1.val sc=SharedObject.sc
2.val spark = SharedObject.spark
其中SharedObject.scala为:
val conf =new SparkConf()
.setMaster("yarn")
.set("deploy-mode","client")
.set("spark.yarn.queue","yjy")
.set("num-excutors","5")
.set("executor-memory","10g")
.set("driver-memory","10g")
.set("excutor-cores","5")
val sc = new SparkContext(conf)
val spark = Spark.Session.builder().config(conf).enableHiveSupport().getOrCreate()
...
3.val bd_rdd = baodan_base.map{line =>
try{
val eles = line.split(sep)
val id_num = try{get_idNum(eles(2),eles(9),eles(12),eles(5),eles(33),eles(27),eles(21),eles(17))}catch{case e:Exception =>""}
....
(id_num,(insurancephone,insuredname,insuredphone,benficiaryname))
}catch{case e:Exception =>("",("","","",""))}
}.filter(eles=>eles._1.length==18).map(eles => BaoDan(eles._1,eles._2._1,eles._2._2,eles._2._3,eles._2._4))
其中BaoDan为:
case class BaoDan(id_num:String,insurerphone:String,insuredname:String,insuredphone:String,beneficiaryname:String)