I need to merge a Pandas df (0.22) to ignore NaN in some of the multiple join Keys

Question

So, I have a (long) rule that needs to be applied from an excel/csv import. The rule comes from another system and contains many criteria columns, but these are not always p