ietf-nntp wildmat implementation updated

[Russ Allbery]

I've updated my wildmat implementation with hopefully full support for
UTF-8.  It's at <http://www.eyrie.org/~eagle/nntp/> as before.  The test
suite is also greatly expanded, including tests for UTF-8 and for handling
of malformed UTF-8 as single characters in a single-octet character set,
as mentioned previously on this list.

I didn't implement the proposed \u syntax.

The test suite includes a few more interesting edge cases, such as [a-]b],
which matches the strings "ab]" and "-b]" and no others, and [^---], which
matches any single character string other than "-".  Note that my
implementation treats the character class [a-c-n] as equivalent to [abcn-].

Please send me any comments, bug reports, or the like.  I'm happy to have
this become the de facto "reference" wildmat implementation and will
gratefully incorporate any fixes or changes.  Currently, it's still
dependent on INN's include file structure (and uses C99 features such as
bool and uint32_t that require configure testing such as that done by INN
for portability); if there's a lot of demand, I can try to find the time
to also support it as a stand-alone distribution.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>