spark编写wordcount程序

object WordCount {

  def main(args: Array[String]): Unit = {

    val lines = Array("hello tom hello jerry", "hello kitty hello tom", "hello tom tom")

    //按空格切割行成词，形成RDD(Array(),Array(),...)，在flatmap压平成RDD(...)
    val words: Array[String] = lines.flatMap(l => l.split(" "))

    //对每个单词映射成元组，value为1
    val wordAndOne: Array[(String, Int)] = words.map(w => (w, 1))

    //按照单词分组，成RDD((word,Array(...)),...)
    val grouped: Map[String, Array[(String, Int)]] = wordAndOne.groupBy(t => t._1)

    //对每个单词计算value
    val wordAndCounts: Map[String, Int] = grouped.map(t => (t._1, t._2.length))

    //按照value排序
    val result: List[(String, Int)] = wordAndCounts.toList.sortBy(t => -t._2)

    //打印结果
    print(result.toBuffer)
  }

}