Java Code Examples for org.apache.hadoop.hive.metastore.api.ColumnStatisticsDesc#setPartName()

@Test
public void typical() throws Exception {
  List<FieldSchema> partitionKeys = Lists.newArrayList(newFieldSchema("a"), newFieldSchema("c"));
  Table table = newTable("t1", "db1", partitionKeys, newStorageDescriptor(new File("bla"), "col1"));
  List<Partition> partitions = Lists.newArrayList(newPartition(table, "b", "d"));
  statisticsPerPartitionName.put("a=b/c=d", columnStats);

  PartitionsAndStatistics partitionsAndStatistics = new PartitionsAndStatistics(partitionKeys, partitions,
      statisticsPerPartitionName);
  List<String> expectedName = Lists.newArrayList("a=b/c=d");

  assertThat(partitionsAndStatistics.getPartitionNames(), is(expectedName));
  assertThat(partitionsAndStatistics.getPartitions(), is(partitions));
  assertThat(partitionsAndStatistics.getPartitionKeys(), is(partitionKeys));
  ColumnStatisticsDesc statsDesc = new ColumnStatisticsDesc(false, "db1", "t1");
  statsDesc.setPartName("a=b/c=d");
  ColumnStatistics expectedStats = new ColumnStatistics(statsDesc, columnStats);
  assertThat(partitionsAndStatistics.getStatisticsForPartition(partitions.get(0)), is(expectedStats));
}
 

@Override
public void setPartitionColumnStatistics(String databaseName, String tableName, String partitionName, List<ColumnStatisticsObj> statistics)
        throws TException
{
    ColumnStatisticsDesc statisticsDescription = new ColumnStatisticsDesc(false, databaseName, tableName);
    statisticsDescription.setPartName(partitionName);
    ColumnStatistics request = new ColumnStatistics(statisticsDescription, statistics);
    client.update_partition_column_statistics(request);
}
 

/**
 * Create columnStatistics from the given Hive column stats of a hive partition.
 */
public static ColumnStatistics createPartitionColumnStats(
		Partition hivePartition,
		String partName,
		Map<String, CatalogColumnStatisticsDataBase> colStats) {
	ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, hivePartition.getDbName(), hivePartition.getTableName());
	desc.setPartName(partName);
	return createHiveColumnStatistics(colStats, hivePartition.getSd(), desc);
}
 

ColumnStatistics newReplicaPartitionStatistics(
    Table replicaTable,
    Partition replicaPartition,
    ColumnStatistics sourcePartitionStatistics) {
  ColumnStatisticsDesc statisticsDesc = new ColumnStatisticsDesc(false, replicaPartition.getDbName(),
      replicaPartition.getTableName());
  try {
    statisticsDesc.setPartName(Warehouse.makePartName(replicaTable.getPartitionKeys(), replicaPartition.getValues()));
  } catch (MetaException e) {
    throw new RuntimeException(e);
  }

  return columnStatisticsTransformation
      .transform(new ColumnStatistics(statisticsDesc, sourcePartitionStatistics.getStatsObj()));
}
 

private ColumnStatistics newPartitionStatistics(String... values) {
  ColumnStatisticsObj columnStatisticsObj1 = new ColumnStatisticsObj(COLUMN_A, "string",
      new ColumnStatisticsData(_Fields.LONG_STATS, new LongColumnStatsData(0, 1)));
  ColumnStatisticsObj columnStatisticsObj2 = new ColumnStatisticsObj(COLUMN_B, "string",
      new ColumnStatisticsData(_Fields.LONG_STATS, new LongColumnStatsData(1, 2)));
  List<ColumnStatisticsObj> columnStatisticsObjs = Arrays.asList(columnStatisticsObj1, columnStatisticsObj2);
  ColumnStatisticsDesc statsDesc = new ColumnStatisticsDesc(false, DB_NAME, TABLE_NAME);
  statsDesc.setPartName(partitionName(values));
  return new ColumnStatistics(statsDesc, columnStatisticsObjs);
}
 

@Override
public ColumnStatistics transform(ColumnStatistics columnStatistics) {
  ColumnStatisticsDesc statsDesc = new ColumnStatisticsDesc(false, "new_db", "new_table");
  statsDesc.setPartName("part=newPart");
  columnStatistics.setStatsDesc(statsDesc);
  return columnStatistics;
}
 

@Test
public void newReplicaPartitionStatistics() throws MetaException {
  sourceTable.setPartitionKeys(
      Arrays.asList(new FieldSchema("one", "string", null), new FieldSchema("two", "string", null)));

