Concatenate two columns with nan based on condition (python)

Question

I have two columns in a pandas DataFrame that look like this:

ColA	ColB
a	a b c
b	a c
c	NaN
NaN	d e f
NaN	NaN

I would like to concatenate ColA and ColB such that

• if ColA is in ColB then retain ColB
• if ColA is not in ColB concatenate them
• if either are NaN, then retain the column with values
• if both are NaN, then retain NaN

How would I code this in Python such that the desired output would look like below:

ColA	ColB	ColC
a	a b c	a b c
b	a c	b a c
c	NaN	c
NaN	d e f	d e f
NaN	NaN	NaN

Note that 'a' represents a word, and 'a b c' represent three words in a text string

Answer 1

UPDATED answer (using pandas DataFrames): Ok, lets assume you did:

import numpy as np
import pandas as pd

and your DataFrame is as follows:

df
  ColA   ColB
0    a  a b c
1    b    a c
2    c    NaN
3  NaN  d e f
4  NaN    NaN

Then you define your combination function:

def concat(row):
    a = row["ColA"]
    b = row["ColB"]
    if not pd.isnull(a) and pd.isnull(b):
        return a
    if pd.isnull(a) and not pd.isnull(b):
        return b
    if pd.isnull(a) and pd.isnull(b):
        return np.nan
    if a in b:
        return b
    else:
        return a + b

and apply it to your DataFrame (to each row):

df.apply(concat, axis="columns")

what gives this result:

0    a b c
1     ba c
2        c
3    d e f
4      NaN
dtype: object

Of course, you can still think about whether you want to do a simple concatenation with a + in concat, or whether you want to add a space, etc. You get your final result by:

df["ColC"] = df.apply(concat, axis="columns")

Over and out.