Question

假设我有一个表单文件（每行一个事件）：

Source,Timestamp aa,2014-05-02 22:12:11 bb,2014-05-02 22:22:11

我想总结按来源分组的事件数，连续时间窗为5分钟。我怎么用Flink做到这一点？

我现在拥有的是：

   final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
    env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);
    DataStreamSource<Event> stream = env.fromCollection(new EventFileReader(new File("path/to/file")), Event.class);

    stream
        .keyBy("getSource()")
        .timeWindow(Time.minutes(5))
        .sum("getTimestamp()");     

    env.execute();

public class Event {
    private final String source;
    private final long timestamp;

    public Event(String source, long timestamp) {
        this.source = source;
        this.timestamp = timestamp;
    }

    public String getSource() {
        return source;
    }

    public long getTimestamp() {
        return timestamp;
    }
}

我错过了两件事。首先，这失败并说Event类不是POJO。其次，我不知道如何计算窗口中的事件数。现在我正在使用.sum("getTimestamp()")，但我确定不是这样。有什么想法吗？

Answer 1

我建议使用fold函数进行窗口聚合。以下代码段应该可以完成这项工作：

public class Job {
    public static void main(String[] args) throws Exception {
        final StreamExecutionEnvironment env = StreamExecutionEnvironment.getExecutionEnvironment();
        env.setStreamTimeCharacteristic(TimeCharacteristic.EventTime);
        DataStream<Event> stream = env.fromElements(new Event("a", 1), new Event("b", 2), new Event("a", 2)).assignTimestampsAndWatermarks(new AssignerWithPunctuatedWatermarks<Event>() {
            @Nullable
            @Override
            public Watermark checkAndGetNextWatermark(Event event, long l) {
                return new Watermark(l);
            }

            @Override
            public long extractTimestamp(Event event, long l) {
                return event.getTimestamp();
            }
        });

        DataStream<Tuple2<String, Integer>> count = stream.keyBy(new KeySelector<Event, String>() {
                @Override
                public String getKey(Event event) throws Exception {
                    return event.getSource();
                }
            })
            .timeWindow(Time.minutes(5))
            .fold(Tuple2.of("", 0), new FoldFunction<Event, Tuple2<String, Integer>>() {
                @Override
                public Tuple2<String, Integer> fold(Tuple2<String, Integer> acc, Event o) throws Exception {
                    return Tuple2.of(o.getSource(), acc.f1 + 1);
                }
            });

        count.print();

        env.execute();
    }

    public static class Event {
        private final String source;
        private final long timestamp;

        public Event(String source, long timestamp) {
            this.source = source;
            this.timestamp = timestamp;
        }

        public String getSource() {
            return source;
        }

        public long getTimestamp() {
            return timestamp;
        }
    }
}

如何使用Apache Flink按属性和时间窗口计算？

1 个答案: