- From: Markku Savela <msa@msa.tte.vtt.fi>
- Date: Thu, 12 Dec 1996 12:39:41 +0200 (EET)
- To: www-lib@w3.org
- CC: msa@hemuli.tte.vtt.fi
Hi,
About a month ago I posted a replacement module for the wwwlib SGML.c
module with some changes, most important being
- simplified into "HTML" tokenizer, no error checking at this level
- skips "HTML comments" correctly, those nasty little things like
<!-- <comment> -- something "<--else-->" -- "comment -->
- passes text upstream in bigger chunks (instead of one char at time).
Now, I found a bug in the implementation of the last point. The parser
may pass the same text chunk twice to upstream, if there is "buffer
break" just in front of entity start (&) or tag start (<). I guess
this doesn't happen very often as I noticed it only after using it for
a month... :-)
So, just in case anyone else is using this module, the fix is to add
some zeroing to the 'count' variables into proper places. Here is the
diff against the version I posted earlier.
*** SGML.c.orig Thu Dec 12 12:07:21 1996
--- SGML.c Thu Dec 12 12:11:29 1996
***************
*** 333,338 ****
--- 333,339 ----
{
if (count > 0)
PUTB(text, count);
+ count = 0;
string->size = 0;
context->state = S_ero;
}
***************
*** 340,345 ****
--- 341,347 ----
{
if (count > 0)
PUTB(text, count);
+ count = 0;
string->size = 0;
/* should scrap LITERAL, and use CDATA and
RCDATA -- msa */
***************
*** 359,364 ****
--- 361,367 ----
{
if (count > 0)
PUTB(text, count);
+ count = 0;
string->size = 0;
context->state =
(context->contents == SGML_LITERAL) ?
--
Markku Savela (msa@hemuli.tte.vtt.fi), Technical Research Centre of Finland
Multimedia Systems, P.O.Box 1203,FIN-02044 VTT,http://www.vtt.fi/tte/staff/msa/
Received on Thursday, 12 December 1996 05:39:49 UTC