flink udtf一行输出多行

import d0708.udtf
import org.apache.flink.streaming.api.scala._
import org.apache.flink.table.api.TableEnvironment
import org.apache.flink.types.Row
//测试
object b {
  def main(args:Array[String]):Unit={
    val env =StreamExecutionEnvironment.getExecutionEnvironment
    val tabEnv =TableEnvironment.getTableEnvironment(env)
    tabEnv.registerFunction("udtf",new udtf)

    var s1 ="dev01,1574944573000,2019/2020"
    var s2 ="dev01,1574944573000,2019/2020/222"
    var s3 ="dev01,1574944573000,"
    var s4 ="dev01,1574944573000,null"

    val ds:DataStream[(String,Long,String)]= env.fromElements(s4)
      .map(x =>{
        val arr = x.split(",",-1)
        Tuple3.apply(arr(0), arr(1).toLong,arr(2))
      })
    ds.print("ds")
    import org.apache.flink.table.api.scala._
    tabEnv.registerDataStream("tbl1", ds,'devId, 'time,'data)
    val rsTab = tabEnv.sqlQuery("select devId,`time`,`value` from tbl1 , LATERAL TABLE(udtf(data,'/')) as t(`value`) ")
    rsTab.toRetractStream[Row].print("result")
    env.execute()
  }
}

package d0708
//自定义udtf函数
import org.apache.flink.api.common.typeinfo.TypeInformation
import org.apache.flink.streaming.api.scala._
import org.apache.flink.table.api.{TableEnvironment, Types}
import org.apache.flink.table.functions.{ScalarFunction, TableFunction}
import org.apache.flink.types.Row

class udtf extends TableFunction[Row]{
  override def getResultType: TypeInformation[Row] = Types.ROW(Types.STRING)
  def eval(s:String,sep:String): Unit ={
   if (s == null || s.isEmpty){
     null
   }else{
     val arr: Array[String] = s.split(sep)
     for(a<- arr){
       collect(Row.of(a))
     }
   }
  }
}