http://netbeans.org/bugzilla/show_bug.cgi?id=192190 is original report. As early as 1.7.1 (did not check earlier; still true in 1.8.2), Locator.decodeUri("file:/tmp/açaí%20berry") produces "file:/tmp/a�a� berry" rather than "file:/tmp/açaí berry" as expected. Note that encodeURI (yes capitalization differs!) produces "file:/tmp/a%C3%A7a%C3%AD%20berry" for the reverse direction, which is also fine and decodes correctly. But new File("/tmp/açaí berry").toURI().toString() produces the problematic "file:/tmp/açaí%20berry"; decodeUri should be more robust and not require toASCIIString() to be used in this case.
Fixed in revision 1055487.