[replica masking] how are masking rules applied on the initial copy?

Hello,

From what I understand, the initial copy of the data with replica masking doesn't seem to rely on the COPY command but rather on a C-level binary protocol. If so, how are masking rules applied at that point in time? I get that following the initial copy, if I'm not mistaken, there are pl/pgsql triggers in place that handle it, but i don't understand how it's done for the initial copy.

Thank you in advance for your help!