本篇我们将使用Java语言来实现Flink的单词统计。

代码开发

环境准备

导入Flink 1.9 pom依赖

<dependencies>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-java</artifactId>
<version>1.9.0</version>
</dependency>
<dependency>
<groupId>org.apache.flink</groupId>
<artifactId>flink-streaming-java_2.11</artifactId>
<version>1.9.0</version>
</dependency>
<dependency>
<groupId>org.apache.commons</groupId>
<artifactId>commons-lang3</artifactId>
<version>3.7</version>
</dependency>
</dependencies>

构建Flink流处理环境

StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

自定义source

每秒生成一行文本

DataStreamSource<String> wordLineDS = env.addSource(new RichSourceFunction<String>() {
private boolean isCanal = false;
private String[] words = {
"important oracle jdk license update",
"the oracle jdk license has changed for releases starting april 16 2019",
"the new oracle technology network license agreement for oracle java se is substantially different from prior oracle jdk licenses the new license permits certain uses such as ",
"personal use and development use at no cost but other uses authorized under prior oracle jdk licenses may no longer be available please review the terms carefully before ",
"downloading and using this product an faq is available here ",
"commercial license and support is available with a low cost java se subscription",
"oracle also provides the latest openjdk release under the open source gpl license at jdk java net"
}; @Override
public void run(SourceContext<String> ctx) throws Exception {
// 每秒发送一行文本
while (!isCanal) {
int randomIndex = RandomUtils.nextInt(0, words.length);
ctx.collect(words[randomIndex]);
Thread.sleep(1000);
}
} @Override
public void cancel() {
isCanal = true;
}
});

单词计算

// 3. 单词统计
// 3.1 将文本行切分成一个个的单词
SingleOutputStreamOperator<String> wordsDS = wordLineDS.flatMap((String line, Collector<String> ctx) -> {
// 切分单词
Arrays.stream(line.split(" ")).forEach(word -> {
ctx.collect(word);
});
}).returns(Types.STRING); //3.2 将单词转换为一个个的元组
SingleOutputStreamOperator<Tuple2<String, Integer>> tupleDS = wordsDS
.map(word -> Tuple2.of(word, 1))
.returns(Types.TUPLE(Types.STRING, Types.INT)); // 3.3 按照单词进行分组
KeyedStream<Tuple2<String, Integer>, String> keyedDS = tupleDS.keyBy(tuple -> tuple.f0); // 3.4 对每组单词数量进行累加
SingleOutputStreamOperator<Tuple2<String, Integer>> resultDS = keyedDS
.timeWindow(Time.seconds(3))
.reduce((t1, t2) -> Tuple2.of(t1.f0, t1.f1 + t2.f1)); resultDS.print();

参考代码

public class WordCount {
public static void main(String[] args) throws Exception {
// 1. 构建Flink流式初始化环境
StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment(); // 2. 自定义source - 每秒发送一行文本
DataStreamSource<String> wordLineDS = env.addSource(new RichSourceFunction<String>() {
private boolean isCanal = false;
private String[] words = {
"important oracle jdk license update",
"the oracle jdk license has changed for releases starting april 16 2019",
"the new oracle technology network license agreement for oracle java se is substantially different from prior oracle jdk licenses the new license permits certain uses such as ",
"personal use and development use at no cost but other uses authorized under prior oracle jdk licenses may no longer be available please review the terms carefully before ",
"downloading and using this product an faq is available here ",
"commercial license and support is available with a low cost java se subscription",
"oracle also provides the latest openjdk release under the open source gpl license at jdk java net"
}; @Override
public void run(SourceContext<String> ctx) throws Exception {
// 每秒发送一行文本
while (!isCanal) {
int randomIndex = RandomUtils.nextInt(0, words.length);
ctx.collect(words[randomIndex]);
Thread.sleep(1000);
}
} @Override
public void cancel() {
isCanal = true;
}
}); // 3. 单词统计
// 3.1 将文本行切分成一个个的单词
SingleOutputStreamOperator<String> wordsDS = wordLineDS.flatMap((String line, Collector<String> ctx) -> {
// 切分单词
Arrays.stream(line.split(" ")).forEach(word -> {
ctx.collect(word);
});
}).returns(Types.STRING); //3.2 将单词转换为一个个的元组
SingleOutputStreamOperator<Tuple2<String, Integer>> tupleDS = wordsDS
.map(word -> Tuple2.of(word, 1))
.returns(Types.TUPLE(Types.STRING, Types.INT)); // 3.3 按照单词进行分组
KeyedStream<Tuple2<String, Integer>, String> keyedDS = tupleDS.keyBy(tuple -> tuple.f0); // 3.4 对每组单词数量进行累加
SingleOutputStreamOperator<Tuple2<String, Integer>> resultDS = keyedDS
.timeWindow(Time.seconds(3))
.reduce((t1, t2) -> Tuple2.of(t1.f0, t1.f1 + t2.f1)); resultDS.print(); env.execute("app");
}
}

Flink对Java Lambda表达式支持情况

Flink支持Java API所有操作符使用Lambda表达式。但是,但Lambda表达式使用Java泛型时,就需要声明类型信息。

我们来看下上述的这段代码:

SingleOutputStreamOperator<String> wordsDS = wordLineDS.flatMap((String line, Collector<String> ctx) -> {
// 切分单词
Arrays.stream(line.split(" ")).forEach(word -> {
ctx.collect(word);
});
}).returns(Types.STRING);

之所以这里将所有的类型信息,因为Flink无法正确自动推断出来Collector中带的泛型。我们来看一下FlatMapFuntion的源代码

@Public
@FunctionalInterface
public interface FlatMapFunction<T, O> extends Function, Serializable {

/**
* The core method of the FlatMapFunction. Takes an element from the input data set and transforms
* it into zero, one, or more elements.
*
* @param value The input value.
* @param out The collector for returning result values.
*
* @throws Exception This method may throw exceptions. Throwing an exception will cause the operation
* to fail and may trigger recovery.
*/
void flatMap(T value, Collector<O> out) throws Exception;
}

我们发现 flatMap的第二个参数是Collector<O>,是一个带参数的泛型。Java编译器编译该代码时会进行参数类型擦除,所以Java编译器会变成成:

void flatMap(T value, Collector out)

这种情况,Flink将无法自动推断类型信息。如果我们没有显示地提供类型信息,将会出现以下错误:

org.apache.flink.api.common.functions.InvalidTypesException: The generic type parameters of 'Collector' are missing.
In many cases lambda methods don't provide enough information for automatic type extraction when Java generics are involved.
An easy workaround is to use an (anonymous) class instead that implements the 'org.apache.flink.api.common.functions.FlatMapFunction' interface.
Otherwise the type has to be specified explicitly using type information.

这种情况下,必须要显示指定类型信息,否则输出将返回值视为Object类型,这将导致Flink无法正确序列化。

所以,我们需要显示地指定Lambda表达式的参数类型信息,并通过returns方法显示指定输出的类型信息

我们再看一段代码:

SingleOutputStreamOperator<Tuple2<String, Integer>> tupleDS = wordsDS
.map(word -> Tuple2.of(word, 1))
.returns(Types.TUPLE(Types.STRING, Types.INT));

为什么map后面也需要指定类型呢?

因为此处map返回的是Tuple2类型,Tuple2是带有泛型参数,在编译的时候同样会被查出泛型参数信息,导致Flink无法正确推断。

更多关于对Java Lambda表达式的支持请参考官网:https://ci.apache.org/projects/flink/flink-docs-release-1.9/dev/java_lambdas.html

「Flink」使用Java lambda表达式实现Flink WordCount的更多相关文章

  1. Java Lambda表达式初探

    Java Lambda表达式初探 前言 本文受启发于Trisha Gee在JavaOne 2016的主题演讲Refactoring to Java 8. Java 8已经发行两年多,但很多人仍然在使用 ...

  2. Java Lambda表达式入门

    Java Lambda表达式入门 http://blog.csdn.net/renfufei/article/details/24600507 Java 8十个lambda表达式案例 http://w ...

  3. Java Lambda表达式入门[转]

    原文链接: Start Using Java Lambda Expressions http://blog.csdn.net/renfufei/article/details/24600507 下载示 ...

