I had a problem with a web app that was not handling Unicode strings (Japanese, Korean) correctly.
The thing was, that the Tomcat (7.0.4x) container I was deploying the app into, was (incorrectly) using the ISO-8859-1 encoding for request parameters.
According to the spec, it could be overwritten in conf/server.xml for each Connector with the
URIEncoding="UTF-8" attribute.
Tried it - had no effect ;( I have also found out some other comments in net that it didn't worked for other people as well.
So, I went for manual setting the encoding with:
ServletRequest.setCharacterEncoding("UTF-8");
the problem was, though, that this has to be invoked BEFORE any request.getParameter() method is invoked.
Unfortunately, I had to build my app on top of some crappy webapp a customer had and I couldn't modify the core code. And, of course, this 'core' was playing with request.getParameter() very early in servlet's lifecycle.
So, the solution I come up was just to create a filter which sets the encoding, which is set up early (before my app) in the servlet chain:
web.xml:
...
<filter>
<filter-name>fixEncodingFilter</filter-name>
<filter-class>pack.age.FixEncodingFilter</filter-class>
</filter>
<filter-mapping>
<filter-name>fixEncodingFilter</filter-name>
<url-pattern>/io/*</url-pattern>
</filter-mapping>
...
This fixed the problem, nice and clean.
