Possible bug in JDK URL class? Certainly the behaviour seems odd. I would expect URL.getFile() to url unencode the url.

gistfile1.java

// Make sure you have a directory which has a space in it's name, e.g:
File f = new File("/home/tim/dir foo/foo.txt");

System.out.println("exists?" + f.exists()); // 'exists?true'

// Convert it to a file URL
URL url = f.toURI().toURL();
System.out.println("url is:" + url); // 'url is:file:/home/tim/dir%20foo/foo.txt' - space has been correctly URL encoded

// Now convert the url back to a file name
String filename = url.getFile(); 
System.out.println("f2 is:" + filename); // 'f2 is:/home/tim/dir%20foo/foo.txt' - whoops hasn't been url unencoded 

File f2 = new File(filename);
System.out.println("exists?" + f2.exists()); // 'exists?false' - ouch

use instead

String filename = url.toURI().getPath();

The behaviour of URL vs URI seems messy and inconsistent, I guess they have to keep it that way though for backward compatibility reasons...

URLs and URIs are different things though.