Flink开发流程
和其他所有的计算框架一样，Flink也有一些基础的开发步骤以及基础，核心的API，从开发步骤的角度来讲，主要分为四大部分
Environment
Flink Job在提交执行计算时，需要首先建立和Flink框架之间的联系，也就指的是当前的Flink运行环境，只有获取了环境信息，才能将task调度到不同的taskManager执行。而这个环境对象的获取方式相对比较简单。
// 批处理环境
val env = ExecutionEnvironment.getExecutionEnvironment
// 流式数据处理环境
val env = StreamExecutionEnvironment.getExecutionEnvironment
Source
示例文件所需添加的依赖
<dependencies>
<dependency>
<groupId>junit</groupId>
<artifactId>junit</artifactId>
<version>4.12</version>
<scope>test</scope>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-scala_2.11</artifactId>
<version>1.7.2</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-streaming-scala_2.11</artifactId>
<version>1.7.2</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-connector-kafka_2.11</artifactId>
<version>1.7.2</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-java</artifactId>
<version>1.7.2</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-streaming-java_2.11</artifactId>
<version>1.7.2</version>
</dependency>
<dependency>
<groupId>org.apache.hadoop</groupId>
<artifactId>hadoop-common</artifactId>
<version>2.6.0</version>
</dependency>
<dependency>
<groupId>org.apache.hadoop</groupId>
<artifactId>hadoop-hdfs</artifactId>
<version>2.6.0</version>
</dependency>
</dependencies>
整个Maven工程的配置项
<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
<modelVersion>4.0.0</modelVersion>
<groupId>nj.zb.kb09</groupId>
<artifactId>finkstu</artifactId>
<version>1.0-SNAPSHOT</version>
<name>finkstu</name>
<!-- FIXME change it to the project's website -->
<url>http://www.example.com</url>
<properties>
<project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
<maven.compiler.source>1.8</maven.compiler.source>
<maven.compiler.target>1.8</maven.compiler.target>
</properties>
<dependencies>
<dependency>
<groupId>junit</groupId>
<artifactId>junit</artifactId>
<version>4.12</version>
<scope>test</scope>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-scala_2.11</artifactId>
<version>1.7.2</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-streaming-scala_2.11</artifactId>
<version>1.7.2</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-connector-kafka_2.11</artifactId>
<version>1.7.2</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-java</artifactId>
<version>1.7.2</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-streaming-java_2.11</artifactId>
<version>1.7.2</version>
</dependency>
<dependency>
<groupId>org.apache.hadoop</groupId>
<artifactId>hadoop-common</artifactId>
<version>2.6.0</version>
</dependency>
<dependency>
<groupId>org.apache.hadoop</groupId>
<artifactId>hadoop-hdfs</artifactId>
<version>2.6.0</version>
</dependency>
</dependencies>
<build>
<pluginManagement><!-- lock down plugins versions to avoid using Maven defaults (may be moved to parent pom) -->
<plugins>
<!-- clean lifecycle, see https://maven.apache.org/ref/current/maven-core/lifecycles.html#clean_Lifecycle -->
<plugin>
<artifactId>maven-clean-plugin</artifactId>
<version>3.1.0</version>
</plugin>
<!-- default lifecycle, jar packaging: see https://maven.apache.org/ref/current/maven-core/default-bindings.html#Plugin_bindings_for_jar_packaging -->
<plugin>
<artifactId>maven-resources-plugin</artifactId>
<version>3.0.2</version>
</plugin>
<plugin>
<artifactId>maven-compiler-plugin</artifactId>
