There are many ways you could do it. What Id suggest to try is to sequence the tables in the order that you need.
Let's say the first table was "Addresses" and you then needed the FK value of a specific address in the Customers table. In this case I would:
- Load the Addresses table
- Load the Customers data from CSV and then include a Lookup to the Addresses table to match on the non-key columns as required and return the PK value as a new column
- Load the Customers table discarding the non-key Address columns and only including the PK value to enable the FK relationship.
It is extra steps in your pipeline but will ensure you keep full control of the loading and enforce the relationships