small spark real case 3 (scala + spark2 version)

demand:

On student achievement in each class, take out the top three. (Packets take TopN) 

Implementation steps: 

• Traversing each group, each group get results
• Converting a set of results into a buffer array
• Descending sort buffer array
• The array buffer sorted whichever first three
• Print output

GroupTop3 {Object 
  DEF main (args: Array [String]): Unit = { 
    ... the Spark Val = SparkSession.builder () appName ( "GroupTop3") Master ( "local") getOrCreate () 
    // Create the initial RDD 
    Val Lines spark.sparkContext.textFile = ( "D: \\ score.txt") 
    // initial segmentation of the text line by RDD space key mapped to key-value pair 
    val pairs = lines.map (line => (line. Split ( "") (0), line.split ( "") (. 1) .toInt)) 
    // pairs of key value pairs packet 
    Val groupedPairs pairs.groupByKey = () 
    // Get each packet 3 before and after results 
    Val top3Score = groupedPairs.map (classScores => { 
      var className = classScores._1 
      // Get results of each group to convert it into a buffer array, in accordance descending order, whichever three 
      var top3 = classScores ._2.toBuffer.sortWith (_> _).take(3)
      Tuple2(className,top3) 
    })
    top3Score.foreach(m => {
      println(m._1)
      for(s <- m._2) println(s)
      println("------------------")
    })
  }