WARC writer: unit tests for conversion of URLs to URIs

[sebastian-nagel]

Nutch uses instances of the class java.net.URL to represent the URLs being crawled. WARC files require URIs for the WARC-Target-URI header. While the conversion to an URI is unproblematic for most of the URLs, there are some issues:

1. there are instances of java.net.URL which fail to be converted to java.net.URI, see URL.toURI(). Note: the URLs were successfully fetched!
2. the conversion of java.net.URI to an ASCII-only URI is not free of pitfalls (see #20)

Would be good to have unit tests to test and verify these issues - of course, ideally with "solutions" to make the conversion from URL to URI succeed. E.g.,

• non-ASCII / Unicode components in URLs, including IDNs
• encoding of white space in the URL path or query
• encoding of characters invalid in URIs but valid in URLs