Java AggregatorFactory类代码示例

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
public static QueryableIndex createDruidSegments() throws IOException {
//  Create druid segments from raw data
	Reader reader = new BufferedReader(new FileReader(new File("./src/test/resources/report.csv")));

    List<String> columns = Arrays.asList("colo", "pool", "report", "URL", "TS", "metric", "value", "count", "min", "max", "sum");
    List<String> exclusions = Arrays.asList("_Timestamp", "_Machine", "_ThreadId", "_Query");
    List<String> metrics = Arrays.asList("value", "count", "min", "max", "sum");
    List<DimensionSchema> dimensions = new ArrayList<DimensionSchema>();
    for(String dim : columns){
    	dimensions.add(new StringDimensionSchema(dim));
    }
    dimensions.removeAll(exclusions);
    dimensions.removeAll(metrics);
    Loader loader = new CSVLoader(reader, columns, columns, "TS");

    DimensionsSpec dimensionsSpec = new DimensionsSpec(dimensions, null, null);
    AggregatorFactory[] metricsAgg = new AggregatorFactory[] {
        new LongSumAggregatorFactory("agg_count", "count"),
        new LongMaxAggregatorFactory("agg_max", "max"),
        new LongMinAggregatorFactory("agg_min", "min"),
        new DoubleSumAggregatorFactory("agg_sum", "sum"),
    };
    IncrementalIndexSchema indexSchema = new IncrementalIndexSchema(0, QueryGranularity.fromString("ALL"), dimensionsSpec, metricsAgg);
    QueryableIndex index = IndexHelper.getQueryableIndex(loader, indexSchema);
    return index;
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
public AggregatorFactoryAdapter(String aggFactorySpec, String metricType) {
  ObjectMapper jsonMapper = HadoopDruidIndexerConfig.JSON_MAPPER;
  try {
    this.aggFactory = jsonMapper.readValue(aggFactorySpec, AggregatorFactory.class);
  } catch(IOException ex) {
    throw new IllegalArgumentException("failed to create aggregator factory", ex);
  }

  if (DruidUtils.isComplex(metricType)) {
    ComplexMetricSerde cms = ComplexMetrics.getSerdeForType(metricType);
    if (cms != null)
      strategy = cms.getObjectStrategy();
    else
      throw new IllegalArgumentException("failed to find object strategy for " + metricType);
  } else {
    strategy = null;
  }
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
@Override
public AggregatorFactory getMergingFactory(AggregatorFactory other) throws AggregatorFactoryNotMergeableException
{
  if (other.getName().equals(this.getName()) && this.getClass() == other.getClass()) {
    return getCombiningFactory();
  } else {
    throw new AggregatorFactoryNotMergeableException(this, other);
  }
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
@Test
public void topNQuery() throws IOException {
	QueryableIndex index = createDruidSegments();
	List<DimFilter> filters = new ArrayList<DimFilter>();
	filters.add(DimFilters.dimEquals("report", "URLTransaction"));
	filters.add(DimFilters.dimEquals("pool", "r1cart"));
	filters.add(DimFilters.dimEquals("metric", "Duration"));
    TopNQuery query =
        new TopNQueryBuilder()
            .threshold(5)
            .metric("agg_count")
            .dataSource("test")
            .intervals(QuerySegmentSpecs.create(new Interval(0, new DateTime().getMillis())))
            .granularity(QueryGranularity.fromString("NONE"))
            .dimension("colo")
            .aggregators(
                Arrays.<AggregatorFactory>asList(
                    new LongSumAggregatorFactory("agg_count", "agg_count"),
                    new LongMaxAggregatorFactory("agg_max", "agg_max"),
                    new LongMinAggregatorFactory("agg_min", "agg_min"),
                    new DoubleSumAggregatorFactory("agg_sum", "agg_sum"))
                    
            )
            .filters(DimFilters.and(filters)).build();
    @SuppressWarnings("unchecked")
    Sequence<Result> sequence = QueryHelper.run(query, index);
    ArrayList<Result> results = Sequences.toList(sequence, Lists.<Result>newArrayList());
    Assert.assertEquals(results.size(), 1);
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
@Override
public Beam<Map<String, Object>> makeBeam(Map<?, ?> conf, IMetricsContext metrics) {

    List<AggregatorFactory> aggregator = getAggregatorList();

    // Tranquility needs to be able to extract timestamps from your object type (in this case, Map<String, Object>).
    final Timestamper<Map<String, Object>> timestamper = new StreamlineTimestamper(timestampField);

    // Tranquility uses ZooKeeper (through Curator) for coordination.
    final CuratorFramework curator = CuratorFrameworkFactory
            .builder()
            .connectString(tranquilityZKconnect) // we can use Storm conf to get config values
            .retryPolicy(new ExponentialBackoffRetry(1000, 20, 30000))
            .build();
    curator.start();

    // The JSON serialization of your object must have a timestamp field in a format that Druid understands. By default,
    // Druid expects the field to be called "timestamp" and to be an ISO8601 timestamp.
    final TimestampSpec timestampSpec = new TimestampSpec(timestampField, "auto", null);

