Python for Bayesian networks, which library do you recommend?

What is considered the best python library to learn Bayesian networks for supervised classification? I see that scikit learn has naive Bayesian networks, but I would also like to define the maximum number of parents a node can have.

p.s. if there's a pretty awesome library for R to do this then I'd be open to using that, but my preference would be python. Is there a scikit learn/weka equivalent in R?

EDIT: I'd like to obtain a network like this for supervised classification (being able to define the maximum number of parents a child node can have). The example is from weka, but I'd really like a python (or R) equivalent.