<version>3.8.0</version>
</plugin>
<plugin>
<artifactId>maven-surefire-plugin</artifactId>
<version>2.22.1</version>
</plugin>
<plugin>
<artifactId>maven-jar-plugin</artifactId>
<version>3.0.2</version>
</plugin>
<plugin>
<artifactId>maven-install-plugin</artifactId>
<version>2.5.2</version>
</plugin>
<plugin>
<artifactId>maven-deploy-plugin</artifactId>
<version>2.8.2</version>
</plugin>
<!-- site lifecycle, see https://maven.apache.org/ref/current/maven-core/lifecycles.html#site_Lifecycle -->
<plugin>
<artifactId>maven-site-plugin</artifactId>
<version>3.7.1</version>
</plugin>
<plugin>
<artifactId>maven-project-info-reports-plugin</artifactId>
<version>3.0.0</version>
</plugin>
</plugins>
</pluginManagement>
</build>
</project>
data.txt
ws_001,1609314670,45.0
ws_002,1609314671,33.0
ws_003,1609314672,32.0
ws_002,1609314673,23.0
ws_003,1609314674,31.0
ws_002,1609314675,45.0
ws_003,1609314676,18.0
ws_002,1609314677,34.0
ws_003,1609314678,47.0
ws_001,1609314679,55.0
ws_001,1609314680,25.0
ws_001,1609314681,25.0
ws_001,1609314682,25.0
ws_001,1609314683,26.0
ws_001,1609314684,21.0
ws_001,1609314685,24.0
ws_001,1609314686,15.0
从集合读取数据
一般情况下，可以将数据临时存储到内存中，形成特殊的数据结构后，作为数据源使用。这里的数据结构采用集合类型是比较普遍的。
代码展示：
package scala
import org.apache.flink.streaming.api.scala.{DataStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.api.scala._
object SourceTest {
//env source transform sink
def main(args: Array[String]): Unit = {
//创建环境
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
//设置并行度
//env.setParallelism(2)
//创建流，从集合中读取数据
val stream1: DataStream[WaterSensor] = env.fromCollection(List(
WaterSensor("ws_001", 1577844001, 45.0),
WaterSensor("ws_002", 1577844015, 43.0),
WaterSensor("ws_003", 1577844020, 42.0)
))
//打印
stream1.print()
//启动流
env.execute("demo")
}
//定义样例类：水位传感器：用于接收空高数据
//id：传感器编号 ts：时间戳 vc：空高
case class WaterSensor(id:String,ts:Long,vc:Double)
}
结果展示：
从文件中读取数据
通常情况下，我们会从存储介质中获取数据，比较常见的就是将日志文件作为数据源。
读取本地文件
代码展示：
package scala
import org.apache.flink.streaming.api.scala.StreamExecutionEnvironment
import org.apache.flink.streaming.api.scala._
object SourceFile {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
//读取本地文件
val fileDS1: DataStream[String] = env.readTextFile("input/data.txt")
fileDS1.print()
env.execute("sensor")
}
}
结果展示：
读取HDFS文件
代码展示：
package scala
import org.apache.flink.streaming.api.scala.StreamExecutionEnvironment
import org.apache.flink.streaming.api.scala._
object SourceFile {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
//读取HDFS文件
val fileDS2: DataStream[String] = env.readTextFile("hdfs://hadoop100:9000/kb09file/word2.txt")
fileDS2.print()
env.execute("sensor")
}
}
结果展示：
注意：读取HDFS文件时一定要添加两个依赖包：hadoop-common和hadoop-hdfs
Kafka读取数据
Kafka作为消息传输队列，是一个分布式的、高吞吐量、易于扩展地基于主题发布/订阅的消息系统。在现今企业级开发中，Kafka和Flink成为构建一个实时的数据处理系统的首选。
代码展示：
package scala
import java.util.Properties
import org.apache.flink.api.common.serialization.SimpleStringSchema
import org.apache.flink.streaming.api.scala.StreamExecutionEnvironment
import org.apache.flink.streaming.api.scala._
import org.apache.flink.streaming.connectors.kafka.FlinkKafkaConsumer
import org.apache.kafka.clients.consumer.ConsumerConfig
object SourceKafka {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val prop = new Properties()
