sparkStreaming程序过滤kafka json数据两种方法

方法1：在foreachRDD之前，就是streams之后过滤
我取的是type值为null（不是"null"）的数据

val sv= stream.map(_.value())
     .filter(jd => {
       val json1 = jd.substring(0, jd.lastIndexOf("}")+1)
       
       JSON.parseObject(json1).getString("type")==null
     })  
     sv.foreachRDD(rdd => {

方法2：在foreachRDD之后，

stream.foreachRDD(rdd => { //rdd是kafkardd
      val ds: Dataset[String] = spark.createDataset(rdd.map(_.value()))
      val offsetRanges: Array[OffsetRange] = rdd.asInstanceOf[HasOffsetRanges].offsetRanges

      val jsondata= ds.filter(JSON.parseObject(_).getString("type")==null)

      jsondata.map(j => {
        val json = JSON.parseObject(j.toString)
        val coupon_batch_id = json.getString("coupon_batch_id")
        val prom_id = json.getString("prom_id")
        val coupon_gt = json.getString("coupon_gt")
        val coupon_minus = json.getString("coupon_minus")
        val prom_msg = json.getString("prom_prom_msg")
        val prom_key = json.getString("prom_key")
        val url = json.getString("url")
        val title = json.getString("title")
        val img_url = json.getString("img_url")
        val brand_name = json.getString("brand_name")
        val cat1_id = json.getString("root_cat_id")
        val cat2_id = json.getString("cat_id")
        val cat3_id = json.getString("cat3_id")
        val cat1_name = json.getString("cat_name_1")
        val cat2_name = json.getString("cat_name_2")
        val cat3_name = json.getString("cat_name_3")
        val price = json.getString("platform_price")
        val vip_price = json.getString("vip_price")
        var cp_type = ""

        /**
          * json串里的"coupon_deal_type": null  这个null不加引号就是空 判断需要coupon_deal_type==null，  加了引号"null"才是字符串才能用equals("null")
          */
        if(prom_msg==null && coupon_gt==null) {

          cp_type=0.toString
        }else{
           cp_type=1.toString
        }
        (coupon_batch_id+"-"+prom_id,coupon_gt+"-"+coupon_minus,prom_msg,prom_key,url,title,img_url,brand_name,price,vip_price,cat1_id,cat2_id,cat3_id,cat1_name,cat2_name,cat3_name,cp_type)
      }).toDF("cp_id", "copn_msg  ".trim, "prom_msg  ".trim, "prom_key  ".trim, "url       ".trim, "title     ".trim, "img_url   ".trim, "brand_name".trim, "price     ".trim, "vip_price ".trim, "cat1_id   ".trim, "cat2_id   ".trim, "cat3_id   ".trim, "cat1_name ".trim, "cat2_name ".trim, "cat3_name ".trim, "cp_type   ".trim)
        .createOrReplaceTempView("cpdata")
        val result = spark.sql("select * from cpdata")
      result.show()
      val props = new Properties()
      props.setProperty("user",load.getString("jdba_user"))
      props.setProperty("password",load.getString("jdbc_password"))
      result.write.mode(SaveMode.Append).jdbc(load.getString("jdbc_url"),load.getString("jdbc_tableName"),props)

    })
    ssc.start()
    ssc.awaitTermination()```