Python and Cyrillic characters in regular expression

[phasma]

Hi, I'm trying extract all alphabetic characters from string.

reg = re.compile('(?u)([\w\s]+)', re.UNICODE)
buf = re.match(string)

But it's doesn't work. If string starts from Cyrillic character, all
works fine. But if string starts from Latin character, match returns
only Latin characters.

Please, help.