prop.setProperty(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG,"192.168.136.100:9092")
prop.setProperty(ConsumerConfig.GROUP_ID_CONFIG,"flink-kafka-demo")
prop.setProperty(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG,"org.apache.kafka.common.serialization.StringDeserializer")
prop.setProperty(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG,"org.apache.kafka.common.serialization.StringDeserializer")
prop.setProperty(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG,"latest")
val kafkaDStream: DataStream[String] = env.addSource(
new FlinkKafkaConsumer[String]("sensor", new SimpleStringSchema(), prop)
)
kafkaDStream.print()
env.execute("kafkademo")
}
}
- 创建所需的topic
[root@hadoop100 ~]# kafka-topics.sh --zookeeper 192.168.136.100:2181 --create --topic sensor --partitions 1 --replication-factor 1
- 启动sensor消息创建者
[root@hadoop100 ~]# kafka-console-producer.sh --topic sensor --broker-list 192.168.136.100:9092
- 运行Scala程序
- 在sensor生产消息
hello flink
hello spark
hello scala
hello kafka
- 相应的在控制台打印出了结果
自定义数据源
大多数情况下，前面的数据源已经能够满足需要，但是难免会存在特殊情况的场合，所以Flink也提供了能自定义数据源的方式
代码展示：
package scala
import scala.util.Random
import org.apache.flink.streaming.api.functions.source.SourceFunction
import org.apache.flink.streaming.api.scala.{DataStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.api.scala._
object SourceMy {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val mydefDStream: DataStream[WaterSensor] = env.addSource(new MySensorSource())
mydefDStream.print()
env.execute("mydefsource")
}
}
case class WaterSensor(id: String, ts: Long, vc: Double)
class MySensorSource extends SourceFunction[WaterSensor] {
var flag=true
override def run(sourceContext: SourceFunction.SourceContext[WaterSensor]): Unit = {
while (flag){
//采集数据
sourceContext.collect(
WaterSensor(
"sensor_"+new Random().nextInt(3),
16091404656011L,
new Random().nextInt(5)+40)
)
Thread.sleep(1000)
}
}
override def cancel(): Unit = {
flag=false
}
}
结果展示：
Transform
在Spark中，算子分为转换算子和行动算子，转换算子的作用可以通过算子方法的调用将一个RDD转换另外一个RDD，Flink中也存在同样的操作，可以将一个数据流转换为其他的数据流。
转换过程中，数据流的类型也会发生变化，那么到底Flink支持什么样的数据类型呢，其实我们常用的数据类型，Flink都是支持的。比如：Long、String、Integer、元组、样例类、List、Map等。
Map
- 映射：将数据流中的数据进行转换，形成新的数据流，消费一个元素并产出一个元素
- 参数：Scala匿名函数或MapFunction
- 返回：DataStream
代码展示：
package transform
import org.apache.flink.streaming.api.scala.{DataStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.api.scala._
object Transform_Map {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val sensorDS: DataStream[WaterSensor] = env.fromCollection(
List(
WaterSensor("sensor_0",16091404656011L,44.0),
WaterSensor("sensor_1",16091404656011L,43.0),
WaterSensor("sensor_2",16091404656011L,45.0)
)
)
val mapDStream: DataStream[(String, Long, Double)] = sensorDS.map(x=>(x.id+"_bak",x.ts+1,x.vc+1))
mapDStream.print()
env.execute("sensor")
}
case class WaterSensor(id: String, ts: Long, vc: Double)
}
结果展示：
MapFunction
Flink为每一个算子的参数都至少提供了Scala匿名函数和函数类两种的方式，其中如果使用函数类作为参数的话，需要让自定义函数继续指定的父类或实现特定的接口。例如：MapFunction
- 匿名函数
代码展示：
package transform
import org.apache.flink.api.common.functions.MapFunction
import org.apache.flink.streaming.api.scala.StreamExecutionEnvironment
import org.apache.flink.streaming.api.scala._
object Transform_mapFunction {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val fileDS: DataStream[String] = env.readTextFile("input/data.txt")
