select nan value

[fullstop]

I have a dataframe which have 2 columns. They have mutually exclusive nan in the cells. I mean the place where col1 is nan, col2 have some numerical value and vice versa. The situation is shown below

df = pd.DataFrame({"A": [1, 'nan'], "B": ['nan', 2]})

I would like to merge them into single columns with values taken from both,

df = pd.DataFrame({"A": [1, 2]})