利用sparksql进行报表分析

package report

import config.ConfigHelper
import org.apache.spark.rdd.RDD
import org.apache.spark.sql.{DataFrame, SparkSession}
import utils.MakeATPKpi

//利用sparksql进行报表分析
object TrainTimeSparkSQLAnalysis {
  def main(args: Array[String]): Unit = {
    //session
    val session = SparkSession
      .builder()
      //设置是否是在本地运行
      //*代表几个线程
      .master("local[*]")
      //设置进程的名称
      .appName(this.getClass.getName)
      //序列化
      .config("spark.serializer", ConfigHelper.serializer)
      .getOrCreate()
    //导入隐式转换
    import session.implicits._
    //读取原始日志
    val frame = session.read.parquet(args(0))
    //读取列车出厂时间的数据
    val trainTimeSource: RDD[String] = session.sparkContext.textFile(args(1))
    //切分并且数据清洗
    val filted = trainTimeSource.map(_.split("\\|",-1)).filter(_.length>=2)
    val trainTimeDF: DataFrame = filted.map(arr=>(arr(0),arr(1))).toDF("trainId","trainTime")
    //注册表
    //临时表
    frame.createTempView("logs")
    //创建临时视图
    trainTimeDF.createTempView("trainTime")
    //注册udf
    session.udf.register("myif",(boolean:Boolean)=>if (boolean) 1 else 0)
    //写sql语句
    //GROUP BY我们可以先从字面上来理解，GROUP表示分组，BY后面写字段名，就表示根据哪个字段进行分组。
    //GROUP BY必须得配合聚合函数来用，分组之后你可以计数（COUNT），求和（SUM），求平均数（AVG）等。
    session.sql(
      "select a.trainTime,sum(c.cnt),sum(c.atpError),sum(c.main),sum(c.wifi),sum(c.balise),sum(c.TCR),sum(c.speed),sum(c.DMI),sum(c.TIU),sum(c.JRU) from trainTime a join (select MPacketHead_TrainID,count(*) as cnt,"+MakeATPKpi.sqlStr+
        " from logs group by MPacketHead_TrainID) c on a.trainId=c.MPacketHead_TrainID group by a.trainTime").show()
    //释放资源
    session.stop()
  }

}

工具类
MakeATPKpi.scala：

package utils

import org.apache.commons.lang.StringUtils
import org.apache.spark.sql.Row

object MakeATPKpi {
  def makeKpi(row:Row)={
    //获取atperror
    val atpError = row.getAs[String]("MATPBaseInfo_AtpError")
    //判断指标
    //如果atpError不为空，前面为1，如果atpError为空，那么前面为0
    //先判断atpError不为空，如果为空，则执行下面的else语句，全部输出为0
    val listAtpError: List[Int] = if (StringUtils.isNotEmpty(atpError)) {
      val listError: List[Int] =
        if (atpError.equals("车载主机")) {
          List[Int](1, 0, 0, 0, 0, 0, 0, 0)
        } else if (atpError.equals("无线传输单元")) {
          List[Int](0, 1, 0, 0, 0, 0, 0, 0)
        } else if (atpError.equals("应答器信息接收单元")) {
          List[Int](0, 0, 1, 0, 0, 0, 0, 0)
        } else if (atpError.equals("轨道电路信息读取器")) {
          List[Int](0, 0, 0, 1, 0, 0, 0, 0)
        } else if (atpError.equals("测速测距单元")) {
          List[Int](0, 0, 0, 0, 1, 0, 0, 0)
        } else if (atpError.equals("人机交互接口单元")) {
          List[Int](0, 0, 0, 0, 0, 1, 0, 0)
        } else if (atpError.equals("列车接口单元")) {
          List[Int](0, 0, 0, 0, 0, 0, 1, 0)
        } else if (atpError.equals("司法记录单元")) {
          List[Int](0, 0, 0, 0, 0, 0, 0, 1)
        } else {
          //这是为了防止获取的atpError的值都不符合上述判断条件的情况
          List[Int](0, 0, 0, 0, 0, 0, 0, 0)
        }
      //两个list的拼接要用++
      List[Int](1) ++ listError
    } else {
      //如果atpError为空的情况，说明这行的这个属性没有值，那么全取为0即可
      List[Int](0, 0, 0, 0, 0, 0, 0, 0, 0)
    }
    //创建一个容器用来存放标签
    //两个list的拼接要用++
    //这个是为了在整合以后统计总共多少条数据用，相当于数量
    val list: List[Int] = List[Int](1) ++ listAtpError
    list
  }

  //sql字符串
  val sqlStr="sum(if(MATPBaseInfo_AtpError != '',1,0)) as atpError,sum(case when MATPBaseInfo_AtpError = '车载主机' then 1 else 0 end) as main,"+
    "sum (myif(MATPBaseInfo_AtpError = '无线传输单元') ) as wifi,"+
    "sum (myif(MATPBaseInfo_AtpError = '应答器信息接收单元') ) as balise,"+
    "sum (myif(MATPBaseInfo_AtpError = '轨道电路信息读取器') ) as TCR,"+
    "sum (myif(MATPBaseInfo_AtpError = '测速测距单元') ) as speed,"+
    "sum (myif(MATPBaseInfo_AtpError = '人机交互接口单元') ) as DMI,"+
    "sum (myif(MATPBaseInfo_AtpError = '列车接口单元') ) as TIU,"+
    "sum (myif(MATPBaseInfo_AtpError = '司法记录单元') ) as JRU"
}

配置文件
application.conf：

#配置文件
#配置压缩格式
parquet.code="snappy"
#配置序列化方式
spark.serializer="org.apache.spark.serializer.KryoSerializer"
#配置jdbc链接
jdbc.url="jdbc:mysql://localhost:3306/test?characterEncoding=UTF-8"
jdbc.driver="com.mysql.jdbc.Driver"
jdbc.user="root"
jdbc.password="000000"
#配置scalikejdbc链接
db.default.url="jdbc:mysql://localhost:3306/test?characterEncoding=UTF-8"
db.default.driver="com.mysql.jdbc.Driver"
db.default.user="root"
db.default.password="000000"

ConfigHelper.scala：

package config

import com.typesafe.config.{Config, ConfigFactory}

object ConfigHelper {
  //加载配置文件
  private lazy val load: Config = ConfigFactory.load()
  //加载压缩格式
  val parquetCode: String = load.getString("parquet.code")
  //加载序列化方式
  val serializer: String = load.getString("spark.serializer")
  //加载jdbc
  val url: String = load.getString("jdbc.url")
  val driver: String = load.getString("jdbc.driver")
  val user: String = load.getString("jdbc.user")
  val password: String = load.getString("jdbc.password")
  //加载scalikejdbc
  val url2: String = load.getString("db.default.url")
  val driver2: String = load.getString("db.default.driver")
  val user2: String = load.getString("db.default.user")
  val password2: String = load.getString("db.default.password")

}