//匿名函数实现map
val waterDStream: DataStream[WaterSensor] = fileDS.map(x => {
val strings: Array[String] = x.split(",")
WaterSensor(strings(0), strings(1).toLong, strings(2).toDouble)
})
waterDStream.print("niminglei")
env.execute("sensor")
}
}
}
case class WaterSensor(id: String, ts: Long, vc: Double)
结果展示：
- 自定义函数
代码展示：
package transform
import org.apache.flink.api.common.functions.MapFunction
import org.apache.flink.streaming.api.scala.StreamExecutionEnvironment
import org.apache.flink.streaming.api.scala._
object Transform_mapFunction {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val fileDS: DataStream[String] = env.readTextFile("input/data.txt")
//自定义map类，实现map操作 实现MapFunction接口，功能同上
val waterDStream2: DataStream[WaterSensor] = fileDS.map(new MyMapFunction)
waterDStream2.print("zidinglei")
env.execute("sensor")
}
}
case class WaterSensor(id: String, ts: Long, vc: Double)
class MyMapFunction extends MapFunction[String,WaterSensor]{
override def map(t: String): WaterSensor = {
val strings: Array[String] = t.split(",")
WaterSensor(strings(0), strings(1).toLong, strings(2).toDouble)
}
}
结果展示：
RichMapFunction
所有Flink函数类都有其Rich版本。它与常规函数的不同在于，可以获取运行环境的上下文，并拥有一些生命周期方法，所以可以实现更复杂的功能。也有意味着提供了更多的、更丰富的功能。例如：RichMapFunction
代码展示：
package transform
import org.apache.flink.api.common.functions.RichMapFunction
import org.apache.flink.streaming.api.scala.{DataStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.api.scala._
object Transform_RichMapFunction {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val sensors: DataStream[String] = env.readTextFile("input/data.txt")
val myMapDS: DataStream[WaterSensor] = sensors.map(new MyRichMapFunction)
myMapDS.print()
env.execute("map")
}
}
class MyRichMapFunction extends RichMapFunction[String,WaterSensor]{
override def map(in: String): WaterSensor = {
val strings: Array[String] = in.split(",")
WaterSensor(strings(0),strings(1).toLong,strings(2).toDouble)
}
//富函数提供了生命周期方法
override def open(parameters: Configuration): Unit = {}
override def close(): Unit = {}
}
case class WaterSensor(id: String, ts: Long, vc: Double)
结果展示：
Rich Function有一个生命周期的概念。典型的生命周期方法有：
- open()方法是rich function的初始化方法，当一个算子例如map或者filter被调用之前open()会被调用
- close()方法是生命周期中的最后一个调用的方法，做一些清理工作
- getRuntimeContext()方法提供了函数的RuntimeContext的一些信息，例如函数执行的并行度，任务的名字，以及state状态
flatMap
- 扁平映射：将数据流中的整体拆分成一个一个的个体使用，消费一个元素并产生零到多个元素
- 参数：Scala匿名函数或FlatMapFunction
- 返回：DataStream
代码展示：
package transform
import org.apache.flink.streaming.api.scala.StreamExecutionEnvironment
import org.apache.flink.streaming.api.scala._
object Transform_FlatMap {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
env.setParallelism(1)
val listDStream: DataStream[List[Int]] = env.fromCollection(
List(
List(1, 2, 3, 4, 5),
List(6, 7, 8, 9, 10)
)
)
val resultStream: DataStream[Int] = listDStream.flatMap(list=>list)
resultStream.print()
env.execute("flatMap")
}
}
结果展示：
filter
- 过滤：根据指定的规则将满足条件（true）的数据保留，不满足条件（false）的数据丢弃
- 参数：Scala匿名函数或FilterFunction
- 返回：DataStream
代码展示：
package transform
import org.apache.flink.streaming.api.scala.StreamExecutionEnvironment
import org.apache.flink.streaming.api.scala._
object Transform_Filter {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
env.setParallelism(1)
val listStream: DataStream[List[Int]] = env.fromCollection(
