A big +1 to the accepted answer, and a little tip if you run into encoding issues with the fdf file. If you generate the fields.fdf and upon running
file -bi fields.fdf
you get
application/octet-stream; charset=binary
then you've most likely run into a UTF-16 character set issue.
Try converting the ftf by means of
cat fields.fdf | sed -e's/\x00//g' | sed -e's/\xFE\xFF//g' > better.fdf
I was then able to edit and import the better.fdf file into my PDF form.
Hopefully this saves someone some Google-ing
