Java Code Examples for org.apache.flink.api.java.utils.DataSetUtils#countElementsPerPartition()
The following examples show how to use
org.apache.flink.api.java.utils.DataSetUtils#countElementsPerPartition() .
You can vote up the ones you like or vote down the ones you don't like,
and go to the original project or source file by following the links above each example. You may check out the related API usage on the sidebar.
Example 1
Source File: SplitBatchOp.java From Alink with Apache License 2.0 | 6 votes |
@Override public SplitBatchOp linkFrom(BatchOperator<?>... inputs) { BatchOperator<?> in = checkAndGetFirst(inputs); final double fraction = getFraction(); if (fraction < 0. || fraction > 1.0) { throw new RuntimeException("invalid fraction " + fraction); } DataSet<Row> rows = in.getDataSet(); DataSet<Tuple2<Integer, Long>> countsPerPartition = DataSetUtils.countElementsPerPartition(rows); DataSet<long[]> numPickedPerPartition = countsPerPartition .mapPartition(new CountInPartition(fraction)) .setParallelism(1) .name("decide_count_of_each_partition"); DataSet<Row> out = rows .mapPartition(new PickInPartition()) .withBroadcastSet(numPickedPerPartition, "counts") .name("pick_in_each_partition"); this.setOutput(out, in.getSchema()); this.setSideOutputTables(new Table[]{in.getOutputTable().minusAll(this.getOutputTable())}); return this; }
Example 2
Source File: BaseComQueue.java From Alink with Apache License 2.0 | 6 votes |
private <T> void createRelationshipAndCachedData(DataSet<T> data, final String key) { final int localSessionId = sessionId; if (cacheDataRel == null) { cacheDataRel = clearObjs( BatchOperator .getExecutionEnvironmentFromDataSets(data) .fromElements(new byte[0]) .mapPartition(new MapPartitionFunction<byte[], byte[]>() { @Override public void mapPartition(Iterable<byte[]> values, Collector<byte[]> out) throws Exception { //pass } }) ); } DataSet<Tuple2<Integer, Long>> rowCount = DataSetUtils.countElementsPerPartition(data); cacheDataRel = data.mapPartition(new PutCachedData<T>(key, localSessionId)) .withBroadcastSet(cacheDataRel, "rel") .withBroadcastSet(rowCount, "rowCount") .name("cachedDataRel@" + key); cacheDataObjNames.add(key); }
Example 3
Source File: DataSetUtilsITCase.java From Flink-CEPplus with Apache License 2.0 | 5 votes |
@Test public void testCountElementsPerPartition() throws Exception { ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment(); long expectedSize = 100L; DataSet<Long> numbers = env.generateSequence(0, expectedSize - 1); DataSet<Tuple2<Integer, Long>> ds = DataSetUtils.countElementsPerPartition(numbers); Assert.assertEquals(env.getParallelism(), ds.count()); Assert.assertEquals(expectedSize, ds.sum(1).collect().get(0).f1.longValue()); }
Example 4
Source File: DataSetUtilsITCase.java From flink with Apache License 2.0 | 5 votes |
@Test public void testCountElementsPerPartition() throws Exception { ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment(); long expectedSize = 100L; DataSet<Long> numbers = env.generateSequence(0, expectedSize - 1); DataSet<Tuple2<Integer, Long>> ds = DataSetUtils.countElementsPerPartition(numbers); Assert.assertEquals(env.getParallelism(), ds.count()); Assert.assertEquals(expectedSize, ds.sum(1).collect().get(0).f1.longValue()); }
Example 5
Source File: DataSetUtilsITCase.java From flink with Apache License 2.0 | 5 votes |
@Test public void testCountElementsPerPartition() throws Exception { ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment(); long expectedSize = 100L; DataSet<Long> numbers = env.generateSequence(0, expectedSize - 1); DataSet<Tuple2<Integer, Long>> ds = DataSetUtils.countElementsPerPartition(numbers); Assert.assertEquals(env.getParallelism(), ds.count()); Assert.assertEquals(expectedSize, ds.sum(1).collect().get(0).f1.longValue()); }
Example 6
Source File: BaseTuning.java From Alink with Apache License 2.0 | 4 votes |
private DataSet<Tuple2<Integer, Row>> split(BatchOperator<?> data, int k) { DataSet<Row> input = shuffle(data.getDataSet()); DataSet<Tuple2<Integer, Long>> counts = DataSetUtils.countElementsPerPartition(input); return input .mapPartition(new RichMapPartitionFunction<Row, Tuple2<Integer, Row>>() { long taskStart = 0L; long totalNumInstance = 0L; @Override public void open(Configuration parameters) throws Exception { List<Tuple2<Integer, Long>> counts1 = getRuntimeContext().getBroadcastVariable("counts"); int taskId = getRuntimeContext().getIndexOfThisSubtask(); for (Tuple2<Integer, Long> cnt : counts1) { if (taskId < cnt.f0) { taskStart += cnt.f1; } totalNumInstance += cnt.f1; } } @Override public void mapPartition(Iterable<Row> values, Collector<Tuple2<Integer, Row>> out) throws Exception { DistributedInfo distributedInfo = new DefaultDistributedInfo(); Tuple2<Integer, Long> split1 = new Tuple2<>(-1, -1L); long lcnt = taskStart; for (int i = 0; i <= k; ++i) { long sp = distributedInfo.startPos(i, k, totalNumInstance); long lrc = distributedInfo.localRowCnt(i, k, totalNumInstance); if (taskStart < sp) { split1.f0 = i - 1; split1.f1 = distributedInfo.startPos(i - 1, k, totalNumInstance) + distributedInfo.localRowCnt(i - 1, k, totalNumInstance); break; } if (taskStart == sp) { split1.f0 = i; split1.f1 = sp + lrc; break; } } for (Row val : values) { if (lcnt >= split1.f1) { split1.f0 += 1; split1.f1 = distributedInfo.localRowCnt(split1.f0, k, totalNumInstance) + lcnt; } out.collect(Tuple2.of(split1.f0, val)); lcnt++; } } }).withBroadcastSet(counts, "counts"); }