org.apache.spark.sql.columnar

Class ColumnType<T extends org.apache.spark.sql.types.DataType,JvmType>

Object

org.apache.spark.sql.columnar.ColumnType<T,JvmType>

Direct Known Subclasses:

ByteArrayColumnType, NativeColumnType

public abstract class ColumnType<T extends org.apache.spark.sql.types.DataType,JvmType>
extends Object

An abstract class that represents type of a column. Used to append/extract Java objects into/from the underlying ByteBuffer of a column.

Constructor Summary

Constructors

Constructor and Description
ColumnType(int typeId, int defaultSize)

Method Summary

Methods

Modifier and Type	Method and Description
int	actualSize(org.apache.spark.sql.Row row, int ordinal) Returns the size of the value row(ordinal).
abstract void	append(JvmType v, java.nio.ByteBuffer buffer) Appends the given value v of type T into the given ByteBuffer.
void	append(org.apache.spark.sql.Row row, int ordinal, java.nio.ByteBuffer buffer) Appends row(ordinal) of type T into the given ByteBuffer.
static ColumnType<?,?>	apply(org.apache.spark.sql.types.DataType dataType)
JvmType	clone(JvmType v) Creates a duplicated copy of the value.
void	copyField(org.apache.spark.sql.Row from, int fromOrdinal, org.apache.spark.sql.catalyst.expressions.MutableRow to, int toOrdinal) Copies from(fromOrdinal) to to(toOrdinal).
int	defaultSize()
abstract JvmType	extract(java.nio.ByteBuffer buffer) Extracts a value out of the buffer at the buffer's current position.
void	extract(java.nio.ByteBuffer buffer, org.apache.spark.sql.catalyst.expressions.MutableRow row, int ordinal) Extracts a value out of the buffer at the buffer's current position and stores in row(ordinal).
abstract JvmType	getField(org.apache.spark.sql.Row row, int ordinal) Returns row(ordinal).
abstract void	setField(org.apache.spark.sql.catalyst.expressions.MutableRow row, int ordinal, JvmType value) Sets row(ordinal) to field.
String	toString()
int	typeId()

Methods inherited from class Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

ColumnType

public ColumnType(int typeId,
          int defaultSize)

Method Detail

apply

public static ColumnType<?,?> apply(org.apache.spark.sql.types.DataType dataType)

typeId

public int typeId()

defaultSize

public int defaultSize()

extract

public abstract JvmType extract(java.nio.ByteBuffer buffer)

Extracts a value out of the buffer at the buffer's current position.

extract

public void extract(java.nio.ByteBuffer buffer,
           org.apache.spark.sql.catalyst.expressions.MutableRow row,
           int ordinal)

Extracts a value out of the buffer at the buffer's current position and stores in row(ordinal). Subclasses should override this method to avoid boxing/unboxing costs whenever possible.

append

public abstract void append(JvmType v,
          java.nio.ByteBuffer buffer)

Appends the given value v of type T into the given ByteBuffer.

append

public void append(org.apache.spark.sql.Row row,
          int ordinal,
          java.nio.ByteBuffer buffer)

Appends row(ordinal) of type T into the given ByteBuffer. Subclasses should override this method to avoid boxing/unboxing costs whenever possible.

actualSize

public int actualSize(org.apache.spark.sql.Row row,
             int ordinal)

Returns the size of the value row(ordinal). This is used to calculate the size of variable length types such as byte arrays and strings.

getField

public abstract JvmType getField(org.apache.spark.sql.Row row,
               int ordinal)

Returns row(ordinal). Subclasses should override this method to avoid boxing/unboxing costs whenever possible.

setField

public abstract void setField(org.apache.spark.sql.catalyst.expressions.MutableRow row,
            int ordinal,
            JvmType value)

Sets row(ordinal) to field. Subclasses should override this method to avoid boxing/unboxing costs whenever possible.

copyField

public void copyField(org.apache.spark.sql.Row from,
             int fromOrdinal,
             org.apache.spark.sql.catalyst.expressions.MutableRow to,
             int toOrdinal)

Copies from(fromOrdinal) to to(toOrdinal). Subclasses should override this method to avoid boxing/unboxing costs whenever possible.

clone

public JvmType clone(JvmType v)

Creates a duplicated copy of the value.

toString

public String toString()

Overrides:

toString in class Object