将输入PCollection的每个元素(文本文件的每一行)切分成单词，统计每个单词个数，返回 内容为(word, count)的PCollection Args: lines (PCollection): 通过读取文本文件构造的PCollection Returns: PCollection: 表示统计结果的PCollection >>> _p = _pipeline.parallelize("to be or not to be") >>> print _p.apply(count_words).get()
(lines)
source not stored for this graph (policy: none)
nothing calls this directly
no test coverage detected