Series.str.cat(',') without "sep" keyword arg raises unintuitive TypeError: len() of unsized object

[hack-c]

I think this easy-to-make mistake should raise a more descriptive error... maybe something like

ValueError: Did you mean to supply a 'sep' keyword?

I might be missing the main usage of str.cat(), but all the uses I've found involve concatenating series together with a sep character.

In [1]: import pandas as pd

In [2]: s = pd.Series(['foo', 'bar', 'baz'])

In [3]: s.str.cat('|')
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-3-8bc5f2a1347b> in <module>()
----> 1 s.str.cat('|')

/home/charles.hack/anaconda/envs/seahydra/lib/python2.7/site-packages/pandas/core/strings.pyc in cat(self, others, sep, na_rep)
   1137     @copy(str_cat)
   1138     def cat(self, others=None, sep=None, na_rep=None):
-> 1139         result = str_cat(self.series, others=others, sep=sep, na_rep=na_rep)
   1140         return self._wrap_result(result)
   1141

/home/charles.hack/anaconda/envs/seahydra/lib/python2.7/site-packages/pandas/core/strings.pyc in str_cat(arr, others, sep, na_rep)
     71         arrays = _get_array_list(arr, others)
     72
---> 73         n = _length_check(arrays)
     74         masks = np.array([isnull(x) for x in arrays])
     75         cats = None

/home/charles.hack/anaconda/envs/seahydra/lib/python2.7/site-packages/pandas/core/strings.pyc in _length_check(others)
    111         if n is None:
    112             n = len(x)
--> 113         elif len(x) != n:
    114             raise ValueError('All arrays must be same length')
    115

TypeError: len() of unsized object

In [4]: s.str.cat(sep='|')
Out[4]: 'foo|bar|baz'

[jreback]

sounds reasonable. pull-request?

[hack-c]

Will do.

[hack-c]

I propose to fix this by changing the order of the keyword args, like so:

def str_cat(arr, sep=None, others=None, na_rep=None):
    """
    Concatenate strings in the Series/Index with given separator.

    Parameters
    ----------
    sep : string or None, default None
    others : list-like, or list of list-likes
      If None, returns str concatenating strings of the Series
    na_rep : string or None, default None
        If None, an NA in any array will propagate

So that sep comes first.

This is an API change though and breaks some tests, because str_cat used to expect the others argument first. (I'm not sure why.)

@jreback, does that sound reasonable?

That way,

    >>> Series(['a', 'b', 'c']).str.cat('|')
    'a|b|c'

    >>> Series(['a','b',np.nan,'c']).str.cat(' ', na_rep='?')
    'a b ? c'

    >>> Series(['a', 'b', 'c']).str.cat(',', others=['A', 'B', 'C'])
    0    a,A
    1    b,B
    2    c,C
    dtype: object


    >>> Series(['a', 'b']).str.cat(others=[['x', 'y'], ['1', '2']], sep=',')
    0    a,x,1
    1    b,y,2
    dtype: object

[jreback]

@hack-c no, that doesn't make sense, why just catch and raise a nice error?

[hack-c]

Ok. :) I just like that API design better but realize it's late in the game for a breaking change like that.

[jreback]

I think you can check for this error conditiion and give a really helpful error (e.g. you can even say, did you mean 'sep')