Add more tests

[mathiasbynens]

• Markus Kuhn’s UTF-8 decoder stress test
• http://www.cl.cam.ac.uk/~mgk25/ucs/examples/quickbrown.txt
• …
• See http://stackoverflow.com/q/1319022/96656 for some more examples.

Note that some of the things mentioned there are explicitly allowed by utf8.js. For example, utf8.js only supports Unicode code points (from 0x0 to 0x10FFFF) as per http://tools.ietf.org/html/rfc3629, so ignore the 5-byte or 6-byte sequences. Also:

There are also some non-characters. U+FFFF is one such - it cannot appear in Unicode, and hence should not appear in UTF-8 encoded data either.

Maybe, but IMHO a good UTF-8 encoder/decoder is able to handle it just fine and convert it to '\uFFFF' in JavaScript. This also seems to be a requirement as per http://encoding.spec.whatwg.org/#utf-8.

[liorean]

Perhaps https://github.com/liorean/util/blob/master/utf-8/valid-boundaries.es and https://github.com/liorean/util/blob/master/utf-8/all-boundaries.es could be of use? I've found them really valuable when developing my own utf-8 libraries, though all my testing have been done in the shell and not in any testing framework.