org.apache.hadoop.hive.serde2.lazybinary.LazyBinarySerDe Java Examples

public static String getDataFormat(StorageDescriptor descriptor) {
  Preconditions.checkNotNull(descriptor);

  String serde = descriptor.getSerdeInfo().getSerializationLib();
  String inputFormat = descriptor.getInputFormat();

  if (LazySimpleSerDe.class.getName().equals(serde)) {
    if (TextInputFormat.class.getName().equals(inputFormat)) {
      return BuiltinStorages.TEXT;
    } else if (SequenceFileInputFormat.class.getName().equals(inputFormat)) {
      return BuiltinStorages.SEQUENCE_FILE;
    } else {
      throw new TajoRuntimeException(new UnknownDataFormatException(inputFormat));
    }
  } else if (LazyBinarySerDe.class.getName().equals(serde)) {
    if (SequenceFileInputFormat.class.getName().equals(inputFormat)) {
      return BuiltinStorages.SEQUENCE_FILE;
    } else {
      throw new TajoRuntimeException(new UnknownDataFormatException(inputFormat));
    }
  } else if (LazyBinaryColumnarSerDe.class.getName().equals(serde) || ColumnarSerDe.class.getName().equals(serde)) {
    if (RCFileInputFormat.class.getName().equals(inputFormat)) {
      return BuiltinStorages.RCFILE;
    } else {
      throw new TajoRuntimeException(new UnknownDataFormatException(inputFormat));
    }
  } else if (ParquetHiveSerDe.class.getName().equals(serde)) {
    return BuiltinStorages.PARQUET;
  } else if (AvroSerDe.class.getName().equals(serde)) {
    return BuiltinStorages.AVRO;
  } else if (OrcSerde.class.getName().equals(serde)) {
    return BuiltinStorages.ORC;
  } else if (RegexSerDe.class.getName().equals(serde)) {
    return BuiltinStorages.REGEX;
  } else {
    throw new TajoRuntimeException(new UnknownDataFormatException(inputFormat));
  }
}
 

@Test
public void testCreateTable() throws Exception {
	String location = warehouse + "/external_location";
	tableEnv.executeSql(String.format(
			"create external table tbl1 (d decimal(10,0),ts timestamp) partitioned by (p string) location '%s' tblproperties('k1'='v1')", location));
	Table hiveTable = hiveCatalog.getHiveTable(new ObjectPath("default", "tbl1"));
	assertEquals(TableType.EXTERNAL_TABLE.toString(), hiveTable.getTableType());
	assertEquals(1, hiveTable.getPartitionKeysSize());
	assertEquals(location, locationPath(hiveTable.getSd().getLocation()));
	assertEquals("v1", hiveTable.getParameters().get("k1"));
	assertFalse(hiveTable.getParameters().containsKey(SqlCreateHiveTable.TABLE_LOCATION_URI));

	tableEnv.executeSql("create table tbl2 (s struct<ts:timestamp,bin:binary>) stored as orc");
	hiveTable = hiveCatalog.getHiveTable(new ObjectPath("default", "tbl2"));
	assertEquals(TableType.MANAGED_TABLE.toString(), hiveTable.getTableType());
	assertEquals(OrcSerde.class.getName(), hiveTable.getSd().getSerdeInfo().getSerializationLib());
	assertEquals(OrcInputFormat.class.getName(), hiveTable.getSd().getInputFormat());
	assertEquals(OrcOutputFormat.class.getName(), hiveTable.getSd().getOutputFormat());

	tableEnv.executeSql("create table tbl3 (m map<timestamp,binary>) partitioned by (p1 bigint,p2 tinyint) " +
			"row format serde 'org.apache.hadoop.hive.serde2.lazybinary.LazyBinarySerDe'");
	hiveTable = hiveCatalog.getHiveTable(new ObjectPath("default", "tbl3"));
	assertEquals(2, hiveTable.getPartitionKeysSize());
	assertEquals(LazyBinarySerDe.class.getName(), hiveTable.getSd().getSerdeInfo().getSerializationLib());

	tableEnv.executeSql("create table tbl4 (x int,y smallint) row format delimited fields terminated by '|' lines terminated by '\n'");
	hiveTable = hiveCatalog.getHiveTable(new ObjectPath("default", "tbl4"));
	assertEquals("|", hiveTable.getSd().getSerdeInfo().getParameters().get(serdeConstants.FIELD_DELIM));
	assertEquals("|", hiveTable.getSd().getSerdeInfo().getParameters().get(serdeConstants.SERIALIZATION_FORMAT));
	assertEquals("\n", hiveTable.getSd().getSerdeInfo().getParameters().get(serdeConstants.LINE_DELIM));

	tableEnv.executeSql("create table tbl5 (m map<bigint,string>) row format delimited collection items terminated by ';' " +
			"map keys terminated by ':'");
	hiveTable = hiveCatalog.getHiveTable(new ObjectPath("default", "tbl5"));
	assertEquals(";", hiveTable.getSd().getSerdeInfo().getParameters().get(serdeConstants.COLLECTION_DELIM));
	assertEquals(":", hiveTable.getSd().getSerdeInfo().getParameters().get(serdeConstants.MAPKEY_DELIM));
}
 

@Test
public void testAlterPartition() throws Exception {
	tableEnv.executeSql("create table tbl (x tinyint,y string) partitioned by (p1 bigint,p2 date)");
	tableEnv.executeSql("alter table tbl add partition (p1=1000,p2='2020-05-01') partition (p1=2000,p2='2020-01-01')");
	CatalogPartitionSpec spec1 = new CatalogPartitionSpec(new LinkedHashMap<String, String>() {{
		put("p1", "1000");
		put("p2", "2020-05-01");
	}});
	CatalogPartitionSpec spec2 = new CatalogPartitionSpec(new LinkedHashMap<String, String>() {{
		put("p1", "2000");
		put("p2", "2020-01-01");
	}});
	ObjectPath tablePath = new ObjectPath("default", "tbl");

	Table hiveTable = hiveCatalog.getHiveTable(tablePath);

	// change location
	String location = warehouse + "/new_part_location";
	tableEnv.executeSql(String.format("alter table tbl partition (p1=1000,p2='2020-05-01') set location '%s'", location));
	Partition partition = hiveCatalog.getHivePartition(hiveTable, spec1);
	assertEquals(location, locationPath(partition.getSd().getLocation()));

	// change file format
	tableEnv.executeSql("alter table tbl partition (p1=2000,p2='2020-01-01') set fileformat rcfile");
	partition = hiveCatalog.getHivePartition(hiveTable, spec2);
	assertEquals(LazyBinaryColumnarSerDe.class.getName(), partition.getSd().getSerdeInfo().getSerializationLib());
	assertEquals(RCFileInputFormat.class.getName(), partition.getSd().getInputFormat());
	assertEquals(RCFileOutputFormat.class.getName(), partition.getSd().getOutputFormat());

	// change serde
	tableEnv.executeSql(String.format("alter table tbl partition (p1=1000,p2='2020-05-01') set serde '%s' with serdeproperties('%s'='%s')",
			LazyBinarySerDe.class.getName(), serdeConstants.LINE_DELIM, "\n"));
	partition = hiveCatalog.getHivePartition(hiveTable, spec1);
	assertEquals(LazyBinarySerDe.class.getName(), partition.getSd().getSerdeInfo().getSerializationLib());
	assertEquals("\n", partition.getSd().getSerdeInfo().getParameters().get(serdeConstants.LINE_DELIM));
}