flex & bison

Chapter 6. A Reference for Bison Specifications

In this chapter we describe the syntax of bison programs, along with the various options and support functions available. POSIX yacc is almost an exact subset of bison, so we note which parts of bison are extensions beyond what POSIX requires.

After the section on the structure of a bison grammar, the sections in this chapter are in alphabetical order by feature.

Structure of a Bison Grammar

A bison grammar consists of three sections: the definition section, the rules section, and the user subroutines section.

    ... definition section ...
    %%
    ... rules section ...
    %%
    ... user subroutines section ...

The sections are separated by lines consisting of two percent signs. The first two sections are required, although a section may be empty. The third section and the preceding %% line may be omitted.

Symbols

A bison grammar is constructed from symbols, which are the “words” of the grammar. Symbols are strings of letters, digits, periods, and underscores that do not start with a digit. The symbol error is reserved for error recovery; otherwise, bison attaches no fixed meaning to any symbol. (Since bison defines a C preprocessor symbol for each token, you also need to be sure token names don’t collide with C reserved words or bison’s own symbols such as yyparse, or strange errors will ensue.)

Symbols produced by the lexer are called terminal symbols or tokens. Those that are defined on the left-hand side of rules are called nonterminal symbols or