How to handle encoding inside ProcessBuilder commands?

Question

A very similar question was posted five years ago at coderanch, but didn't get any answer: http://www.coderanch.com/t/463055/java-io/java/handle-charset-ProcessBuilder

Here is my code:

new ProcessBuilder("/bin/bash", "-c", "echo тест | hexdump").start();

Here is what I'm expecting:

0000000 d1 82 d0 b5 d1 81 d1 82

This is what I'm getting on another Linux machine:

0000000 f2 e5 f1 f2

What is it? How to resolve?

Answer 1

The encoding in ProcessBuilder refers to the file.encoding system property.

For example, when you have the following source:

import java.io.*;

class Main {
    public static void main(String args[]) throws Exception {
        ProcessBuilder pb = new ProcessBuilder("/bin/bash", "-c", "echo -n тест | hexdump -C");
        Process p = pb.start();
        InputStream is = p.getInputStream();
        printInputStream(is);
        p.waitFor();
    }

    public static void printInputStream(InputStream is) throws IOException {
        BufferedReader br = new BufferedReader(new InputStreamReader(is));
        try {
            for (;;) {
                String line = br.readLine();
                if (line == null) break;
                System.out.println(line);
            }
        } finally {
            br.close();
        }
    }
}

You will get f2 e5 f1 f2 if the encoding like Windows-1251 is the default encoding in your runtime environment (or you have set Windows-1251 to file.encoding explicitly).

$ java -Dfile.encoding=Windows-1251 Main
00000000  f2 e5 f1 f2                                       |....|
00000004

You should set UTF-8 to file.encoding explicitly if you want to get d1 82 d0 b5 d1 81 d1 82 as a result.

$ java -Dfile.encoding=UTF-8 Main
00000000  d1 82 d0 b5 d1 81 d1 82                           |........|
00000008