In a Bayesian analysis, suppose the prior for $\theta$ is $\pi(\theta)$ and the likelihood is $L(D|\theta)$. What is the correct expression for the marginal likelihood $p(D)$?