  4. Java Lambda表达式教程与示例

    Lambda表达式是Java 8中引入的一个新特性.一个lambda表达式是一个匿名函数,而且这个函数没有名称且不属于任何类.lambda表达式的概念最初是在LISP编程语言中引入的. Java La ...

  5. Java Lambda表达式forEach无法跳出循环的解决思路

    Java Lambda表达式forEach无法跳出循环的解决思路 如果你使用过forEach方法来遍历集合,你会发现在lambda表达式中的return并不会终止循环,这是由于lambda的底层实现导 ...

  6. java lambda表达式学习笔记

    lambda是函数式编程(FP,functional program),在java8中引入,而C#很早之前就有了.在java中lambda表达式是'->',在C#中是‘=>’. 杜甫说:射 ...

  7. 《Java基础知识》Java Lambda表达式

    接触Lambda表达式的时候,第一感觉就是,这个是啥?我居然看不懂,于是开始寻找资料,必须弄懂它. 先来看一个案例: @FunctionalInterface public interface MyL ...

  8. Java lambda 表达式常用示例

    实体类 package com.lkb.java_lambda.dto; import lombok.Data; /** * @program: java_lambda * @description: ...

  9. Java lambda 表达式详解(JDK 8 新特性)

    什么是 lambda 表达式 lambda 表达式(拉姆达表达式)是 JAVA 8 中提供的一种新的特性,它使 Java 也能进行简单的"函数式编程". lambda 表达式的本质 ...

随机推荐

  1. 你可能不知道的 Python 技巧

    英文 | Python Tips and Trick, You Haven't Already Seen 原作 | Martin Heinz (https://martinheinz.dev) 译者 ...

  2. VSCODE更改文件时,提示EACCES permission denied的解决办法(mac电脑系统)

    permission denied:权限问题 具体解决办法: 1.在项目文件夹右键-显示简介-点击右下角解锁 2.权限全部设置为读与写 3.最关键一步:点击"应用到包含的项目",这 ...

  3. 双指针,BFS与图论(一)

    (一)双指针 1.日志统计 小明维护着一个程序员论坛.现在他收集了一份”点赞”日志,日志共有 N 行. 其中每一行的格式是: ts id 表示在 ts 时刻编号 id 的帖子收到一个”赞”. 现在小明 ...

  4. LeetCode-指针法

    LeetCode刷题总结-指针法 方法介绍:指针法主要使用在一组按从小到大排好序的数组中,当按照条件查找对应元素时,在数组的前后定义两个指针,当两个指针代表的元素进行运算时:若结果大于目标值,则左移右 ...

  5. Windows PHP 开启opcache的方法

    PHP opcache可以提升性能.Windows PHP 配置 opcache 的方法如下: 1.先检查PHP目录下ext目录中有没有php_opcache.dll,没有的话自己下载(PHP 5.5 ...

  6. Shell之信号捕获

    前言 当我们在运行某一段代码的时候,希望有类似事物一样的操作,要么成功,要么失败:一般的shell脚本都是自上而下,从左之后运行,碰到异常信号就会出错,从而终止脚本的运行,这个时候脚本可能运行到某一处 ...

  7. Java博客专栏

    1. Java23种设计模式 2. JVM虚拟机 3. 设计模式6大原则 4. Java代码性能优化总结 5. 三种代理 6. iText操作PDF 7. 解析XML的4种方式 8. 面向对象思想 9 ...

  8. Java8新特性一点通 | 回顾文件操作和watchService文件变更监视操作

    文件操作 WatchService 文件操作 今天重温了一些文件操作: - Files.list() 遍历文件和目录 //List all files and sub-directories usin ...

  9. Kafka源码工程examples项目配置log4j

    examples项目启动想知道有哪些错误,通过日志了解代码执行逻辑,但是启动SimpleConsumerDemo了报错如下: log4j.proproties也配置了 log4j.proproties ...

  10. C++基类和派生类的析构函数

    1.派生类也不能继承基类的析构函数. 2.与构造函数不同的是,在派生类的析构函数中不用显式地调用基类的析构函数,因为每个类只有一个析构函数,编译器知道如何选择,无需程序员干涉. 3.构造函数和虚构函数 ...