Javalambda表达式如何实现FlinkWordCount

小编给大家分享一下Javalambda表达式如何实现FlinkWordCount，相信大部分人都还不怎么了解，因此分享这篇文章给大家参考一下，希望大家阅读完这篇文章后大有收获，下面让我们一起去了解一下吧！

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@FunctionalInterfacepublic 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。

以上是“Javalambda表达式如何实现FlinkWordCount”这篇文章的所有内容，感谢各位的阅读！相信大家都有了一定的了解，希望分享的内容对大家有所帮助，如果还想学习更多知识，欢迎关注亿速云行业资讯频道！