Shannon's sampling theorem in a distributional setting

The classical Shannon sampling theorem states that a signal f with Fourier transform F in L^2® having its support contained in (-\pi,\pi) can be recovered from the sequence of samples (f(n)){n in Z} via f(t)=\sum{n in Z} f(n) (sin(\pi (t -n)))/(\pi (t-n)) (t in R). In this article we prove a generalization of this result under the assumption that F is a compactly supported distribution with its support contained in (-\pi,\pi).