org.apache.hadoop.hive.ql.exec.vector

Class VectorDeserializeRow<T extends DeserializeRead>

java.lang.Object

org.apache.hadoop.hive.ql.exec.vector.VectorDeserializeRow<T>

public final class VectorDeserializeRow<T extends DeserializeRead>
extends Object

This class deserializes a serialization format into a row of a VectorizedRowBatch. The caller provides the hive type names and output column numbers in the order desired to deserialize. This class uses an provided DeserializeRead object to directly deserialize by reading field-by-field from a serialization format into the primitive values of the VectorizedRowBatch.

Field Summary

Fields

Modifier and Type	Field and Description
protected org.apache.hadoop.hive.common.type.DataTypePhysicalVariation[]	dataTypePhysicalVariations

Constructor Summary

Constructors

Constructor and Description
VectorDeserializeRow(T deserializeRead)

Method Summary

Modifier and Type	Method and Description
void	deserialize(org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatch batch, int batchIndex) Deserialize a row from the range of bytes specified by setBytes.
void	deserializeByRef(org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatch batch, int batchIndex) Deserialize a row from the range of bytes specified by setBytes.
String	getDetailedReadPositionString()
void	init()
void	init(boolean[] columnsToIncludeTruncated)
void	init(int startColumn)
void	init(int[] outputColumns)
void	init(List<Integer> outputColumns)
void	initConversion(TypeInfo[] targetTypeInfos, boolean[] columnsToIncludeTruncated) Initialize for converting the source data type that are going to be read with the DeserializedRead interface passed to the constructor to the target data types desired in the VectorizedRowBatch.
void	setBytes(byte[] bytes, int offset, int length) Specify the range of bytes to deserialize in the next call to the deserialize method.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

dataTypePhysicalVariations

protected org.apache.hadoop.hive.common.type.DataTypePhysicalVariation[] dataTypePhysicalVariations

Constructor Detail

VectorDeserializeRow

public VectorDeserializeRow(T deserializeRead)

Parameters:

deserializeRead - Set useExternalBuffer to true to avoid buffer copying and to get more efficient reading.

Method Detail

init

public void init(int[] outputColumns)
          throws HiveException

Throws:

HiveException

init

public void init(List<Integer> outputColumns)
          throws HiveException

Throws:

HiveException

init

public void init(int startColumn)
          throws HiveException

Throws:

HiveException

init

public void init(boolean[] columnsToIncludeTruncated)
          throws HiveException

Throws:

HiveException

initConversion

public void initConversion(TypeInfo[] targetTypeInfos,
                           boolean[] columnsToIncludeTruncated)
                    throws HiveException

Initialize for converting the source data type that are going to be read with the DeserializedRead interface passed to the constructor to the target data types desired in the VectorizedRowBatch. No projection -- using the column range 0 .. columnCount-1

Parameters:

targetTypeInfos -

columnsToIncludeTruncated -

Throws:

HiveException

init

public void init()
          throws HiveException

Throws:

HiveException

setBytes

public void setBytes(byte[] bytes,
                     int offset,
                     int length)

Specify the range of bytes to deserialize in the next call to the deserialize method.

Parameters:

bytes -

offset -

length -

deserialize

public void deserialize(org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatch batch,
                        int batchIndex)
                 throws IOException

Deserialize a row from the range of bytes specified by setBytes. Use getDetailedReadPositionString to get detailed read position information to help diagnose exceptions that are thrown... This version of deserialize does not keep byte references to string/char/varchar/binary data type field. The bytes are copied into the BytesColumnVector buffer with setVal. (See deserializeByRef below if keep references is safe).

Parameters:

batch -

batchIndex -

Throws:

IOException

deserializeByRef

public void deserializeByRef(org.apache.hadoop.hive.ql.exec.vector.VectorizedRowBatch batch,
                             int batchIndex)
                      throws IOException

Deserialize a row from the range of bytes specified by setBytes. Use this method instead of deserialize when it is safe to retain references to the bytes source for DeserializeRead. I.e. the STRING, CHAR/VARCHAR data can be set in BytesColumnVector with setRef instead of with setVal which copies data. An example of a safe usage: Referring to bytes in a hash table entry that is immutable. An example of a unsafe usage: Referring to bytes in a reduce receive buffer that will be overwritten with new data. Use getDetailedReadPositionString to get detailed read position information to help diagnose exceptions that are thrown...

Parameters:

batch -

batchIndex -

Throws:

IOException

getDetailedReadPositionString

public String getDetailedReadPositionString()