When selecting fields with the "select" parameter of cf.read with PP/UM file inputs, the currently selection occurs after all of the input files have been processed. This can be made much more efficient by avoiding processing the non-selected fields.

E.g.

import re
import cf
f = cf.read('~/aaadpa#pg000003385c1+.pp', select=["stash_code=3237", "stash_code=1"])
g = cf.read('~/aaadpa#pg000003385c1+.pp', select=re.compile('^stash_code=\d\d\d\d$'))