Internationalization problems

Question

Sorry if this has been already answered, but there are tons of similar (but not same) questions around about internationalization and character sets and I haven't been able to solve the problem.
First, a summary about translating messages and internationalization - correct me if I'm wrong:
I have to translate the following:
1) Jhs Exception messages (in JhsUserMessages.java)
2) messages.js messages
3) user interface texts
For point 1) I created my own JhsUserMessages_gr.java with the translations of Jhs messages. It works, but I had to include it seperately as a property of the MessageResources factory in struts_config.xml . This doesn't scale well and doesn't seem correct. Am I doing something wrong? Anyway, the correct messages do appear in greek.
For 2), I have a problem. messages.js states that special characters have to be inserted with their codes - so I'll have to use an automated tool to generate Unicode codes for my greek sentences and then include these unicode codes in messages.js. Am I right? Is there any simpler way?
In 3) I have the most serious problem:
I saw that JHeadstart uses ${nls.SOMETHING} in its generated code. I tried the same in my project (which is JHS-enabled, but not automatically generated).
I had the following problem: if I used greek in application.properties_gr (or was it application_gr.properties?) the characters were appearing incorrectly. I read that I should avoid special characters in application.properties files (what's their use, then? Translating from english to american? ;)). So I created a custom bundle class (Tags.java) and included it in the page as a bundle. Of course, this doesn't solve the problem correctly (because I cannot find a way to dynamically choose between Tags_en and Tags_gr for example). Is there such a way? I suppose that it should be some special declarations of the bundles in configuration files or overriding some function, but I don't know more.

Answer 1

Hello
I guess your problem is encoding. There's many possible points of failure if you're using many 3rd party components. However there's a couple of things you should try at first:
Set the content-type of your response to UTF-8:
     response.setContentType("text/html; charset=UTF-8");
You can add that line in the begin of your servlet's execution point - doGet-method - or any preprosessor filter you've any. Point is that you set content-type before you've committed any data to output stream.
The other trick is to set system property "file.encoding". You can use this from command line using flag "-Dfile.encoding=UTF8" when launching java-application (see tomcat's launching bat to set this). Or you can tell this explicitly from you application:
     System.setProperty("file.encoding", "UTF-8");
With JSP-pages you should use page-directive to set content-type to utf8.
I really hope this helps. Encoding is pretty tricky with different servers. Oh, and of course there are couple encoding-related bugs in the specs... :o
Latka