List(
List(1, 2, 3, 4),
List(5, 6, 7, 8, 9, 10),
List(11, 12, 13, 14),
List(15, 16, 17, 18, 19, 20, 21),
List(21, 22),
List(23, 24, 25, 26, 27, 28, 29, 30)
)
)
listStream.filter(list=>list.size>6).print("filter")
env.execute("filter")
}
}
结果展示：
keyBy
在Spark中有一个GroupBy的算子，用于根据指定的规则将数据进行分组，在Flink中也有类似的功能，那就是keyBy，根据指定的key对数据进行分流
- 分流：根据指定的key将元素发送到不同的分区，相同的key会被分到一个分区（这里分区指的就是下游算子多个并行节点的其中一个）。keyBy()是通过哈希来分区的。
- 参数：Scala匿名函数或POJO属性或元组索引，不能使用数组
- 返回：KeyedStream
代码展示：
package transform
import org.apache.flink.api.java.tuple.Tuple
import org.apache.flink.streaming.api.scala.{DataStream, KeyedStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.api.scala._
object Transform_KeyBy {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val sensorDS: DataStream[WaterSensor] = env.fromCollection(
List(
WaterSensor("sensor_0",16091404656011L,44.0),
WaterSensor("sensor_1",16091404656011L,43.0),
WaterSensor("sensor_2",16091404656011L,45.0)
)
)
val mapDStream: DataStream[(String, Long, Double)] = sensorDS.map(x=>(x.id+"_bak",x.ts+1,x.vc+1))
//mapDStream.print()
val keyStream: KeyedStream[(String, Long, Double), String] = mapDStream.keyBy(_._1)
/*
使用keyBy进行分组
TODO关于返回的key的类型
1.如果是位置索引或字段名称，程序无法推断出key的类型，所以给一个java的Tuple类型
2.如果是匿名函数或函数类的方式，可以推断出key的类型，比较推荐使用
分组的概念：分组只是逻辑上进行分组，打上了记号（标签），跟并行度没有绝对的关系
同一个分组的数据在一起
同一个分区里可以有多个不同的组
*/
/*val value1: KeyedStream[WaterSensor, String] = sensorDS.keyBy(x=>x.id)
val value2: KeyedStream[WaterSensor, Tuple] = sensorDS.keyBy("id")
val value3: KeyedStream[WaterSensor, Tuple] = sensorDS.keyBy(0)*/
keyStream.print()
env.execute("sensor")
}
}
结果展示：
shuffle
- 打乱重组（洗牌）：将数据按照均匀分布打散到下游
- 参数：无
- 返回：DataStream
代码展示：
package transform
import org.apache.flink.streaming.api.scala.{DataStream, StreamExecutionEnvironment}
object Transform_Shuffle {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
env.setParallelism(3)
val fileDStream: DataStream[String] = env.readTextFile("input/data.txt")
val shuffleStream: DataStream[String] = fileDStream.shuffle
fileDStream.print("data")
shuffleStream.print("shuffle")
env.execute("shuffledemo")
}
}
结果展示：
split
在某些情况下，我们需要将数据流根据某些特征拆分成两个或者多个数据流，给不同数据流增加标记以便于从流中取出
代码展示：
package transform
import org.apache.flink.streaming.api.scala.{DataStream, SplitStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.api.scala._
object Transform_Split {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val fileDS: DataStream[String] = env.readTextFile("input/data.txt")
val waterDStream: DataStream[WaterSensor] = fileDS.map(x => {
val strings: Array[String] = x.split(",")
WaterSensor(strings(0), strings(1).toLong, strings(2).toDouble)
})
val splitStream: SplitStream[WaterSensor] = waterDStream.split(sensor => {
if (sensor.vc <30) {
Seq("normal")
} else if (sensor.vc < 40) {
Seq("warn")
} else {
Seq("alarm")
}
})
env.execute("splitdemo")
}
}
select
将数据流进行切分后，如何从流中讲不同的标价取出呢，这时就需要使用select算子了。
代码展示：
package transform
import org.apache.flink.streaming.api.scala.{DataStream, SplitStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.api.scala._
object Transform_Split {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val fileDS: DataStream[String] = env.readTextFile("input/data.txt")
val waterDStream: DataStream[WaterSensor] = fileDS.map(x => {
val strings: Array[String] = x.split(",")
