class AnalyzerDemo(object): examples = ["http://www.baidu.com/ www.baidu.com", "联系 本站 版权 所有 上海 交通 大学BBS 饮水思源 站 沪ICP备020861".decode('gbk')] analyzers = [WhitespaceAnalyzer(), SimpleAnalyzer(), StopAnalyzer(Version.LUCENE_CURRENT), StandardAnalyzer(Version.LUCENE_CURRENT), CJKAnalyzer(Version.LUCENE_CURRENT)] def main(cls, argv): # Use the embedded example strings, unless # command line arguments are specified, then use those. strings = cls.examples if len(argv) > 1: strings = argv[1:] for string in strings: cls.analyze(string) def analyze(cls, text): print 'Analyzing "%s"' %(text) for analyzer in cls.analyzers: name = type(analyzer).__name__ print " %s:" %(name), AnalyzerUtils.displayTokens(analyzer, text) print print main = classmethod(main) analyze = classmethod(analyze)