wildcard state \000 rules makes lexer behave weard

[skvadrik]

Hi,

Using a <*> "\000" rule make the parser behave weard : It matches the rule at a place where it shouldn't. I've make a testcase that shows the behaviour. Tarball is available on http://www.palsenberg.com/plain/test-re2c.tar.gz

If you need more info : Please let me know.

Original comment by: palsenberg

[skvadrik]

It seems to go wrong with all single-character matches : \x01 also gives wrong results.

Original comment by: palsenberg

[skvadrik]

Please attach a fail case, the smaller the better. I tried to download your url but got a 404. I am dropping the priority until a fail case is supplied.

Original comment by: paulmcq

[skvadrik]

• priority: 5 --> 1

Original comment by: paulmcq

[skvadrik]

Can't reproduce. This works fine with current master:

/*!re2c
  <c1> *      { default }
  <*>  "\000" { rule-01 }
*/

Run as:

$ ./re2c -ci 3.re 
/* Generated by re2c 1.1.1 on Tue Mar 19 17:52:10 2019 */

{
        YYCTYPE yych;
        switch (YYGETCONDITION()) {
        case yycc1: goto yyc_c1;
        }
/* *********************************** */
yyc_c1:
        if (YYLIMIT <= YYCURSOR) YYFILL(1);
        yych = *YYCURSOR;
        if (yych >= 0x01) goto yy5;
        ++YYCURSOR;
        { rule-01 }
yy5:
        ++YYCURSOR;
        { default }
}

Seems the correct output to me.