WaterSensor(strings(0), strings(1).toLong, strings(2).toDouble)
})
val splitStream: SplitStream[WaterSensor] = waterDStream.split(sensor => {
if (sensor.vc <30) {
Seq("normal")
} else if (sensor.vc < 40) {
Seq("warn")
} else {
Seq("alarm")
}
})
val normalStream: DataStream[WaterSensor] = splitStream.select("normal")
val warnStream: DataStream[WaterSensor] = splitStream.select("warn")
val alarmStream: DataStream[WaterSensor] = splitStream.select("alarm")
normalStream.print("normal")
warnStream.print("warn")
alarmStream.print("alarm")
env.execute("splitdemo")
}
}
结果展示：
connect
在某些情况下，我们需要将两个不同来源的数据流进行连接，实现数据匹配，比如订单支付和第三方交易信息，这两个信息的数据就来自于不同数据源，连接后，将订单支付和第三方交易信息进行对账，此时，才能算真正的支付完成。
Flink中的connect算子可以连接两个保持他们类型的数据流，两个数据流被connect之后，只是被放在了一个同一个流中，内部依然保持各自的数据和形式不发生任何变化，两个流相互独立。
代码展示：
package transform
import org.apache.flink.streaming.api.scala.{ConnectedStreams, DataStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.api.scala._
object Transform_Connect {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val fileDS: DataStream[String] = env.readTextFile("input/data.txt")
val waterDStream: DataStream[WaterSensor] = fileDS.map(x => {
val strings: Array[String] = x.split(",")
WaterSensor(strings(0), strings(1).toLong, strings(2).toDouble)
})
val numStream: DataStream[Int] = env.fromCollection(List(1,2,3,4,5,6))
val connStream: ConnectedStreams[WaterSensor, Int] = waterDStream.connect(numStream)
connStream.map(sensor=>sensor.id,num=>num+1).print("connect")
env.execute("connectdemo")
}
}
结果展示：
union
对两个或者两个以上的DataStream进行union操作，产生一个包含所有DataStream元素的新DataStream
connect与union区别：
1.union之前两个流的类型必须是一样的，connect可以不一样。
2.connect只能操作两个流，union可以操作多个。
代码展示：
package transform
import org.apache.flink.streaming.api.scala.{ConnectedStreams, DataStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.api.scala._
object Transform_Connect {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val num1Stream: DataStream[Int] = env.fromCollection(List(1,2,3))
val num2Stream: DataStream[Int] = env.fromCollection(List(4,5,6))
val num3Stream: DataStream[Int] = env.fromCollection(List(7,8,9))
val unionStream: DataStream[Int] = num1Stream.union(num2Stream).union(num3Stream)
//unionStream.print()
env.execute("uniondemo")
}
}
结果展示：
Operator
Flink作为计算框架，主要应用于数据计算处理上，所以在keyBy对数据进行分流后，可以对数据进行相应的统计分析
滚动聚合算子（Rolling 、Aggregation）
这些算子可以针对KeyedStream的每一个支流做聚合。执行完成后，会将聚合的结果合成一个流返回，所以结果都是DataStream
- sum()
- min()
- max()
reduce
一个分组数据流的聚合操作，合并当前的元素和上次聚合的结果，产生一个新的值，返回的流中包含每一次聚合的结果，而不是只返回最后一次聚合的最终结果。
代码展示：
package transform
import org.apache.flink.streaming.api.scala.{DataStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.api.scala._
object Transform_Reduce {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val fileDS: DataStream[String] = env.readTextFile("input/data.txt")
val waterDStream: DataStream[WaterSensor] = fileDS.map(x => {
val strings: Array[String] = x.split(",")
WaterSensor(strings(0), strings(1).toLong, strings(2).toDouble)
})
val keyStream: KeyedStream[WaterSensor, String] = waterDStream.keyBy(_.id)
val reduceStream: DataStream[WaterSensor] = keyStream.reduce((x, y) => {
println(x.id, x.vc, y.id, y.vc)
WaterSensor(x.id, System.currentTimeMillis(), x.vc + y.vc)
})
reduceStream.print()
env.execute("reducedemo")
}
}
结果展示：
process
