Strange character encoding

Question

I have a project that runs on WebSphere Liberty on Linux. My language is brazilian portuguese and we have some accented words. My java code sets some user messages like below:

...
ErroResponse erroResponse = new ErroResponse();
erroResponse.setMensagem("Esse grupo não pode ser criado. Já existe um grupo criado com esse nome.");
response = Response.status(Status.BAD_REQUEST).entity(erroResponse).build();
...

When the same message is show to the user, it looks this way:

I don't think it's a browser encoding problem because the message looks the same on my server logs.

I've tried to set JVM encoding using -Dclient.encoding.override -Dfile.encoding to ISO-8859-1 and UTF-8, without success.

The same project, running on a Windows server, runs without problem, showing the messages with the correct accents.

Only message that were directly written in the source code has this problem. Accented words that came from database query result are correctly presented.

I'm using Suse 11.4.

I really appreciate any help.

Thanks

One (here unlikely) cause could be that the java source is edited as UTF-8 but compiled as for instance ISO-8859-1. (More often seen with JSPs.) Try by using `\u00E3` instead of `ã` to check this. — Joop Eggen, Jul 24 '17 at 23:04
how do I force it to compile as UTF-8 instead of ISO-8859-1? — Ranieri Mazili, Jul 25 '17 at 16:59
@JoopEggen you're right. After your comment I could find a solution. Thanks. — Ranieri Mazili, Jul 25 '17 at 18:31

score 0 · Answer 1 · answered Jul 24 '17 at 22:49

0

Try to check your Suse encoding and language, by doing:

$ echo $LC_CTYPE
ISO-8859-1

$ echo $LANG
pt_BR

Get all languages:

$ locale -a

Change to pt_PT.utf8:

$ export LC_ALL=pt_PT.utf8 
$ export LANG="$LC_ALL"

answered Jul 24 '17 at 22:49

andreybleme

689
9
23

I've tried to set to pt_BR.utf8 and en_US.iso885915. Still without success. – Ranieri Mazili Jul 25 '17 at 13:45

score 0 · Answer 2 · answered Jul 24 '17 at 22:52

It is likely that before adding those JVM encoding properties, you were getting UTF-8 encoding of that character in the response, but your browser was unwilling to interpret them as UTF-8. I'd suggest deleting them and keeping that within your JAX-RS resources.

A useful debugging step is to look at the respone with a command like client and hex editor (like od -t x1) because there are many things that can obscure what data is actually being transmitted (browser, terminal emulator, etc).

If you explicitly tell the server to use a charset (UTF-8 or a local codepage like ISO8859-1) your strings will be transformed to the specified codepage. The browser will also see the charset in the Content-Type header.

The simplest way is to use @Produces in JAX-RS as in:

@Produces("text/html; charset=UTF-8")

or

@Produces("text/html; charset=ISO8859-1")

score 0 · Accepted Answer · answered Jul 25 '17 at 18:30

0

The problem was solved changing the project text encoding. By default eclipse was saving the source code as UTF-8 and javac was compiling it as CP1252 (Windows Default).

To make this change I had to right click on Project -> Properties -> Resource then changed Text file encoding to Other (ISO-8859-1).

answered Jul 25 '17 at 18:30

Ranieri Mazili

723
6
21

If you use Maven or a similar build tool, the declaration of the correct encoding belongs in the pom.xml (Maven) or build.properties (Gradle) or build.xml (Ant) file so that it survives a reinstallation of Eclipse. – Roland Illig Feb 26 '19 at 02:34

score 0 · Answer 4 · answered Mar 14 '19 at 13:20

I've had a similar problem with maven and jax-ws service which returned bad characters (of text inside source code).

Resolved it by adding this to the parent maven project:

<properties>
    <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
</properties>

Strange character encoding

4 Answers4