--- spambayes-1.0.4/spambayes/classifier.py.orig 2006-10-31 15:56:23.000000000 +0900 +++ spambayes-1.0.4/spambayes/classifier.py 2006-10-31 20:21:12.000000000 +0900 @@ -490,9 +490,17 @@ clues = [] push = clues.append for word in Set(wordstream): - tup = self._worddistanceget(word) - if tup[0] >= mindist: - push(tup) + # 2006-10-31: fixed UnicodeDecodeError in clues.sort(), by owa. + valid = 1 + if isinstance(word, str): + try: + unicode(word) + except: + valid = 0 + if valid: + tup = self._worddistanceget(word) + if tup[0] >= mindist: + push(tup) clues.sort() if len(clues) > options["Classifier", "max_discriminators"]: