Suggestion for filtering by calendar?

I have a time-series dataset for which I need to make several plots;
some of them will use all of the data, but others need not only to
show just the usual business hours, but also take business holidays
into account. I thoght this might be sufficiently common in this
community that somebody could easily point me to a python idiom/recipe
they've used that I could extend to use masked arrays.

FWIW, the idea is not that I want to skip those time periods on the
visualizations--and thus the recipes which treat the data as
non-timeseries and construct custom ticks are not the solution for me.
One of the tasks will be to use a fill to shade in these "working hours"
for some of the plots. Another will be to calculate and plot running
averages that exclude non-working hours.