lambda x: x[i]
--> 563     return _na_map(f, arr)
    564 
    565 

/Users/wesm/code/pandas/pandas/core/strings.pyc in _na_map(f, arr, na_result)
     87 def _na_map(f, arr, na_result=np.nan):
     88     # should really _check_ for NA
---> 89     return _map(f, arr, na_mask=True, na_value=na_result)
     90 
     91 

/Users/wesm/code/pandas/pandas/core/strings.pyc in _map(f, arr, na_mask, na_value)
     96         mask = isnull(arr)
     97         try:
---> 98             result = lib.map_infer_mask(arr, f, mask.view(np.uint8))
     99         except (TypeError, AttributeError):
    100             def g(x):

/Users/wesm/code/pandas/pandas/lib.so in pandas.lib.map_infer_mask (pandas/lib.c:41222)()

/Users/wesm/code/pandas/pandas/core/strings.pyc in <lambda>(x)
    560     items : array
    561     """
--> 562     f = lambda x: x[i]
    563     return _na_map(f, arr)
    564 

IndexError: tuple index out of range

[jreback]

I agree, should default to return na, as you are viewing the Series effectiviely like a Frame here

also note that this is another case of the above (just slightly different exception)

In [6]: s = pd.Series(["foo","b","bar"])

In [7]: s
Out[7]: 
0    foo
1      b
2    bar

In [8]: s.str[0]
Out[8]: 
0    f
1    b
2    b

In [9]: s.str[1]

IndexError: string index out of range