Flink算子通用状态应用测试样例

Flink算子通用状态应用测试样例

1. 获取Flink执行环境

        final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setParallelism(1);
    

---

2. 创建数据源，生成随机数据

        DataStream<Map<String, String>> source = env.addSource(new SourceFunction<Map<String, String>>() {
            @Override
            public void run(SourceContext<Map<String, String>> ctx) throws Exception {
                while (true) {
                    HashMap<String, String> hashMap = new HashMap<>();
                    hashMap.put("ID", new Random().nextInt(3) + 1 + "");
                    hashMap.put("AMT", "1");
                    System.out.println("------");
                    System.out.println("生产数据：" + hashMap);
                    ctx.collect(hashMap);
                    Thread.sleep(1000);
                }
            }
            @Override
            public void cancel() {}
        });

---

3. 按照ID和星期几进行分组

        KeyedStream<Map<String, String>, String> keyedStream = source.keyBy(new KeySelector<Map<String, String>, String>() {
            @Override
            public String getKey(Map<String, String> value) throws Exception {
                return value.get("ID") + LocalDate.now().getDayOfWeek();
            }
        });

---

4. 处理函数，实现状态初始化和元素处理逻辑

        SingleOutputStreamOperator<Map<String, String>> process = keyedStream.process(new KeyedProcessFunction<String, Map<String, String>, Map<String, String>>() {
            private AggregatingState<Map<String, String>, Map<String, String>> aggState;

            @Override
            public void open(Configuration parameters) throws Exception {
                // 配置状态的TTL
                StateTtlConfig ttlConfig = StateTtlConfig
                        .newBuilder(Time.days(1))
                        .setUpdateType(StateTtlConfig.UpdateType.OnCreateAndWrite) // 仅在创建和写入时清除，另一个读和写时清除
                        .setStateVisibility(StateTtlConfig.StateVisibility.NeverReturnExpired) // 不退回过期值
                        .build();
                // 初始化状态
                AggregatingStateDescriptor<Map<String, String>, Map<String, String>, Map<String, String>> aggRes = new AggregatingStateDescriptor<>("aggRes", new AggregateFunction<Map<String, String>, Map<String, String>, Map<String, String>>() {
                    @Override
                    public Map<String, String> createAccumulator() {
                        return new HashMap<>();
                    }

                    @Override
                    public Map<String, String> add(Map<String, String> in, Map<String, String> acc) {
                        String amt = acc.get("AMT");
                        if (amt == null) {
                            acc.put("ID", in.get("ID"));
                            acc.put("AMT", in.get("AMT"));
                        } else {
                            acc.put("AMT", Integer.valueOf(in.get("AMT")) + Integer.valueOf(amt) + "");
                        }
                        return acc;
                    }

                    @Override
                    public Map<String, String> getResult(Map<String, String> acc) {
                        return acc;
                    }

                    @Override
                    public Map<String, String> merge(Map<String, String> a, Map<String, String> b) {
                        return null;
                    }
                }, TypeInformation.of(new TypeHint<Map<String, String>>() {
                }));
                aggRes.enableTimeToLive(ttlConfig);
                aggState = getRuntimeContext().getAggregatingState(aggRes);
            }

            @Override
            public void processElement(Map<String, String> value, KeyedProcessFunction<String, Map<String, String>, Map<String, String>>.Context ctx, Collector<Map<String, String>> out) throws Exception {
                aggState.add(value);
                out.collect(aggState.get());
            }
        });

---

5. 打印聚合结果

        process.map((MapFunction<Map<String, String>, Object>) value -> {
            System.out.println("聚合结果：" + value);
            return null;
        });

---

6. 执行作业

        env.execute("Flink Common State Test");

---

7. 执行结果

------
生产数据：{AMT=1, ID=2}
聚合结果：{AMT=1, ID=2}
------
生产数据：{AMT=1, ID=3}
聚合结果：{AMT=1, ID=3}
------
生产数据：{AMT=1, ID=3}
聚合结果：{AMT=2, ID=3}
------
生产数据：{AMT=1, ID=1}
聚合结果：{AMT=1, ID=1}
------
生产数据：{AMT=1, ID=1}
聚合结果：{AMT=2, ID=1}
------
生产数据：{AMT=1, ID=1}
聚合结果：{AMT=3, ID=1}
...

这段代码实现了一个 Flink 作业，生成随机数据并对数据进行状态聚合处理。其中包括数据源生成、按键分区、状态初始化、元素聚合处理和结果输出。可以作为多场景下通用的实时数据处理模型。