Flink在数据流通过keyBy进行分流处理后，如果想要处理过程中获取环境相关信息，可以采用process算子自定义继承KeyedProcessFunction抽象类，并定义泛型[KEY,IN,OUT]
代码展示：
package transform
import org.apache.flink.streaming.api.functions.KeyedProcessFunction
import org.apache.flink.streaming.api.scala.{DataStream, StreamExecutionEnvironment}
import org.apache.flink.util.Collector
import org.apache.flink.streaming.api.scala._
object Transform_Process {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val fileDS: DataStream[String] = env.readTextFile("input/data.txt")
val waterDStream: DataStream[WaterSensor] = fileDS.map(x => {
val strings: Array[String] = x.split(",")
WaterSensor(strings(0), strings(1).toLong, strings(2).toDouble)
})
val keyStream: KeyedStream[WaterSensor, String] = waterDStream.keyBy(_.id)
val outStream: DataStream[String] = keyStream.process(new MyKeyedProcessFunction)
outStream.print("myprocess")
env.execute("myprocess")
}
}
class MyKeyedProcessFunction extends KeyedProcessFunction[String,WaterSensor,String]{
override def processElement(i: WaterSensor, context: KeyedProcessFunction[String, WaterSensor, String]#Context, collector: Collector[String]): Unit = {
val key: String = context.getCurrentKey
println(key)
collector.collect("process key:"+key+" value:"+i)
}
}
结果展示：
Sink
Sink有下沉的意思，在Flink中所谓的Sink其实可以表示为将数据存储起来的意思，也可以将范围扩大，表示将处理完的数据发送到指定的存储系统的输出操作。
代码展示：
package sink
import java.util.Properties
import org.apache.flink.api.common.serialization.SimpleStringSchema
import org.apache.flink.streaming.api.scala.{DataStream, StreamExecutionEnvironment}
import org.apache.flink.streaming.connectors.kafka.{FlinkKafkaConsumer, FlinkKafkaProducer}
import org.apache.kafka.clients.consumer.ConsumerConfig
import org.apache.flink.streaming.api.scala._
object SinkKafka {
def main(args: Array[String]): Unit = {
val env: StreamExecutionEnvironment = StreamExecutionEnvironment.getExecutionEnvironment
val prop = new Properties()
prop.setProperty(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG,"192.168.136.100:9092")
prop.setProperty(ConsumerConfig.GROUP_ID_CONFIG,"flink-kafka-demo")
prop.setProperty(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG,"org.apache.kafka.common.serialization.StringDeserializer")
prop.setProperty(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG,"org.apache.kafka.common.serialization.StringDeserializer")
prop.setProperty(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG,"latest")
val kafkaDStream: DataStream[String] = env.addSource(
new FlinkKafkaConsumer[String]("sensor", new SimpleStringSchema(), prop)
)
kafkaDStream.addSink(new FlinkKafkaProducer[String]("192.168.136.100:9092","sensorout",new SimpleStringSchema()))
kafkaDStream.print()
env.execute("kafkademo")
}
}
- 创建两个相应的topic
[root@hadoop100 ~]# kafka-topics.sh --zookeeper hadoop100:2181 --create --topic sensor --partitions 1 --replication-factor 1
[root@hadoop100 ~]# kafka-topics.sh --zookeeper hadoop100:2181 --create --topic sensorout --partitions 1 --replication-factor 1
- 创建生产信息
[root@hadoop100 ~]# kafka-console-producer.sh --topic sensor --broker-list hadoop100:9092
- 创建消费信息
[root@hadoop100 ~]# kafka-console-consumer.sh --topic sensorout --bootstrap-server hadoop100:9092
- 启动Scala程序
- 在生产者sensor生产信息
hello
java
hello
spark
hello
flink
- 相应的在控制台打印出了信息和sensorout消费到了消息
1327
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
