spark sql 统计pv uv

话不多说，在开始之前先介绍下pv uv
uv：user views，count（distinct guid）
pv：page views，count（url）
直接上代码
import com.alibaba.fastjson.JSON
import org.apache.spark.sql.SQLContext
import org.apache.spark.{SparkContext, SparkConf}
case class Log(appv:String, userid:String)
object DailyUV {
def main(args: Array[String]) {
val conf = new SparkConf().setAppName(“DailyUV”).setMaster(“local”)
val sc = new SparkContext(conf)
val sqlContext = new SQLContext(sc)
//过滤产品id为3的数据
val data = sc.textFile(“D:\work\data\row.dat”).filter(line=>line.contains(“\”pid\”:\”3”))
val mapData = data.map(line=>{
val json = JSON.parseObject(line)
//appv版本号
val appv = json.getString(“appv”)
//aid 用户唯一标示
val userid = json.getString(“aid”)
(appv,userid)
})
import sqlContext.implicits._
//将RDD转换为DF
val DF = mapData.map(log => Log(log._1,log._2)).toDF()
DF.registerTempTable(“log”)
val sqlStr =
s”“”
|select appv,
|count(userid) as pv ,
|count(distinct userid) as uv
| from log
| group by appv
“”“.stripMargin

val result = sqlContext.sql(sqlStr)result.show()sc.stop()

}
}

运行结果：

+——+—+—+
| appv| pv| uv|
+——+—+—+
|4.35.1| 1| 1|
|4.35.2| 1| 1|
| | 1| 1|
+——+—+—+