This chapter explains the steps in designing and coding a working parser. The core design of a parser is the same for all the parsers in this book: recognize a lan-guage and build a result.

* Construct LR(0) sets of items and generate lookahead information for each of those states. * May perform REDUCE rather than ERROR like LR(1), but will catch error before any more input is processed.

Basic idea: LR parser has a stack and input Given contents of stack and k tokens look-ahead parser does one of following operations: Shift: move first input token to top of stack Reduce: top of stack matches …

Basic idea: parser keeps track, simultaneously, of all possible productions that could be matched given what it’s seen so far. When it sees a full production, match it.

We have just discussed the Earley recognizer, not the Earley parser. Right now, we can only detect whether a string is valid by seeing if S → E· @1 is in the last item set. We need to discuss how to …

8.2 Shift/Reduce Parsers A shift/reduce parser is a modified kind of DPDA. Firstly, push moves, called shift moves , are restricted so that exactly one symbol is pushed on top of the stack.

As an application, we build a formally-veri ed parser for the C99 language. Parsing remains an essential component of compilers and other programs that input textual representations of structured data.