[CENTOS] Japanese file name is garbled when setting LANG environment variable when running Java program on Docker container

Overview

When I was running a Java program on a CentOS Docker container, a mysterious garbled character occurred when I got a list of files containing Japanese filenames.

Sample.java

import java.io.*;

public class Sample {
   public static void main(String[] args) {
      //File name is Japanese file "/sample/AIUEO.Place "csv"
      new File("/sample").listFiles(new FilenameFilter() {
         public boolean accept(File dir, String name) {
            System.out.println(name);   // =>Japanese file name is garbled when getting file list
            return false;
         }
      });
   }
}

By the way, it has been confirmed that garbled characters do not occur when the LANG environment variable is set to ʻen_US.UTF-8, and garbled characters occur when the LANG environment variable is set to ja_JP.UTF-8`.

This article describes the causes of garbled Japanese file names and how to deal with them.

Cause and remedy

First, setting ja_JP.UTF-8 in the LANG environment variable causes garbled characters, which is because the Japanese locale is not registered in the ** CentOS image of Docker **.

You can check the locales that can be specified in the LANG environment variable with the locale -a command. Try running the command inside the container of the CentOS image to check.

# locale -a
C
POSIX
en_US.utf8

As mentioned above, the Japanese locale is not included in the Docker CentOS image container. If you try to get the file list from a Java program by specifying the LANG environment variable as shown below in this container, the Japanese file name will be garbled.

LANG=ja_JP.UTF-8
export LANG

java Sample
=>Garbled Japanese file name.csv

As a workaround, use the localedef command to ** add a Japanese locale ** to eliminate the garbled characters. Add the following command as a RUN instruction in the Dockerfile or run it inside the container.

# localedef -f UTF-8 -i ja_JP ja_JP.UTF-8

Check the locales that can be specified with the locale -a command again.

# locale -a
C
POSIX
en_US.utf8
ja_JP.utf8

The localedef command added ja_JP.utf8. Now, even if you set the LANG environment variable, you can handle Japanese file names without garbled characters.

Conclusion

• Japanese locale is not included in the Docker CentOS image container
• Japanese locale can be added with the localedef command
• If a locale that cannot be specified (does not exist in the environment) is specified from the LANG environment variable, the Japanese file name will be garbled in the Java program.