I have heard that from a given divisor $D \hookrightarrow X$ (where $X$ is a projective scheme and $D = \sum n_i Y_i$ such that each $Y_i$ is smooth and $Y_i \hookrightarrow X$ is a closed immersion), one can construct an exact sequence $0 \rightarrow \mathcal{O}(-D) \rightarrow \mathcal{O}_X \rightarrow i_{*} \mathcal{O}_D \rightarrow 0$, however what´s $\mathcal{O}_D$ in this case?
For an ordinary irreducible hypersurface, the sequence is clear, but, for a linear combination of such hypersurfaces, I don´t know exactly how to generalize this concept.
Thanks in advance.
One requires that $D$ be an effective divisor, i.e. that $n_i \geq 0$ for all $i$. Then one can identify $D$ with a closed subscheme of $X$. Thus $\mathcal{O}_D$ makes sense. Note that each $Y_i$ need not be smooth, although perhaps in particular applications one might add this in.