I want to perform a NVL join with pandas, e.g. in SQL:

select * from TA 
join TB on
 TA.column = NVL(TB.column, TA.column)

TA and TB don’t have the same number of rows. Maybe I should use something like combine_first but I don’t know how. What’s the simpliest & effective way to do this in pandas, if possible?

What if there are several NVL conditions?

Thanks