@WebServlet - encoding - @MultipartConfig?

[rsoika]

Hi,

I have a strange problem with a WebServlet running in Wildfly 8 and 9.

The servlet is used to extract multipart fileupload requests. The servlet works well in general. It extracts the file content and stores the file in a session parameter. (You can see the sources here.)

The problem is the extraction of the filename from the parts of a mulitpart http request. In case the filename contains german umlaute (ä,ü,ö...) the filename in the header of the corresponding javax.servlet.http.Part has the wrong encoding.

I checked this with firebug and the name is send form the browser with UTF-8 encoding. But my servlet did not receife this encoding and it looks like the encoding of the header has change. 

Is there a known bug with the transmission of multiparts in a servlet or did I missed a configuration isseue?

The same servlet runs well in GlassFish3

Thanks for any hints helping me to find a soution

====

Ralph

[rsoika]

I found this thread Rich:fileUpload and character encoding describing the same problem.

Currently I am running on Wildfly 9.0.1 And adding

<default-encoding>UTF-8</default-encoding>

to jboss-web.xml did not solve the problem.

Only translating the filename with:

fileName = new String(fileName.getBytes("iso-8859-1"), "utf-8"); 

did result in the correct format. But I don't think that this is a platform neutral solution? :-/

===
Ralph

[rsoika]

I now try to solve the problem by guessing the ISO-8859-1 encoding

byte fileNameISOBytes[] = fileName.getBytes("iso-8859-1");
  String fileNameUTF8 = new String(fileNameISOBytes, "UTF-8");
  if (fileName.length() != fileNameUTF8.length()) {
  // convert to utf-8
  logger.fine("filename seems to be ISO-8859-1 encoded");
  fileName = new String(fileName.getBytes("iso-8859-1"),
  "utf-8");
  }

A little bit ugly but it works on Wildfly and GlassFish.

[rouvas]

Try to set this:

request.setCharacterEncoding("UTF8");

in your servlet before actual processing of the request.

[rsoika]

setting

request.setCharacterEncoding("UTF8"); 

was my first try. But this have had no effect. The problem is that the header is receifed and processed by undertow in ISO-8859-1 encoding instead of UTF-8, as expected.