+ try:
+ self._breaker = _icu.WordBreaker(icu_locale)
+ except ValueError:
+ # Thai is the only language with a special break locale,
+ # so this is a harmless error.
+ self._breaker = _icu.WordBreaker("root")
+
+ def words(self, string):
+ """Split the string along word boundries."""
+ if isinstance(string, str):
+ string = string.decode(self.encoding)
+ words = self._breaker.words(string)
+ return [w for w in words if not w.isspace()]
+