GNU/Linux 2.6.35-23, openjdk6
There's no recognizer for CP866 (DOS russian encoding) in tika yet.
I've used ngrams from cp1251 and wrote custom byteMap. All russian letters, used in cp1251 are present in cp866, so no changes in NGrams needed.
Added inner static class in CharsetRecog_sbcs and CharsetDetector#createRecognizers modified to register this class.
Thank you. I added unit-test for this issue
Thank you. Commited in r1050348