IIRC you might have some luck adding

df.coalesce(1)

to the last part of your node before it gets returned, that might generate only one file