Spark读HBase写MySQL

标签：tst   返回   单表   singleton   导致   park   sso   去重   date   

1 Spark读HBase

Spark读HBase黑名单数据，过滤出当日新增userid，并与mysql黑名单表内userid去重后，写入mysql。

def main(args: Array[String]): Unit = {
  @volatile var broadcastMysqlUserids: Broadcast[Array[String]] = null

  val today = args(0)
  val sourceHBaseTable = PropertiesUtil.getProperty("anticheat.blacklist.hbase.tbale")
  val sinkMysqlTable = PropertiesUtil.getProperty("anticheat.blacklist.mysql.dbtable")
  val zookeeper = PropertiesUtil.getProperty("anticheat.blacklist.zookeeper.quorum")
  val zkport = PropertiesUtil.getProperty("anticheat.blacklist.zookeeper.port")
  val znode = PropertiesUtil.getProperty("anticheat.blacklist.zookeeper.znode")

  //创建SparkSession
  val sparkconf = new SparkConf().setAppName("").set("spark.serializer", "org.apache.spark.serializer.KryoSerializer")
  val sc = new SparkContext(sparkconf)
  val spark = AnticheatUtil.SparkSessionSingleton.getInstance(sc.getConf)

  //配置hbase参数
  val conf = HBaseConfiguration.create
  conf.set("hbase.zookeeper.quorum", zookeeper)
  conf.set("hbase.zookeeper.property.clientPort", zkport)
  conf.set("zookeeper.znode.parent", znode)
  conf.set(TableInputFormat.INPUT_TABLE, sourceHBaseTable)

  // 从数据源获取数据
  val hbaseRDD = sc.newAPIHadoopRDD(conf,classOf[TableInputFormat],classOf[ImmutableBytesWritable],classOf[Result])

  //读取mysql表,并将mysql表中的userid广播出去,用于去重
  broadcastMysqlUserids = get_mysql_user_blacklist(spark,sinkMysqlTable)

  //获取当日新增userid数据组装成与mysql表结构一致的对象rdd
  val records_userid_rdd = get_new_blacklist_rdd(hbaseRDD,today,broadcastMysqlUserids)

  //将当日新增userid数据存入mysql
  save_blacklist_to_mysql(records_userid_rdd,today,spark,sinkMysqlTable)
}

2 Spark读MySQL表广播出去

/**
  * Spark读Mysql用户黑名单表,将黑名单中所有userid赋予广播变量
  * @param spark
  * @return
  */
def get_mysql_user_blacklist(spark: SparkSession,table :String) :Broadcast[Array[String]] = {
  @volatile var broadcastMysqlUserids: Broadcast[Array[String]] = null
  val url = PropertiesUtil.getProperty("anticheat.blacklist.mysql.url")
  val user = PropertiesUtil.getProperty("anticheat.blacklist.mysql.user")
  val password = PropertiesUtil.getProperty("anticheat.blacklist.mysql.password")

  import spark.implicits._
  val mysql_userids_rdd = spark.sqlContext.read
    .format("jdbc")
    .option("url",url)
    .option("dbtable",table)
    .option("user",user)
    .option("password",password)
    .load()
    .map(record => {
     val userid = record.getString(0)
     userid
  })

  if(broadcastMysqlUserids !=null){
    broadcastMysqlUserids.unpersist()
  }
  broadcastMysqlUserids = spark.sparkContext.broadcast(mysql_userids_rdd.collect())
  println(s"broadcastMysqlUserids.size= ${broadcastMysqlUserids.value.size}")
  broadcastMysqlUserids
}

3 构建黑名单数据对象rdd

/**
  * 构建新增userid数据写入mysql
  * @param hbaseRDD
  * @param today
  * @return
  */
def get_new_blacklist_rdd(hbaseRDD: RDD[(ImmutableBytesWritable, Result)],today: String,broadcastMysqlUserids: Broadcast[Array[String]]): RDD[BlackList] = {

  val records_userid_rdd : RDD[BlackList] = hbaseRDD.filter(line =>{
    //过滤出当日新增userid
    var flag = false  //默认非当日新增
    val userid = Bytes.toString(line._2.getRow)
    val dt = Bytes.toString(line._2.getValue(Bytes.toBytes("user"), Bytes.toBytes("dt")))
    val did_dt = Bytes.toString(line._2.getValue(Bytes.toBytes("user"), Bytes.toBytes("did_dt")))

    /* 判断为当日新增userid同时需满足三个条件：
    1. 用户维度加入时间dt=today
    2. 或者用户维度加入时间dt=null 且设备维度加入时间did_dt=today
    3. 并且不在mysql黑名单表中
     */
    if(today.equals(dt) || (dt==null && today.equals(did_dt))){
      //broadcastMysqlUserids.value.search(userid).isInstanceOf[InsertionPoint]调用scala 二分查找函数,注意此函数找到返回false
      if(broadcastMysqlUserids.value.search(userid).isInstanceOf[InsertionPoint]){
        //以上三个条件全满足,表示为当日新增,flag 赋值为 true
        flag = true
      }
    }
    flag
  }).map(record =>{
    //获取新增用户userid,加入黑名单时间设为today,其余字段设为默认值
    val userid = Bytes.toString(record._2.getRow)
    val day = Integer.parseInt(today)
    BlackList(userid,day,null,0,"system")
  })
  records_userid_rdd
}

case class BlackList(userid: String, dt: Int, update_time: Timestamp,delete_flag: Int,operator : String)

4 Spark写MySQL

/**
  * 将userid黑名单数据写入mysql
  * @param blacklist_rdd
  * @param today
  * @param spark
  */
def save_blacklist_to_mysql(blacklist_rdd: RDD[BlackList],today: String,spark: SparkSession,table :String): Unit ={
  val url = PropertiesUtil.getProperty("anticheat.blacklist.mysql.url")
  val user = PropertiesUtil.getProperty("anticheat.blacklist.mysql.user")
  val password = PropertiesUtil.getProperty("anticheat.blacklist.mysql.password")

  import spark.implicits._
  val records_userid_dataset = blacklist_rdd.toDS()
  records_userid_dataset.write
    .format("jdbc")
    .option("url",url)
    .option("dbtable",table)
    .option("user",user)
    .option("password",password)
    .mode(SaveMode.Append)
    .save()
}

5 注意问题

数据存入Mysql注意事项
尽量先设置好存储模式
默认为SaveMode.ErrorIfExists模式，该模式下，如果数据库中已经存在该表，则会直接报异常，导致数据不能存入数据库.另外三种模式如下：
SaveMode.Append 如果表已经存在，则追加在该表中；若该表不存在，则会先创建表，再插入数据；
SaveMode.Overwrite 重写模式，其实质是先将已有的表及其数据全都删除，再重新创建该表，最后插入新的数据；
SaveMode.Ignore 若表不存在，则创建表，并存入数据；在表存在的情况下，直接跳过数据的存储，不会报错。

Spark读HBase写MySQL

标签：tst   返回   单表   singleton   导致   park   sso   去重   date   

原文地址：https://www.cnblogs.com/xiaodf/p/10710831.html