Hey! Depending on the complexity and encodings of the PDFs you might have some luck adding a column to simply read the file contents, which Coda search happily picks up on
PDF decoding is hard. To do this properly I’d try finding a node package that parses PDFs to use in a custom Coda pack