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

Class ColumnVector

java.lang.Object

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

Direct Known Subclasses:

BytesColumnVector, DecimalColumnVector, DoubleColumnVector, IntervalDayTimeColumnVector, LongColumnVector, MultiValuedColumnVector, StructColumnVector, TimestampColumnVector, UnionColumnVector, VoidColumnVector

public abstract class ColumnVector
extends Object

ColumnVector contains the shared structure for the sub-types, including NULL information, and whether this vector repeats, i.e. has all values the same, so only the first one is set. This is used to accelerate query performance by handling a whole vector in O(1) time when applicable. The fields are public by design since this is a performance-critical structure that is used in the inner loop of query execution.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	ColumnVector.Type The current kinds of column vectors.

Field Summary

Fields

Modifier and Type	Field and Description
boolean[]	isNull
boolean	isRepeating
boolean	noNulls
ColumnVector.Type	type

Constructor Summary

Constructors

Constructor and Description
ColumnVector(ColumnVector.Type type, int len) Constructor for super-class ColumnVector.

Method Summary

Modifier and Type	Method and Description
abstract void	copySelected(boolean selectedInUse, int[] sel, int size, ColumnVector outputColVector)
int	decRef()
void	ensureSize(int size, boolean preserveData) Ensure the ColumnVector can hold at least size values.
abstract void	flatten(boolean selectedInUse, int[] sel, int size)
protected void	flattenNoNulls(boolean selectedInUse, int[] sel, int size)
protected void	flattenPush()
protected void	flattenRepeatingNulls(boolean selectedInUse, int[] sel, int size)
int	getRef()
void	incRef()
void	init() Initialize the column vector.
void	reset() Resets the column to default state - fills the isNull array with false - sets noNulls to true - sets isRepeating to false
abstract void	setElement(int outputElementNum, int inputElementNum, ColumnVector inputColVector) Set the element in this column vector from the given input vector.
void	setRepeating(boolean isRepeating) Sets the isRepeating flag.
void	shallowCopyTo(ColumnVector otherCv) Shallow copy of the contents of this vector to the other vector; replaces other vector's values.
abstract void	stringifyValue(StringBuilder buffer, int row) Print the value for this column into the given string builder.
void	unFlatten() Restore the state of isRepeating and noNulls to what it was before flattening.

Methods inherited from class java.lang.Object

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

Field Detail

type

public final ColumnVector.Type type

isNull

public boolean[] isNull

noNulls

public boolean noNulls

isRepeating

public boolean isRepeating

Constructor Detail

ColumnVector

public ColumnVector(ColumnVector.Type type,
                    int len)

Constructor for super-class ColumnVector. This is not called directly, but used to initialize inherited fields.

Parameters:

len - Vector length

Method Detail

reset

public void reset()

Resets the column to default state - fills the isNull array with false - sets noNulls to true - sets isRepeating to false

incRef

public final void incRef()

getRef

public final int getRef()

decRef

public final int decRef()

setRepeating

public void setRepeating(boolean isRepeating)

Sets the isRepeating flag. Recurses over structs and unions so that the flags are set correctly.

Parameters:

isRepeating - flag for repeating value.

flatten

public abstract void flatten(boolean selectedInUse,
                             int[] sel,
                             int size)

flattenRepeatingNulls

protected void flattenRepeatingNulls(boolean selectedInUse,
                                     int[] sel,
                                     int size)

flattenNoNulls

protected void flattenNoNulls(boolean selectedInUse,
                              int[] sel,
                              int size)

unFlatten

public void unFlatten()

Restore the state of isRepeating and noNulls to what it was before flattening. This must only be called just after flattening and then evaluating a VectorExpression on the column vector. It is an optimization that allows other operations on the same column to continue to benefit from the isRepeating and noNulls indicators.

flattenPush

protected void flattenPush()

setElement

public abstract void setElement(int outputElementNum,
                                int inputElementNum,
                                ColumnVector inputColVector)

Set the element in this column vector from the given input vector. The inputElementNum will be adjusted to 0 if the input column has isRepeating set. On the other hand, the outElementNum must have been adjusted to 0 in ADVANCE when the output has isRepeating set. IMPORTANT: if the output entry is marked as NULL, this method will do NOTHING. This supports the caller to do output NULL processing in advance that may cause the output results operation to be ignored. Thus, make sure the output isNull entry is set in ADVANCE. The inputColVector noNulls and isNull entry will be examined. The output will only be set if the input is NOT NULL. I.e. noNulls || !isNull[inputElementNum] where inputElementNum may have been adjusted to 0 for isRepeating. If the input entry is NULL or out-of-range, the output will be marked as NULL. I.e. set output noNull = false and isNull[outElementNum] = true. An example of out-of-range is the DecimalColumnVector which can find the input decimal does not fit in the output precision/scale. (Since we return immediately if the output entry is NULL, we have no need and do not mark the output entry to NOT NULL).

copySelected

public abstract void copySelected(boolean selectedInUse,
                                  int[] sel,
                                  int size,
                                  ColumnVector outputColVector)

init

public void init()

Initialize the column vector. This method can be overridden by specific column vector types. Use this method only if the individual type of the column vector is not known, otherwise its preferable to call specific initialization methods.

ensureSize

public void ensureSize(int size,
                       boolean preserveData)

Ensure the ColumnVector can hold at least size values. This method is deliberately *not* recursive because the complex types can easily have more (or less) children than the upper levels.

Parameters:

size - the new minimum size

preserveData - should the old data be preserved?

stringifyValue

public abstract void stringifyValue(StringBuilder buffer,
                                    int row)

Print the value for this column into the given string builder.

Parameters:

buffer - the buffer to print into

row - the id of the row to print

shallowCopyTo

public void shallowCopyTo(ColumnVector otherCv)

Shallow copy of the contents of this vector to the other vector; replaces other vector's values.