I would also recommend M. Reed and B. Simon, Methods of modern mathematical physics, vol I, chapter VIII (and maybe vol II, chapter X). The theorem itself is VIII.14 with Corollary after that. The reference is useful also due to counterexample, demonstrating that the idea about “one-to-one correspondence” used in Wikipedia and some other places is not very good one.