  Partition replicaPartition = new Partition(sourcePartition);
  replicaPartition.setDbName(MAPPED_DB_NAME);
  replicaPartition.setTableName(MAPPED_TABLE_NAME);
  replicaPartition.setValues(Arrays.asList("A", "B"));

  ColumnStatisticsObj columnStatisticsObj1 = new ColumnStatisticsObj();
  ColumnStatisticsObj columnStatisticsObj2 = new ColumnStatisticsObj();
  List<ColumnStatisticsObj> columnStatisticsObjs = Arrays.asList(columnStatisticsObj1, columnStatisticsObj2);

  ColumnStatisticsDesc columnStatisticsDesc = new ColumnStatisticsDesc(false, DB_NAME, TABLE_NAME);
  columnStatisticsDesc
      .setPartName(Warehouse.makePartName(sourceTable.getPartitionKeys(), replicaPartition.getValues()));

  ColumnStatistics sourcePartitionStatistics = new ColumnStatistics(columnStatisticsDesc, columnStatisticsObjs);

  ColumnStatistics replicaPartitionStatistics = factory.newReplicaPartitionStatistics(sourceTable, replicaPartition,
      sourcePartitionStatistics);

  assertThat(replicaPartitionStatistics.getStatsDesc().getDbName(), is(MAPPED_DB_NAME));
  assertThat(replicaPartitionStatistics.getStatsDesc().getTableName(), is(MAPPED_TABLE_NAME));
  assertThat(replicaPartitionStatistics.getStatsDesc().getPartName(), is("one=A/two=B"));
  assertThat(replicaPartitionStatistics.getStatsObj().size(), is(2));
  assertThat(replicaPartitionStatistics.getStatsObj().get(0), is(columnStatisticsObj1));
  assertThat(replicaPartitionStatistics.getStatsObj().get(1), is(columnStatisticsObj2));
}
 

@Test
public void newReplicaPartitionStatisticsWithTransformation() throws MetaException {
  sourceTable.setPartitionKeys(
      Arrays.asList(new FieldSchema("one", "string", null), new FieldSchema("two", "string", null)));

  Partition replicaPartition = new Partition(sourcePartition);
  replicaPartition.setDbName(MAPPED_DB_NAME);
  replicaPartition.setTableName(MAPPED_TABLE_NAME);
  replicaPartition.setValues(Arrays.asList("A", "B"));

  ColumnStatisticsObj columnStatisticsObj1 = new ColumnStatisticsObj();
  ColumnStatisticsObj columnStatisticsObj2 = new ColumnStatisticsObj();
  List<ColumnStatisticsObj> columnStatisticsObjs = Arrays.asList(columnStatisticsObj1, columnStatisticsObj2);

  ColumnStatisticsDesc columnStatisticsDesc = new ColumnStatisticsDesc(false, DB_NAME, TABLE_NAME);
  columnStatisticsDesc
      .setPartName(Warehouse.makePartName(sourceTable.getPartitionKeys(), replicaPartition.getValues()));

  ColumnStatistics sourcePartitionStatistics = new ColumnStatistics(columnStatisticsDesc, columnStatisticsObjs);

  ReplicaTableFactory factory = new ReplicaTableFactory(SOURCE_META_STORE_URIS, TableTransformation.IDENTITY,
      PartitionTransformation.IDENTITY, COLUMN_STATISTICS_TRANSFORMATION);

  ColumnStatistics replicaPartitionStatistics = factory.newReplicaPartitionStatistics(sourceTable, replicaPartition,
      sourcePartitionStatistics);

  assertThat(replicaPartitionStatistics.getStatsDesc().getDbName(), is("new_db"));
  assertThat(replicaPartitionStatistics.getStatsDesc().getTableName(), is("new_table"));
  assertThat(replicaPartitionStatistics.getStatsDesc().getPartName(), is("part=newPart"));
  assertThat(replicaPartitionStatistics.getStatsObj().size(), is(2));
  assertThat(replicaPartitionStatistics.getStatsObj().get(0), is(columnStatisticsObj1));
  assertThat(replicaPartitionStatistics.getStatsObj().get(1), is(columnStatisticsObj2));
}
 

/**
 * Create columnStatistics from the given Hive column stats of a hive partition.
 */
public static ColumnStatistics createPartitionColumnStats(
		Partition hivePartition,
		String partName,
		Map<String, CatalogColumnStatisticsDataBase> colStats,
		String hiveVersion) {
	ColumnStatisticsDesc desc = new ColumnStatisticsDesc(false, hivePartition.getDbName(), hivePartition.getTableName());
	desc.setPartName(partName);
	return createHiveColumnStatistics(colStats, hivePartition.getSd(), desc, hiveVersion);
}