- From: Jungshik Shin <jshin@i18nl10n.com>
- Date: Sat, 17 Jul 2004 06:25:13 +0900
- To: www-international@w3.org
Hi,
I've been wrestling with a mysterious problem for the last few hours. I
made a patch to
the web search front-end of 'Nutch' (http://www.nutch.org an open
source search engine
that strives be an open source google [1]) so that query strings made of
characters outside ISO-8859-1
character repertoire can work.
Following the standard-step of adding contentType and pageEncoding
directives at the beginning
of jsp files (I also added request.setCharacterEncoding("UTF-8"); along
with making sure that
that's honored because recent versions of Apache tomcat by default
ignores that for GET),
I expected everything to work. To my great surprise, all the JSP files with
'contentType="text/html; charset=UTF-8"' directive still emit
'Content-Type:text/html; charset=ISO-8859-1'
in HTTP header. Even more surprsing is that cached versions of
translated java source files for
those jsp files have the following line:
response.setContenttype("text/html; charset=UTF-8");
It's completely beyond me how I've been getting 'text/html;
charset=ISO-8859-1' despite that.
You can try it at http://pippin.kaist.ac.kr:8080. I ran nutch crawler to
fetch a small number (about
4000) of pages in several different scripts (if you give '1234' as a
query, you'll get 4 hits). The
search result page(handled by search.jsp) is supposed to be in UTF-8
with the correct C-T header
emitted in HTTP header.
Is there anyone who's been beaten by this bizzare problem? It'd be great
to know how that was solved.
Thank you tons in advance,
Jungshik
[1] Needless to say, there are a number of things to improve in I18N as
well as in other aspects before Nutch can compete with Google.
Received on Friday, 16 July 2004 17:25:17 UTC