sparkSql 随机打散分配

UDF 函数

val randomNew = (str :String, n: Int) => {
        val buffer = ArrayBuffer[Int](-1)
        var length:Int = 1
        var index:Int = -1
        for(a <- str){
          index = index + 1
          if (a ==':') {
            length = length+1
            buffer +=index
          }
        }
        val routeKey = length <= n
        routeKey match {
          case true => str
          case _ => {
            val index = (new Random).nextInt(length)
            if(index+n < length){
              str.slice(buffer(index)+1,buffer(index+n)-1)
            }else if(index+n ==length){
              str.slice(buffer(index)+1,str.length)
            }else{
              val index2 =n-(length-index)
              str.slice(buffer(index)+1,str.length)+":"+str.slice(buffer(0),buffer(index2)-1)
            }
          }
        }
      }
      hive.udf.register("randomNew", randomNew)

假如有一串以 冒号 : 分割的字符串 val arr:String = “adfa:adfa:adag:asdf:hgfh:fgjgfj:fgjfjk:fjfkfjk”
随机获取三个子串的拼接字符串 randomNew(arr,3) adag:asdf:hgfh