yeah, there definitely has to be a way to identify pieces considering they’re most often in one file and we don’t want to write a whole document parsing service for the initial release haha