Deterministic Top-Down Parsing

In Chapter 6 we discussed two general top-down methods: one using breadth-first search and one using depth-first search. These methods have in common the need to search to find derivations, and thus are not efficient. In this chapter and the next we will concentrate on parsers that do not have to search: there will always be only one possibility to choose from. Parsers with this property are called deterministic. Deterministic parsers have several advantages over non-deterministic ones: they are much faster; they produce only one parse tree, so ambiguity is no longer a problem; and this parse tree can be constructed on the fly rather than having to be retrieved afterwards. But there is a penalty: the class of grammars that the deterministic parsing methods are suitable for, while depending on the method chosen, is more restricted than that of the grammars suitable for non-deterministic parsing methods. In particular, only non-ambiguous grammars can be used.

In this chapter we will focus on deterministic top-down methods. As has been explained in Section 3.5.5, there is only one such method, this in contrast with the deterministic bottom-up methods, which will be discussed in the next chapter. From Chapters 3 and 6 we know that in a top-down parser we have a prediction for the rest of the input, and that this prediction has either a terminal symbol in front, in which case we “match”, or a non-terminal, in which case we “predict”.