gistfile1

def get_canonical_plexname(plexname):
    """Return a canonicalised form of a plexname."""

    if not isinstance(plexname, unicode):
        plexname = unicode(plexname, 'utf-8')

    # @/@ this has lotsa skope for optimisation. also, need to choose between:

    #   NFKD(toCasefold(NFKD(toCasefold(NFD(X)))))
    #   NFD(toCasefold(NFD(X)))

    if u'\u0345' in plexname: # COMBINING GREEK YPOGEGRAMMENI
        plexname = normalise_unicode('NFD', plexname)

    canonised = []; out = canonised.append
    space = False

    for char in plexname:

        # @/@ http://www.fileformat.info/info/unicode/category/index.htm

        if (category(char) in ['Cc', 'Zs']) or (ord(char) in SPECIALS):
            space = True
            continue

        if space:
            space = False
            if canonised:
                out(u'-')

        out(CASE_MAP.get(char, char))

    plexname = normalise_unicode('NFKD', u''.join(canonised))
    canonised[:] = []

    for char in plexname:
        out(CASE_MAP.get(char, char))

    # @/@ we're using NFKC sinse it looks prettier
    # @/@ see FC_NFKC_Closure to do properly -- DerivedNormalizationProps.txt

    return normalise_unicode('NFKC', u''.join(canonised))