unparsing a term

For textual output, we can use unparse rules, specified using pattern matching.

For example:

Const( i )              -> [: i ];
Plus( e1, e2 )          -> [:     e1  " + "  e2     ];
Times( e1, e2 )         -> [:     e1  " * "  e2     ];

Times( e1 = Plus(*, *),
       e2             ) -> [: "(" e1 ") * "  e2     ];
Times( e1,
       e2 = Plus(*, *)) -> [:     e1 "  * (" e2 ")" ];
Times( e1 = Plus(*, *),
       e2 = Plus(*, *)) -> [: "(" e1 ") * (" e2 ")" ];

The first rule can actually be omitted, since it only 'copies' the default behaviour (unparse each subtree in left-to-right order).
Note how the pattern matching is used here to insert parentheses when necessary.

The unparse rules can be used in the following way:


%{ /* make the include files available in the generated code, a la yacc */
#include "unpk.h"         /* unparse (view) defs (generated by kimwitu) */
#include "printer.h"   /* printer fn declaration (generated by kimwitu) */
%}
int main() {
    expr e = Times(
           Plus( Const( 1 ), Const( 1 )),
           Plus( Const( 1 ), Const( 1 )) );
    unparse_expr( e, printer, base_view );
}

views are used to group unparse rules; base_view is the default view.

printer is a user-defined function that takes care of outputting strings. A simple printer function can be:

%{
#include "unpk.h"         /* unparse (view) defs (generated by kimwitu) */
%}
void printer( char*s, uview v)
{ fprintf( stdout, "%s", s ); }

next overview up

Last updated at 03 February '98 by kimwitu@cs.utwente.nl