[KARAF-3327] Incorrect LC_CTYPE parsing

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.0.2
• Fix Version/s: 3.0.3, 2.4.1, 4.0.0.M2, 4.0.0.M3
• Component/s: karaf

Environment

Linux

• Assignee: Jean-Baptiste Onofré
• Reporter: Nicolas Dutertry
• Created: 20/Oct/14 13:31
• Updated: 15/Sep/18 16:35
• Resolved: 04/Nov/14 05:17

Description

The code for resolving encoding from LC_CTYPE is not correct.
For instance I have LC_CTYPE=fr_FR@euro and karaf ssh connection is not working. I have found the following exception in the log:

java.lang.RuntimeException: Error opening console reader
	at org.apache.karaf.shell.console.impl.jline.ConsoleImpl.<init>(ConsoleImpl.java:116)
	at org.apache.karaf.shell.console.impl.jline.ConsoleFactoryService.create(ConsoleFactoryService.java:72)
	at Proxy282c46ee_31a7_4e49_9bec_93cb7b913686.create(Unknown Source)
	at Proxy8c51f02f_b7a1_49de_a5fd_3d68070575b3.create(Unknown Source)
	at org.apache.karaf.shell.ssh.ShellFactoryImpl$ShellImpl$1.runConsole(ShellFactoryImpl.java:136)[57:org.apache.karaf.shell.ssh:3.0.2]
	at org.apache.karaf.shell.ssh.ShellFactoryImpl$ShellImpl$1$1.run(ShellFactoryImpl.java:115)
	at java.security.AccessController.doPrivileged(Native Method)[:1.7.0_65]
	at org.apache.karaf.jaas.modules.JaasHelper.doAs(JaasHelper.java:57)[26:org.apache.karaf.jaas.modules:3.0.2]
	at org.apache.karaf.shell.ssh.ShellFactoryImpl$ShellImpl$1.run(ShellFactoryImpl.java:111)[57:org.apache.karaf.shell.ssh:3.0.2]
Caused by: java.io.UnsupportedEncodingException: fr_FR@euro
	at sun.nio.cs.StreamEncoder.forOutputStreamWriter(StreamEncoder.java:61)[:1.7.0_65]
	at java.io.OutputStreamWriter.<init>(OutputStreamWriter.java:100)[:1.7.0_65]
	at jline.console.ConsoleReader.<init>(ConsoleReader.java:231)
	at org.apache.karaf.shell.console.impl.jline.ConsoleImpl.<init>(ConsoleImpl.java:110)
	... 8 more

The issue is in the classes org.apache.karaf.shell.ssh.ShellFactoryImpl and org.apache.karaf.shell.console.impl.jline.LocalConsoleManager. The following code is erroneous:

String encoding = env.getEnv().get("LC_CTYPE");
if (encoding != null && encoding.indexOf('.') > 0) {
    encoding = encoding.substring(encoding.indexOf('.') + 1);
}

You should instead use this kind of code (extracted from jline) :

/**
 * Get the default encoding.  Will first look at the LC_CTYPE environment variable, then the input.encoding
 * system property, then the default charset according to the JVM.
 *
 * @return The default encoding to use when none is specified.
 */
public static String getEncoding() {
	// LC_CTYPE is usually in the form en_US.UTF-8
	String envEncoding = extractEncodingFromCtype(System.getenv("LC_CTYPE"));
	if (envEncoding != null) {
		return envEncoding;
	}
	return System.getProperty("input.encoding", Charset.defaultCharset().name());
}

/**
 * Parses the LC_CTYPE value to extract the encoding according to the POSIX standard, which says that the LC_CTYPE
 * environment variable may be of the format <code>[language[_territory][.codeset][@modifier]]</code>
 *
 * @param ctype The ctype to parse, may be null
 * @return The encoding, if one was present, otherwise null
 */
static String extractEncodingFromCtype(String ctype) {
	if (ctype != null && ctype.indexOf('.') > 0) {
		String encodingAndModifier = ctype.substring(ctype.indexOf('.') + 1);
		if (encodingAndModifier.indexOf('@') > 0) {
			return encodingAndModifier.substring(0, encodingAndModifier.indexOf('@'));
		} else {
			return encodingAndModifier;
		}
	}
	return null;
}