Migrating Java application from HP/UX to Linux

JayMac2013 · 04-07-2013, 11:59 AM

I have a Java application which is currently running on an HP/UX box and is being migrated to Suse Linux. The application must translate data from ASCII to EBCDIC to send to a mainframe legacy app. If I have the code line below in in HP/UX, it works fine:

String transactionName = new String(transNameByteArray);

However, executing this same code in Linux produces a String which is neither ASCII nor EBCDIC valid data. I understand that I may need to force the code page used as it may be different on HP/UX than on Linux but I am not sure what the code page would be. The line below doesn't correct the issue:

String transactionName = new String(transNameByteArray, "ISO-8859-1");

ntubski · 04-07-2013, 01:30 PM

Quote:

However, executing this same code in Linux produces a String which is neither ASCII nor EBCDIC valid data.

A String is Unicode.

ASCII <-> EBCDIC conversion (I used IBM500 because the EBCDIC Wikipedia page uses it in the example):

Code:

import java.math.*;
import java.nio.charset.*;

public class Ascii2Ebcdic {
    public static byte[] conv(Charset from, Charset to, byte[] bytes) {
        return new String(bytes, from).getBytes(to);
    }
    public static void main(String[] args) throws Exception {
        Charset ascii = Charset.forName("US-ASCII");
        Charset ebcdic = Charset.forName("IBM500");

        byte[] abc_ascii = { 0x61, 0x62, 0x63 };
        byte[] xyz_ebcdic = { (byte)0xA7, (byte)0xA8, (byte)0xA9 };
        System.out.write(conv(ascii, ebcdic, abc_ascii));
        System.out.write(conv(ebcdic, ascii, xyz_ebcdic));
    }
}

Output (which is "abc" in EBCDIC followed by "xyz" in ASCII):

Code:

% java Ascii2Ebcdic | od -t x1
0000000 81 82 83 78 79 7a
0000006