Bug in lexer_v3
David Relson
relson at osagesoftware.com
Tue Jan 18 13:30:27 CET 2005
Evgeny,
Your modification _does_ change the behavior of the parser. It now
works properly for _your_ test case, but works improperly for other
cases. Using the good.mbx and spam.mbx files in src/tests/inputs try
this:
bogolexer-official -p < good.mbx > test.1
bogolexer-evgeny -p < good.mbx > test.2
diff test.1 test.2
Regards,
David
On Tue, 18 Jan 2005 11:36:30 +0300 Evgeny Kotsuba wrote:
> David Relson wrote:
>
> >Evgeny,
> >
> >Moving "msg_header = false" has no effect in my test. Both old and new
> >versions of bogolexer produce the same tokens.
> >
> >
> yes, I have to comment yy_set_state_initial and put BEGIN INITIAL as
> part of yy_set_state_initial (right in lexer_v3.c, but the same is in
> lexer_v3.l)
>
> case 20:
> YY_RULE_SETUP
> { enum mimetype type = get_content_type();
> have_body = true;
> // msg_header = false;
> clr_tag();
> switch (type) {
> case MIME_TEXT_HTML: BEGIN HTML; break;
> case MIME_MESSAGE:
>
> BEGIN INITIAL;
> /* yy_set_state_initial();*/ break;
>
> default: BEGIN TEXT;
> }
> msg_header = false;
> return EOH;
> }
> YY_BREAK
>
>
> SY,
> EK
--
David Relson Osage Software Systems, Inc.
relson at osagesoftware.com Ann Arbor, MI 48103
www.osagesoftware.com tel: 734.821.8800
More information about the bogofilter-dev
mailing list