    // Tranquility needs to be able to serialize your object type to JSON for transmission to Druid. By default this is
    // done with Jackson. If you want to provide an alternate serializer, you can provide your own via ```.objectWriter(...)```.
    // In this case, we won't provide one, so we're just using Jackson.
    final Beam<Map<String, Object>> beam = DruidBeams
            .builder(timestamper)
            .curator(curator)
            .discoveryPath(discoveryPath)
            .location(DruidLocation.create(indexService, dataSource))
            .timestampSpec(timestampSpec)
            .rollup(DruidRollup.create(DruidDimensions.specific(getTrimmedDimensions(dimensions)), aggregator, getQueryGranularity()))
            .tuning(
                    ClusteredBeamTuning
                            .builder()
                            .segmentGranularity(getSegmentGranularity())
                            .windowPeriod(new Period(windowPeriod))
                            .partitions(clusterPartitions)
                            .replicants(clusterReplication)
                            .build()
            )
            .druidBeamConfig(
                    DruidBeamConfig
                            .builder()
                            .indexRetryPeriod(new Period(indexRetryPeriod))
                            .build())
            .buildBeam();

    return beam;
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
public static List<AggregatorFactory> build(String rawAggregators) {
//        final List<String> splittedAggregators = Arrays.asList(rawAggregators.split(","));
        //        List<AggregatorFactory> list = new ArrayList<AggregatorFactory>();
        //        for (String aggregator : aggregators) {
        //            list.add(null);
        //        }
        //TODO pending implementation
        List<AggregatorFactory> aggregators = new ArrayList<AggregatorFactory>();
        aggregators.add(new CountAggregatorFactory("count"));
        return aggregators;

    }
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
@Override
public AggregatorFactory getCombiningFactory()
{
  return new ExampleSumAggregatorFactory(name, name);
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
@Override
public List<AggregatorFactory> getRequiredColumns()
{
  return ImmutableList.of(new ExampleSumAggregatorFactory(fieldName, fieldName));
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
@Override
public void configure(Context context) {
    indexService = context.getString(INDEX_SERVICE);
    discoveryPath = context.getString(DISCOVERY_PATH);
    dimensions = Arrays.asList(context.getString(DIMENSIONS).split(","));
    firehosePattern = context.getString(FIREHOSE_PATTERN, DEFAULT_FIREHOSE);
    dataSource = context.getString(DATA_SOURCE, DEFAUL_DATASOURCE);
    aggregators = AggregatorsHelper.build(context.getString(AGGREGATORS));
    queryGranularity = QueryGranularity.fromString(context.getString(QUERY_GRANULARITY, DEFAULT_QUERY_GRANULARITY));
    segmentGranularity = Granularity.valueOf(context.getString(SEGMENT_GRANULARITY, DEFAULT_SEGMENT_GRANULARITY));
    windowPeriod = context.getString(WINDOW_PERIOD, DEFAULT_PERIOD);
    partitions = context.getInteger(PARTITIONS, DEFAULT_PARTITIONS);
    replicants = context.getInteger(REPLICANTS, DEFAULT_REPLICANTS);
    timestampField = context.getString(TIMESTAMP_FIELD, DEFAULT_TIMESTAMP_FIELD);
    timestampFormat = context.getString(TIMESTAMP_FORMAT, null);
    zookeeperLocation = context.getString(ZOOKEEPER_LOCATION, DEFAULT_ZOOKEEPER_LOCATION);
    baseSleepTime = context.getInteger(ZOOKEEPPER_BASE_SLEEP_TIME, DEFAULT_ZOOKEEPER_BASE_SLEEP);
    maxRetries = context.getInteger(ZOOKEEPER_MAX_RETRIES, DEFAULT_ZOOKEEPER_MAX_RETRIES);
    maxSleep = context.getInteger(ZOOKEEPER_MAX_SLEEP, DEFAULT_ZOOKEEPER_MAX_SLEEP);
    batchSize = context.getInteger(BATCH_SIZE, DEFAULT_BATCH_SIZE);

    druidService = sinkStrategy.getDruidService();
    if (druidService != null) {
        sinkCounter = new SinkCounter(this.getName());
        if (timestampFormat.equals("auto"))
            dateTimeFormatter = DateTimeFormat.forPattern("yyyy-MM-dd'T'HH:mm:ss'Z'");
        else if (timestampFormat.equals("millis"))
            dateTimeFormatter = null;
        else
            dateTimeFormatter = DateTimeFormat.forPattern(timestampFormat);

        // Filter defined fields
        Set<String> filter = new HashSet<String>();
        filter.add(timestampField);
        filter.addAll(dimensions);
        for (AggregatorFactory aggregatorFactory : aggregators) {
            filter.addAll(aggregatorFactory.requiredFields());
        }

        eventParser = new FlumeEventParser(timestampField, dateTimeFormatter, filter);
    }
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
private AggregatorFactory getLongMinAggregator(Map<String, String> map) {
    return new LongMinAggregatorFactory(map.get("name"), map.get("fieldName"));
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
private AggregatorFactory getLongMaxAggregator(Map<String, String> map) {
    return new LongMaxAggregatorFactory(map.get("name"), map.get("fieldName"));
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
private AggregatorFactory getLongSumAggregator(Map<String, String> map) {
    return new LongSumAggregatorFactory(map.get("name"), map.get("fieldName"));
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
private AggregatorFactory getDoubleMinAggregator(Map<String, String> map) {
    return new DoubleMinAggregatorFactory(map.get("name"), map.get("fieldName"));
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
private AggregatorFactory getDoubleMaxAggregator(Map<String, String> map) {
    return new DoubleMaxAggregatorFactory(map.get("name"), map.get("fieldName"));
}
 

import io.druid.query.aggregation.AggregatorFactory; //导入依赖的package包/类
private AggregatorFactory getDoubleSumAggregator(Map<String, String> map) {
    return new DoubleSumAggregatorFactory(map.get("name"), map.get("fieldName"));
}