From: "Anthony Lipworth" <LIP AT odie DOT ee DOT wits DOT ac DOT za>
To: jarmo AT ksvltd DOT fi (Jarmo Raiha), djgpp AT sun DOT soe DOT clarkson DOT edu
Date:          Fri, 14 Oct 1994 11:34:46 SAT
Subject:       Re: yyless Lex/Yacc function

First thanks to all who have responded. The byacc (Berkeley yacc) is 
available on Simtel mirrors in .../pgmutil/byacc19.zip  or perhaps a 
more recent version. Unfortunately the problem seems to stem from the 
Flex output as illustarted below.

The makefile contains:

lex.yy.c: lexsource
    lex lexsource
    ex lex.yy.c < fixer

y.tab.c: lex.yy.c yaccsource
    yacc yaccsource

y.tab.o: lex.yy.c y.tab.c

Note the inter-dependencies. I replaced lex.yy.c with lexyy.c and 
y.tab.c with y_tab.c, lex with "flex -l" and yacc with bison. Fixer 
is:

1a
extern int get_sig_char();
.
.,$s/getc(yyin)/get_sig_char()
:wq

which seems to replace getc(yyin) with get_sig_char()

> However ,I doubt that there can be a general library function for 
it. > yyless() interacts heavily with scanner internals.
> Flex implements it as a macro and it is thus available only within flex file.
> See also flex docs for yyless() and example of an infinite loop.

This is an excerpt from the lexyy.c generated by "flex -l lexsource"
(DJGPP port of flex 2.4.7)
The ....... indicates stuff cut out to conserve bandwidth.

----------------------lexyy.c-------------
...........
..........
/* The funky do-while in the following #define is used to turn the 
definition
 * int a single C statement (which needs a semi-colon terminator).  
This
 * avoids problems with code like:
 *
 *  if ( condition_holds )
 *      yyless( 5 );
 *  else
 *      do_something_else();
 *
 * Prior to using the do-while the compiler would get upset at the
 * "else" because it interpreted the "if" statement as being all
 * done when it reached the ';' after the yyless() call.
 */

/* Return all but the first 'n' matched characters back to the input 
stream. */

#define yyless(n) \
    do \
        { \
        /* Undo effects of setting up yytext. */ \
        *yy_cp = yy_hold_char; \
        yy_c_buf_p = yy_cp = yy_bp + n - YY_MORE_ADJ; \
        YY_DO_BEFORE_ACTION; /* set up yytext again */ \
        } \
    while ( 0 )
.......................
.......................
    while ( 1 )     /* loops until end-of-file is reached */
        {
...................
...................
case 37:
YY_USER_ACTION
# line 61 "lexsource"
{
        yyless(yyleng - 1); /* we'll want to send the '(' as it's own 
token */
        strcpy(yylval.sval, yytext);
        return(MATH_FUNC1);
        }
    YY_BREAK
....................
....................
        } /* end of scanning one token */
    } /* end of yylex */
....................
....................
/* Redefine yyless() so it works in section 3 code. */

#undef yyless
#define yyless(n) \
    do \
        { \
        /* Undo effects of setting up yytext. */ \
        yytext[yyleng] = yy_hold_char; \
        yy_c_buf_p = yytext + n - YY_MORE_ADJ; \
        yy_hold_char = *yy_c_buf_p; \
        *yy_c_buf_p = '\0'; \
        yyleng = n; \
        } \
    while ( 0 )
......................
......................
> 

The inline definitions geneertaed by Flex don't look right to me. Any 
comments from the experts? (is a patch required for the lexyy.c 
output from Flex?)

> BTW: do you scan text with 8:th bit set. If so and you don't use
> flex 8 bit option flex used to jam.

Can you elaborate on this. 

Tony.