Comments on: Workaround for bug id 6753651 – find path to jar in cache under webstart

By: Nick

Nick — Thu, 06 Nov 2008 16:22:27 +0000

Looks about right John, the getJarFilePath() should give you the path to the jar file downloaded into the webstart cache.

However it would be worth checking out David’s solution above if you need the path to the jar. I haven’t yet validated this myself, but in theory it uses a public API to get the same information. My solution is a rather nasty workaround to get this same info – it does work fine currently, but since it uses reflection rather than a public API it will be more prone to break in future versions of webstart/the jdk. It uses parts of the webstart library which are not public. If David’s getProtectionDomain() method works it should be less likely to break in the future in this way.

By: john

john — Thu, 06 Nov 2008 10:05:24 +0000

sorry I mean
urlString=URLDecoder.decode(getJarFilePath(myUrl),”UTF8″);

By: john

john — Thu, 06 Nov 2008 10:04:17 +0000

Hi,
Thanks for your code, I think it can help me but I can´t see how, because I´m not an java expert.
I had
URL myUrl = this.getClass().getResource(“Pant.class”);
String urlString=URLDecoder.decode(myUrl.toString(),”UTF8″);

To replace this with your code should I use now
URL myUrl= this.getClass().getResource(“Pant.class”);
String urlString=getJarFilePath(myUrl);
??

Thanks

By: Nick

Nick — Fri, 03 Oct 2008 15:16:00 +0000

thanks for that code David, that looks like another approach worth investigating. Does it also work with 1.6.0_07 I wonder?

By: David Ehrlich

David Ehrlich — Fri, 03 Oct 2008 14:51:15 +0000

While not returning exactly the same result as what you suggest, the following code seems to address the JNLP issue Sun introduced with JDK 1.5.0_16:

public static URL getResource(Class clazz, String name) {
// Get the URL for the resource using the standard behavior
URL result = clazz.getResource(name);

// Check to see that the URL is not null and that it's a JAR URL.
if (result != null && "jar".equalsIgnoreCase(result.getProtocol())) {
// Get the URL to the "clazz" itself. In a JNLP environment, the "getProtectionDomain" call should succeed only with properly signed JARs.
URL classSourceLocationURL = clazz.getProtectionDomain().getCodeSource().getLocation();
// Create a String which embeds the classSourceLocationURL in a JAR URL referencing the desired resource.
String urlString = MessageFormat.format("jar:{0}!/{1}/{2}", classSourceLocationURL.toExternalForm(), packageNameOfClass(clazz).replaceAll("\.", "/"), name);

// Check to see that new URL differs. There's no reason to instantiate a new URL if the external forms are identical (as happens on pre-1.5.0_16 builds of the JDK).
if (urlString.equals(result.toExternalForm()) == false) {
// The URLs are different, try instantiating the new URL.
try {
result = new URL(urlString);
} catch (MalformedURLException malformedURLException) {
throw new RuntimeException(malformedURLException);
}
}
}
return result;
}

public static String packageNameOfClass(Class clazz) {
String result = "";
String className = clazz.getName();
int lastPeriod = className.lastIndexOf(".");

if (lastPeriod > -1) {
result = className.substring(0, lastPeriod);
}
return result;
}