how to apply Butterwoth filtering on data with missing entries represented by some NaN?

Question

I want to apply Butterworth filtering to some time series that have dropouts. I work with python / numpy / scipy. I am well aware of the Butterworth filters implemented in Pytho