Object–Oriented Tokens for the Parser Generator jay

jay is a parser generator for JAVA derived from ATTs’ parser generator yacc, which was designed for the C programming language. Being based on yacc, jay is not well adapted for object-orientation, because this principle is not known in the C-world. To correct this flaw jay was redesigned to cope with objects. 1 The Parser Generator jay The parser generator jay [1] is a tool which generates for a given grammar in BNF a respective parser in JAVA. The specification of jay is very similar to that of yacc [2]. This means that it is very easy to learn jay for a yacc user. But there are some significant differences, which arises from the fact that JAVA is object–oriented and the original yacc bases on C: The generated JAVA program is parameterized by the JAVA interface yyInput: boolean advance () throws java.io.IOException; int token (); Object value (); The method advance() reads the next lexem and gives false if end-of-file is reached. The method token() gives the corresponding token. Finally, the method value() is used to get the corresponding object for the jay’s value–stack. This interface must be implemented by the user, himself. For the implementation of advance() and token() a scanner is needed. The simplest possibility is to implement a subclass of the JAVA’s stringtokenizer. More comfortable is for example to use JLex [3]. JLex is a generator for a lexical analyzer. It bases on the well-known lex [2] but produces JAVA code and is written in JAVA. The design approach of JLex is object–oriented, which means that the produced tokens can be real objects (instances of classes, not only ints). In contrast jay accepts, as yacc, only int values as tokens.