Flink 1.9 Table API & SQL 2

Flink 1.9 Table API & SQL 2

Table API是用于流和批处理的统一关系的API，以下为对同一种需求的三种简单的不同使用形式

需求：求每个用户的订单总额

1.使用tableEnv.registerTableSource()进行表的注册

import org.apache.flink.api.common.typeinfo.{TypeInformation, Types}
import org.apache.flink.table.api.scala._
import org.apache.flink.streaming.api.scala.{ StreamExecutionEnvironment}
import org.apache.flink.table.api.{EnvironmentSettings}
import org.apache.flink.table.descriptors.{FileSystem, OldCsv, Schema}
import org.apache.flink.types.Row

object FlinkTableSQL {
  def main(args: Array[String]): Unit = {
    val fsSettings = EnvironmentSettings.newInstance().useOldPlanner().inStreamingMode().build()
    val fsEnv = StreamExecutionEnvironment.getExecutionEnvironment
    val tEnv = StreamTableEnvironment.create(fsEnv, fsSettings)
    //从项目根目录中加载数据
    val path = this.getClass.getClassLoader.getResource("Orders.txt").getPath
    tEnv.connect(new FileSystem().path(path))
      .withFormat(new OldCsv()    //数据格式
        .field("cID", Types.STRING)
        .field("cName",Types.STRING)
        .field("cCountry",Types.STRING)
        .field("revenue",Types.DOUBLE).lineDelimiter("\n"))
      .withSchema(new Schema()   //数据类型
        .field("cID", Types.STRING)
        .field("cName",Types.STRING)
        .field("cCountry",Types.STRING)
        .field("revenue",Types.DOUBLE))
      .inAppendMode()
      .registerTableSource("Orders") //Orders表的注册
    val orders = tEnv.scan("Orders")
    //以单个记号（'）开头来引用表的属性
    val resultTable = orders
      .groupBy('cID, 'cName)                          //分组
      .select('cID, 'cName, 'revenue.sum as 'revSum) //查询计算

    // ******************
    //2.使用SQL进行表的查询
    // ******************
    val revenue1 = tEnv.sqlQuery(
      """
        |SELECT cID, cName, SUM(revenue) AS revSum
        |FROM Orders
        |GROUP BY cID, cName
      """
        .stripMargin) //stripMargin 删除由空格或控制字符组成的前导前缀
     //控制台打印结果
    implicit val tpe: TypeInformation[Row] = Types.ROW(Types.STRING,Types.STRING, Types.DOUBLE)
    resultTable.toRetractStream[Row].print()
     // 启动
    fsEnv.execute()
  }
}

2.使用 tableEnv.registerTable()进行表的注册

import org.apache.flink.streaming.api.scala.{DataStream, StreamExecutionEnvironment}
import org.apache.flink.table.api.{EnvironmentSettings, Table}
import org.apache.flink.table.api.scala.StreamTableEnvironment
import org.apache.flink.table.api.scala._
import org.apache.flink.streaming.api.scala._
object FlinkTableSQL  {
  def main(args: Array[String]): Unit = {
    val fsSettings = EnvironmentSettings.newInstance().useOldPlanner().inStreamingMode().build()
    val fsEnv = StreamExecutionEnvironment.getExecutionEnvironment
    val tEnv = StreamTableEnvironment.create(fsEnv, fsSettings)
    //从本地读取数据并将流转化为table
    val orders: Table = fsEnv.readTextFile("file:///E:Orders.txt")
      .map(line => line.split(","))
      .map(o => Orders(o(0), o(1), o(2), o(3).toDouble))
      .toTable(tEnv, 'cID, 'cName, 'cCountry, 'revenue)
    //进行表的注册
    tEnv.registerTable("orders",orders)
    val resultTable = tEnv.sqlQuery(
      """
        |SELECT cID, cName, SUM(revenue) AS revSum
        |FROM orders
        |GROUP BY cID, cName
      """
        .stripMargin) //stripMargin 删除由空格或控制字符组成的前导前缀

    /*
    	val resultTable = orders
      	.groupBy('cID, 'cName) //分组
      	.select('cID, 'cName, 'revenue.sum as 'revSum) //查询计算
    */
    tEnv.toRetractStream[(String, String, Double)](resultTable).print()
    fsEnv.execute()
  }
}

3.使用BLINK的Planner进行流式处理

//需创建样例类
case class Orders(cID:String,cName:String,cCountry:String,revenue:Double)


import org.apache.flink.streaming.api.scala.{DataStream, StreamExecutionEnvironment}
import org.apache.flink.table.api.EnvironmentSettings
import org.apache.flink.table.api.scala.StreamTableEnvironment
import org.apache.flink.table.api.scala._
import org.apache.flink.streaming.api.scala._

object FlinkTableSQL  {
  def main(args: Array[String]): Unit = {
      val bsEnv = StreamExecutionEnvironment.getExecutionEnvironment
      val bsS = EnvironmentSettings.newInstance().useBlinkPlanner().inStreamingMode().build()
      val bsTEnv = StreamTableEnvironment.create(bsEnv, bsS)
      //从本地读取数据作为数据源
      val value: DataStream[Orders] = bsEnv.readTextFile("file:///E:Orders.txt")
          .map(line => line.split(","))
          .map(o => Orders(o(0), o(1), o(2), o(3).toDouble))
      val orders = bsTEnv.registerDataStream("orders",value,'cID,'cName,'cCountry,'revenue)
      val resultTable =  bsTEnv.scan("orders")
          .groupBy('cID, 'cName)                          //分组
          .select('cID, 'cName, 'revenue.sum as 'revSum) //查询计算
      bsTEnv.toRetractStream[(String, String,Double)](resultTable).print()
       bsEnv.execute()
 }
}