# Can I do seasonal decomposition on a series on a rolling window basis?

I've been experimenting with the seasonal_decompose() feature from Python's statsmodels (equivalent to the same in R). The problem is that the 'seasonal' series it extracts uses the whole timeseries, where in reality, at some point in the past, such data would not have been available.

Is it possible to produce this seasonal curve without looking to the future?

I am guessing that Python statsmodels can't do this. Is there any other software that could help?