Replacing various types of empty lists with np.nan

Question

I have the following dataframe which contains a list of strings, an empty list, an empty string within a list, and a string of empty brackets (I guess).

df2 =