python 3.x - rsplit on pandas series with regular expression not working

Question

rsplit on pandas series using regular expression not working. I want to split the series based on separator without removing separator.

df2= pd.Series(['Series of Class A','Series of Class B part of Class C','Class D','Class'])
seperator='Class'
data = df2.str.split(r'.(?='+seperator+')', n = 2, expand=True)

result is:

 0                1        2
0  Series of          Class A     None
1  Series of  Class B part of  Class C
2    Class D             None     None
3      Class             None     None

I want to do same thing using rsplit

I tried

data = df2.str.rsplit(r'.(?='+seperator+')', n = 2, expand=True)

Expecting same result using rsplit

 0                1        2
0  Series of          Class A     None
1  Series of  Class B part of  Class C
2    Class D             None     None
3      Class             None     None

Answer 1

Unfortunately, pd.Series.str.rsplit does not work as documented (v0.25, stable/v1+). The project's GitHub issue tracker has an open bug from Nov. 2019 that repots that rsplit is not working with regex patterns (v 0.24.2 and 0.25.2). Internally, the method is calling str.rsplit which does not support regular expressions.

Luckily, the reporter jamespreed added a (homegrown) alternative function:

def str_rsplit(arr, pat=None, n=None):

    if pat is None or len(pat) == 1:
        if n is None or n == 0:
            n = -1
        f = lambda x: x.rsplit(pat, n)
    else:
        if n is None or n == -1:
            n = 0
        regex = re.compile(pat)
        def f(x):
            s = regex.split(x)
            a, b = s[:-n], s[-n:]
            if not a:
                return b
            ix = 0
            for a_ in a:
                ix = x.find(a_, ix) + len(a_)
            x_ = [x[:ix]]
            return x_ + b
    return f
    res = _na_map(f, arr)
    return res