在Apache flink中，如何获取过滤函数中不匹配的值的输出。

Question

我是Apache flink的新手，我试图过滤以字母 "N "开头的单词，我得到了输出，但我如何能得到不以单词 "N "开头的单词，下面是我使用的代码。

package DataStream;

import org.apache.flink.api.common.functions.FilterFunction;
import org.apache.flink.api.common.functions.FlatMapFunction;
import org.apache.flink.api.java.tuple.Tuple2;
import org.apache.flink.streaming.api.datastream.DataStream;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.util.Collector;

public class WordStream {

    public static void main(String[] args) throws Exception {

        final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

        DataStream<String> inputData = env.socketTextStream("localhost", 9999);

        DataStream<String> filterData = inputData.filter(new FilterFunction<String>() {

            /**
             * 
             */
            private static final long serialVersionUID = 1L;

            @Override
            public boolean filter(String value) throws Exception {
                return value.startsWith("N");
            }
        });

        DataStream<Tuple2<String, Integer>> tokenize = filterData
                .flatMap(new FlatMapFunction<String, Tuple2<String, Integer>>() {

                    @Override
                    public void flatMap(String value, Collector<Tuple2<String, Integer>> out) throws Exception {
                        out.collect(new Tuple2<String, Integer>(value, Integer.valueOf(1)));

                    }
                });

        DataStream<Tuple2<String, Integer>> counts = tokenize.keyBy(0).sum(1);

        counts.print();

        env.execute("WordStream");

    }

}

你能建议如何捕获不匹配的单词到另一个流。

Answer 1

我想你可以使用 侧产出 来实现。只需要在实际采集器中发出匹配的元素，用ProcessFunction发出带有侧输出标签的未匹配元素，然后从主流中获取侧输出元素。

举个例子，你的代码可以改成这样。

package datastream;


import org.apache.flink.api.java.tuple.Tuple2;
import org.apache.flink.streaming.api.datastream.DataStream;
import org.apache.flink.streaming.api.datastream.SingleOutputStreamOperator;
import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment;
import org.apache.flink.streaming.api.functions.ProcessFunction;
import org.apache.flink.util.Collector;
import org.apache.flink.util.OutputTag;

public class WordStream {

    public static void main(String[] args) throws Exception {

        final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();

        DataStream<String> inputData = env.socketTextStream("localhost", 9999);

        // Initialize side-output tag to collect the un-matched elements 
        OutputTag<Tuple2<String, Integer>> unMatchedSideOutput = new OutputTag<Tuple2<String, Integer>>("unmatched-side-output") {};

        SingleOutputStreamOperator<Tuple2<String, Integer>> tokenize = inputData
                .process(new ProcessFunction<String, Tuple2<String, Integer>>() {
                    @Override
                    public void processElement(String value, Context ctx, Collector<Tuple2<String, Integer>> out) {
                        if (value.startsWith("N")) {
                            // Emit the data to actual collector
                            out.collect(new Tuple2<>("Matched=" + value, Integer.valueOf(1)));
                        } else {
                            // Emit the un-matched data to side output
                            ctx.output(unMatchedSideOutput, new Tuple2<>("UnMatched=" + value, Integer.valueOf(1)));
                        }
                    }
                });

        DataStream<Tuple2<String, Integer>> count = tokenize.keyBy(0).sum(1);

        // Fetch the un-matched element using side-output tag and process it
        DataStream<Tuple2<String, Integer>> unMatchedCount = tokenize.getSideOutput(unMatchedSideOutput).keyBy(0).sum(1);

        count.print();

        unMatchedCount.print();

        env.execute("WordStream");

    }
}

注 我稍微改变了发射值的前缀。Matched= 和 UnMatched= 以便在输出中得到清晰的理解。

对于下面的输入。

Hello
Nevermind
Hello

我得到以下输出。

3> (UnMatched=Hello,1)
4> (Matched=Nevermind,1)
3> (UnMatched=Hello,2)

Answer 2

一个更简单的解决方案。

DataStream<String> nwords = input.filter(s -> startsWith("N"));
DataStream<String> others = input.filter(s -> !startsWith("N"));

我相信这比使用边输出的解决方案效率稍低，但它仍将在一个任务中运行，使用操作者链，所以它也不需要serde开销，或网络。

不要误会我的意思--一般来说，侧输出是分割流的最佳方式。